System

Abstract

To provide a system.SOLUTION: A system includes: means for reading a user's habit of replies from message history of the user; means for analyzing flow of talking between the user and a partner; means for generating a prompt statement for directing creation of a user-like reply plan on the basis of the habit of the replies and the results from the analysis of flow of talking; a generative AI model for inputting the prompt statement and generating the user-like reply plan; and means for displaying the reply plan generated by the generative AI model onto a device of the user and enabling the user to select.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The technology of the present disclosure relates to a system. [Background technology]

[0002] Patent document 1 discloses a persona chatbot control method performed by at least one processor, the method including the steps of receiving a user utterance, adding the user utterance to a prompt including a description and related instruction sentence regarding the chatbot character, encoding the prompt, and inputting the encoded prompt into a language model to generate a chatbot utterance in response to the user utterance. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2022-180282 A Summary of the Invention [Problem to be solved by the invention]

[0004] In messaging applications, there is a problem that users spend a long time thinking about replying, making it difficult to communicate efficiently. [Means for solving the problem]

[0005] To solve this problem, we provide a system that reads the user's reply habits from their messaging history and analyzes the flow of conversation between the user and the other person. We then provide a system that generates reply suggestions that are appropriate for the user based on the analysis results and presents the generated reply suggestions to the user. This allows users to significantly reduce the time they spend struggling to reply, and achieves efficient communication. [Brief description of the drawings]

[0006] [Figure 1] 1 is a conceptual diagram showing an example of the configuration of a data processing system according to a first embodiment. [Diagram 2] 1 is a conceptual diagram showing an example of main functions of a data processing device and a smart device according to a first embodiment. FIG. [Diagram 3] FIG. 11 is a conceptual diagram showing an example of the configuration of a data processing system according to a second embodiment. [Figure 4] FIG. 11 is a conceptual diagram showing an example of main functions of a data processing device and smart glasses according to a second embodiment. [Diagram 5] FIG. 13 is a conceptual diagram showing an example of the configuration of a data processing system according to a third embodiment. [Figure 6] FIG. 11 is a conceptual diagram showing an example of main functions of a data processing device and a headset-type terminal according to a third embodiment. [Figure 7] FIG. 13 is a conceptual diagram showing an example of the configuration of a data processing system according to a fourth embodiment. [Figure 8] FIG. 13 is a conceptual diagram showing an example of main functions of a data processing device and a robot according to a fourth embodiment. [Figure 9] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 10] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 11] FIG. 2 is a sequence diagram showing a process flow of the data processing system in the first embodiment of the first form example. [Figure 12] FIG. 11 is a sequence diagram showing a process flow of the data processing system in the application example 1 of the first embodiment. [Figure 13] FIG. 11 is a sequence diagram showing a process flow of the data processing system in the second embodiment of the second embodiment. [Figure 14] FIG. 13 is a sequence diagram showing a process flow of the data processing system in application example 2 of embodiment 2. [Figure 15] FIG. 13 is a sequence diagram showing a process flow of a data processing system in a third embodiment of the present invention. [Figure 16]FIG. 13 is a sequence diagram showing a process flow of the data processing system in the third application example of the third embodiment. [Figure 17] FIG. 13 is a sequence diagram showing the flow of processing in the data processing system in the first embodiment of the first form example when an emotion engine is combined. [Figure 18] FIG. 13 is a sequence diagram showing the flow of processing in the data processing system in Application Example 1 of Form Example 1 when combined with an emotion engine. [Figure 19] FIG. 13 is a sequence diagram showing the flow of processing in the data processing system according to the embodiment 2 of the second embodiment in the case where an emotion engine is combined. [Figure 20] FIG. 13 is a sequence diagram showing the flow of processing in the data processing system in Application Example 2 of Form Example 2 when combined with an emotion engine. [Figure 21] FIG. 13 is a sequence diagram showing the flow of processing in the data processing system in the embodiment 3 of the form example 3 in the case where an emotion engine is combined. [Figure 22] FIG. 13 is a sequence diagram showing the flow of processing in the data processing system in Application Example 3 of Form Example 3 when combined with an emotion engine. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0007] Hereinafter, an example of an embodiment of a system according to the technology of the present disclosure will be described with reference to the accompanying drawings.

[0008] First, the terms used in the following description will be explained.

[0009] In the following embodiments, a signed processor (hereinafter simply referred to as a "processor") may be one arithmetic device or a combination of multiple arithmetic devices. Furthermore, a processor may be one type of arithmetic device or a combination of multiple types of arithmetic devices. Examples of arithmetic devices include a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), a GPGPU (General-Purpose computing on Graphics Processing Units), an APU (Accelerated Processing Unit), or a TPU (TENSOR PROCESSING UNIT (registered trademark)).

[0010] In the following embodiments, a signed RAM (Random Access Memory) is a memory in which information is temporarily stored and is used as a working memory by a processor.

[0011] In the following embodiments, the coded storage is one or more non-volatile storage devices that store various programs, various parameters, etc. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), magnetic disks (e.g., hard disks), and magnetic tapes.

[0012] In the following embodiments, a communication I / F (Interface) with a code is an interface including a communication processor and an antenna. The communication I / F controls communication between multiple computers. An example of a communication standard applied to the communication I / F is a wireless communication standard including 5G (5th Generation Mobile Communication System), Wi-Fi (registered trademark), Bluetooth (registered trademark), etc.

[0013] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B." In other words, "A and / or B" means that it may be only A, only B, or a combination of A and B. In addition, in this specification, the same idea as "A and / or B" is also applied when three or more things are expressed by connecting them with "and / or."

[0014] [First embodiment]

[0015] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0016] 1, a data processing system 10 includes a data processing device 12 and a smart device 14. An example of the data processing device 12 is a server.

[0017] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a wide area network (WAN) and / or a local area network (LAN).

[0018] The smart device 14 includes a computer 36, a reception device 38, an output device 40, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The reception device 38, the output device 40, and the camera 42 are also connected to the bus 52.

[0019] The reception device 38 includes a touch panel 38A and a microphone 38B, and receives user input. The touch panel 38A detects contact with a pointer (e.g., a pen or a finger) to receive user input by the touch of the pointer. The microphone 38B detects the user's voice to receive user input by voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the data indicating the user input.

[0020] The output device 40 includes a display 40A and a speaker 40B, and presents data to the user 20 by outputting the data in a form of expression that the user 20 can perceive (e.g., voice and / or text). The display 40A displays visible information such as text and images according to instructions from the processor 46. The speaker 40B outputs voice according to instructions from the processor 46. The camera 42 is a compact digital camera equipped with an optical system including a lens, an aperture, and a shutter, and an imaging element such as a Complementary Metal-Oxide-Semiconductor (CMOS) image sensor or a Charge Coupled Device (CCD) image sensor.

[0021] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for transmitting and receiving various types of information between the processor 46 and the processor 28 via the network 54.

[0022] FIG. 2 shows an example of main functions of the data processing device 12 and the smart device 14.

[0023] As shown in Fig. 2, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32. The specific process program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific process program 56 from the storage 32, and executes the read specific process program 56 on the RAM 30. The specific process is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific process program 56 executed on the RAM 30.

[0024] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0025] In the smart device 14, the processor 46 performs the reception output processing. The storage 50 stores a reception output program 60. The reception output program 60 is used together with the specific processing program 56 by the data processing system 10. The processor 46 reads out the reception output program 60 from the storage 50, and executes the read reception output program 60 on the RAM 48. The reception output processing is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0026] Next, the specific processing by the specific processing unit 290 of the data processing device 12 will be described.

[0027] "Example 1"

[0028] In one embodiment of the present invention, a means is provided for reading a user's own reply habits from the user's messaging history. Specifically, the contents, sending times, sending frequency, etc. of messages sent by the user in the past are analyzed to extract the user's reply habits. For example, if the user frequently uses the phrase "thank you," the use of the phrase "thank you" is extracted as the user's reply habits.

[0029] "Example 2"

[0030] It also provides a means to analyze the flow of conversation between users and other people. Specifically, it analyzes the patterns of message exchanges between users and other people, changes in topics, changes in emotions, etc. to understand the flow of the conversation. For example, if a user always expresses gratitude after receiving a question from the other person, this pattern can be extracted and the flow of the conversation analyzed.

[0031] "Example 3"

[0032] Furthermore, a means is provided for generating reply suggestions that are appropriate for the user based on the analysis results. Specifically, reply suggestions that are appropriate for the user are automatically generated based on the analysis results of the user's reply habits and the flow of the conversation. For example, if a user frequently uses the phrase "thank you" and has a pattern of expressing gratitude after receiving a question from the other person, a reply suggestion of "Thank you, about that..." is generated based on that information.

[0033] "Example 4"

[0034] Finally, a means is provided for presenting the generated reply suggestions to the user. Specifically, the generated reply suggestions are displayed on the user's device and the user can select one. For example, the generated reply suggestions are displayed at the bottom of the message input field, and the user can insert the reply suggestion into the message input field by tapping it. This allows the user to significantly reduce the time spent thinking about replying and achieve efficient communication.

[0035] The process flow of each embodiment will be described below.

[0036] "Example 1"

[0037] Step 1: Obtain the user's messaging history. Specifically, obtain information from the database about the user's past messages, such as the content of the messages, the time they were sent, and the frequency of their messages.

[0038] Step 2: Extract the user's own reply habits from the acquired messaging history. Specifically, the system analyzes the phrases the user frequently uses, messages sent at specific times, and reply patterns to specific people to extract the user's reply habits.

[0039] "Example 2"

[0040] Step 1: Obtain the messaging history between the user and the other party. Specifically, obtain message exchange information between the user and the other party from the database.

[0041] Step 2: Analyze the flow of conversation from the acquired messaging history. Specifically, analyze message exchange patterns, topic changes, and emotional changes to understand the flow of conversation.

[0042] "Example 3"

[0043] Step 1: Obtain the analysis results of the user's reply habits and the flow of the conversation. Specifically, obtain the information extracted and analyzed using the methods described above.

[0044] Step 2: Generate reply suggestions that are appropriate for the user based on the obtained analysis results. Specifically, the system automatically generates reply suggestions that are appropriate for the user, taking into account the user's reply habits and the flow of the conversation.

[0045] "Example 4"

[0046] Step 1: Obtain the generated reply proposal. Specifically, obtain the reply proposal generated by the above-mentioned means.

[0047] Step 2: Present the retrieved reply suggestions to the user. Specifically, the generated reply suggestions are displayed on the user's device and the user can select one.

[0048] Example 1

[0049] Next, a description will be given of Example 1 of the embodiment 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal".

[0050] In conventional messaging systems, it was difficult to automatically generate reply suggestions that took into account the user's reply habits. As a result, users had to think up the reply content themselves every time, which was time-consuming and caused problems such as inconsistent replies. Furthermore, there was a lack of a way to effectively utilize the user's messaging history, making it impossible to provide personalized reply suggestions based on the user's past behavior.

[0051] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0052] In this invention, the server includes means for reading a user's own reply habits from the user's messaging history, means for analyzing the flow of a dialogue between the user and the other party, means for generating reply suggestions that suit the user based on the analysis results, means for presenting the generated reply suggestions to the user, means for collecting the user's messaging history, means for preprocessing the collected messaging history, means for analyzing the preprocessed message contents, means for extracting the user's reply habits based on the analysis results, and means for saving the extraction results. This makes it possible to automatically generate personalized reply suggestions based on the user's past messaging history and present them to the user.

[0053] "User's messaging history" is a collection of data about messages the user has sent in the past, such as the content of those messages, the time they were sent, and the frequency with which they were sent.

[0054] "Reply habits" are patterns such as phrases that a user frequently uses when replying to messages or a tendency to reply at certain times of the day.

[0055] "Dialogue flow" refers to the order and content relevance of messages exchanged between a user and another person.

[0056] "Analysis results" are the results of analyzing users' reply habits and conversation flow based on collected and preprocessed messaging history.

[0057] "Suggested replies" are candidate messages for a user to reply to, generated based on the user's past messaging history and analysis results.

[0058] "Preprocessing" refers to the process of removing unnecessary spaces and special characters from the collected messaging history, and performing tokenization and removing stop words.

[0059] A "natural language processing library" is a software tool for tokenizing message content, analyzing sentiment, extracting frequent phrases, and so on.

[0060] "Database" refers to an information management system for storing users' messaging history and analysis results.

[0061] A "prompt sentence" is an instruction sentence to be input into a generative AI model, which clearly indicates the message history to be analyzed and the purpose of the analysis.

[0062] A "generative AI model" is an artificial intelligence model that generates reply suggestions that are appropriate for a user based on the user's messaging history and analysis results.

[0063] The present invention relates to a system for analyzing a user's messaging history, extracting the user's reply habits, and generating reply suggestions that are suited to the user. A specific embodiment of this system will be described below.

[0064] The server first collects the user's messaging history, including the content of messages the user has sent in the past, the time they were sent, how often they were sent, etc. The collected data is then stored in a database.

[0065] The server then pre-processes the collected messaging history, which includes cleaning the message content (removing unnecessary spaces and special characters), tokenizing (splitting into words), and removing stop words (removing frequent but meaningless words). This pre-processing is done using Python, with natural language processing libraries such as NLTK and spaCy.

[0066] Once preprocessed, the message content is then analyzed. The server uses natural language processing libraries to extract common phrases and words. It also performs sentiment analysis on the message and classifies it as positive, negative, or neutral.

[0067] Based on the analysis results, the server extracts the user's reply habits. For example, if a user frequently uses the phrase "thank you," that phrase is identified as the user's reply habits. Also, if there are many replies during a certain time period, that time period is also extracted as a habit.

[0068] The extracted user reply habits are stored in a database, and this information is then fed into a generative AI model as prompts to generate reply suggestions that are more likely to be user-friendly.

[0069] As a concrete example, consider the case where you want to analyze the history of messages sent by user A. Suppose user A's message history contains the following data:

[0070] Message 1: "Thank you, that was helpful."

[0071] Message 2: "Thank you for your hard work."

[0072] Message 3: "Thank you, I look forward to working with you again."

[0073] The server analyzes these messages and extracts the frequent use of the phrase "Thank you." As a result, the use of the phrase "Thank you" is identified as a reply habit of User A.

[0074] Examples of prompts to input to a generative AI model might include the following:

[0075] Analyze the user's messaging history and extract frequent phrases. The following message histories will be analyzed:

[0076] 1. "Thank you, that was helpful."

[0077] 2. "Thank you for your hard work."

[0078] 3. "Thank you, I look forward to working with you again."

[0079] By inputting this prompt into a generative AI model, it is possible to automatically extract the user's reply habits and generate reply suggestions that are appropriate for the user.

[0080] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0081] Step 1:

[0082] The server collects the messaging history of users. As input, a user ID is given. The server retrieves data such as message content, sending time, and sending frequency for that user from the database. As output, the collected messaging history is obtained. Specifically, the server executes SQL queries to extract the required information from the database.

[0083] Step 2:

[0084] The server preprocesses the collected messaging history. As input, it is given the collected messaging history. The server cleans the message content (removes unnecessary spaces and special characters), tokenizes it (splits it into words), and removes stop words. As output, it gets the preprocessed message content. Specifically, the server runs a Python script to perform preprocessing using natural language processing libraries (NLTK and spaCy).

[0085] Step 3:

[0086] The server analyzes the preprocessed message content. The preprocessed message content is given as input. The server uses a natural language processing library to extract frequent phrases and words and perform sentiment analysis. The analysis results are obtained as output. Specifically, the server runs a Python script to extract frequent phrases and perform sentiment analysis.

[0087] Step 4:

[0088] The server extracts the user's reply habits based on the analysis results. The analysis results are given as input. The server identifies frequent phrases and tendencies to reply at specific times of the day. The output is the user's reply habits. In concrete terms, the server analyzes the analysis results and runs an algorithm to identify the user's reply habits.

[0089] Step 5:

[0090] The server saves the extracted user's reply habits in a database. As input, the user's reply habits are given. The server performs an insert operation on the database and saves the extracted results. As output, it is confirmed that the saving has been completed. As a specific operation, the server executes a SQL query to insert data into the database.

[0091] Step 6:

[0092] The server inputs a prompt sentence into the generative AI model to generate a reply proposal that is appropriate for the user. The user's reply habits and the prompt sentence are given as input. The server inputs the prompt sentence into the generative AI model to generate a reply proposal. The generated reply proposal is obtained as output. In concrete terms, the server sends a prompt sentence to the generative AI model and obtains a reply proposal.

[0093] Step 7:

[0094] The server presents the generated reply proposal to the user. As input, the generated reply proposal is given. The server sends the reply proposal to the user's device and presents it to the user. As output, a state is obtained where the user can confirm the reply proposal. As a specific operation, the server sends a notification to the user's device and displays the reply proposal.

[0095] (Application example 1)

[0096] Next, a description will be given of Application Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0097] Conventional messaging systems do not recommend personalized content that takes into account the user's reply habits. This makes it difficult to efficiently provide content that is of interest to the user. In addition, there has been no system that integrates the generation of reply suggestions based on the user's reply habits and content recommendation.

[0098] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0099] In this invention, the server includes means for reading the user's own reply habits from the user's messaging history, means for analyzing the flow of a conversation between the user and another person, means for generating reply suggestions that are suited to the user based on the analysis results, means for presenting the generated reply suggestions to the user, means for recommending personalized content based on the user's reply habits, and means for presenting the recommended content to the user. This makes it possible to recommend personalized content that takes into account the user's reply habits, and efficiently provide content that is more interesting to the user.

[0100] "User's messaging history" is a collection of data about messages the user has sent in the past, such as the content of those messages, the time they were sent, and the frequency with which they were sent.

[0101] "Reply habits" are patterns of phrases or words that a user frequently uses when replying to messages.

[0102] A "talk flow" is a process of continuous exchange of messages between a user and another person.

[0103] "Analysis Results" refers to information or data obtained after analyzing a user's messaging history and conversation flow.

[0104] A "suggested reply" is a suggested message that a user can use as a reference when replying.

[0105] "Personalized Content" means information or entertainment content that is specifically selected based on a user's individual characteristics and interests.

[0106] A "recommendation means" is a method or technology for selecting and presenting appropriate content or information to a user.

[0107] A system for implementing this invention includes means for reading a user's own reply habits from the user's messaging history, means for analyzing the flow of conversation between the user and another party, means for generating reply suggestions that are typical of the user based on the analysis results, means for presenting the generated reply suggestions to the user, means for recommending personalized content based on the user's reply habits, and means for presenting the recommended content to the user.

[0108] The server runs software to collect and analyze users' messaging history. Specifically, it uses programming languages ​​such as Python to analyze the content, time and frequency of users' messages and extracts their reply habits. This analysis is carried out using data processing tools such as collections.Counter.

[0109] The server then generates a reply that is appropriate for the user based on the analysis results, and presents the reply to the user, who can use it as is or customize it in his or her own way.

[0110] In addition, the server recommends personalized content based on the user's reply habits. For this recommendation, the server retrieves content that corresponds to the user's reply habits from the database and presents it to the user. For example, if the user frequently uses the phrase "thank you," the server recommends content such as "How to express gratitude" and "10 ways to say thank you."

[0111] This system is implemented using a smartphone. An application is installed on the smartphone to collect the user's messaging history and send it to a server. The application periodically sends the user's messaging history to the server, receives reply suggestions and recommended content from the server, and presents them to the user.

[0112] As a concrete example, by inputting the following prompt sentence into a generative AI model, personalized content recommendations based on the user’s reply habits can be realized:

[0113] "Build an application that extracts reply habits from a user's messaging history and recommends personalized content based on those habits. For example, if a user frequently uses the phrase 'Thank you,' recommend content like 'How to express gratitude' or '10 ways to say thank you.'"

[0114] In this way, it is possible to leverage a user's messaging history to provide them with the most relevant content.

[0115] The flow of the specific process in the application example 1 will be described with reference to FIG.

[0116] Step 1:

[0117] The server collects the user's messaging history. It receives data such as message content, sending time, and sending frequency from the user's device and stores them in a database. The input is the user's messaging history, and the output is the stored message data.

[0118] Step 2:

[0119] The server analyzes the collected messaging history. Using a programming language such as Python, it counts the frequency of message content using collections.Counter. The input is the stored message data, and the output is data that indicates the user's reply habits.

[0120] Step 3:

[0121] The server generates reply suggestions that are appropriate for the user based on the analysis results. Using a generative AI model, it generates multiple reply suggestions based on the user's reply habits. The input is data that indicates the user's reply habits, and the output is the generated multiple reply suggestions.

[0122] Step 4:

[0123] The server presents the generated reply suggestions to the user's terminal. The server transmits reply suggestions to the user's terminal for the user to select or arrange. The input is the generated multiple reply suggestions, and the output is the reply suggestions presented to the user.

[0124] Step 5:

[0125] The server recommends personalized content based on the user's reply habits. It retrieves content corresponding to the user's reply habits from a database and creates a recommendation list. The input is data indicating the user's reply habits, and the output is a recommended content list.

[0126] Step 6:

[0127] The server presents the recommended content to the user's terminal. The server transmits the recommended content to the user's terminal so that the user can view it. The input is the recommended content list, and the output is the content presented to the user.

[0128] Step 7:

[0129] The user reviews the suggested replies and recommended content, selects or arranges a suggested reply as necessary, and views the recommended content. The input is the suggested replies and content presented to the user, and the output is the user's selected or arranged reply and viewed content.

[0130] Example 2

[0131] Next, a second embodiment of the second embodiment will be described. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0132] Conventional messaging systems generate reply suggestions without considering the user's reply patterns or changes in emotions, so they often present replies that are unnatural to the user. In addition, there is a lack of means to analyze the flow of conversation between the user and the other person, making it difficult to generate reply suggestions that appropriately reflect changes in topic or emotions.

[0133] The identification process by the identification processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means. In this invention, the server includes a means for reading a user's own reply pattern from the user's messaging history, a means for collecting message exchanges between the user and the other party, a means for pre-processing the collected message data, a means for tokenizing the pre-processed messages and tagging them with parts of speech, a means for analyzing the sentiment of each message, a means for analyzing message exchange patterns, topic changes, and sentiment changes, a means for generating reply suggestions that are appropriate for the user based on the analysis results, and a means for presenting the generated reply suggestions to the user. This makes it possible to generate natural reply suggestions that take into account the user's reply patterns and sentiment changes.

[0134] "User's messaging history" means a record of messages that a User has sent or received in the past.

[0135] "Reply patterns" are the tendencies of the reply format and content that a user repeatedly uses in response to a particular situation or topic.

[0136] "Message exchange" refers to the sending and receiving of a series of messages between a User and another person.

[0137] "Preprocessing" refers to the process of removing unnecessary spaces and special characters from the message data and preparing the data in a format suitable for analysis.

[0138] "Tokenization" is the process of dividing a message into its smallest units, such as words or phrases.

[0139] "Part-of-speech tagging" is the process of identifying and assigning a part of speech (noun, verb, adjective, etc.) to a tokenized word or phrase.

[0140] "Sentiment analysis" is the process of extracting emotions (happiness, sadness, anger, etc.) from the content of a message and evaluating their intensity.

[0141] "Topic changes" refers to how the topic changes during a message exchange.

[0142] "Analysis results" refers to information obtained from analysis of message exchange patterns, changes in topics, changes in sentiment, etc.

[0143] A "suggested reply" is a suggested message that a user can use to reply to the other person.

[0144] "Presenting" is the act of showing the generated reply suggestions to the user.

[0145] This invention is a system that reads a user's own reply patterns from the user's messaging history, analyzes message exchanges between the user and other parties, generates reply suggestions that are appropriate for the user based on the analysis results, and presents the generated reply suggestions to the user.

[0146] The server first collects message data between users and their contacts. This data is retrieved from a database such as MySQL or MongoDB. The server then pre-processes the collected message data. Pre-processing includes removing unnecessary whitespace and special characters.

[0147] The preprocessed message data is tokenized using natural language processing libraries such as SpaCy and NLTK, and each token is tagged with a part of speech. For example, the message "Hello, how are you today?" is tagged as "Hello / NOUN How are you today / ADV How are you today / VERB AUX".

[0148] The server then uses a sentiment analysis tool such as VADER or TextBlob to calculate a sentiment score for each message. For example, the message "Hello, how are you today?" is calculated to have a sentiment score of 0.5.

[0149] The server then analyzes messaging patterns, topic changes, and sentiment shifts — for example, finding that a user always says "Thank you" after receiving a question from the other person.

[0150] Based on the analysis results, the server generates a reply that is appropriate for the user. This reply is generated taking into account the user's past reply patterns and changes in emotions. The generated reply is presented to the user. The user can use the proposed reply as is or arrange it in their own way.

[0151] As a concrete example, consider the following message exchange:

[0152] Other person: "Hello, how are you today?"

[0153] User: "Hi, I'm fine. How about you?"

[0154] The other person: "I'm fine. Has anything new happened to you lately?"

[0155] User: "Yes, I've started a new project. Thank you."

[0156] An example prompt to parse this interaction is:

[0157] Analyze the messages exchanged between users and their contacts and extract the following information:

[0158] 1. Message exchange patterns

[0159] 2. Change the topic

[0160] 3. Emotional changes

[0161] Message exchange:

[0162] Other person: "Hello, how are you today?"

[0163] User: "Hi, I'm fine. How about you?"

[0164] The other person: "I'm fine. Has anything new happened to you lately?"

[0165] User: "Yes, I've started a new project. Thank you."

[0166] By feeding this prompt into a generative AI model, results can be obtained that analyze message exchange patterns, topic changes, and emotional shifts.

[0167] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0168] Step 1: Collect message data

[0169] The server retrieves the message data between the user and the other party from the database. Specifically, it executes the query "SELECT FROM messages WHERE user_id = '12345'" to the MySQL database to collect the message exchanges between the user and the other party. The input is the user ID, and the output is the corresponding message data.

[0170] Step 2: Preprocessing the message data

[0171] The server pre-processes the collected message data. Specifically, it removes unnecessary whitespace and special characters to clean the message. For example, it converts the message "Hello! How are you today?" to "Hello, how are you today?". The input is the collected message data, and the output is the pre-processed message data.

[0172] Step 3: Tokenize the message and tag parts of speech

[0173] The server tokenizes the preprocessed message and tags each token with a part-of-speech tag. Specifically, it uses SpaCy and NLTK to tag the message "Hello, how are you today?" as "Hello / NOUN How are you today / ADV How are you today / VERB AUX." The input is the preprocessed message data, and the output is the tokenized and part-of-speech tagged data.

[0174] Step 4: Sentiment analysis

[0175] The server analyzes the sentiment of each message. Specifically, it uses VADER and TextBlob to calculate a sentiment score for each message. For example, the sentiment score of the message "Hello, how are you today" is calculated to be 0.5. The input is the tokenized and part-of-speech tagged data, and the output is the sentiment score.

[0176] Step 5: Analyze the flow of conversation

[0177] The server analyzes message exchange patterns, topic changes, and sentiment changes. Specifically, it finds a pattern where a user always says "Thank you" after receiving a question from the other party. The input is the sentiment score and tokenized data, and the output is the analysis result.

[0178] Step 6: Generate suggested replies

[0179] The server generates reply suggestions that are appropriate for the user based on the analysis results. Specifically, it generates reply suggestions by taking into account the user's past reply patterns and changes in emotions. The input is the analysis results, and the output is the generated reply suggestions.

[0180] Step 7: Propose a response

[0181] The server presents the generated reply to the user. Specifically, the reply is displayed to the user through a web browser. The user can use the proposed reply as is or arrange it in their own way. The input is the generated reply, and the output is the reply presented to the user.

[0182] (Application example 2)

[0183] Next, a description will be given of Application Example 2 of Form Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0184] Conventional messaging systems were unable to analyze users' reply habits or conversation flow, making it difficult to generate natural reply suggestions for users. They also lacked the ability to detect signs of fraud and misconduct, making it difficult to ensure user safety.

[0185] The identification process by the identification processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means. In this invention, the server includes a means for reading the reply habits of the user from the messaging history of the user, a means for analyzing the flow of the conversation between the user and the other party, a means for generating reply suggestions that are suitable for the user based on the analysis results, a means for presenting the generated reply suggestions to the user, a means for analyzing message patterns, topic changes, and emotional changes, and a means for detecting signs of misconduct or fraud. This makes it possible to generate reply suggestions that are natural to the user and to detect signs of misconduct or fraud at an early stage.

[0186] "User's messaging history" means a record of messages that a User has sent or received in the past.

[0187] "Reply habits" are the tendencies or patterns of how a user responds to messages.

[0188] "Chat flow" refers to the progress and changes in the message exchange between a user and the other person.

[0189] "Analysis results" refers to the information and data obtained after analyzing message exchanges.

[0190] "Suggested replies" refers to suggestions or candidates for a user to reply to a message.

[0191] A "message pattern" refers to a recurring format or structure in a message exchange.

[0192] "Topic changes" refers to how the topic changes during a message exchange.

[0193] "Changes in emotion" refers to how emotions change during message exchange.

[0194] "Fraud" refers to committing fraud or malicious acts against users.

[0195] "Fraud indicators" are signs or patterns in messaging that indicate the possibility of fraudulent activity.

[0196] In order to implement this invention, it is necessary to build a system in which the elements of the server, terminals, and users work together.

[0197] First, the server collects the user's messaging history and reads the user's own reply habits. This requires software to store and analyze the messaging history using a database management system (DBMS). Specifically, it uses a programming language such as Python or R to analyze message patterns and changes in sentiment using natural language processing (NLP) libraries (e.g., NLTK, spaCy).

[0198] Next, the server analyzes the flow of the conversation between the user and the other party. To do this, a generative AI model is used to detect changes in the topic of the message and changes in sentiment. Advanced natural language processing models such as GPT-3 (registered trademark) and BERT are suitable as generative AI models. These models are used to analyze message exchanges in real time and understand the flow of the conversation.

[0199] Furthermore, the server generates reply suggestions that are appropriate for the user based on the analysis results. The generated reply suggestions are customized based on the user's past reply patterns. In order to generate multiple reply suggestions and present them to the user, a user interface (UI) needs to be designed. The UI is often implemented as a smartphone app or a web application.

[0200] The server also analyzes message patterns, topic changes, and sentiment changes to detect signs of misconduct or fraud. To do this, we use machine learning algorithms (e.g., Naive Bayes, SVM) to build a system that detects anomalous message patterns. If an anomaly is detected, we also need the ability to alert users.

[0201] As a concrete example, consider the following message exchange:

[0202] "Hello, how are you?"

[0203] "Yeah, I'm fine. How about you?"

[0204] "I'm fine too. By the way, how are you doing lately?"

[0205] "Very well. By the way, can you give me your bank details?"

[0206] In this case, a sudden change in emotion is detected and deemed to be an indication of fraudulent activity.

[0207] Example of a prompt to be input to a generative AI model:

[0208] It analyzes messaging between users and their contacts to provide a way to understand the flow of conversations, specifically by analyzing message patterns, topic changes, and sentiment changes to detect signs of fraud or misconduct.

[0209] In this way, it is possible to realize a system that can generate reply suggestions that are natural to the user and can also detect signs of misconduct or fraud at an early stage.

[0210] The flow of the specific process in the application example 2 will be described with reference to FIG.

[0211] Step 1:

[0212] The server collects the messaging history of users. As input, it takes the message data that the user has sent and received in the past and stores it in a database. Specifically, it uses a database management system (DBMS) to store the messaging history and build a dataset for analysis. As output, it obtains the messaging history data in a format that can be analyzed.

[0213] Step 2:

[0214] The server reads the user's own reply habits from the user's messaging history. As input, it uses the messaging history data collected in step 1. Specifically, it uses a natural language processing (NLP) library (e.g., NLTK, spaCy) to analyze message patterns and reply trends. As output, it obtains data about the user's reply habits.

[0215] Step 3:

[0216] The server analyzes the flow of conversation between the user and the other party. As input, it receives message data sent and received in real time. Specifically, it uses generative AI models (e.g. GPT-3, BERT) to detect changes in the topic and sentiment of messages. As output, it obtains the analysis results on the flow of conversation.

[0217] Step 4:

[0218] The server generates reply suggestions that are appropriate for the user based on the analysis results. As input, it uses the data obtained in steps 2 and 3. Specifically, it uses a generative AI model to generate reply suggestions based on the user's past reply patterns. As output, it obtains multiple reply suggestions.

[0219] Step 5:

[0220] The server presents the generated reply suggestions to the user. As input, it uses the reply suggestions generated in step 4. Specifically, it displays the reply suggestions in a smartphone app or web application through a user interface (UI). As output, it presents the user with reply suggestions that they can select from.

[0221] Step 6:

[0222] The server analyzes message patterns, topic changes, and sentiment changes to detect signs of misconduct or fraud. As input, it uses real-time messaging data. Specifically, it uses machine learning algorithms (e.g., Naive Bayes, SVM) to detect anomalous message patterns. As output, it provides warnings about signs of misconduct or fraud.

[0223] Step 7:

[0224] The server alerts the user if any indication of misconduct or fraud is detected. As input, it uses the alert data obtained in step 6. Specifically, it displays a warning message through a user interface in a smartphone app or web application. As output, a warning is presented to the user.

[0225] Example 3

[0226] Next, a third embodiment of the third embodiment will be described. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0227] Conventional messaging systems have the problem that it takes time for users to think about what to reply, making efficient communication difficult. It is also difficult to generate automatic reply suggestions that take into account the user's reply habits and the flow of the conversation. This increases the time users spend thinking about how to reply, which reduces the efficiency of communication.

[0228] The specific processing by the specific processing unit 290 of the data processing device 12 in the third embodiment is realized by the following means.

[0229] In this invention, the server includes a means for reading a user's own reply habits from the user's messaging history, a means for analyzing the flow of conversation between the user and the other party, a means for generating reply suggestions that are suitable for the user based on the analysis results, a means for displaying the generated reply suggestions on the user's device so that the user can select one, a means for generating reply suggestions using a generative AI model, and a means for inputting prompt sentences into the generative AI model. This allows the user to significantly reduce the time spent struggling with replying, enabling efficient communication.

[0230] "User's messaging history" means a record of messages that a User has sent or received in the past.

[0231] "Reply habits" are reply patterns or phrases that a user repeatedly uses in particular situations or with particular people.

[0232] "Chat flow" refers to the continuity and context of messages exchanged between a user and the other person.

[0233] "Analysis Results" refers to data or information obtained after analyzing a User's messaging history and conversation flow.

[0234] "Suggested replies" are suggested messages that users can use as a reference when replying.

[0235] A "generative AI model" is a model that uses artificial intelligence to analyze data and generate new data and information.

[0236] A "prompt" is an instruction or question that is input into a generative AI model.

[0237] "User Device" refers to any electronic device used by a User, such as a computer, smartphone, or tablet.

[0238] The present invention is a system that analyzes a user's messaging history and automatically generates reply suggestions based on the user's reply habits and conversation flow. A specific embodiment of this system will be described below.

[0239] Hardware and Software Used

[0240] The server includes hardware and software for collecting and analyzing the messaging history of users. Specifically, the server may have the following configuration:

[0241] Hardware: High-performance server computer

[0242] Software: Analytical software using natural language processing (NLP) techniques, generative AI models (e.g., Google®'s BERT, OpenAI®'s GPT-3)

[0243] The terminal is a device that allows the user to check and select reply suggestions. Specifically, the following configuration is possible.

[0244] Hardware: Smartphones, tablets, computers

[0245] Software: Messaging application

[0246] Data processing and calculation

[0247] The server collects the user's messaging history and analyzes it using natural language processing technology. As a result of the analysis, frequently used phrases and reply patterns are extracted from the user. Based on this, reply suggestions are generated using a generative AI model.

[0248] The specific processing flow is as follows:

[0249] 1. Message Data Collection:

[0250] The server retrieves past message data through the API of the user's messaging application, and stores the retrieved data in a database.

[0251] 2. Message data analysis:

[0252] The server inputs prompts into the generative AI model and analyzes the user's message data. As a result of the analysis, frequently used phrases and reply patterns are extracted.

[0253] 3. Generate reply suggestions:

[0254] The server generates reply suggestions based on the analysis results, using a generative AI model to generate reply suggestions that are suited to the user's reply habits and the flow of the conversation.

[0255] 4. Proposal of a response:

[0256] The device presents the generated reply suggestions to the user by displaying the reply suggestions below the message input field so that the user can select one.

[0257] Examples and prompts

[0258] If a user frequently uses the phrase "thank you" and has a pattern of expressing gratitude after being asked a question, the following prompt sentence can be input into the generative AI model:

[0259] Example prompt:

[0260] User's previous messages:

[0261] 1. "Thank you for answering my question."

[0262] 2. "Thanks, about that..."

[0263] 3. "Thanks, next time..."

[0264] The other person asks: "What do you think about this?"

[0265] Generate reply: "Thanks, about that..."

[0266] The server inputs this prompt sentence into a generative AI model to generate a reply proposal. The generated reply proposal is sent to the terminal and displayed at the bottom of the message input field. When the user taps on a reply proposal, the content is inserted into the message input field. This allows the user to significantly reduce the time spent thinking about a reply, and enables efficient communication. The flow of the identification process in the third embodiment will be described with reference to FIG. 15.

[0267] Step 1:

[0268] The server collects the user's messaging history.

[0269] Input: Historical messaging data obtained through the API of the user's messaging application.

[0270] Data processing: The acquired message data is stored in a database.

[0271] Output: The saved message data.

[0272] What happens: The server calls the messaging application's API to retrieve the user's past messages and save them in a database.

[0273] Step 2:

[0274] The server analyzes the collected message data.

[0275] Input: The saved message data.

[0276] Data calculation: Input the prompt sentence into the generative AI model and analyze the user's message data.

[0277] Output: A pattern of frequent phrases or replies used by users.

[0278] How it works: The server inputs prompts into the generative AI model and analyzes the user's message data. As a result of the analysis, it extracts frequently used phrases and reply patterns from the user.

[0279] Step 3:

[0280] The server generates a proposed reply based on the analysis.

[0281] Input: A pattern of frequently used phrases or replies by users.

[0282] Data calculation: Using a generative AI model, we generate reply suggestions that are suited to the user's reply habits and the flow of the conversation.

[0283] Output: The generated suggested replies.

[0284] Specific operation: Based on the analysis results, the server inputs the prompt sentence into the generative AI model to generate reply suggestions, which are then stored in a database.

[0285] Step 4:

[0286] The server transmits the generated reply proposal to the user's device.

[0287] Input: The generated suggested reply.

[0288] Data processing: Converting the proposed reply into a data format for sending to the user's device.

[0289] Output: The suggested reply sent to the user's device.

[0290] Specific behavior: The server converts the generated reply proposal into an appropriate data format and sends it to the user's device.

[0291] Step 5:

[0292] The terminal presents the generated reply suggestions to the user.

[0293] Input: The suggested reply sent to the user's device.

[0294] Data processing: Display reply suggestions at the bottom of the message input field.

[0295] Output: Suggested replies displayed in the message input field.

[0296] Specific operation: The device displays reply suggestions received from the server at the bottom of the message input field. When the user taps a reply suggestion, the reply is inserted into the message input field.

[0297] (Application example 3)

[0298] Next, a description will be given of Application Example 3 of Form Example 3. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0299] In virtual stores, responding to customer inquiries quickly and appropriately is important to improve customer satisfaction. However, it takes time and effort for store clerks to respond to each customer individually, making efficient operation difficult. Furthermore, if the store clerk's responses are inconsistent, it may undermine customer trust. To solve these problems, a system is needed that can automatically generate replies from store clerks and provide efficient and consistent customer service.

[0300] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 3 is realized by the following means.

[0301] In this invention, the server includes means for reading a user's own reply habits from the user's messaging history, means for analyzing the flow of conversation between the user and the other party, means for generating reply suggestions that are characteristic of the user based on the analysis results, means for presenting the generated reply suggestions to the user, means for automatically generating replies for store clerks to streamline customer service in the virtual store, and means for displaying the generated reply suggestions on a smartphone application so that the user can select one. This makes it possible to automatically generate quick and appropriate replies to customer inquiries, reduce the burden on store clerks, and provide consistent customer service.

[0302] "User's messaging history" means a record of messages that a User has sent or received in the past.

[0303] "Reply habits" are patterns such as words and phrases that a user frequently uses when replying to messages, as well as the timing and order of replies.

[0304] "Chat flow" refers to the progress and context of a series of messages exchanged between a user and another person.

[0305] "Analysis Results" refers to information or data obtained after analyzing a User's messaging history and conversation flow.

[0306] A "suggested reply" is an automatically generated reply as a candidate for a message that the user may send.

[0307] A "virtual store" is a virtual store operated on the Internet where customers can purchase goods and services online.

[0308] "Customer service" refers to the task of responding to inquiries and requests from customers in the virtual store.

[0309] A "smartphone application" is a software program that runs on a smartphone.

[0310] "Automatic generation" means that the system generates suggested responses based on user input and past data, without manual intervention.

[0311] "Consistency" means always maintaining the same quality and standards in customer service.

[0312] To put this invention into practice, it is necessary to build a system that reads reply habits from a user's messaging history, analyzes the flow of the conversation, and generates reply suggestions that suit the user based on the analysis results. This system has the function of automatically generating replies for store clerks and displaying the generated reply suggestions on a smartphone application to streamline customer service in a virtual store.

[0313] Hardware and software used

[0314] Hardware: Smartphone

[0315] Software: OpenAI API, Python

[0316] Data processing and calculation

[0317] 1. Data Collection: The server collects the user's messaging history and identifies the user's reply habits, including the words and phrases the user uses most frequently, and the timing and sequence of replies.

[0318] 2. Conversation flow analysis: The server analyzes the conversation flow between the user and the other party to understand the context and progress.

[0319] 3. Reply generation: Based on the analysis results, the server uses a generative AI model to generate reply suggestions that are appropriate for the user. At this time, the prompt sentence is input to the generative AI model.

[0320] 4. Presentation of reply suggestions: The server displays the generated reply suggestions on the smartphone application, allowing the user to select one.

[0321] Examples

[0322] For example, if a customer asks "Do you have this item in stock?" in a virtual store, the server will generate a suggested reply such as "Thank you. We do have this item in stock..." based on the store clerk's past reply history and patterns.

[0323] Examples of prompt statements

[0324] User response history: Thank you. This item is in stock...

[0325] User response pattern: Question followed by a thank you

[0326] Generate the following reply:

[0327] In this way, the server can automatically generate prompt and appropriate replies to customer inquiries, reducing the burden on store clerks and achieving consistent customer service.

[0328] The flow of the specific process in the application example 3 will be described with reference to FIG.

[0329] Step 1:

[0330] The server collects the user's messaging history, i.e., records of messages the user has sent and received in the past, from a database. This input data includes the words and phrases the user frequently uses, the timing and sequence of replies, etc. As an output, a dataset is obtained that can be used to identify the user's reply habits.

[0331] Step 2:

[0332] The server analyzes the conversation flow between the user and the other party by analyzing the collected messaging history to understand the context and progress. The input data includes the sequence of messages exchanged between the user and the other party. The output is the analysis result of the conversation flow.

[0333] Step 3:

[0334] Based on the analysis results, the server uses a generative AI model to generate reply suggestions that are appropriate for the user. Specifically, the prompt sentence is input into the generative AI model, which generates appropriate reply suggestions. This input data includes the user's reply habits and the analysis results of the conversation flow. The output is a reply suggestion that is appropriate for the user.

[0335] Step 4:

[0336] The server displays the generated reply suggestions on the smartphone application. Specifically, the server transmits the generated reply suggestions to the user's device and displays them below the message input field of the application. The input data includes the generated reply suggestions. As an output, the smartphone screen displays the reply suggestions that the user can select.

[0337] Step 5:

[0338] The user can select a suggested reply or arrange it as they like. Specifically, the user can either insert the reply into the message input field by tapping on it, or edit the message based on the reply. The input data includes the reply selected by the user and the edited message. The output is the final reply message.

[0339] Furthermore, an emotion engine that estimates the emotion of the user may be combined. That is, the identification processing unit 290 may estimate the emotion of the user using the emotion identification model 59, and perform identification processing using the emotion of the user.

[0340] "Example 1"

[0341] As one embodiment of the present invention, a system is provided that includes a means for reading a user's own reply habits from the user's messaging history, a means for analyzing the flow of conversation between the user and another person, a means for generating reply suggestions that are typical of the user based on the analysis results, a means for presenting the generated reply suggestions to the user, and an emotion engine for recognizing the user's emotions.

[0342] "Example 2"

[0343] The emotion engine extracts changes in a user's emotions from the user's messaging history, specifically from the context and wording of messages, and the use of emojis and stamps.

[0344] "Example 3"

[0345] The emotion engine generates reply suggestions that take into account the changes in the extracted emotions. Specifically, if the user is expressing emotions such as happiness or sadness, the engine generates reply suggestions that match that emotion. For example, if the user is expressing happiness, the engine generates reply suggestions that also express happiness.

[0346] "Example 4"

[0347] The emotion engine extracts changes in a user's emotions in real time and generates reply suggestions in real time that take those changes into account. Specifically, every time a user sends a message, the emotion is extracted from the message and a reply suggestion based on that emotion is instantly generated.

[0348] The process flow of each embodiment will be described below.

[0349] "Example 1"

[0350] Step 1: Read the user's own reply habits from their messaging history.

[0351] Step 2: Analyze the flow of conversation between the user and the other party.

[0352] Step 3: Generate appropriate replies based on the analysis results.

[0353] Step 4: Present the generated reply suggestions to the user.

[0354] Step 5: Activate the emotion engine to recognize the user's emotions.

[0355] "Example 2"

[0356] Step 1: Extract user sentiment changes from users' messaging history.

[0357] Step 2: Generate reply suggestions that take into account the extracted emotion changes.

[0358] "Example 3"

[0359] Step 1: Extract user sentiment changes from users' messaging history.

[0360] Step 2: Generate reply suggestions that take into account the extracted emotion changes.

[0361] "Example 4"

[0362] Step 1: Extract real-time changes in user sentiment from users’ messaging history.

[0363] Step 2: Generate reply suggestions in real time that take into account the extracted emotion changes.

[0364] Example 1

[0365] Next, a description will be given of Example 1 of the embodiment 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal".

[0366] In conventional messaging systems, it was difficult for users to respond consistently, and it was especially difficult to generate appropriate replies when they were busy or emotional. In addition, there was no system that provided reply suggestions that took the user's emotions into account, so it was not possible to respond in accordance with the user's intentions and emotions. This could lead to a decline in the quality of communication and the occurrence of misunderstandings and problems.

[0367] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0368] In this invention, the server includes a means for reading a user's own reply habits from the user's messaging history, a means for analyzing the flow of conversation between the user and the other party, a means for generating reply suggestions that are appropriate for the user based on the analysis results, a means for presenting the generated reply suggestions to the user, and an emotion engine for recognizing the user's emotions. This allows the user to reply consistently, and further makes it possible to provide appropriate reply suggestions that take the user's emotions into consideration.

[0369] "User messaging history" refers to data such as the content of messages a user has sent in the past, the time they were sent, and the frequency of those messages.

[0370] "Reply habits" refers to characteristics such as the phrases and wording that a user frequently uses when replying to messages, as well as the timing and frequency of replies.

[0371] "Conversation flow" refers to information such as the continuity, context, and tone of messages exchanged between a user and the other person.

[0372] "Means for analyzing" refers to methods or devices that use natural language processing techniques and / or generative AI models to understand the content and context of a message and extract certain patterns or characteristics.

[0373] "Suggested replies" refers to sentences that are generated as candidates for messages that the user may send.

[0374] An "emotion engine" refers to technology or devices that analyze emotions contained in a user's messages and suggested replies and recognize the user's emotional state.

[0375] A "generative AI model" is a model for generating text using artificial intelligence technology, such as one that uses natural language processing technology.

[0376] "Presenting means" refers to a method or apparatus for displaying the generated suggested replies to a user.

[0377] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0378] This invention is a system that analyzes a user's messaging history, extracts the user's reply habits, and further analyzes the flow of a conversation between the user and the other party to generate reply suggestions that are appropriate for the user and present them to the user. A specific embodiment of this system will be described below.

[0379] Collecting messaging history

[0380] The server retrieves message data from the user's messaging application (e.g., a general messaging application). The retrieved data is stored in a database. This data includes the content of messages the user has sent in the past, the time they were sent, and how frequently they were sent.

[0381] Analyzing messaging history

[0382] The server analyzes the stored message data. It uses natural language processing (NLP) techniques to extract frequently used phrases and vocabulary used by users. For example, it uses Python's NLTK library for text analysis. This reveals the user's reply habits.

[0383] Talk flow analysis

[0384] The server uses a generative AI model (e.g., a general generative AI model) to analyze the flow of the conversation between the user and the other party, thereby obtaining basic data for understanding the context and tone of the conversation and generating appropriate reply suggestions.

[0385] Generate user-friendly reply suggestions

[0386] The server generates reply suggestions that are appropriate for the user based on the analysis results. Using a generative AI model, it creates reply suggestions that take into account the user's reply habits and the flow of the conversation. For example, if a user often uses the phrase "thank you," a reply suggestion that includes that phrase will be generated.

[0387] Suggested replies

[0388] The server sends the reply proposal it has generated to the user's device. The user can review the proposed reply and make corrections or approvals as necessary. If the user approves the reply proposal, the reply is sent to the other party.

[0389] Emotion recognition by emotion engine

[0390] The server uses an emotion engine to recognize the user's emotions. The emotion engine analyzes the emotions contained in the user's messages and reply suggestions to understand the user's emotional state, which provides feedback to generate more appropriate reply suggestions.

[0391] Examples

[0392] For example, if a user frequently uses the phrase "Otsukaresama desu" (Thank you for your hard work), the server will extract this phrase as the user's reply habit. If the user receives a message saying "How was the meeting today?", the server will generate a reply suggestion of "Otsukaresama desu. The meeting went well" and present it to the user.

[0393] Examples of prompt statements

[0394] "Analyze the user's messaging history and extract the user's reply habits. Then, analyze the conversation flow between the user and the other party and generate reply suggestions that are appropriate for the user. Finally, present the generated reply suggestions to the user and provide feedback using an emotion engine that recognizes the user's emotions."

[0395] The system allows users to respond more naturally and consistently, and also provides appropriate reply suggestions that take the user's feelings into account.

[0396] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0397] Step 1:

[0398] The server retrieves message data from the user's messaging application. As input, it receives data such as the content of the user's past messages, the time of sending, and the frequency of sending. It then stores this data in a database. Specifically, it uses an API to collect data from the messaging application and stores it in a database.

[0399] Step 2:

[0400] The server analyzes the stored message data. As input, it uses the message data stored in the database. It uses natural language processing (NLP) techniques to extract phrases and wording frequently used by users. As output, it obtains data showing the user's reply habits. Specifically, it uses Python's NLTK library to perform text analysis.

[0401] Step 3:

[0402] The server uses a generative AI model to analyze the flow of the conversation between the user and the other party. As input, it uses message data between the user and the other party. It uses a generative AI model (e.g., a general generative AI model) to understand the context and tone of the conversation. As output, it obtains data showing the flow of the conversation. Specifically, it calls the generative AI model via an API and obtains the analysis results.

[0403] Step 4:

[0404] The server generates reply suggestions that are appropriate for the user based on the analysis results. As input, it uses data showing the user's reply habits and conversation flow. It uses a generative AI model to create reply suggestions that take into account the user's reply habits and conversation flow. As output, it obtains multiple reply suggestions. In concrete terms, it inputs a prompt sentence into the generative AI model and generates reply suggestions.

[0405] Step 5:

[0406] The server sends the generated reply suggestions to the user's terminal, using the generated reply suggestions as input. The user reviews the proposed reply suggestions and modifies or approves them if necessary. As output, the reply approved by the user is obtained. The specific operations include displaying the reply suggestions through a user interface and receiving the user's selection.

[0407] Step 6:

[0408] The server uses the emotion engine to recognize the user's emotions. It uses the user's message and suggested replies as input. The emotion engine analyzes the emotions contained in the user's message and suggested replies to understand the user's emotional state. As output, it obtains data indicating the user's emotional state. Specifically, it calls the emotion engine via an API and obtains the emotion analysis results.

[0409] (Application example 1)

[0410] Next, a description will be given of Application Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0411] Conventional messaging systems were unable to generate personalized reply suggestions that took into account the user's reply habits and emotions. They also lacked the ability to recommend appropriate content based on the user's interests and emotions. This made it difficult for users to quickly respond in a way that suited them, and it took a long time to find content that matched their interests.

[0412] The identification process by the identification processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means. In this invention, the server includes a means for reading the user's own reply habits from the user's messaging history, a means for analyzing the flow of the conversation between the user and the other party, a means for generating a reply proposal that suits the user based on the analysis result, a means for presenting the generated reply proposal to the user, a means for recommending personalized content based on the user's interests and emotions, and a means for calculating the similarity between the user's message and content. This enables the user to quickly reply in a way that suits them, and further enables the user to efficiently find content that matches their interests.

[0413] "User's messaging history" means a record of messages that a User has previously sent or received.

[0414] "Reply habits" are the tendencies or patterns of how a user responds to messages.

[0415] A "talk flow" refers to a series of messages exchanged between a user and another person.

[0416] "Analysis Results" refers to data or information obtained after analyzing a User's messaging history and conversation flow.

[0417] "Suggested Responses" are suggestions or sample messages that a User may use to reply to a Message.

[0418] "Personalized content" refers to information and entertainment that is individually tailored to a user's interests and emotions.

[0419] "Similarity" is an indicator that shows the relevance or degree of match between a user's message and content.

[0420] A system for implementing this invention includes means for reading a user's own reply habits from the user's messaging history, means for analyzing the flow of conversation between the user and the other party, means for generating reply suggestions that are typical of the user based on the analysis results, means for presenting the generated reply suggestions to the user, means for recommending personalized content based on the user's interests and emotions, and means for calculating the similarity between the user's message and content.

[0421] The server stores the user's messaging history in a database and uses an analysis engine to extract the user's reply habits. The analysis engine uses TextBlob, a natural language processing library, and scikit-learn, a machine learning library. This allows the content, sending time, and sending frequency of messages sent by the user in the past to be analyzed, and the user's reply habits to be extracted.

[0422] Next, the server analyzes the flow of the conversation between the user and the other person and generates a reply that suits the user based on the analysis results. The generated reply is displayed on the user's smartphone or other device. The user can use the proposed reply as is or arrange it in their own way.

[0423] In addition, the server recommends personalized content based on the user's interests and emotions, which includes a method to calculate the similarity between the user's messages and content. The similarity calculation involves vectorizing the messages and content using CountVectorizer, and calculating the similarity using cosine_similarity.

[0424] For example, if a user sends a message such as "What do you think of recent movies?" or "I want to listen to new music," the system will recommend content such as "latest movies" and "popular music." In this way, users can quickly respond in a way that suits them, and more efficiently find content that matches their interests.

[0425] Examples of prompts include:

[0426] "Write a program that analyzes a user's message history and recommends personalized content based on the user's interests. The user's message history contains messages about movies and music. The content database contains the latest movie and music information. Calculate the similarity between the user's messages and the content, and recommend the most relevant content."

[0427] The flow of the specific process in the application example 1 will be described with reference to FIG.

[0428] Step 1:

[0429] The server retrieves the user's messaging history from a database: as input, a user ID is given, and as output, a list of the user's past messages, including information such as the message content, the time sent, and how often they were sent.

[0430] Step 2:

[0431] The server analyzes the retrieved messaging history and extracts the user's reply habits. As input, it is given the list of messages obtained in step 1, and as output, it obtains patterns that indicate the user's reply habits. Specifically, it uses TextBlob to analyze the message content and identify frequently used phrases and expressions.

[0432] Step 3:

[0433] The server analyzes the talk flow between the user and the other party: as input, it is given a sequence of messages exchanged between the user and the other party, and as output, it obtains data describing the talk flow, including the order of messages and their relationships to each other.

[0434] Step 4:

[0435] The server generates reply suggestions that are appropriate for the user based on the analysis results. As input, the data obtained in steps 2 and 3 are given, and as output, reply suggestions that are appropriate for the user are generated. A generative AI model is used to create natural reply suggestions based on the user's reply habits and the flow of the conversation.

[0436] Step 5:

[0437] The server presents the generated reply suggestions to the user's terminal. The reply suggestions generated in step 4 are given as input, and are displayed on the user's terminal as output. The user can use the presented reply suggestions as they are, or arrange them in their own way.

[0438] Step 6:

[0439] The server recommends personalized content based on the user's interests and emotions. As input, the user's messaging history and current talk stream are given, and as output, a list of content suitable for the user is obtained. Messages and content are vectorized using CountVectorizer, and similarity is calculated using cosine_similarity.

[0440] Step 7:

[0441] The server recommends the most relevant content to the user based on the calculated similarity. As input, the list of content obtained in step 6 and the similarity data are given, and as output, the content to be displayed to the user is determined. The user can browse the recommended content and efficiently obtain information that matches his / her interests.

[0442] Example 2

[0443] Next, a second embodiment of the second embodiment will be described. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0444] Conventional messaging systems have difficulty analyzing users' reply habits and conversation flow, and therefore cannot provide optimal reply suggestions for users. In addition, they do not fully understand the conversation flow taking into account changes in the user's emotions, and therefore cannot improve the quality of communication. This results in users having to spend a lot of time thinking of appropriate replies, which reduces the efficiency of communication.

[0445] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0446] In this invention, the server includes means for reading a user's own reply habits from the user's messaging history, means for analyzing the flow of a conversation between the user and another person, means for generating reply suggestions that are appropriate for the user based on the analysis results, means for presenting the generated reply suggestions to the user, means for analyzing the message context, wording, and usage of emoticons and stamps to extract changes in emotion, and means for understanding the flow of the conversation based on the extracted changes in emotion. This makes it possible to provide optimal reply suggestions that take into account the user's reply habits and changes in emotion.

[0447] "User's messaging history" means a record of messages that a User has sent or received in the past.

[0448] A "reply habit" is a user's tendency to reply to messages in a particular pattern or style.

[0449] "Chat flow" refers to the progress and changes in the message exchange between a user and another person.

[0450] "Analysis results" are the results of analysis conducted based on collected data.

[0451] A "Suggested Replies" is a suggestion for a User to reply to a Message.

[0452] "Context" is the situation or background information in which a message is sent or received.

[0453] "Language" refers to the choice of words and expressions that a user uses in their messages.

[0454] "Emoji" are small pictures or icons used to express emotions or intent in messages.

[0455] A "stamp" is a large image or illustration used to convey emotion or meaning in a message.

[0456] "Emotional changes" refers to the change in a user's emotions over time and depending on the content of the message.

[0457] An "emotion engine" is software that analyzes a user's emotions from the content and expression of a message.

[0458] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0459] The present invention is a system that analyzes a user's messaging history and provides optimal reply suggestions that take into account the user's reply habits and changes in emotions. A specific embodiment of this system is described below.

[0460] 1. Program Generation

[0461] The server generates a program to analyze the flow of the conversation between the user and the other party. This program analyzes the exchange of messages between the user and the other party and understands changes in topic and sentiment.

[0462] 2. Program Processing Explanation

[0463] The server collects and analyzes the messages exchanged between the user and the other party. Specifically, it performs the following processes:

[0464] Data collection: The server collects the message history between the user and the other party. This data includes text messages, emojis, stamps, etc. The database used is, for example, MySQL or MongoDB.

[0465] Conversation flow analysis: The server analyzes the collected message data and understands the conversation flow. For example, NLTK or spaCy is used as a natural language processing (NLP) library. The server extracts patterns of users expressing gratitude after receiving a question from the other party.

[0466] Emotion change analysis: The server uses an emotion engine to extract changes in the user's emotions. Examples of emotion engines include IBM Watson (registered trademark) Tone Analyzer and Google Cloud Natural Language API. This allows the server to analyze the context and wording of messages, as well as the use of emojis and stamps.

[0467] Output of results: The server provides the analysis results to the user. The analysis results are generated in report format and sent to the user's terminal. For example, a report in PDF format is generated and sent by email.

[0468] 3. Examples of concrete examples and prompts

[0469] As a concrete example, consider the following scenario.

[0470] scenario:

[0471] User A and User B are exchanging messages. User A asks, "How was your day?" and User B replies, "It was great, thanks!" In this exchange, we can see a pattern where User B expresses gratitude.

[0472] Example prompt:

[0473] "Please generate a program that analyzes the message history of users A and B and understands the flow of the conversation and changes in emotions. In particular, please extract the pattern of user B expressing gratitude and analyze the changes in emotions using an emotion engine."

[0474] In this way, the server analyzes the flow of conversation between the user and the other person and executes a program to understand changes in emotions, making it possible to provide optimal reply suggestions that take into account the user's reply habits and changes in emotions.

[0475] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0476] Step 1:

[0477] The server collects message history between the user and the recipient.

[0478] Input: User ID and contact ID

[0479] What happens: The server retrieves the user and their contacts' past messages from a database (e.g. MySQL or MongoDB).

[0480] Output: Collected messaging data (text messages, emojis, stamps, etc.)

[0481] Step 2:

[0482] The server analyzes the collected message data and understands the flow of the conversation.

[0483] Input: Collected message data

[0484] What it does: The server uses a natural language processing (NLP) library (e.g., NLTK or spaCy) to analyze the content of the messages and extract patterns where the user expresses gratitude after receiving a question from the other person.

[0485] Output: Analysis of conversation flow (patterns and topic changes)

[0486] Step 3:

[0487] The server uses an emotion engine to extract changes in the user's emotions.

[0488] Input: Collected message data

[0489] What it does: The server uses an emotion engine (e.g., IBM Watson Tone Analyzer or Google Cloud Natural Language API) to analyze the message's context, wording, and use of emojis and stickers.

[0490] Output: Emotion change analysis results

[0491] Step 4:

[0492] The server generates suggested replies that are likely to be appropriate for the user based on the analysis results.

[0493] Input: Analysis results of conversation flow, analysis results of emotion changes

[0494] How it works: The server uses a generative AI model to generate reply suggestions that take into account the user's reply habits and changes in sentiment.

[0495] Output: The generated reply proposal

[0496] Step 5:

[0497] The server presents the generated reply suggestions to the user.

[0498] Input: Generated reply proposal

[0499] What happens: The server sends the suggested reply to the user's device, for example by sending a notification to the user's messaging app.

[0500] Output: The suggested replies offered to the user

[0501] In this way, the server analyzes the flow of conversation between the user and the other person and executes a program to understand changes in emotions, making it possible to provide optimal reply suggestions that take into account the user's reply habits and changes in emotions.

[0502] (Application example 2)

[0503] Next, a description will be given of Application Example 2 of Form Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0504] Conventional messaging systems can analyze a user's reply habits and the flow of conversation, but they cannot detect changes in emotions or abnormal conversation patterns. Therefore, they cannot issue appropriate warnings even when the user may be in a dangerous situation. To solve this problem, a system that can analyze changes in a user's emotions in real time, detect abnormal conversation patterns, and issue warnings is required.

[0505] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[0506] In this invention, the server includes means for reading a user's own reply habits from the user's messaging history, means for analyzing the flow of a conversation between the user and another party, means for generating reply suggestions that are characteristic of the user based on the analysis results, means for presenting the generated reply suggestions to the user, means for extracting changes in emotion from the user's messaging history, means for detecting abnormal talk patterns based on the changes in emotion, and means for issuing a warning when an abnormal talk pattern is detected. This makes it possible to analyze changes in a user's emotions in real time, detect abnormal talk patterns, and issue a warning.

[0507] "User's messaging history" means a record of messages that a User has sent or received in the past.

[0508] "Reply habits" are patterns or tendencies in how a user responds to messages.

[0509] "Chat flow" refers to the progress and changes in the message exchange between a user and the other person.

[0510] "Analysis results" refer to information and data obtained after analyzing the flow of conversation and changes in emotions.

[0511] "User-friendly replies" are reply suggestions generated based on a user's past messaging history and replying habits.

[0512] "Changes in emotions" refers to fluctuations in emotions that can be read from the content of a user's message and the way it is expressed.

[0513] An "unusual talk pattern" is a pattern of messaging that differs from normal interactions and indicates a particular risk or problem.

[0514] A "warning" is a message issued to the user to warn them when an abnormal talk pattern is detected.

[0515] A system for implementing the present invention has a function of analyzing a user's messaging history and detecting changes in emotions and abnormal talk patterns. A specific embodiment of this system will be described below.

[0516] First, the server collects the user's messaging history, which includes records of messages the user has sent and received in the past. The server analyzes this history to understand the user's reply habits and conversation flow.

[0517] The server then generates personalized replies based on the analysis, based on the user's messaging history and replying habits, and presents the replies to the user.

[0518] In addition, the server extracts changes in user emotions from the user's messaging history. Specifically, it analyzes changes in user emotions from the context of messages, wording, and the use of emojis and stamps. For this analysis, it uses TextBlob, a natural language processing library.

[0519] Based on the changes in emotion, the server detects anomalous talk patterns, which are patterns of messaging that are different from normal interactions and indicate a particular risk or problem, such as a sudden change in emotion or a series of negative emotions.

[0520] If an abnormal talk pattern is detected, the server will issue a warning to the user, letting them know that they may be in a dangerous situation.

[0521] As a concrete example, consider the following message exchange:

[0522] Message 1: "Today was such a fun day!"

[0523] Message 2: "I suddenly felt really sad."

[0524] In this case, due to the large change in emotions, the server detects the abnormal talk pattern and issues a warning to the user.

[0525] Here is an example of a prompt to be input to the generative AI model:

[0526] Analyze a user's message history to detect changes in sentiment. The following messages will be analyzed:

[0527] 1. Today was such a fun day!

[0528] 2. I suddenly felt very sad.

[0529] If there are big changes in emotion, raise a warning.

[0530] In this way, a security chat monitoring system can be realized to protect the safety of users.

[0531] The flow of the identification process in the application example 2 will be described with reference to FIG.

[0532] Step 1:

[0533] The server collects the messaging history of the user. The input is a record of messages sent and received by the user in the past, and the output is the collected messaging history data. In particular, the server obtains the message data through the API of the messaging application.

[0534] Step 2:

[0535] The server analyzes the collected messaging history to understand the user's reply habits and the flow of the conversation. The input is the collected messaging history data, and the output is the analysis results regarding the user's reply habits and the flow of the conversation. In concrete terms, the server analyzes message patterns using a natural language processing algorithm.

[0536] Step 3:

[0537] The server generates reply suggestions that are appropriate for the user based on the analysis results. The input is the analysis results on the user's reply habits and the flow of the conversation, and the output is the generated reply suggestions that are appropriate for the user. In concrete terms, the server generates multiple reply suggestions using a generative AI model.

[0538] Step 4:

[0539] The server presents the generated reply suggestions to the user. The input is the generated user-like reply suggestions, and the output is the reply suggestions presented to the user. In particular, the server sends a notification to the user's device and displays the reply suggestions.

[0540] Step 5:

[0541] The server extracts emotion changes from a user's messaging history. The input is the collected messaging history data, and the output is data on emotion changes. Specifically, the server uses TextBlob to analyze the emotion of a message.

[0542] Step 6:

[0543] The server detects anomalous talk patterns based on changes in emotion. The input is data on changes in emotion, and the output is information on the detected anomalous talk patterns. In specific operations, the server calculates the magnitude of the change in emotion and determines whether there is an anomalous change.

[0544] Step 7:

[0545] The server issues a warning when it detects an abnormal talk pattern. The input is information about the detected abnormal talk pattern, and the output is a warning message for the user. In particular, the server sends a warning notification to the user's terminal.

[0546] Example 3

[0547] Next, a third embodiment of the third embodiment will be described. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0548] Conventional messaging systems have the problem that it takes time for users to think of what to reply, making it difficult to communicate efficiently. In addition, they lack the functionality to automatically generate appropriate replies according to the user's emotions, making it difficult to communicate in a way that reflects the user's emotions. Furthermore, there is insufficient means for users to easily select suggested replies and insert them into the message input field.

[0549] The specific processing by the specific processing unit 290 of the data processing device 12 in the third embodiment is realized by the following means.

[0550] In this invention, the server includes means for reading a user's own reply habits from the user's messaging history, means for analyzing the flow of conversation between the user and the other party, means for generating reply suggestions that suit the user based on the analysis results, means for extracting emotions from the user's message and generating reply suggestions based on the emotions, means for presenting the generated reply suggestions to the user, and means for the user to select the presented reply suggestions and insert them into the message input field. This allows the user to significantly reduce the time spent struggling with replying, enabling efficient communication that reflects emotions.

[0551] "User's messaging history" means a record of messages that a User has previously sent and received.

[0552] "Reply habits" are patterns of what words and phrases a user frequently uses in response to messages.

[0553] "Chat flow" refers to the sequence and content of messages exchanged between a user and another person.

[0554] "Analysis Results" refers to information or data obtained after analyzing a User's messaging history and conversation flow.

[0555] A "suggested reply" is a suggested reply that a user can make to a message.

[0556] "Sentiment extraction" is the process of identifying emotions such as happiness, sadness, anger, etc. from a user's message.

[0557] "Sentiment-based reply suggestions" are reply suggestions that reflect the user's emotions.

[0558] "Presenting" is the act of showing the generated reply suggestions to the user.

[0559] The "message input field" is a text field in which the user can enter a message.

[0560] "Selection" is the act of the user choosing one of the suggested replies presented.

[0561] "Insert" refers to the act of adding the selected reply proposal to the message input field.

[0562] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0563] The present invention is a system that analyzes a user's messaging history and automatically generates reply suggestions that are appropriate for the user. A specific embodiment of this system is described below.

[0564] 1. Program Generation

[0565] The server runs a program to collect and analyze users' messaging history. The program uses natural language processing (NLP) techniques to analyze users' reply habits and conversation flow. Specifically, it uses generative AI models such as Google's BERT and OpenAI's GPT-3.

[0566] 2. Program Processing Explanation

[0567] The server collects the user's past message data, including messages sent and received by the user, for example by obtaining message history from a chat application database.

[0568] Next, the server analyzes the collected message data. It uses a natural language processing library (e.g., spaCy or NLTK) to extract frequently used phrases and reply patterns by users. Based on the analysis results, it generates reply suggestions that are appropriate for the user. It inputs prompt sentences that reflect the user's reply patterns into the generative AI model to generate appropriate reply suggestions.

[0569] In addition, the server extracts emotions from the user's message. Using an emotion engine (e.g., IBM Watson's Tone Analyzer), it analyzes the emotions (happiness, sadness, anger, etc.) expressed by the user in real time. It then generates a reply proposal based on that emotion. For example, if a user sends a message saying, "I had a lot of fun today!", the emotion engine extracts the emotion of "happiness" and generates a reply proposal saying, "I had a great time too! Let's go together again!"

[0570] 3. Examples of concrete examples and prompts

[0571] As a concrete example, if a user sends a message saying, "I had so much fun today!", the emotion engine will extract the emotion of "joy" and generate a suggested reply saying, "I had fun too! Let's go together again!"

[0572] Example prompt:

[0573] User message: "Today was so much fun!"

[0574] Prompt for generative AI model: "Generate suggested replies when the user shows delight."

[0575] Showing generated reply suggestions

[0576] The device presents the generated reply suggestions to the user. Specifically, the generated reply suggestions are displayed at the bottom of the message input field, and the user can insert the reply suggestion into the message input field by tapping it. This significantly reduces the time users spend thinking about replying, enabling efficient communication that reflects their feelings.

[0577] This system allows the user to reply quickly and appropriately, improving the quality of communication. The flow of the identification process in the third embodiment will be described with reference to FIG.

[0578] Step 1:

[0579] The server collects the user's messaging history, specifically, retrieves past message data from the chat application's database. The input is the user ID and associated message history, and the output is the collected message data, including messages sent and received by the user.

[0580] Step 2:

[0581] The server analyzes the collected message data. It uses a natural language processing library (e.g., spaCy or NLTK) to extract frequently used phrases and reply patterns by users. The input is the collected message data, and the output is the analysis results on the user's reply habits and the flow of the conversation. Specific operations include tokenizing messages, tagging parts of speech, and extracting frequently used phrases.

[0582] Step 3:

[0583] The server generates a reply proposal that is appropriate for the user based on the analysis results. A prompt sentence that reflects the user's reply patterns is input to a generative AI model (e.g., GPT-3), which generates appropriate reply proposals. The input is the analysis results and the prompt sentence, and the output is the generated reply proposal. Specific operations include generating a prompt sentence, inputting it into the AI ​​model, and generating reply proposals.

[0584] Step 4:

[0585] The server extracts emotions from the user's message. Using an emotion engine (for example, IBM Watson's Tone Analyzer), it analyzes the emotions expressed by the user (happiness, sadness, anger, etc.) in real time. The input is the user's message, and the output is the extracted emotion information. Specific operations include analyzing the emotion of the message and adding emotion tags.

[0586] Step 5:

[0587] The server generates reply suggestions based on emotions. Based on the output of the emotion engine, a prompt that reflects the emotion is input to the generative AI model, and an appropriate reply suggestion is generated. The input is emotion information and a prompt, and the output is a reply suggestion based on the emotion. Specific operations include generating a prompt according to the emotion, inputting it into the AI ​​model, and generating a reply suggestion.

[0588] Step 6:

[0589] The device presents the generated reply suggestions to the user. Specifically, the generated reply suggestions are displayed at the bottom of the message input field, and the user can insert the reply suggestions into the message input field by tapping on them. The input is the generated reply suggestions, and the output is the reply suggestions presented to the user. Specific operations include updating the UI, displaying reply suggestions, and supporting user selection.

[0590] (Application example 3)

[0591] Next, a description will be given of Application Example 3 of Form Example 3. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0592] Conventional chat support systems have the problem that it is difficult for customer support staff to respond quickly and appropriately when communicating with customers. In particular, it is not possible to automatically generate reply suggestions that take into account the staff's reply habits and changes in emotions, making it difficult to achieve efficient communication.

[0593] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 3 is realized by the following means. In this invention, the server includes a means for reading the reply habits of the user from the messaging history of the user, a means for analyzing the flow of the conversation between the user and the other party, a means for generating a reply proposal that is suitable for the user based on the analysis result, a means for presenting the generated reply proposal to the user, a means for extracting changes in the user's emotions in real time and generating a reply proposal that takes into account the changes in the emotions, a means for generating a reply proposal based on a prompt sentence using a generative AI model, and a means for displaying the generated reply proposal at the bottom of the chat window so that the user can select and use it. This enables a customer support representative to quickly and appropriately reply to a customer in communication with the customer.

[0594] "User's messaging history" means a record of messages that a User has sent or received in the past.

[0595] "Reply habits" are phrases or patterns that a user frequently uses when replying to messages.

[0596] "Chat flow" refers to the sequence and content of messages exchanged between a user and another person.

[0597] "Analysis Results" refers to the information and data obtained after analyzing a User's messaging history and conversation flow.

[0598] "User-friendly reply suggestions" are suggestions of replies that a user would naturally use, generated based on the user's reply habits and the flow of the conversation.

[0599] "Sentiment change" refers to the change in the emotional state extracted from a user's messages over time.

[0600] A "generative AI model" is a model that has been trained using artificial intelligence to perform a specific task, and is used here to generate reply suggestions.

[0601] A "prompt" is an instruction given to a generative AI model that serves as the basis for the model to generate appropriate reply suggestions.

[0602] A "chat window" is an interface through which users can send and receive messages.

[0603] "Real-time" refers to data processing and analysis being carried out immediately.

[0604] A system for implementing this invention includes means for reading a user's own reply habits from the user's messaging history, means for analyzing the flow of conversation between the user and another party, means for generating reply suggestions that are typical of the user based on the analysis results, means for presenting the generated reply suggestions to the user, means for extracting changes in the user's emotions in real time and generating reply suggestions taking into account the changes in emotion, means for generating reply suggestions based on prompt sentences using a generative AI model, and means for displaying the generated reply suggestions at the bottom of the chat window so that the user can select and use them.

[0605] The server stores the user's messaging history in a database and uses an analysis engine to extract the user's reply habits. The analysis engine identifies phrases and patterns frequently used by the user and generates reply suggestions that are appropriate for the user based on those. In addition, a sentiment analysis module is used to extract changes in emotions from the user's messages in real time and generate reply suggestions based on those emotions.

[0606] We use OpenAI's GPT-3 as the generative AI model. GPT-3 generates appropriate reply suggestions based on the prompt text. The prompt text is constructed based on information including the user's reply habits, current emotional state, and the flow of the conversation. For example, the following prompt text is generated:

[0607] User reply habits: ["Thank you", "Okay", "Nice to meet you"]

[0608] Emotion: neutral

[0609] Message: "How do I return an item?"

[0610] Generate a proposed reply.

[0611] The device displays the generated reply suggestions at the bottom of the chat window for the user to select and use. The user can use the suggested reply suggestions as is or arrange them in their own way.

[0612] As a concrete example, when a customer support representative receives a message from a customer saying, "Please tell me about returning an item," the system analyzes the representative's past response history and current emotional state to generate a response suggestion like the following:

[0613] "Thank you. Please follow the steps below to return your item."

[0614] This way, customer support representatives can respond quickly and appropriately.

[0615] The flow of the specific process in the application example 3 will be described with reference to FIG.

[0616] Step 1:

[0617] The server retrieves the user's messaging history from a database. The input is a user ID, and the output is the user's past messaging history. Specifically, the server executes a database query to retrieve the user's messaging history.

[0618] Step 2:

[0619] The server analyzes the retrieved messaging history and extracts the user's reply habits. The input is the messaging history, and the output is patterns that indicate the user's reply habits. Specifically, the server uses natural language processing algorithms to identify frequently used phrases and patterns.

[0620] Step 3:

[0621] The server analyzes the flow of conversation between the user and the other party in real time. The input is the current chat messages, and the output is the analysis result of the conversation flow. Specifically, the server analyzes the order and content of messages to understand the conversation flow.

[0622] Step 4:

[0623] The server extracts the user's emotional changes in real time. The input is the current chat message, and the output is data indicating the emotional changes. Specifically, the server uses an emotion analysis module to extract emotions from the messages.

[0624] Step 5:

[0625] The server generates a prompt sentence based on the analysis results. The input is the user's reply habits, the flow of the conversation, and changes in emotions, and the output is the prompt sentence. In concrete terms, the server combines these data to build a prompt sentence to input into the generative AI model.

[0626] Step 6:

[0627] The server uses a generative AI model to generate a reply proposal based on the prompt sentence. The input is the prompt sentence, and the output is the reply proposal. Specifically, the server sends the prompt sentence to OpenAI's GPT-3 and obtains the reply proposal.

[0628] Step 7:

[0629] The server sends the generated reply proposal to the terminal. The input is the reply proposal, and the output is the reply proposal displayed on the terminal. In concrete terms, the server sends the reply proposal to the terminal and displays it at the bottom of the chat window.

[0630] Step 8:

[0631] The user selects and uses the suggested reply. The input is the displayed reply, and the output is the selected reply. Specifically, the user taps a reply displayed at the bottom of the chat window to insert it into the message input field.

[0632] The specific processing unit 290 transmits the result of the specific processing to the smart device 14. In the smart device 14, the control unit 46A causes the output device 40 to output the result of the specific processing. The microphone 38B acquires a voice indicating a user input for the result of the specific processing. The control unit 46A transmits the voice data indicating the user input acquired by the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the voice data.

[0633] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (registered trademark) (Internet search<URL: https: / / openai.com / blog / chatgpt> ) and other generation AIs. The data generation model 58 is obtained by making a neural network perform deep learning. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating a voice, text data indicating a text, and image data indicating an image is input. The data generation model 58 performs inference on the input inference data according to the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0634] Another example of generative AI is Gemini (registered trademark) (Internet search <url: https: gemini.google.com ?hl="ja">) are mentioned.

[0635] In the above embodiment, an example was given in which the specific process was performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific process may be performed by the smart device 14.

[0636] [Second embodiment]

[0637] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0638] 3, the data processing system 210 includes the data processing device 12 and smart glasses 214. An example of the data processing device 12 is a server.

[0639] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a wide area network (WAN) and / or a local area network (LAN).

[0640] The smart glasses 214 include a computer 36, a microphone 238, a speaker 240, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, and the camera 42 are also connected to the bus 52.

[0641] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs the voice according to instructions from the processor 46.

[0642] Camera 42 is a small digital camera equipped with an optical system including a lens, an aperture, and a shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (e.g., an imaging range defined by an angle of view equivalent to the width of the field of vision of an average healthy person).

[0643] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for transmitting and receiving various types of information between the processor 46 and the processor 28 via the network 54. The transmission and reception of various types of information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is performed in a secure state.

[0644] Fig. 4 shows an example of main functions of the data processing device 12 and the smart glasses 214. As shown in Fig. 4, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0645] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32, and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0646] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0647] In the smart glasses 214, the reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50, and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0648] Next, the specific processing by the specific processing unit 290 of the data processing device 12 will be described.

[0649] "Example 1"

[0650] In one embodiment of the present invention, a means is provided for reading a user's own reply habits from the user's messaging history. Specifically, the contents, sending times, sending frequency, etc. of messages sent by the user in the past are analyzed to extract the user's reply habits. For example, if the user frequently uses the phrase "thank you," the use of the phrase "thank you" is extracted as the user's reply habits.

[0651] "Example 2"

[0652] It also provides a means to analyze the flow of conversation between users and other people. Specifically, it analyzes the patterns of message exchanges between users and other people, changes in topics, changes in emotions, etc. to understand the flow of the conversation. For example, if a user always expresses gratitude after receiving a question from the other person, this pattern can be extracted and the flow of the conversation analyzed.

[0653] "Example 3"

[0654] Furthermore, a means is provided for generating reply suggestions that are appropriate for the user based on the analysis results. Specifically, reply suggestions that are appropriate for the user are automatically generated based on the analysis results of the user's reply habits and the flow of the conversation. For example, if a user frequently uses the phrase "thank you" and has a pattern of expressing gratitude after receiving a question from the other person, a reply suggestion of "Thank you, about that..." is generated based on that information.

[0655] "Example 4"

[0656] Finally, a means is provided for presenting the generated reply suggestions to the user. Specifically, the generated reply suggestions are displayed on the user's device and the user can select one. For example, the generated reply suggestions are displayed at the bottom of the message input field, and the user can insert the reply suggestion into the message input field by tapping it. This allows the user to significantly reduce the time spent thinking about replying and achieve efficient communication.

[0657] The process flow of each embodiment will be described below.

[0658] "Example 1"

[0659] Step 1: Obtain the user's messaging history. Specifically, obtain information from the database about the user's past messages, such as the content of the messages, the time they were sent, and the frequency of their messages.

[0660] Step 2: Extract the user's own reply habits from the acquired messaging history. Specifically, the system analyzes the phrases the user frequently uses, messages sent at specific times, and reply patterns to specific people to extract the user's reply habits.

[0661] "Example 2"

[0662] Step 1: Obtain the messaging history between the user and the other party. Specifically, obtain message exchange information between the user and the other party from the database.

[0663] Step 2: Analyze the flow of conversation from the acquired messaging history. Specifically, analyze message exchange patterns, topic changes, and emotional changes to understand the flow of conversation.

[0664] "Example 3"

[0665] Step 1: Obtain the analysis results of the user's reply habits and the flow of the conversation. Specifically, obtain the information extracted and analyzed using the methods described above.

[0666] Step 2: Generate reply suggestions that are appropriate for the user based on the obtained analysis results. Specifically, the system automatically generates reply suggestions that are appropriate for the user, taking into account the user's reply habits and the flow of the conversation.

[0667] "Example 4"

[0668] Step 1: Obtain the generated reply proposal. Specifically, obtain the reply proposal generated by the above-mentioned means.

[0669] Step 2: Present the retrieved reply suggestions to the user. Specifically, the generated reply suggestions are displayed on the user's device and the user can select one.

[0670] Example 1

[0671] Next, a description will be given of Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal".

[0672] In conventional messaging systems, it was difficult to automatically generate reply suggestions that took into account the user's reply habits. As a result, users had to think up the reply content themselves every time, which was time-consuming and caused problems such as inconsistent replies. Furthermore, there was a lack of a way to effectively utilize the user's messaging history, making it impossible to provide personalized reply suggestions based on the user's past behavior.

[0673] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0674] In this invention, the server includes means for reading a user's own reply habits from the user's messaging history, means for analyzing the flow of a dialogue between the user and the other party, means for generating reply suggestions that suit the user based on the analysis results, means for presenting the generated reply suggestions to the user, means for collecting the user's messaging history, means for preprocessing the collected messaging history, means for analyzing the preprocessed message contents, means for extracting the user's reply habits based on the analysis results, and means for saving the extraction results. This makes it possible to automatically generate personalized reply suggestions based on the user's past messaging history and present them to the user.

[0675] "User's messaging history" is a collection of data about messages the user has sent in the past, such as the content of those messages, the time they were sent, and the frequency with which they were sent.

[0676] "Reply habits" are patterns such as phrases that a user frequently uses when replying to messages or a tendency to reply at certain times of the day.

[0677] "Dialogue flow" refers to the order and content relevance of messages exchanged between a user and another person.

[0678] "Analysis results" are the results of analyzing users' reply habits and conversation flow based on collected and preprocessed messaging history.

[0679] "Suggested replies" are candidate messages for a user to reply to, generated based on the user's past messaging history and analysis results.

[0680] "Preprocessing" refers to the process of removing unnecessary spaces and special characters from the collected messaging history, and performing tokenization and removing stop words.

[0681] A "natural language processing library" is a software tool for tokenizing message content, analyzing sentiment, extracting frequent phrases, and so on.

[0682] "Database" refers to an information management system for storing users' messaging history and analysis results.

[0683] A "prompt sentence" is an instruction sentence to be input into a generative AI model, which clearly indicates the message history to be analyzed and the purpose of the analysis.

[0684] A "generative AI model" is an artificial intelligence model that generates reply suggestions that are appropriate for a user based on the user's messaging history and analysis results.

[0685] The present invention relates to a system for analyzing a user's messaging history, extracting the user's reply habits, and generating reply suggestions that are suited to the user. A specific embodiment of this system will be described below.

[0686] The server first collects the user's messaging history, including the content of messages the user has sent in the past, the time they were sent, how often they were sent, etc. The collected data is then stored in a database.

[0687] The server then pre-processes the collected messaging history, which includes cleaning the message content (removing unnecessary spaces and special characters), tokenizing (splitting into words), and removing stop words (removing frequent but meaningless words). This pre-processing is done using Python, with natural language processing libraries such as NLTK and spaCy.

[0688] Once preprocessed, the message content is then analyzed. The server uses natural language processing libraries to extract common phrases and words. It also performs sentiment analysis on the message and classifies it as positive, negative, or neutral.

[0689] Based on the analysis results, the server extracts the user's reply habits. For example, if a user frequently uses the phrase "thank you," that phrase is identified as the user's reply habits. Also, if there are many replies during a certain time period, that time period is also extracted as a habit.

[0690] The extracted user reply habits are stored in a database, and this information is then fed into a generative AI model as prompts to generate reply suggestions that are more likely to be user-friendly.

[0691] As a concrete example, consider the case where you want to analyze the history of messages sent by user A. Suppose user A's message history contains the following data:

[0692] Message 1: "Thank you, that was helpful."

[0693] Message 2: "Thank you for your hard work."

[0694] Message 3: "Thank you, I look forward to working with you again."

[0695] The server analyzes these messages and extracts the frequent use of the phrase "Thank you." As a result, the use of the phrase "Thank you" is identified as a reply habit of User A.

[0696] Examples of prompts to input to a generative AI model might include the following:

[0697] Analyze the user's messaging history and extract frequent phrases. The following message histories will be analyzed:

[0698] 1. "Thank you, that was helpful."

[0699] 2. "Thank you for your hard work."

[0700] 3. "Thank you, I look forward to working with you again."

[0701] By inputting this prompt into a generative AI model, it is possible to automatically extract the user's reply habits and generate reply suggestions that are appropriate for the user.

[0702] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0703] Step 1:

[0704] The server collects the messaging history of users. As input, a user ID is given. The server retrieves data such as message content, sending time, and sending frequency for that user from the database. As output, the collected messaging history is obtained. Specifically, the server executes SQL queries to extract the required information from the database.

[0705] Step 2:

[0706] The server preprocesses the collected messaging history. As input, it is given the collected messaging history. The server cleans the message content (removes unnecessary spaces and special characters), tokenizes it (splits it into words), and removes stop words. As output, it gets the preprocessed message content. Specifically, the server runs a Python script to perform preprocessing using natural language processing libraries (NLTK and spaCy).

[0707] Step 3:

[0708] The server analyzes the preprocessed message content. The preprocessed message content is given as input. The server uses a natural language processing library to extract frequent phrases and words and perform sentiment analysis. The analysis results are obtained as output. Specifically, the server runs a Python script to extract frequent phrases and perform sentiment analysis.

[0709] Step 4:

[0710] The server extracts the user's reply habits based on the analysis results. The analysis results are given as input. The server identifies frequent phrases and tendencies to reply at specific times of the day. The output is the user's reply habits. In concrete terms, the server analyzes the analysis results and runs an algorithm to identify the user's reply habits.

[0711] Step 5:

[0712] The server saves the extracted user's reply habits in a database. As input, the user's reply habits are given. The server performs an insert operation on the database and saves the extracted results. As output, it is confirmed that the saving has been completed. As a specific operation, the server executes a SQL query to insert data into the database.

[0713] Step 6:

[0714] The server inputs a prompt sentence into the generative AI model to generate a reply proposal that is appropriate for the user. The user's reply habits and the prompt sentence are given as input. The server inputs the prompt sentence into the generative AI model to generate a reply proposal. The generated reply proposal is obtained as output. In concrete terms, the server sends a prompt sentence to the generative AI model and obtains a reply proposal.

[0715] Step 7:

[0716] The server presents the generated reply proposal to the user. As input, the generated reply proposal is given. The server sends the reply proposal to the user's device and presents it to the user. As output, a state is obtained where the user can confirm the reply proposal. As a specific operation, the server sends a notification to the user's device and displays the reply proposal.

[0717] (Application example 1)

[0718] Next, a description will be given of Application Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal".

[0719] Conventional messaging systems do not recommend personalized content that takes into account the user's reply habits. This makes it difficult to efficiently provide content that is of interest to the user. In addition, there has been no system that integrates the generation of reply suggestions based on the user's reply habits and content recommendation.

[0720] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0721] In this invention, the server includes means for reading the user's own reply habits from the user's messaging history, means for analyzing the flow of a conversation between the user and another person, means for generating reply suggestions that are suited to the user based on the analysis results, means for presenting the generated reply suggestions to the user, means for recommending personalized content based on the user's reply habits, and means for presenting the recommended content to the user. This makes it possible to recommend personalized content that takes into account the user's reply habits, and efficiently provide content that is more interesting to the user.

[0722] "User's messaging history" is a collection of data about messages the user has sent in the past, such as the content of those messages, the time they were sent, and the frequency with which they were sent.

[0723] "Reply habits" are patterns of phrases or words that a user frequently uses when replying to messages.

[0724] A "talk flow" is a process of continuous exchange of messages between a user and another person.

[0725] "Analysis Results" refers to information or data obtained after analyzing a user's messaging history and conversation flow.

[0726] A "suggested reply" is a suggested message that a user can use as a reference when replying.

[0727] "Personalized Content" means information or entertainment content that is specifically selected based on a user's individual characteristics and interests.

[0728] A "recommendation means" is a method or technology for selecting and presenting appropriate content or information to a user.

[0729] A system for implementing this invention includes means for reading a user's own reply habits from the user's messaging history, means for analyzing the flow of conversation between the user and another party, means for generating reply suggestions that are typical of the user based on the analysis results, means for presenting the generated reply suggestions to the user, means for recommending personalized content based on the user's reply habits, and means for presenting the recommended content to the user.

[0730] The server runs software to collect and analyze users' messaging history. Specifically, it uses programming languages ​​such as Python to analyze the content, time and frequency of users' messages and extracts their reply habits. This analysis is carried out using data processing tools such as collections.Counter.

[0731] The server then generates a reply that is appropriate for the user based on the analysis results, and presents the reply to the user, who can use it as is or customize it in his or her own way.

[0732] In addition, the server recommends personalized content based on the user's reply habits. For this recommendation, the server retrieves content that corresponds to the user's reply habits from the database and presents it to the user. For example, if the user frequently uses the phrase "thank you," the server recommends content such as "How to express gratitude" and "10 ways to say thank you."

[0733] This system is implemented using a smartphone. An application is installed on the smartphone to collect the user's messaging history and send it to a server. The application periodically sends the user's messaging history to the server, receives reply suggestions and recommended content from the server, and presents them to the user.

[0734] As a concrete example, by inputting the following prompt sentence into a generative AI model, personalized content recommendations based on the user’s reply habits can be realized:

[0735] "Build an application that extracts reply habits from a user's messaging history and recommends personalized content based on those habits. For example, if a user frequently uses the phrase 'Thank you,' recommend content like 'How to express gratitude' or '10 ways to say thank you.'"

[0736] In this way, it is possible to leverage a user's messaging history to provide them with the most relevant content.

[0737] The flow of the specific process in the application example 1 will be described with reference to FIG.

[0738] Step 1:

[0739] The server collects the user's messaging history. It receives data such as message content, sending time, and sending frequency from the user's device and stores them in a database. The input is the user's messaging history, and the output is the stored message data.

[0740] Step 2:

[0741] The server analyzes the collected messaging history. Using a programming language such as Python, it counts the frequency of message content using collections.Counter. The input is the stored message data, and the output is data that indicates the user's reply habits.

[0742] Step 3:

[0743] The server generates reply suggestions that are appropriate for the user based on the analysis results. Using a generative AI model, it generates multiple reply suggestions based on the user's reply habits. The input is data that indicates the user's reply habits, and the output is the generated multiple reply suggestions.

[0744] Step 4:

[0745] The server presents the generated reply suggestions to the user's terminal. The server transmits reply suggestions to the user's terminal for the user to select or arrange. The input is the generated multiple reply suggestions, and the output is the reply suggestions presented to the user.

[0746] Step 5:

[0747] The server recommends personalized content based on the user's reply habits. It retrieves content corresponding to the user's reply habits from a database and creates a recommendation list. The input is data indicating the user's reply habits, and the output is a recommended content list.

[0748] Step 6:

[0749] The server presents the recommended content to the user's terminal. The server transmits the recommended content to the user's terminal so that the user can view it. The input is the recommended content list, and the output is the content presented to the user.

[0750] Step 7:

[0751] The user reviews the suggested replies and recommended content, selects or arranges a suggested reply as necessary, and views the recommended content. The input is the suggested replies and content presented to the user, and the output is the user's selected or arranged reply and viewed content.

[0752] Example 2

[0753] Next, a description will be given of Example 2 of the embodiment 2. In the following description, the data processing device 12 is referred to as a "server" and the smart glasses 214 are referred to as a "terminal".

[0754] Conventional messaging systems generate reply suggestions without considering the user's reply patterns or changes in emotions, so they often present replies that are unnatural to the user. In addition, there is a lack of means to analyze the flow of conversation between the user and the other person, making it difficult to generate reply suggestions that appropriately reflect changes in topic or emotions.

[0755] The identification process by the identification processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means. In this invention, the server includes a means for reading a user's own reply pattern from the user's messaging history, a means for collecting message exchanges between the user and the other party, a means for pre-processing the collected message data, a means for tokenizing the pre-processed messages and tagging them with parts of speech, a means for analyzing the sentiment of each message, a means for analyzing message exchange patterns, topic changes, and sentiment changes, a means for generating reply suggestions that are appropriate for the user based on the analysis results, and a means for presenting the generated reply suggestions to the user. This makes it possible to generate natural reply suggestions that take into account the user's reply patterns and sentiment changes.

[0756] "User's messaging history" means a record of messages that a User has sent or received in the past.

[0757] "Reply patterns" are the tendencies of the reply format and content that a user repeatedly uses in response to a particular situation or topic.

[0758] "Message exchange" refers to the sending and receiving of a series of messages between a User and another person.

[0759] "Preprocessing" refers to the process of removing unnecessary spaces and special characters from the message data and preparing the data in a format suitable for analysis.

[0760] "Tokenization" is the process of dividing a message into its smallest units, such as words or phrases.

[0761] "Part-of-speech tagging" is the process of identifying and assigning a part of speech (noun, verb, adjective, etc.) to a tokenized word or phrase.

[0762] "Sentiment analysis" is the process of extracting emotions (happiness, sadness, anger, etc.) from the content of a message and evaluating their intensity.

[0763] "Topic changes" refers to how the topic changes during a message exchange.

[0764] "Analysis results" refers to information obtained from analysis of message exchange patterns, changes in topics, changes in sentiment, etc.

[0765] A "suggested reply" is a suggested message that a user can use to reply to the other person.

[0766] "Presenting" is the act of showing the generated reply suggestions to the user.

[0767] This invention is a system that reads a user's own reply patterns from the user's messaging history, analyzes message exchanges between the user and other parties, generates reply suggestions that are appropriate for the user based on the analysis results, and presents the generated reply suggestions to the user.

[0768] The server first collects user and recipient message data. This data is retrieved from a database such as MySQL or MongoDB. It then pre-processes the collected message data. Pre-processing includes removing unnecessary whitespace and special characters.

[0769] The preprocessed message data is tokenized using natural language processing libraries such as SpaCy and NLTK, and each token is tagged with a part of speech. For example, the message "Hello, how are you today?" is tagged as "Hello / NOUN How are you today / ADV How are you today / VERB AUX".

[0770] The server then uses a sentiment analysis tool such as VADER or TextBlob to calculate a sentiment score for each message. For example, the message "Hello, how are you today?" is calculated to have a sentiment score of 0.5.

[0771] The server then analyzes messaging patterns, topic changes, and sentiment shifts — for example, finding that a user always says "Thank you" after receiving a question from the other person.

[0772] Based on the analysis results, the server generates a reply that is appropriate for the user. This reply is generated taking into account the user's past reply patterns and changes in emotions. The generated reply is presented to the user. The user can use the proposed reply as is or arrange it in their own way.

[0773] As a concrete example, consider the following message exchange:

[0774] Other person: "Hello, how are you today?"

[0775] User: "Hi, I'm fine. How about you?"

[0776] The other person: "I'm fine. Has anything new happened to you lately?"

[0777] User: "Yes, I've started a new project. Thank you."

[0778] An example prompt to parse this interaction is:

[0779] Analyze the messages exchanged between users and their contacts and extract the following information:

[0780] 1. Message exchange patterns

[0781] 2. Change the topic

[0782] 3. Emotional changes

[0783] Message exchange:

[0784] Other person: "Hello, how are you today?"

[0785] User: "Hi, I'm fine. How about you?"

[0786] The other person: "I'm fine. Has anything new happened to you lately?"

[0787] User: "Yes, I've started a new project. Thank you."

[0788] By feeding this prompt into a generative AI model, results can be obtained that analyze message exchange patterns, topic changes, and emotional shifts.

[0789] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0790] Step 1: Collect message data

[0791] The server retrieves the message data between the user and the other party from the database. Specifically, it executes the query "SELECT FROM messages WHERE user_id = '12345'" to the MySQL database to collect the message exchanges between the user and the other party. The input is the user ID, and the output is the corresponding message data.

[0792] Step 2: Preprocessing the message data

[0793] The server pre-processes the collected message data. Specifically, it removes unnecessary whitespace and special characters to clean the message. For example, it converts the message "Hello! How are you today?" to "Hello, how are you today?". The input is the collected message data, and the output is the pre-processed message data.

[0794] Step 3: Tokenize the message and tag parts of speech

[0795] The server tokenizes the preprocessed message and tags each token with a part-of-speech tag. Specifically, it uses SpaCy and NLTK to tag the message "Hello, how are you today?" as "Hello / NOUN How are you today / ADV How are you today / VERB AUX." The input is the preprocessed message data, and the output is the tokenized and part-of-speech tagged data.

[0796] Step 4: Sentiment analysis

[0797] The server analyzes the sentiment of each message. Specifically, it uses VADER and TextBlob to calculate a sentiment score for each message. For example, the sentiment score of the message "Hello, how are you today" is calculated to be 0.5. The input is the tokenized and part-of-speech tagged data, and the output is the sentiment score.

[0798] Step 5: Analyze the flow of conversation

[0799] The server analyzes message exchange patterns, topic changes, and sentiment changes. Specifically, it finds a pattern where a user always says "Thank you" after receiving a question from the other party. The input is the sentiment score and tokenized data, and the output is the analysis result.

[0800] Step 6: Generate suggested replies

[0801] The server generates reply suggestions that are appropriate for the user based on the analysis results. Specifically, it generates reply suggestions by taking into account the user's past reply patterns and changes in emotions. The input is the analysis results, and the output is the generated reply suggestions.

[0802] Step 7: Propose a response

[0803] The server presents the generated reply to the user. Specifically, the reply is displayed to the user through a web browser. The user can use the proposed reply as is or arrange it in their own way. The input is the generated reply, and the output is the reply presented to the user.

[0804] (Application example 2)

[0805] Next, a description will be given of Application Example 2 of Form Example 2. In the following description, the data processing device 12 will be referred to as a "server", and the smart glasses 214 will be referred to as a "terminal".

[0806] Conventional messaging systems were unable to analyze users' reply habits or conversation flow, making it difficult to generate natural reply suggestions for users. They also lacked the ability to detect signs of fraud and misconduct, making it difficult to ensure user safety.

[0807] The identification process by the identification processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means. In this invention, the server includes a means for reading the reply habits of the user from the messaging history of the user, a means for analyzing the flow of the conversation between the user and the other party, a means for generating reply suggestions that are suitable for the user based on the analysis results, a means for presenting the generated reply suggestions to the user, a means for analyzing message patterns, topic changes, and emotional changes, and a means for detecting signs of misconduct or fraud. This makes it possible to generate reply suggestions that are natural to the user and to detect signs of misconduct or fraud at an early stage.

[0808] "User's messaging history" means a record of messages that a User has sent or received in the past.

[0809] "Reply habits" are the tendencies or patterns of how a user responds to messages.

[0810] "Chat flow" refers to the progress and changes in the message exchange between a user and the other person.

[0811] "Analysis results" refers to the information and data obtained after analyzing message exchanges.

[0812] "Suggested replies" refers to suggestions or candidates for a user to reply to a message.

[0813] A "message pattern" refers to a recurring format or structure in a message exchange.

[0814] "Topic changes" refers to how the topic changes during a message exchange.

[0815] "Changes in emotion" refers to how emotions change during message exchange.

[0816] "Fraud" refers to committing fraud or malicious acts against users.

[0817] "Fraud indicators" are signs or patterns in messaging that indicate the possibility of fraudulent activity.

[0818] In order to implement this invention, it is necessary to build a system in which the elements of the server, terminal, and user work in cooperation with each other.

[0819] First, the server collects the user's messaging history and reads the user's own reply habits. This requires software to store and analyze the messaging history using a database management system (DBMS). Specifically, it uses a programming language such as Python or R to analyze message patterns and changes in sentiment using natural language processing (NLP) libraries (e.g., NLTK, spaCy).

[0820] Next, the server analyzes the flow of conversation between the user and the other party. To do this, a generative AI model is used to detect changes in the topic of messages and changes in sentiment. Advanced natural language processing models such as GPT-3 and BERT are suitable as generative AI models. These models are used to analyze message exchanges in real time and understand the flow of the conversation.

[0821] Furthermore, the server generates reply suggestions that are appropriate for the user based on the analysis results. The generated reply suggestions are customized based on the user's past reply patterns. In order to generate multiple reply suggestions and present them to the user, a user interface (UI) needs to be designed. The UI is often implemented as a smartphone app or a web application.

[0822] The server also analyzes message patterns, topic changes, and sentiment changes to detect signs of misconduct or fraud. This involves using machine learning algorithms (e.g., Naive Bayes, SVM) to build a system that detects anomalous message patterns. If an anomaly is detected, the server must also be able to alert the user.

[0823] As a concrete example, consider the following message exchange:

[0824] "Hello, how are you?"

[0825] "Yeah, I'm fine. How about you?"

[0826] "I'm fine too. By the way, how are you doing lately?"

[0827] "Very well. By the way, can you give me your bank details?"

[0828] In this case, a sudden change in emotion is detected and deemed to be an indication of fraudulent activity.

[0829] Example of a prompt to be input to a generative AI model:

[0830] It analyzes messaging between users and their contacts to provide a way to understand the flow of conversations, specifically by analyzing message patterns, topic changes, and sentiment changes to detect signs of fraud or misconduct.

[0831] In this way, it is possible to realize a system that can generate reply suggestions that are natural to the user and can also detect signs of misconduct or fraud at an early stage.

[0832] The flow of the specific process in the application example 2 will be described with reference to FIG.

[0833] Step 1:

[0834] The server collects the messaging history of users. As input, it takes the message data that the user has sent and received in the past and stores it in a database. Specifically, it uses a database management system (DBMS) to store the messaging history and build a dataset for analysis. As output, it obtains the messaging history data in a format that can be analyzed.

[0835] Step 2:

[0836] The server reads the user's own reply habits from the user's messaging history. As input, it uses the messaging history data collected in step 1. Specifically, it uses a natural language processing (NLP) library (e.g., NLTK, spaCy) to analyze message patterns and reply trends. As output, it obtains data about the user's reply habits.

[0837] Step 3:

[0838] The server analyzes the flow of conversation between the user and the other party. As input, it receives message data sent and received in real time. Specifically, it uses generative AI models (e.g. GPT-3, BERT) to detect changes in the topic and sentiment of messages. As output, it obtains the analysis results on the flow of conversation.

[0839] Step 4:

[0840] The server generates reply suggestions that are appropriate for the user based on the analysis results. As input, it uses the data obtained in steps 2 and 3. Specifically, it uses a generative AI model to generate reply suggestions based on the user's past reply patterns. As output, it obtains multiple reply suggestions.

[0841] Step 5:

[0842] The server presents the generated reply suggestions to the user. As input, it uses the reply suggestions generated in step 4. Specifically, it displays the reply suggestions in a smartphone app or web application through a user interface (UI). As output, it presents the user with reply suggestions that they can select from.

[0843] Step 6:

[0844] The server analyzes message patterns, topic changes, and sentiment changes to detect signs of misconduct or fraud. As input, it uses real-time messaging data. Specifically, it uses machine learning algorithms (e.g., Naive Bayes, SVM) to detect anomalous message patterns. As output, it provides warnings about signs of misconduct or fraud.

[0845] Step 7:

[0846] The server alerts the user if any indication of misconduct or fraud is detected. As input, it uses the alert data obtained in step 6. Specifically, it displays a warning message through a user interface in a smartphone app or web application. As output, a warning is presented to the user.

[0847] Example 3

[0848] Next, a third embodiment of the third embodiment will be described. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal".

[0849] Conventional messaging systems have the problem that it takes time for users to think about what to reply, making efficient communication difficult. It is also difficult to generate automatic reply suggestions that take into account the user's reply habits and the flow of the conversation. This increases the time users spend thinking about how to reply, which reduces the efficiency of communication.

[0850] The specific processing by the specific processing unit 290 of the data processing device 12 in the third embodiment is realized by the following means.

[0851] In this invention, the server includes a means for reading a user's own reply habits from the user's messaging history, a means for analyzing the flow of conversation between the user and the other party, a means for generating reply suggestions that are suitable for the user based on the analysis results, a means for displaying the generated reply suggestions on the user's device so that the user can select one, a means for generating reply suggestions using a generative AI model, and a means for inputting prompt sentences into the generative AI model. This allows the user to significantly reduce the time spent struggling with replying, enabling efficient communication.

[0852] "User's messaging history" means a record of messages that a User has sent or received in the past.

[0853] "Reply habits" are reply patterns or phrases that a user repeatedly uses in particular situations or with particular people.

[0854] "Chat flow" refers to the continuity and context of messages exchanged between a user and the other person.

[0855] "Analysis Results" refers to data or information obtained after analyzing a User's messaging history and conversation flow.

[0856] "Suggested replies" are suggested messages that users can use as a reference when replying.

[0857] A "generative AI model" is a model that uses artificial intelligence to analyze data and generate new data and information.

[0858] A "prompt" is an instruction or question that is input into a generative AI model.

[0859] "User Device" refers to any electronic device used by a User, such as a computer, smartphone, or tablet.

[0860] The present invention is a system that analyzes a user's messaging history and automatically generates reply suggestions based on the user's reply habits and conversation flow. A specific embodiment of this system will be described below.

[0861] Hardware and Software Used

[0862] The server includes hardware and software for collecting and analyzing the messaging history of users. Specifically, the server may have the following configuration:

[0863] Hardware: High-performance server computer

[0864] Software: Analytical software using natural language processing (NLP) techniques, generative AI models (e.g. Google's BERT, OpenAI's GPT-3)

[0865] The terminal is a device that allows the user to check and select reply suggestions. Specifically, the following configuration is possible.

[0866] Hardware: Smartphones, tablets, computers

[0867] Software: Messaging application

[0868] Data processing and calculation

[0869] The server collects the user's messaging history and analyzes it using natural language processing technology. As a result of the analysis, frequently used phrases and reply patterns are extracted from the user. Based on this, reply suggestions are generated using a generative AI model.

[0870] The specific processing flow is as follows:

[0871] 1. Message Data Collection:

[0872] The server retrieves past message data through the API of the user's messaging application, and stores the retrieved data in a database.

[0873] 2. Message data analysis:

[0874] The server inputs prompts into the generative AI model and analyzes the user's message data. As a result of the analysis, frequently used phrases and reply patterns are extracted.

[0875] 3. Generate reply suggestions:

[0876] The server generates reply suggestions based on the analysis results, using a generative AI model to generate reply suggestions that are suited to the user's reply habits and the flow of the conversation.

[0877] 4. Proposal of a response:

[0878] The device presents the generated reply suggestions to the user by displaying the reply suggestions below the message input field so that the user can select one.

[0879] Examples and prompts

[0880] If a user frequently uses the phrase "thank you" and has a pattern of expressing gratitude after being asked a question, the following prompt sentence can be input into the generative AI model:

[0881] Example prompt:

[0882] User's previous messages:

[0883] 1. "Thank you for answering my question."

[0884] 2. "Thanks, about that..."

[0885] 3. "Thanks, next time..."

[0886] The other person asks: "What do you think about this?"

[0887] Generate reply: "Thanks, about that..."

[0888] The server inputs this prompt sentence into a generative AI model to generate a reply proposal. The generated reply proposal is sent to the terminal and displayed at the bottom of the message input field. When the user taps on a reply proposal, the content is inserted into the message input field. This allows the user to significantly reduce the time spent thinking about a reply, and enables efficient communication. The flow of the specific process in the third embodiment will be described with reference to FIG. 15.

[0889] Step 1:

[0890] The server collects the user's messaging history.

[0891] Input: Historical messaging data obtained through the API of the user's messaging application.

[0892] Data processing: The acquired message data is stored in a database.

[0893] Output: The saved message data.

[0894] What happens: The server calls the messaging application's API to retrieve the user's past messages and save them in a database.

[0895] Step 2:

[0896] The server analyzes the collected message data.

[0897] Input: The saved message data.

[0898] Data calculation: Input the prompt sentence into the generative AI model and analyze the user's message data.

[0899] Output: A pattern of frequent phrases or replies used by users.

[0900] How it works: The server inputs prompts into the generative AI model and analyzes the user's message data. As a result of the analysis, it extracts frequently used phrases and reply patterns from the user.

[0901] Step 3:

[0902] The server generates a proposed reply based on the analysis.

[0903] Input: A pattern of frequently used phrases or replies by users.

[0904] Data calculation: Using a generative AI model, we generate reply suggestions that are suited to the user's reply habits and the flow of the conversation.

[0905] Output: The generated suggested replies.

[0906] Specific operation: Based on the analysis results, the server inputs the prompt sentence into the generative AI model to generate reply suggestions, which are then stored in a database.

[0907] Step 4:

[0908] The server transmits the generated reply proposal to the user's device.

[0909] Input: The generated suggested reply.

[0910] Data processing: Converting the proposed reply into a data format for sending to the user's device.

[0911] Output: The suggested reply sent to the user's device.

[0912] Specific behavior: The server converts the generated reply proposal into an appropriate data format and sends it to the user's device.

[0913] Step 5:

[0914] The terminal presents the generated reply suggestions to the user.

[0915] Input: The suggested reply sent to the user's device.

[0916] Data processing: Display reply suggestions at the bottom of the message input field.

[0917] Output: Suggested replies displayed in the message input field.

[0918] Specific operation: The device displays reply suggestions received from the server at the bottom of the message input field. When the user taps a reply suggestion, the reply is inserted into the message input field.

[0919] (Application example 3)

[0920] Next, a description will be given of Application Example 3 of Form Example 3. In the following description, the data processing device 12 will be referred to as a "server", and the smart glasses 214 will be referred to as a "terminal".

[0921] In virtual stores, responding to customer inquiries quickly and appropriately is important to improve customer satisfaction. However, it takes time and effort for store clerks to respond to each customer individually, making efficient operation difficult. Furthermore, if the store clerk's responses are inconsistent, it may undermine customer trust. To solve these problems, a system is needed that can automatically generate replies from store clerks and provide efficient and consistent customer service.

[0922] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 3 is realized by the following means.

[0923] In this invention, the server includes means for reading a user's own reply habits from the user's messaging history, means for analyzing the flow of conversation between the user and the other party, means for generating reply suggestions that are characteristic of the user based on the analysis results, means for presenting the generated reply suggestions to the user, means for automatically generating replies for store clerks to streamline customer service in the virtual store, and means for displaying the generated reply suggestions on a smartphone application so that the user can select one. This makes it possible to automatically generate quick and appropriate replies to customer inquiries, reduce the burden on store clerks, and provide consistent customer service.

[0924] "User's messaging history" means a record of messages that a User has sent or received in the past.

[0925] "Reply habits" are patterns such as words and phrases that a user frequently uses when replying to messages, as well as the timing and order of replies.

[0926] "Chat flow" refers to the progress and context of a series of messages exchanged between a user and another person.

[0927] "Analysis Results" refers to information or data obtained after analyzing a User's messaging history and conversation flow.

[0928] A "suggested reply" is an automatically generated reply as a candidate for a message that the user may send.

[0929] A "virtual store" is a virtual store operated on the Internet where customers can purchase goods and services online.

[0930] "Customer service" refers to the task of responding to inquiries and requests from customers in the virtual store.

[0931] A "smartphone application" is a software program that runs on a smartphone.

[0932] "Automatic generation" means that the system generates suggested responses based on user input and past data, without manual intervention.

[0933] "Consistency" means always maintaining the same quality and standards in customer service.

[0934] To put this invention into practice, it is necessary to build a system that reads reply habits from a user's messaging history, analyzes the flow of the conversation, and generates reply suggestions that suit the user based on the analysis results. This system has the function of automatically generating replies for store clerks and displaying the generated reply suggestions on a smartphone application to streamline customer service in a virtual store.

[0935] Hardware and software used

[0936] Hardware: Smartphone

[0937] Software: OpenAI API, Python

[0938] Data processing and calculation

[0939] 1. Data Collection: The server collects the user's messaging history and identifies the user's reply habits, including the words and phrases the user uses most frequently, and the timing and sequence of replies.

[0940] 2. Conversation flow analysis: The server analyzes the conversation flow between the user and the other party to understand the context and progress.

[0941] 3. Reply generation: Based on the analysis results, the server uses a generative AI model to generate reply suggestions that are appropriate for the user. At this time, the prompt sentence is input to the generative AI model.

[0942] 4. Presentation of reply suggestions: The server displays the generated reply suggestions on the smartphone application, allowing the user to select one.

[0943] Examples

[0944] For example, if a customer asks "Do you have this item in stock?" in a virtual store, the server will generate a suggested reply such as "Thank you. We do have this item in stock..." based on the store clerk's past reply history and patterns.

[0945] Examples of prompt statements

[0946] User response history: Thank you. This item is in stock...

[0947] User response pattern: Question followed by a thank you

[0948] Generate the following reply:

[0949] In this way, the server can automatically generate prompt and appropriate replies to customer inquiries, reducing the burden on store clerks and achieving consistent customer service.

[0950] The flow of the specific process in the application example 3 will be described with reference to FIG.

[0951] Step 1:

[0952] The server collects the user's messaging history, i.e., records of messages the user has sent and received in the past, from a database. This input data includes the words and phrases the user frequently uses, the timing and sequence of replies, etc. As an output, a dataset is obtained that can be used to identify the user's reply habits.

[0953] Step 2:

[0954] The server analyzes the conversation flow between the user and the other party by analyzing the collected messaging history to understand the context and progress. The input data includes the sequence of messages exchanged between the user and the other party. The output is the analysis result of the conversation flow.

[0955] Step 3:

[0956] Based on the analysis results, the server uses a generative AI model to generate reply suggestions that are appropriate for the user. Specifically, the prompt sentence is input into the generative AI model, which generates appropriate reply suggestions. This input data includes the user's reply habits and the analysis results of the conversation flow. The output is a reply suggestion that is appropriate for the user.

[0957] Step 4:

[0958] The server displays the generated reply suggestions on the smartphone application. Specifically, the server transmits the generated reply suggestions to the user's device and displays them below the message input field of the application. The input data includes the generated reply suggestions. As an output, the smartphone screen displays the reply suggestions that the user can select.

[0959] Step 5:

[0960] The user can select a suggested reply or arrange it as they like. Specifically, the user can either insert the reply into the message input field by tapping on it, or edit the message based on the reply. The input data includes the reply selected by the user and the edited message. The output is the final reply message.

[0961] In addition, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[0962] "Example 1"

[0963] As one embodiment of the present invention, a system is provided that includes a means for reading a user's own reply habits from the user's messaging history, a means for analyzing the flow of conversation between the user and another person, a means for generating reply suggestions that are typical of the user based on the analysis results, a means for presenting the generated reply suggestions to the user, and an emotion engine for recognizing the user's emotions.

[0964] "Example 2"

[0965] The emotion engine extracts changes in a user's emotions from the user's messaging history, specifically from the context and wording of messages, and the use of emojis and stamps.

[0966] "Example 3"

[0967] The emotion engine generates reply suggestions that take into account the changes in the extracted emotions. Specifically, if the user is expressing emotions such as happiness or sadness, the engine generates reply suggestions that match that emotion. For example, if the user is expressing happiness, the engine generates reply suggestions that also express happiness.

[0968] "Example 4"

[0969] The emotion engine extracts changes in a user's emotions in real time and generates reply suggestions in real time that take those changes into account. Specifically, every time a user sends a message, the emotion is extracted from the message and a reply suggestion based on that emotion is instantly generated.

[0970] The process flow of each embodiment will be described below.

[0971] "Example 1"

[0972] Step 1: Read the user's own reply habits from their messaging history.

[0973] Step 2: Analyze the flow of conversation between the user and the other party.

[0974] Step 3: Generate appropriate replies based on the analysis results.

[0975] Step 4: Present the generated reply suggestions to the user.

[0976] Step 5: Activate the emotion engine to recognize the user's emotions.

[0977] "Example 2"

[0978] Step 1: Extract user sentiment changes from users' messaging history.

[0979] Step 2: Generate reply suggestions that take into account the extracted emotion changes.

[0980] "Example 3"

[0981] Step 1: Extract user sentiment changes from users' messaging history.

[0982] Step 2: Generate reply suggestions that take into account the extracted emotion changes.

[0983] "Example 4"

[0984] Step 1: Extract real-time changes in user sentiment from users’ messaging history.

[0985] Step 2: Generate reply suggestions in real time that take into account the extracted emotion changes.

[0986] Example 1

[0987] Next, a description will be given of Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal".

[0988] In conventional messaging systems, it was difficult for users to respond consistently, and it was especially difficult to generate appropriate replies when they were busy or emotional. In addition, there was no system that provided reply suggestions that took the user's emotions into account, so it was not possible to respond in accordance with the user's intentions and emotions. This could lead to a decline in the quality of communication and the occurrence of misunderstandings and problems.

[0989] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0990] In this invention, the server includes a means for reading a user's own reply habits from the user's messaging history, a means for analyzing the flow of conversation between the user and the other party, a means for generating reply suggestions that are appropriate for the user based on the analysis results, a means for presenting the generated reply suggestions to the user, and an emotion engine for recognizing the user's emotions. This allows the user to reply consistently, and further makes it possible to provide appropriate reply suggestions that take the user's emotions into consideration.

[0991] "User messaging history" refers to data such as the content of messages a user has sent in the past, the time they were sent, and the frequency of those messages.

[0992] "Reply habits" refers to characteristics such as the phrases and wording that a user frequently uses when replying to messages, as well as the timing and frequency of replies.

[0993] "Conversation flow" refers to information such as the continuity, context, and tone of messages exchanged between a user and the other person.

[0994] "Means for analyzing" refers to methods or devices that use natural language processing techniques and / or generative AI models to understand the content and context of a message and extract certain patterns or characteristics.

[0995] "Suggested replies" refers to sentences that are generated as candidates for messages that the user may send.

[0996] An "emotion engine" refers to technology or devices that analyze emotions contained in a user's messages and suggested replies and recognize the user's emotional state.

[0997] A "generative AI model" is a model for generating text using artificial intelligence technology, such as one that uses natural language processing technology.

[0998] "Presenting means" refers to a method or apparatus for displaying the generated suggested replies to a user.

[0999] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[1000] This invention is a system that analyzes a user's messaging history, extracts the user's reply habits, and further analyzes the flow of a conversation between the user and the other party to generate reply suggestions that are appropriate for the user and present them to the user. A specific embodiment of this system will be described below.

[1001] Collecting messaging history

[1002] The server retrieves message data from the user's messaging application (e.g., a general messaging application). The retrieved data is stored in a database. This data includes the content of messages the user has sent in the past, the time they were sent, and how frequently they were sent.

[1003] Analyzing messaging history

[1004] The server analyzes the stored message data. It uses natural language processing (NLP) techniques to extract frequently used phrases and vocabulary used by users. For example, it uses Python's NLTK library for text analysis. This reveals the user's reply habits.

[1005] Talk flow analysis

[1006] The server uses a generative AI model (e.g., a general generative AI model) to analyze the flow of the conversation between the user and the other party, thereby obtaining basic data for understanding the context and tone of the conversation and generating appropriate reply suggestions.

[1007] Generate user-friendly reply suggestions

[1008] The server generates reply suggestions that are appropriate for the user based on the analysis results. Using a generative AI model, it creates reply suggestions that take into account the user's reply habits and the flow of the conversation. For example, if a user often uses the phrase "thank you," a reply suggestion that includes that phrase will be generated.

[1009] Suggested replies

[1010] The server sends the reply proposal it has generated to the user's device. The user can review the proposed reply and make corrections or approvals as necessary. If the user approves the reply proposal, the reply is sent to the other party.

[1011] Emotion recognition by emotion engine

[1012] The server uses an emotion engine to recognize the user's emotions. The emotion engine analyzes the emotions contained in the user's messages and reply suggestions to understand the user's emotional state, which provides feedback to generate more appropriate reply suggestions.

[1013] Examples

[1014] For example, if a user frequently uses the phrase "Otsukaresama desu" (Thank you for your hard work), the server will extract this phrase as the user's reply habit. If the user receives a message saying "How was the meeting today?", the server will generate a reply suggestion of "Otsukaresama desu. The meeting went well" and present it to the user.

[1015] Examples of prompt statements

[1016] "Analyze the user's messaging history and extract the user's reply habits. Then, analyze the conversation flow between the user and the other party and generate reply suggestions that are appropriate for the user. Finally, present the generated reply suggestions to the user and provide feedback using an emotion engine that recognizes the user's emotions."

[1017] The system allows users to respond more naturally and consistently, and also provides appropriate reply suggestions that take the user's feelings into account.

[1018] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1019] Step 1:

[1020] The server retrieves message data from the user's messaging application. As input, it receives data such as the content of the user's past messages, the time of sending, and the frequency of sending. It then stores this data in a database. Specifically, it uses an API to collect data from the messaging application and stores it in a database.

[1021] Step 2:

[1022] The server analyzes the stored message data. As input, it uses the message data stored in the database. It uses natural language processing (NLP) techniques to extract phrases and wording frequently used by users. As output, it obtains data showing the user's reply habits. Specifically, it uses Python's NLTK library to perform text analysis.

[1023] Step 3:

[1024] The server uses a generative AI model to analyze the flow of the conversation between the user and the other party. As input, it uses message data between the user and the other party. It uses a generative AI model (e.g., a general generative AI model) to understand the context and tone of the conversation. As output, it obtains data showing the flow of the conversation. Specifically, it calls the generative AI model via an API and obtains the analysis results.

[1025] Step 4:

[1026] The server generates reply suggestions that are appropriate for the user based on the analysis results. As input, it uses data showing the user's reply habits and conversation flow. It uses a generative AI model to create reply suggestions that take into account the user's reply habits and conversation flow. As output, it obtains multiple reply suggestions. In concrete terms, it inputs a prompt sentence into the generative AI model and generates reply suggestions.

[1027] Step 5:

[1028] The server sends the generated reply suggestions to the user's terminal, using the generated reply suggestions as input. The user reviews the proposed reply suggestions and modifies or approves them if necessary. As output, the reply approved by the user is obtained. The specific operations include displaying the reply suggestions through a user interface and receiving the user's selection.

[1029] Step 6:

[1030] The server uses the emotion engine to recognize the user's emotions. It uses the user's message and suggested replies as input. The emotion engine analyzes the emotions contained in the user's message and suggested replies to understand the user's emotional state. As output, it obtains data indicating the user's emotional state. Specifically, it calls the emotion engine via an API and obtains the emotion analysis results.

[1031] (Application example 1)

[1032] Next, a description will be given of Application Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal".

[1033] Conventional messaging systems were unable to generate personalized reply suggestions that took into account the user's reply habits and emotions. They also lacked the ability to recommend appropriate content based on the user's interests and emotions. This made it difficult for users to quickly respond in a way that suited them, and it took a long time to find content that matched their interests.

[1034] The identification process by the identification processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means. In this invention, the server includes a means for reading the user's own reply habits from the user's messaging history, a means for analyzing the flow of the conversation between the user and the other party, a means for generating a reply proposal that suits the user based on the analysis result, a means for presenting the generated reply proposal to the user, a means for recommending personalized content based on the user's interests and emotions, and a means for calculating the similarity between the user's message and content. This enables the user to quickly reply in a way that suits them, and further enables the user to efficiently find content that matches their interests.

[1035] "User's messaging history" means a record of messages that a User has previously sent or received.

[1036] "Reply habits" are the tendencies or patterns of how a user responds to messages.

[1037] A "talk flow" refers to a series of messages exchanged between a user and another person.

[1038] "Analysis Results" refers to data or information obtained after analyzing a User's messaging history and conversation flow.

[1039] "Suggested Responses" are suggestions or sample messages that a User may use to reply to a Message.

[1040] "Personalized content" refers to information and entertainment that is individually tailored to a user's interests and emotions.

[1041] "Similarity" is an indicator that shows the relevance or degree of match between a user's message and content.

[1042] A system for implementing this invention includes means for reading a user's own reply habits from the user's messaging history, means for analyzing the flow of conversation between the user and the other party, means for generating reply suggestions that are typical of the user based on the analysis results, means for presenting the generated reply suggestions to the user, means for recommending personalized content based on the user's interests and emotions, and means for calculating the similarity between the user's message and content.

[1043] The server stores the user's messaging history in a database and uses an analysis engine to extract the user's reply habits. The analysis engine uses TextBlob, a natural language processing library, and scikit-learn, a machine learning library. This allows the content, sending time, and sending frequency of messages sent by the user in the past to be analyzed, and the user's reply habits to be extracted.

[1044] Next, the server analyzes the flow of the conversation between the user and the other person and generates a reply that suits the user based on the analysis results. The generated reply is displayed on the user's smartphone or other device. The user can use the proposed reply as is or arrange it in their own way.

[1045] In addition, the server recommends personalized content based on the user's interests and emotions, which includes a method to calculate the similarity between the user's messages and content. The similarity calculation involves vectorizing the messages and content using CountVectorizer, and calculating the similarity using cosine_similarity.

[1046] For example, if a user sends a message such as "What do you think of recent movies?" or "I want to listen to new music," the system will recommend content such as "latest movies" and "popular music." In this way, users can quickly respond in a way that suits them, and more efficiently find content that matches their interests.

[1047] Examples of prompts include:

[1048] "Write a program that analyzes a user's message history and recommends personalized content based on the user's interests. The user's message history contains messages about movies and music. The content database contains the latest movie and music information. Calculate the similarity between the user's messages and the content, and recommend the most relevant content."

[1049] The flow of the specific process in the application example 1 will be described with reference to FIG.

[1050] Step 1:

[1051] The server retrieves the user's messaging history from a database: as input, a user ID is given, and as output, a list of the user's past messages, including information such as the message content, the time sent, and how often they were sent.

[1052] Step 2:

[1053] The server analyzes the retrieved messaging history and extracts the user's reply habits. As input, it is given the list of messages obtained in step 1, and as output, it obtains patterns that indicate the user's reply habits. Specifically, it uses TextBlob to analyze the message content and identify frequently used phrases and expressions.

[1054] Step 3:

[1055] The server analyzes the talk flow between the user and the other party: as input, it is given a sequence of messages exchanged between the user and the other party, and as output, it obtains data describing the talk flow, including the order of messages and their relationships to each other.

[1056] Step 4:

[1057] The server generates reply suggestions that are appropriate for the user based on the analysis results. As input, the data obtained in steps 2 and 3 are given, and as output, reply suggestions that are appropriate for the user are generated. A generative AI model is used to create natural reply suggestions based on the user's reply habits and the flow of the conversation.

[1058] Step 5:

[1059] The server presents the generated reply suggestions to the user's terminal. The reply suggestions generated in step 4 are given as input, and are displayed on the user's terminal as output. The user can use the presented reply suggestions as they are, or arrange them in their own way.

[1060] Step 6:

[1061] The server recommends personalized content based on the user's interests and emotions. As input, the user's messaging history and current talk stream are given, and as output, a list of content suitable for the user is obtained. Messages and content are vectorized using CountVectorizer, and similarity is calculated using cosine_similarity.

[1062] Step 7:

[1063] The server recommends the most relevant content to the user based on the calculated similarity. As input, the list of content obtained in step 6 and the similarity data are given, and as output, the content to be displayed to the user is determined. The user can browse the recommended content and efficiently obtain information that matches his / her interests.

[1064] Example 2

[1065] Next, a description will be given of Example 2 of the embodiment 2. In the following description, the data processing device 12 is referred to as a "server" and the smart glasses 214 are referred to as a "terminal".

[1066] Conventional messaging systems have difficulty analyzing users' reply habits and conversation flow, and therefore cannot provide optimal reply suggestions for users. In addition, they do not fully understand the conversation flow taking into account changes in the user's emotions, and therefore cannot improve the quality of communication. This results in users having to spend a lot of time thinking of appropriate replies, which reduces the efficiency of communication.

[1067] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[1068] In this invention, the server includes means for reading a user's own reply habits from the user's messaging history, means for analyzing the flow of a conversation between the user and another person, means for generating reply suggestions that are appropriate for the user based on the analysis results, means for presenting the generated reply suggestions to the user, means for analyzing the message context, wording, and usage of emoticons and stamps to extract changes in emotion, and means for understanding the flow of the conversation based on the extracted changes in emotion. This makes it possible to provide optimal reply suggestions that take into account the user's reply habits and changes in emotion.

[1069] "User's messaging history" means a record of messages that a User has sent or received in the past.

[1070] A "reply habit" is a user's tendency to reply to messages in a particular pattern or style.

[1071] "Chat flow" refers to the progress and changes in the message exchange between a user and another person.

[1072] "Analysis results" are the results of analysis conducted based on collected data.

[1073] A "Suggested Replies" is a suggestion for a User to reply to a Message.

[1074] "Context" is the situation or background information in which a message is sent or received.

[1075] "Language" refers to the choice of words and expressions that a user uses in their messages.

[1076] "Emoji" are small pictures or icons used to express emotions or intent in messages.

[1077] A "stamp" is a large image or illustration used to convey emotion or meaning in a message.

[1078] "Emotional changes" refers to the change in a user's emotions over time and depending on the content of the message.

[1079] An "emotion engine" is software that analyzes a user's emotions from the content and expression of a message.

[1080] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[1081] The present invention is a system that analyzes a user's messaging history and provides optimal reply suggestions that take into account the user's reply habits and changes in emotions. A specific embodiment of this system is described below.

[1082] 1. Program Generation

[1083] The server generates a program to analyze the flow of the conversation between the user and the other party. This program analyzes the exchange of messages between the user and the other party and understands changes in topic and sentiment.

[1084] 2. Program Processing Explanation

[1085] The server collects and analyzes the messages exchanged between the user and the other party. Specifically, it performs the following processes:

[1086] Data collection: The server collects the message history between the user and the other party. This data includes text messages, emojis, stamps, etc. The database used is, for example, MySQL or MongoDB.

[1087] Conversation flow analysis: The server analyzes the collected message data and understands the conversation flow. For example, NLTK or spaCy is used as a natural language processing (NLP) library. The server extracts patterns of users expressing gratitude after receiving a question from the other party.

[1088] Emotion change analysis: The server uses an emotion engine to extract changes in the user's emotions. For example, IBM Watson Tone Analyzer or Google Cloud Natural Language API can be used as the emotion engine. This allows the server to analyze the context and wording of messages, as well as the use of emojis and stamps.

[1089] Output of results: The server provides the analysis results to the user. The analysis results are generated in report format and sent to the user's terminal. For example, a report in PDF format is generated and sent by email.

[1090] 3. Examples of concrete examples and prompts

[1091] As a concrete example, consider the following scenario.

[1092] scenario:

[1093] User A and User B are exchanging messages. User A asks, "How was your day?" and User B replies, "It was great, thanks!" In this exchange, we can see a pattern where User B expresses gratitude.

[1094] Example prompt:

[1095] "Please generate a program that analyzes the message history of users A and B and understands the flow of the conversation and changes in emotions. In particular, please extract the pattern of user B expressing gratitude and analyze the changes in emotions using an emotion engine."

[1096] In this way, the server analyzes the flow of conversation between the user and the other person and executes a program to understand changes in emotions, making it possible to provide optimal reply suggestions that take into account the user's reply habits and changes in emotions.

[1097] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1098] Step 1:

[1099] The server collects message history between the user and the recipient.

[1100] Input: User ID and contact ID

[1101] What happens: The server retrieves the user and their contacts' past messages from a database (e.g. MySQL or MongoDB).

[1102] Output: Collected messaging data (text messages, emojis, stamps, etc.)

[1103] Step 2:

[1104] The server analyzes the collected message data and understands the flow of the conversation.

[1105] Input: Collected message data

[1106] What it does: The server uses a natural language processing (NLP) library (e.g., NLTK or spaCy) to analyze the content of the messages and extract patterns where the user expresses gratitude after receiving a question from the other person.

[1107] Output: Analysis of conversation flow (patterns and topic changes)

[1108] Step 3:

[1109] The server uses an emotion engine to extract changes in the user's emotions.

[1110] Input: Collected message data

[1111] What it does: The server uses an emotion engine (e.g., IBM Watson Tone Analyzer or Google Cloud Natural Language API) to analyze the message's context, wording, and use of emojis and stickers.

[1112] Output: Emotion change analysis results

[1113] Step 4:

[1114] The server generates suggested replies that are likely to be appropriate for the user based on the analysis results.

[1115] Input: Analysis results of conversation flow, analysis results of emotion changes

[1116] How it works: The server uses a generative AI model to generate reply suggestions that take into account the user's reply habits and changes in sentiment.

[1117] Output: The generated reply proposal

[1118] Step 5:

[1119] The server presents the generated reply suggestions to the user.

[1120] Input: Generated reply proposal

[1121] What happens: The server sends the suggested reply to the user's device, for example by sending a notification to the user's messaging app.

[1122] Output: The suggested replies offered to the user

[1123] In this way, the server analyzes the flow of conversation between the user and the other person and executes a program to understand changes in emotions, making it possible to provide optimal reply suggestions that take into account the user's reply habits and changes in emotions.

[1124] (Application example 2)

[1125] Next, a description will be given of Application Example 2 of Form Example 2. In the following description, the data processing device 12 will be referred to as a "server", and the smart glasses 214 will be referred to as a "terminal".

[1126] Conventional messaging systems can analyze a user's reply habits and the flow of conversation, but they cannot detect changes in emotions or abnormal conversation patterns. Therefore, they cannot issue appropriate warnings even when the user may be in a dangerous situation. To solve this problem, a system that can analyze changes in a user's emotions in real time, detect abnormal conversation patterns, and issue warnings is required.

[1127] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[1128] In this invention, the server includes means for reading a user's own reply habits from the user's messaging history, means for analyzing the flow of a conversation between the user and another party, means for generating reply suggestions that are characteristic of the user based on the analysis results, means for presenting the generated reply suggestions to the user, means for extracting changes in emotion from the user's messaging history, means for detecting abnormal talk patterns based on the changes in emotion, and means for issuing a warning when an abnormal talk pattern is detected. This makes it possible to analyze changes in a user's emotions in real time, detect abnormal talk patterns, and issue a warning.

[1129] "User's messaging history" means a record of messages that a User has sent or received in the past.

[1130] "Reply habits" are patterns or tendencies in how a user responds to messages.

[1131] "Chat flow" refers to the progress and changes in the message exchange between a user and the other person.

[1132] "Analysis results" refer to information and data obtained after analyzing the flow of conversation and changes in emotions.

[1133] "User-friendly replies" are reply suggestions generated based on a user's past messaging history and replying habits.

[1134] "Changes in emotions" refers to fluctuations in emotions that can be read from the content of a user's message and the way it is expressed.

[1135] An "unusual talk pattern" is a pattern of messaging that differs from normal interactions and indicates a particular risk or problem.

[1136] A "warning" is a message issued to the user to warn them when an abnormal talk pattern is detected.

[1137] A system for implementing the present invention has a function of analyzing a user's messaging history and detecting changes in emotions and abnormal talk patterns. A specific embodiment of this system will be described below.

[1138] First, the server collects the user's messaging history, which includes records of messages the user has sent and received in the past. The server analyzes this history to understand the user's reply habits and conversation flow.

[1139] The server then generates personalized replies based on the analysis, based on the user's messaging history and replying habits, and presents the replies to the user.

[1140] In addition, the server extracts changes in user emotions from the user's messaging history. Specifically, it analyzes changes in user emotions from the context of messages, wording, and the use of emojis and stamps. For this analysis, it uses TextBlob, a natural language processing library.

[1141] Based on the changes in emotion, the server detects anomalous talk patterns, which are patterns of messaging that are different from normal interactions and indicate a particular risk or problem, such as a sudden change in emotion or a series of negative emotions.

[1142] If an abnormal talk pattern is detected, the server will issue a warning to the user, letting them know that they may be in a dangerous situation.

[1143] As a concrete example, consider the following message exchange:

[1144] Message 1: "Today was such a fun day!"

[1145] Message 2: "I suddenly felt really sad."

[1146] In this case, due to the large change in emotions, the server detects the abnormal talk pattern and issues a warning to the user.

[1147] Here is an example of a prompt to be input to the generative AI model:

[1148] Analyze a user's message history to detect changes in sentiment. The following messages will be analyzed:

[1149] 1. Today was such a fun day!

[1150] 2. I suddenly felt very sad.

[1151] If there are big changes in emotion, raise a warning.

[1152] In this way, a security chat monitoring system can be realized to protect the safety of users.

[1153] The flow of the identification process in the application example 2 will be described with reference to FIG.

[1154] Step 1:

[1155] The server collects the messaging history of the user. The input is a record of messages sent and received by the user in the past, and the output is the collected messaging history data. In particular, the server obtains the message data through the API of the messaging application.

[1156] Step 2:

[1157] The server analyzes the collected messaging history to understand the user's reply habits and the flow of the conversation. The input is the collected messaging history data, and the output is the analysis results regarding the user's reply habits and the flow of the conversation. In concrete terms, the server analyzes message patterns using a natural language processing algorithm.

[1158] Step 3:

[1159] The server generates reply suggestions that are appropriate for the user based on the analysis results. The input is the analysis results on the user's reply habits and the flow of the conversation, and the output is the generated reply suggestions that are appropriate for the user. In concrete terms, the server generates multiple reply suggestions using a generative AI model.

[1160] Step 4:

[1161] The server presents the generated reply suggestions to the user. The input is the generated user-like reply suggestions, and the output is the reply suggestions presented to the user. In particular, the server sends a notification to the user's device and displays the reply suggestions.

[1162] Step 5:

[1163] The server extracts emotion changes from a user's messaging history. The input is the collected messaging history data, and the output is data on emotion changes. Specifically, the server uses TextBlob to analyze the emotion of a message.

[1164] Step 6:

[1165] The server detects anomalous talk patterns based on changes in emotion. The input is data on changes in emotion, and the output is information on the detected anomalous talk patterns. In specific operations, the server calculates the magnitude of the change in emotion and determines whether there is an anomalous change.

[1166] Step 7:

[1167] The server issues a warning when it detects an abnormal talk pattern. The input is information about the detected abnormal talk pattern, and the output is a warning message for the user. In particular, the server sends a warning notification to the user's terminal.

[1168] Example 3

[1169] Next, a third embodiment of the third embodiment will be described. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal".

[1170] Conventional messaging systems have the problem that it takes time for users to think of what to reply, making it difficult to communicate efficiently. In addition, they lack the functionality to automatically generate appropriate replies according to the user's emotions, making it difficult to communicate in a way that reflects the user's emotions. Furthermore, there is insufficient means for users to easily select suggested replies and insert them into the message input field.

[1171] The specific processing by the specific processing unit 290 of the data processing device 12 in the third embodiment is realized by the following means.

[1172] In this invention, the server includes means for reading a user's own reply habits from the user's messaging history, means for analyzing the flow of conversation between the user and the other party, means for generating reply suggestions that suit the user based on the analysis results, means for extracting emotions from the user's message and generating reply suggestions based on the emotions, means for presenting the generated reply suggestions to the user, and means for the user to select the presented reply suggestions and insert them into the message input field. This allows the user to significantly reduce the time spent struggling with replying, enabling efficient communication that reflects emotions.

[1173] "User's messaging history" means a record of messages that a User has previously sent and received.

[1174] "Reply habits" are patterns of what words and phrases a user frequently uses in response to messages.

[1175] "Chat flow" refers to the sequence and content of messages exchanged between a user and another person.

[1176] "Analysis Results" refers to information or data obtained after analyzing a User's messaging history and conversation flow.

[1177] A "suggested reply" is a suggested reply that a user can make to a message.

[1178] "Sentiment extraction" is the process of identifying emotions such as happiness, sadness, anger, etc. from a user's message.

[1179] "Sentiment-based reply suggestions" are reply suggestions that reflect the user's emotions.

[1180] "Presenting" is the act of showing the generated reply suggestions to the user.

[1181] The "message input field" is a text field in which the user can enter a message.

[1182] "Selection" is the act of the user choosing one of the suggested replies presented.

[1183] "Insert" refers to the act of adding the selected reply proposal to the message input field.

[1184] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[1185] The present invention is a system that analyzes a user's messaging history and automatically generates reply suggestions that are appropriate for the user. A specific embodiment of this system is described below.

[1186] 1. Program Generation

[1187] The server runs a program to collect and analyze users' messaging history. The program uses natural language processing (NLP) techniques to analyze users' reply habits and conversation flow. Specifically, it uses generative AI models such as Google's BERT and OpenAI's GPT-3.

[1188] 2. Program Processing Explanation

[1189] The server collects the user's past message data, including messages sent and received by the user, for example by obtaining message history from a chat application database.

[1190] Next, the server analyzes the collected message data. It uses a natural language processing library (e.g., spaCy or NLTK) to extract frequently used phrases and reply patterns by users. Based on the analysis results, it generates reply suggestions that are appropriate for the user. It inputs prompt sentences that reflect the user's reply patterns into the generative AI model to generate appropriate reply suggestions.

[1191] In addition, the server extracts emotions from the user's message. Using an emotion engine (e.g., IBM Watson's Tone Analyzer), it analyzes the emotions (happiness, sadness, anger, etc.) expressed by the user in real time. It then generates a reply proposal based on that emotion. For example, if a user sends a message saying, "I had a lot of fun today!", the emotion engine extracts the emotion of "happiness" and generates a reply proposal saying, "I had a great time too! Let's go together again!"

[1192] 3. Examples of concrete examples and prompts

[1193] As a concrete example, if a user sends a message saying, "I had so much fun today!", the emotion engine will extract the emotion of "joy" and generate a suggested reply saying, "I had fun too! Let's go together again!"

[1194] Example prompt:

[1195] User message: "Today was so much fun!"

[1196] Prompt for generative AI model: "Generate suggested replies when the user shows delight."

[1197] Showing generated reply suggestions

[1198] The device presents the generated reply suggestions to the user. Specifically, the generated reply suggestions are displayed at the bottom of the message input field, and the user can insert the reply suggestion into the message input field by tapping it. This significantly reduces the time users spend thinking about replying, enabling efficient communication that reflects their feelings.

[1199] This system allows the user to reply quickly and appropriately, improving the quality of communication. The flow of the identification process in the third embodiment will be described with reference to FIG.

[1200] Step 1:

[1201] The server collects the user's messaging history, specifically, retrieves past message data from the chat application's database. The input is the user ID and associated message history, and the output is the collected message data, including messages sent and received by the user.

[1202] Step 2:

[1203] The server analyzes the collected message data. It uses a natural language processing library (e.g., spaCy or NLTK) to extract frequently used phrases and reply patterns by users. The input is the collected message data, and the output is the analysis results on the user's reply habits and the flow of the conversation. Specific operations include tokenizing messages, tagging parts of speech, and extracting frequently used phrases.

[1204] Step 3:

[1205] The server generates a reply proposal that is appropriate for the user based on the analysis results. A prompt sentence that reflects the user's reply patterns is input to a generative AI model (e.g., GPT-3), which generates appropriate reply proposals. The input is the analysis results and the prompt sentence, and the output is the generated reply proposal. Specific operations include generating a prompt sentence, inputting it into the AI ​​model, and generating reply proposals.

[1206] Step 4:

[1207] The server extracts emotions from the user's message. Using an emotion engine (for example, IBM Watson's Tone Analyzer), it analyzes the emotions expressed by the user (happiness, sadness, anger, etc.) in real time. The input is the user's message, and the output is the extracted emotion information. Specific operations include analyzing the emotion of the message and adding emotion tags.

[1208] Step 5:

[1209] The server generates reply suggestions based on emotions. Based on the output of the emotion engine, a prompt that reflects the emotion is input to the generative AI model, and an appropriate reply suggestion is generated. The input is emotion information and a prompt, and the output is a reply suggestion based on the emotion. Specific operations include generating a prompt according to the emotion, inputting it into the AI ​​model, and generating a reply suggestion.

[1210] Step 6:

[1211] The device presents the generated reply suggestions to the user. Specifically, the generated reply suggestions are displayed at the bottom of the message input field, and the user can insert the reply suggestions into the message input field by tapping on them. The input is the generated reply suggestions, and the output is the reply suggestions presented to the user. Specific operations include updating the UI, displaying reply suggestions, and supporting user selection.

[1212] (Application example 3)

[1213] Next, a description will be given of Application Example 3 of Form Example 3. In the following description, the data processing device 12 will be referred to as a "server", and the smart glasses 214 will be referred to as a "terminal".

[1214] Conventional chat support systems have the problem that it is difficult for customer support staff to respond quickly and appropriately when communicating with customers. In particular, it is not possible to automatically generate reply suggestions that take into account the staff's reply habits and changes in emotions, making it difficult to achieve efficient communication.

[1215] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 3 is realized by the following means. In this invention, the server includes a means for reading the reply habits of the user from the messaging history of the user, a means for analyzing the flow of the conversation between the user and the other party, a means for generating a reply proposal that is suitable for the user based on the analysis result, a means for presenting the generated reply proposal to the user, a means for extracting changes in the user's emotions in real time and generating a reply proposal that takes into account the changes in the emotions, a means for generating a reply proposal based on a prompt sentence using a generative AI model, and a means for displaying the generated reply proposal at the bottom of the chat window so that the user can select and use it. This enables a customer support representative to quickly and appropriately reply to a customer in communication with the customer.

[1216] "User's messaging history" means a record of messages that a User has sent or received in the past.

[1217] "Reply habits" are phrases or patterns that a user frequently uses when replying to messages.

[1218] "Chat flow" refers to the sequence and content of messages exchanged between a user and another person.

[1219] "Analysis Results" refers to the information and data obtained after analyzing a User's messaging history and conversation flow.

[1220] "User-friendly reply suggestions" are suggestions of replies that a user would naturally use, generated based on the user's reply habits and the flow of the conversation.

[1221] "Sentiment change" refers to the change in the emotional state extracted from a user's messages over time.

[1222] A "generative AI model" is a model that has been trained using artificial intelligence to perform a specific task, and is used here to generate reply suggestions.

[1223] A "prompt" is an instruction given to a generative AI model that serves as the basis for the model to generate appropriate reply suggestions.

[1224] A "chat window" is an interface through which users can send and receive messages.

[1225] "Real-time" refers to data processing and analysis being carried out immediately.

[1226] A system for implementing this invention includes means for reading a user's own reply habits from the user's messaging history, means for analyzing the flow of conversation between the user and another party, means for generating reply suggestions that are typical of the user based on the analysis results, means for presenting the generated reply suggestions to the user, means for extracting changes in the user's emotions in real time and generating reply suggestions taking into account the changes in emotion, means for generating reply suggestions based on prompt sentences using a generative AI model, and means for displaying the generated reply suggestions at the bottom of the chat window so that the user can select and use them.

[1227] The server stores the user's messaging history in a database and uses an analysis engine to extract the user's reply habits. The analysis engine identifies phrases and patterns frequently used by the user and generates reply suggestions that are appropriate for the user based on those. In addition, a sentiment analysis module is used to extract changes in emotions from the user's messages in real time and generate reply suggestions based on those emotions.

[1228] We use OpenAI's GPT-3 as the generative AI model. GPT-3 generates appropriate reply suggestions based on the prompt text. The prompt text is constructed based on information including the user's reply habits, current emotional state, and the flow of the conversation. For example, the following prompt text is generated:

[1229] User reply habits: ["Thank you", "Okay", "Nice to meet you"]

[1230] Emotion: neutral

[1231] Message: "How do I return an item?"

[1232] Generate a proposed reply.

[1233] The device displays the generated reply suggestions at the bottom of the chat window for the user to select and use. The user can use the suggested reply suggestions as is or arrange them in their own way.

[1234] As a concrete example, when a customer support representative receives a message from a customer saying, "Please tell me about returning an item," the system analyzes the representative's past response history and current emotional state to generate a response suggestion like the following:

[1235] "Thank you. Please follow the steps below to return your item."

[1236] This way, customer support representatives can respond quickly and appropriately.

[1237] The flow of the specific process in the application example 3 will be described with reference to FIG.

[1238] Step 1:

[1239] The server retrieves the user's messaging history from a database. The input is a user ID, and the output is the user's past messaging history. Specifically, the server executes a database query to retrieve the user's messaging history.

[1240] Step 2:

[1241] The server analyzes the retrieved messaging history and extracts the user's reply habits. The input is the messaging history, and the output is patterns that indicate the user's reply habits. Specifically, the server uses natural language processing algorithms to identify frequently used phrases and patterns.

[1242] Step 3:

[1243] The server analyzes the flow of conversation between the user and the other party in real time. The input is the current chat messages, and the output is the analysis result of the conversation flow. Specifically, the server analyzes the order and content of messages to understand the conversation flow.

[1244] Step 4:

[1245] The server extracts the user's emotional changes in real time. The input is the current chat message, and the output is data indicating the emotional changes. Specifically, the server uses an emotion analysis module to extract emotions from the messages.

[1246] Step 5:

[1247] The server generates a prompt sentence based on the analysis results. The input is the user's reply habits, the flow of the conversation, and changes in emotions, and the output is the prompt sentence. In concrete terms, the server combines these data to build a prompt sentence to input into the generative AI model.

[1248] Step 6:

[1249] The server uses a generative AI model to generate a reply proposal based on the prompt sentence. The input is the prompt sentence, and the output is the reply proposal. Specifically, the server sends the prompt sentence to OpenAI's GPT-3 and obtains the reply proposal.

[1250] Step 7:

[1251] The server sends the generated reply proposal to the terminal. The input is the reply proposal, and the output is the reply proposal displayed on the terminal. In concrete terms, the server sends the reply proposal to the terminal and displays it at the bottom of the chat window.

[1252] Step 8:

[1253] The user selects and uses the suggested reply. The input is the displayed reply, and the output is the selected reply. In concrete terms, the user taps on a reply displayed at the bottom of the chat window to insert it into the message input field.

[1254] The specific processing unit 290 transmits the result of the specific processing to the smart glasses 214. In the smart glasses 214, the control unit 46A causes the speaker 240 to output the result of the specific processing. The microphone 238 acquires a voice indicating a user input for the result of the specific processing. The control unit 46A transmits the voice data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the voice data.

[1255] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet search<URL: https: / / openai.com / blog / chatgpt> ) and other generation AIs. The data generation model 58 is obtained by making a neural network perform deep learning. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating a voice, text data indicating a text, and image data indicating an image is input. The data generation model 58 performs inference on the input inference data according to the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1256] Another example of generative AI is Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) are mentioned.

[1257] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the smart glasses 214.

[1258] [Third embodiment]

[1259] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[1260] 5, the data processing system 310 includes the data processing device 12 and a headset type terminal 314. An example of the data processing device 12 is a server.

[1261] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a wide area network (WAN) and / or a local area network (LAN).

[1262] The headset type terminal 314 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a display 343. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the display 343 are also connected to the bus 52.

[1263] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs the voice according to instructions from the processor 46.

[1264] Camera 42 is a small digital camera equipped with an optical system including a lens, an aperture, and a shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (e.g., an imaging range defined by an angle of view equivalent to the width of the field of vision of an average healthy person).

[1265] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for transmitting and receiving various types of information between the processor 46 and the processor 28 via the network 54. The transmission and reception of various types of information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is performed in a secure state.

[1266] Fig. 6 shows an example of main functions of the data processing device 12 and the headset type terminal 314. As shown in Fig. 6, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[1267] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32, and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[1268] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[1269] In the headset type terminal 314, the reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50, and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[1270] Next, the specific processing by the specific processing unit 290 of the data processing device 12 will be described.

[1271] "Example 1"

[1272] In one embodiment of the present invention, a means is provided for reading a user's own reply habits from the user's messaging history. Specifically, the contents, sending times, sending frequency, etc. of messages sent by the user in the past are analyzed to extract the user's reply habits. For example, if the user frequently uses the phrase "thank you," the use of the phrase "thank you" is extracted as the user's reply habits.

[1273] "Example 2"

[1274] It also provides a means to analyze the flow of conversation between users and other people. Specifically, it analyzes the patterns of message exchanges between users and other people, changes in topics, changes in emotions, etc. to understand the flow of the conversation. For example, if a user always expresses gratitude after receiving a question from the other person, this pattern can be extracted and the flow of the conversation analyzed.

[1275] "Example 3"

[1276] Furthermore, a means is provided for generating reply suggestions that are appropriate for the user based on the analysis results. Specifically, reply suggestions that are appropriate for the user are automatically generated based on the analysis results of the user's reply habits and the flow of the conversation. For example, if a user frequently uses the phrase "thank you" and has a pattern of expressing gratitude after receiving a question from the other person, a reply suggestion of "Thank you, about that..." is generated based on that information.

[1277] "Example 4"

[1278] Finally, a means is provided for presenting the generated reply suggestions to the user. Specifically, the generated reply suggestions are displayed on the user's device and the user can select one. For example, the generated reply suggestions are displayed at the bottom of the message input field, and the user can insert the reply suggestion into the message input field by tapping it. This allows the user to significantly reduce the time spent thinking about replying and achieve efficient communication.

[1279] The process flow of each embodiment will be described below.

[1280] "Example 1"

[1281] Step 1: Obtain the user's messaging history. Specifically, obtain information from the database about the user's past messages, such as the content of the messages, the time they were sent, and the frequency of their messages.

[1282] Step 2: Extract the user's own reply habits from the acquired messaging history. Specifically, the system analyzes the phrases the user frequently uses, messages sent at specific times, and reply patterns to specific people to extract the user's reply habits.

[1283] "Example 2"

[1284] Step 1: Obtain the messaging history between the user and the other party. Specifically, obtain message exchange information between the user and the other party from the database.

[1285] Step 2: Analyze the flow of conversation from the acquired messaging history. Specifically, analyze message exchange patterns, topic changes, and emotional changes to understand the flow of conversation.

[1286] "Example 3"

[1287] Step 1: Obtain the analysis results of the user's reply habits and the flow of the conversation. Specifically, obtain the information extracted and analyzed using the methods described above.

[1288] Step 2: Generate reply suggestions that are appropriate for the user based on the obtained analysis results. Specifically, the system automatically generates reply suggestions that are appropriate for the user, taking into account the user's reply habits and the flow of the conversation.

[1289] "Example 4"

[1290] Step 1: Obtain the generated reply proposal. Specifically, obtain the reply proposal generated by the above-mentioned means.

[1291] Step 2: Present the retrieved reply suggestions to the user. Specifically, the generated reply suggestions are displayed on the user's device and the user can select one.

[1292] Example 1

[1293] Next, a description will be given of Example 1 of the embodiment 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal".

[1294] In conventional messaging systems, it was difficult to automatically generate reply suggestions that took into account the user's reply habits. As a result, users had to think up the reply content themselves every time, which was time-consuming and caused problems such as inconsistent replies. Furthermore, there was a lack of a way to effectively utilize the user's messaging history, making it impossible to provide personalized reply suggestions based on the user's past behavior.

[1295] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[1296] In this invention, the server includes means for reading a user's own reply habits from the user's messaging history, means for analyzing the flow of a dialogue between the user and the other party, means for generating reply suggestions that suit the user based on the analysis results, means for presenting the generated reply suggestions to the user, means for collecting the user's messaging history, means for preprocessing the collected messaging history, means for analyzing the preprocessed message contents, means for extracting the user's reply habits based on the analysis results, and means for saving the extraction results. This makes it possible to automatically generate personalized reply suggestions based on the user's past messaging history and present them to the user.

[1297] "User's messaging history" is a collection of data about messages the user has sent in the past, such as the content of those messages, the time they were sent, and the frequency with which they were sent.

[1298] "Reply habits" are patterns such as phrases that a user frequently uses when replying to messages or a tendency to reply at certain times of the day.

[1299] "Dialogue flow" refers to the order and content relevance of messages exchanged between a user and another person.

[1300] "Analysis results" are the results of analyzing users' reply habits and conversation flow based on collected and preprocessed messaging history.

[1301] "Suggested replies" are candidate messages for a user to reply to, generated based on the user's past messaging history and analysis results.

[1302] "Preprocessing" refers to the process of removing unnecessary spaces and special characters from the collected messaging history, and performing tokenization and removing stop words.

[1303] A "natural language processing library" is a software tool for tokenizing message content, analyzing sentiment, extracting frequent phrases, and so on.

[1304] "Database" refers to an information management system for storing users' messaging history and analysis results.

[1305] A "prompt sentence" is an instruction sentence to be input into a generative AI model, which clearly indicates the message history to be analyzed and the purpose of the analysis.

[1306] A "generative AI model" is an artificial intelligence model that generates reply suggestions that are appropriate for a user based on the user's messaging history and analysis results.

[1307] The present invention relates to a system for analyzing a user's messaging history, extracting the user's reply habits, and generating reply suggestions that are suited to the user. A specific embodiment of this system will be described below.

[1308] The server first collects the user's messaging history, including the content of messages the user has sent in the past, the time they were sent, how often they were sent, etc. The collected data is then stored in a database.

[1309] The server then pre-processes the collected messaging history, which includes cleaning the message content (removing unnecessary spaces and special characters), tokenizing (splitting into words), and removing stop words (removing frequent but meaningless words). This pre-processing is done using Python, with natural language processing libraries such as NLTK and spaCy.

[1310] Once preprocessed, the message content is then analyzed. The server uses natural language processing libraries to extract common phrases and words. It also performs sentiment analysis on the message and classifies it as positive, negative, or neutral.

[1311] Based on the analysis results, the server extracts the user's reply habits. For example, if a user frequently uses the phrase "thank you," that phrase is identified as the user's reply habits. Also, if there are many replies during a certain time period, that time period is also extracted as a habit.

[1312] The extracted user reply habits are stored in a database, and this information is then fed into a generative AI model as prompts to generate reply suggestions that are more likely to be user-friendly.

[1313] As a concrete example, consider the case where you want to analyze the history of messages sent by user A. Suppose user A's message history contains the following data:

[1314] Message 1: "Thank you, that was helpful."

[1315] Message 2: "Thank you for your hard work."

[1316] Message 3: "Thank you, I look forward to working with you again."

[1317] The server analyzes these messages and extracts the frequent use of the phrase "Thank you." As a result, the use of the phrase "Thank you" is identified as a reply habit of User A.

[1318] Examples of prompts to input to a generative AI model might include the following:

[1319] Analyze the user's messaging history and extract frequent phrases. The following message histories will be analyzed:

[1320] 1. "Thank you, that was helpful."

[1321] 2. "Thank you for your hard work."

[1322] 3. "Thank you, I look forward to working with you again."

[1323] By inputting this prompt into a generative AI model, it is possible to automatically extract the user's reply habits and generate reply suggestions that are appropriate for the user.

[1324] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1325] Step 1:

[1326] The server collects the messaging history of users. As input, a user ID is given. The server retrieves data such as message content, sending time, and sending frequency for that user from the database. As output, the collected messaging history is obtained. Specifically, the server executes SQL queries to extract the required information from the database.

[1327] Step 2:

[1328] The server preprocesses the collected messaging history. As input, it is given the collected messaging history. The server cleans the message content (removes unnecessary spaces and special characters), tokenizes it (splits it into words), and removes stop words. As output, it gets the preprocessed message content. Specifically, the server runs a Python script to perform preprocessing using natural language processing libraries (NLTK and spaCy).

[1329] Step 3:

[1330] The server analyzes the preprocessed message content. The preprocessed message content is given as input. The server uses a natural language processing library to extract frequent phrases and words and perform sentiment analysis. The analysis results are obtained as output. Specifically, the server runs a Python script to extract frequent phrases and perform sentiment analysis.

[1331] Step 4:

[1332] The server extracts the user's reply habits based on the analysis results. The analysis results are given as input. The server identifies frequent phrases and tendencies to reply at specific times of the day. The output is the user's reply habits. In concrete terms, the server analyzes the analysis results and runs an algorithm to identify the user's reply habits.

[1333] Step 5:

[1334] The server saves the extracted user's reply habits in a database. As input, the user's reply habits are given. The server performs an insert operation on the database and saves the extracted results. As output, it is confirmed that the saving has been completed. As a specific operation, the server executes a SQL query to insert data into the database.

[1335] Step 6:

[1336] The server inputs a prompt sentence into the generative AI model to generate a reply proposal that is appropriate for the user. The user's reply habits and the prompt sentence are given as input. The server inputs the prompt sentence into the generative AI model to generate a reply proposal. The generated reply proposal is obtained as output. In concrete terms, the server sends a prompt sentence to the generative AI model and obtains a reply proposal.

[1337] Step 7:

[1338] The server presents the generated reply proposal to the user. As input, the generated reply proposal is given. The server sends the reply proposal to the user's device and presents it to the user. As output, a state is obtained where the user can confirm the reply proposal. As a specific operation, the server sends a notification to the user's device and displays the reply proposal.

[1339] (Application example 1)

[1340] Next, a description will be given of Application Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as a "server", and the headset type terminal 314 will be referred to as a "terminal".

[1341] Conventional messaging systems do not recommend personalized content that takes into account the user's reply habits. This makes it difficult to efficiently provide content that is of interest to the user. In addition, there has been no system that integrates the generation of reply suggestions based on the user's reply habits and content recommendation.

[1342] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[1343] In this invention, the server includes means for reading the user's own reply habits from the user's messaging history, means for analyzing the flow of a conversation between the user and another person, means for generating reply suggestions that are suited to the user based on the analysis results, means for presenting the generated reply suggestions to the user, means for recommending personalized content based on the user's reply habits, and means for presenting the recommended content to the user. This makes it possible to recommend personalized content that takes into account the user's reply habits, and efficiently provide content that is more interesting to the user.

[1344] "User's messaging history" is a collection of data about messages the user has sent in the past, such as the content of those messages, the time they were sent, and the frequency with which they were sent.

[1345] "Reply habits" are patterns of phrases or words that a user frequently uses when replying to messages.

[1346] A "talk flow" is a process of continuous exchange of messages between a user and another person.

[1347] "Analysis Results" refers to information or data obtained after analyzing a user's messaging history and conversation flow.

[1348] A "suggested reply" is a suggested message that a user can use as a reference when replying.

[1349] "Personalized Content" means information or entertainment content that is specifically selected based on a user's individual characteristics and interests.

[1350] A "recommendation means" is a method or technology for selecting and presenting appropriate content or information to a user.

[1351] A system for implementing this invention includes means for reading a user's own reply habits from the user's messaging history, means for analyzing the flow of conversation between the user and another party, means for generating reply suggestions that are typical of the user based on the analysis results, means for presenting the generated reply suggestions to the user, means for recommending personalized content based on the user's reply habits, and means for presenting the recommended content to the user.

[1352] The server runs software to collect and analyze users' messaging history. Specifically, it uses programming languages ​​such as Python to analyze the content, time and frequency of users' messages and extracts their reply habits. This analysis is carried out using data processing tools such as collections.Counter.

[1353] The server then generates a reply that is appropriate for the user based on the analysis results, and presents the reply to the user, who can use it as is or customize it in his or her own way.

[1354] In addition, the server recommends personalized content based on the user's reply habits. For this recommendation, the server retrieves content that corresponds to the user's reply habits from the database and presents it to the user. For example, if the user frequently uses the phrase "thank you," the server recommends content such as "How to express gratitude" and "10 ways to say thank you."

[1355] This system is implemented using a smartphone. An application is installed on the smartphone to collect the user's messaging history and send it to a server. The application periodically sends the user's messaging history to the server, receives reply suggestions and recommended content from the server, and presents them to the user.

[1356] As a concrete example, by inputting the following prompt sentence into a generative AI model, personalized content recommendations based on the user’s reply habits can be realized:

[1357] "Build an application that extracts reply habits from a user's messaging history and recommends personalized content based on those habits. For example, if a user frequently uses the phrase 'Thank you,' recommend content like 'How to express gratitude' or '10 ways to say thank you.'"

[1358] In this way, it is possible to leverage a user's messaging history to provide them with the most relevant content.

[1359] The flow of the specific process in the application example 1 will be described with reference to FIG.

[1360] Step 1:

[1361] The server collects the user's messaging history. It receives data such as message content, sending time, and sending frequency from the user's device and stores them in a database. The input is the user's messaging history, and the output is the stored message data.

[1362] Step 2:

[1363] The server analyzes the collected messaging history. Using a programming language such as Python, it counts the frequency of message content using collections.Counter. The input is the stored message data, and the output is data that indicates the user's reply habits.

[1364] Step 3:

[1365] The server generates reply suggestions that are appropriate for the user based on the analysis results. Using a generative AI model, it generates multiple reply suggestions based on the user's reply habits. The input is data that indicates the user's reply habits, and the output is the generated multiple reply suggestions.

[1366] Step 4:

[1367] The server presents the generated reply suggestions to the user's terminal. The server transmits reply suggestions to the user's terminal for the user to select or arrange. The input is the generated multiple reply suggestions, and the output is the reply suggestions presented to the user.

[1368] Step 5:

[1369] The server recommends personalized content based on the user's reply habits. It retrieves content corresponding to the user's reply habits from a database and creates a recommendation list. The input is data indicating the user's reply habits, and the output is a recommended content list.

[1370] Step 6:

[1371] The server presents the recommended content to the user's terminal. The server transmits the recommended content to the user's terminal so that the user can view it. The input is the recommended content list, and the output is the content presented to the user.

[1372] Step 7:

[1373] The user reviews the suggested replies and recommended content, selects or arranges a suggested reply as necessary, and views the recommended content. The input is the suggested replies and content presented to the user, and the output is the user's selected or arranged reply and viewed content.

[1374] Example 2

[1375] Next, a second embodiment of the second form will be described. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal".

[1376] Conventional messaging systems generate reply suggestions without considering the user's reply patterns or changes in emotions, so they often present replies that are unnatural to the user. In addition, there is a lack of means to analyze the flow of conversation between the user and the other person, making it difficult to generate reply suggestions that appropriately reflect changes in topic or emotions.

[1377] The identification process by the identification processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means. In this invention, the server includes a means for reading a user's own reply pattern from the user's messaging history, a means for collecting message exchanges between the user and the other party, a means for pre-processing the collected message data, a means for tokenizing the pre-processed messages and tagging them with parts of speech, a means for analyzing the sentiment of each message, a means for analyzing message exchange patterns, topic changes, and sentiment changes, a means for generating reply suggestions that are appropriate for the user based on the analysis results, and a means for presenting the generated reply suggestions to the user. This makes it possible to generate natural reply suggestions that take into account the user's reply patterns and sentiment changes.

[1378] "User's messaging history" means a record of messages that a User has sent or received in the past.

[1379] "Reply patterns" are the tendencies of the reply format and content that a user repeatedly uses in response to a particular situation or topic.

[1380] "Message exchange" refers to the sending and receiving of a series of messages between a User and another person.

[1381] "Preprocessing" refers to the process of removing unnecessary spaces and special characters from the message data and preparing the data in a format suitable for analysis.

[1382] "Tokenization" is the process of dividing a message into its smallest units, such as words or phrases.

[1383] "Part-of-speech tagging" is the process of identifying and assigning a part of speech (noun, verb, adjective, etc.) to a tokenized word or phrase.

[1384] "Sentiment analysis" is the process of extracting emotions (happiness, sadness, anger, etc.) from the content of a message and evaluating their intensity.

[1385] "Topic changes" refers to how the topic changes during a message exchange.

[1386] "Analysis results" refers to information obtained from analysis of message exchange patterns, changes in topics, changes in sentiment, etc.

[1387] A "suggested reply" is a suggested message that a user can use to reply to the other person.

[1388] "Presenting" is the act of showing the generated reply suggestions to the user.

[1389] This invention is a system that reads a user's own reply patterns from the user's messaging history, analyzes message exchanges between the user and other parties, generates reply suggestions that are appropriate for the user based on the analysis results, and presents the generated reply suggestions to the user.

[1390] The server first collects user and recipient message data. This data is retrieved from a database such as MySQL or MongoDB. It then pre-processes the collected message data. Pre-processing includes removing unnecessary whitespace and special characters.

[1391] The preprocessed message data is tokenized using natural language processing libraries such as SpaCy and NLTK, and each token is tagged with a part of speech. For example, the message "Hello, how are you today?" is tagged as "Hello / NOUN How are you today / ADV How are you today / VERB AUX".

[1392] The server then uses a sentiment analysis tool such as VADER or TextBlob to calculate a sentiment score for each message. For example, the message "Hello, how are you today?" is calculated to have a sentiment score of 0.5.

[1393] The server then analyzes messaging patterns, topic changes, and sentiment shifts — for example, finding that a user always says "Thank you" after receiving a question from the other person.

[1394] Based on the analysis results, the server generates a reply that is appropriate for the user. This reply is generated taking into account the user's past reply patterns and changes in emotions. The generated reply is presented to the user. The user can use the proposed reply as is or arrange it in their own way.

[1395] As a concrete example, consider the following message exchange:

[1396] Other person: "Hello, how are you today?"

[1397] User: "Hi, I'm fine. How about you?"

[1398] The other person: "I'm fine. Has anything new happened to you lately?"

[1399] User: "Yes, I've started a new project. Thank you."

[1400] An example prompt to parse this interaction is:

[1401] Analyze the messages exchanged between users and their contacts and extract the following information:

[1402] 1. Message exchange patterns

[1403] 2. Change the topic

[1404] 3. Emotional changes

[1405] Message exchange:

[1406] Other person: "Hello, how are you today?"

[1407] User: "Hi, I'm fine. How about you?"

[1408] The other person: "I'm fine. Has anything new happened to you lately?"

[1409] User: "Yes, I've started a new project. Thank you."

[1410] By feeding this prompt into a generative AI model, results can be obtained that analyze message exchange patterns, topic changes, and emotional shifts.

[1411] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1412] Step 1: Collect message data

[1413] The server retrieves the message data between the user and the other party from the database. Specifically, it executes the query "SELECT FROM messages WHERE user_id = '12345'" to the MySQL database to collect the message exchanges between the user and the other party. The input is the user ID, and the output is the corresponding message data.

[1414] Step 2: Preprocessing the message data

[1415] The server pre-processes the collected message data. Specifically, it removes unnecessary whitespace and special characters to clean the message. For example, it converts the message "Hello! How are you today?" to "Hello, how are you today?". The input is the collected message data, and the output is the pre-processed message data.

[1416] Step 3: Tokenize the message and tag the part of speech

[1417] The server tokenizes the preprocessed message and tags each token with a part-of-speech tag. Specifically, it uses SpaCy and NLTK to tag the message "Hello, how are you today?" as "Hello / NOUN How are you today / ADV How are you today / VERB AUX." The input is the preprocessed message data, and the output is the tokenized and part-of-speech tagged data.

[1418] Step 4: Sentiment analysis

[1419] The server analyzes the sentiment of each message. Specifically, it uses VADER and TextBlob to calculate a sentiment score for each message. For example, the sentiment score of the message "Hello, how are you today" is calculated to be 0.5. The input is the tokenized and part-of-speech tagged data, and the output is the sentiment score.

[1420] Step 5: Analyze the flow of conversation

[1421] The server analyzes message exchange patterns, topic changes, and sentiment changes. Specifically, it finds a pattern where a user always says "Thank you" after receiving a question from the other party. The input is the sentiment score and tokenized data, and the output is the analysis result.

[1422] Step 6: Generate suggested replies

[1423] The server generates reply suggestions that are appropriate for the user based on the analysis results. Specifically, it generates reply suggestions by taking into account the user's past reply patterns and changes in emotions. The input is the analysis results, and the output is the generated reply suggestions.

[1424] Step 7: Propose a response

[1425] The server presents the generated reply to the user. Specifically, the reply is displayed to the user through a web browser. The user can use the proposed reply as is or arrange it in their own way. The input is the generated reply, and the output is the reply presented to the user.

[1426] (Application example 2)

[1427] Next, a description will be given of Application Example 2 of Form Example 2. In the following description, the data processing device 12 will be referred to as a "server", and the headset type terminal 314 will be referred to as a "terminal".

[1428] Conventional messaging systems were unable to analyze users' reply habits or conversation flow, making it difficult to generate natural reply suggestions for users. They also lacked the ability to detect signs of fraud and misconduct, making it difficult to ensure user safety.

[1429] The identification process by the identification processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means. In this invention, the server includes a means for reading the reply habits of the user from the messaging history of the user, a means for analyzing the flow of the conversation between the user and the other party, a means for generating reply suggestions that are suitable for the user based on the analysis results, a means for presenting the generated reply suggestions to the user, a means for analyzing message patterns, topic changes, and emotional changes, and a means for detecting signs of misconduct or fraud. This makes it possible to generate reply suggestions that are natural to the user and to detect signs of misconduct or fraud at an early stage.

[1430] "User's messaging history" means a record of messages that a User has sent or received in the past.

[1431] "Reply habits" are the tendencies or patterns of how a user responds to messages.

[1432] "Chat flow" refers to the progress and changes in the message exchange between a user and the other person.

[1433] "Analysis results" refers to the information and data obtained after analyzing message exchanges.

[1434] "Suggested replies" refers to suggestions or candidates for a user to reply to a message.

[1435] A "message pattern" refers to a recurring format or structure in a message exchange.

[1436] "Topic changes" refers to how the topic changes during a message exchange.

[1437] "Changes in emotion" refers to how emotions change during message exchange.

[1438] "Fraud" refers to committing fraud or malicious acts against users.

[1439] "Fraud indicators" are signs or patterns in messaging that indicate the possibility of fraudulent activity.

[1440] In order to implement this invention, it is necessary to build a system in which the elements of the server, terminals, and users work together.

[1441] First, the server collects the user's messaging history and reads the user's own reply habits. This requires software to store and analyze the messaging history using a database management system (DBMS). Specifically, it uses a programming language such as Python or R to analyze message patterns and changes in sentiment using natural language processing (NLP) libraries (e.g., NLTK, spaCy).

[1442] Next, the server analyzes the flow of conversation between the user and the other party. To do this, a generative AI model is used to detect changes in the topic of messages and changes in sentiment. Advanced natural language processing models such as GPT-3 and BERT are suitable as generative AI models. These models are used to analyze message exchanges in real time and understand the flow of the conversation.

[1443] Furthermore, the server generates reply suggestions that are appropriate for the user based on the analysis results. The generated reply suggestions are customized based on the user's past reply patterns. In order to generate multiple reply suggestions and present them to the user, a user interface (UI) needs to be designed. The UI is often implemented as a smartphone app or a web application.

[1444] The server also analyzes message patterns, topic changes, and sentiment changes to detect signs of misconduct or fraud. To do this, we use machine learning algorithms (e.g., Naive Bayes, SVM) to build a system that detects anomalous message patterns. If an anomaly is detected, we also need the ability to alert users.

[1445] As a concrete example, consider the following message exchange:

[1446] "Hello, how are you?"

[1447] "Yeah, I'm fine. How about you?"

[1448] "I'm fine too. By the way, how are you doing lately?"

[1449] "Very well. By the way, can you give me your bank details?"

[1450] In this case, a sudden change in emotion is detected and deemed to be an indication of fraudulent activity.

[1451] Example of a prompt to be input to a generative AI model:

[1452] It analyzes messaging between users and their contacts to provide a way to understand the flow of conversations, specifically by analyzing message patterns, topic changes, and sentiment changes to detect signs of fraud or misconduct.

[1453] In this way, it is possible to realize a system that can generate reply suggestions that are natural to the user and can also detect signs of misconduct or fraud at an early stage.

[1454] The flow of the specific process in the application example 2 will be described with reference to FIG.

[1455] Step 1:

[1456] The server collects the messaging history of users. As input, it takes the message data that the user has sent and received in the past and stores it in a database. Specifically, it uses a database management system (DBMS) to store the messaging history and build a dataset for analysis. As output, it obtains the messaging history data in a format that can be analyzed.

[1457] Step 2:

[1458] The server reads the user's own reply habits from the user's messaging history. As input, it uses the messaging history data collected in step 1. Specifically, it uses a natural language processing (NLP) library (e.g., NLTK, spaCy) to analyze message patterns and reply trends. As output, it obtains data about the user's reply habits.

[1459] Step 3:

[1460] The server analyzes the flow of conversation between the user and the other party. As input, it receives message data sent and received in real time. Specifically, it uses generative AI models (e.g. GPT-3, BERT) to detect changes in the topic and sentiment of messages. As output, it obtains the analysis results on the flow of conversation.

[1461] Step 4:

[1462] The server generates reply suggestions that are appropriate for the user based on the analysis results. As input, it uses the data obtained in steps 2 and 3. Specifically, it uses a generative AI model to generate reply suggestions based on the user's past reply patterns. As output, it obtains multiple reply suggestions.

[1463] Step 5:

[1464] The server presents the generated reply suggestions to the user. As input, it uses the reply suggestions generated in step 4. Specifically, it displays the reply suggestions in a smartphone app or web application through a user interface (UI). As output, it presents the user with reply suggestions that they can select from.

[1465] Step 6:

[1466] The server analyzes message patterns, topic changes, and sentiment changes to detect signs of misconduct or fraud. As input, it uses real-time messaging data. Specifically, it uses machine learning algorithms (e.g., Naive Bayes, SVM) to detect anomalous message patterns. As output, it provides warnings about signs of misconduct or fraud.

[1467] Step 7:

[1468] The server alerts the user if any indication of misconduct or fraud is detected. As input, it uses the alert data obtained in step 6. Specifically, it displays a warning message through a user interface in a smartphone app or web application. As output, a warning is presented to the user.

[1469] Example 3

[1470] Next, a third embodiment of the third embodiment will be described. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal".

[1471] Conventional messaging systems have the problem that it takes time for users to think about what to reply, making efficient communication difficult. It is also difficult to generate automatic reply suggestions that take into account the user's reply habits and the flow of the conversation. This increases the time users spend thinking about how to reply, which reduces the efficiency of communication.

[1472] The specific processing by the specific processing unit 290 of the data processing device 12 in the third embodiment is realized by the following means.

[1473] In this invention, the server includes a means for reading a user's own reply habits from the user's messaging history, a means for analyzing the flow of conversation between the user and the other party, a means for generating reply suggestions that are suitable for the user based on the analysis results, a means for displaying the generated reply suggestions on the user's device so that the user can select one, a means for generating reply suggestions using a generative AI model, and a means for inputting prompt sentences into the generative AI model. This allows the user to significantly reduce the time spent struggling with replying, enabling efficient communication.

[1474] "User's messaging history" means a record of messages that a User has sent or received in the past.

[1475] "Reply habits" are reply patterns or phrases that a user repeatedly uses in particular situations or with particular people.

[1476] "Chat flow" refers to the continuity and context of messages exchanged between a user and the other person.

[1477] "Analysis Results" refers to data or information obtained after analyzing a User's messaging history and conversation flow.

[1478] "Suggested replies" are suggested messages that users can use as a reference when replying.

[1479] A "generative AI model" is a model that uses artificial intelligence to analyze data and generate new data and information.

[1480] A "prompt" is an instruction or question that is input into a generative AI model.

[1481] "User Device" refers to any electronic device used by a User, such as a computer, smartphone, or tablet.

[1482] The present invention is a system that analyzes a user's messaging history and automatically generates reply suggestions based on the user's reply habits and conversation flow. A specific embodiment of this system will be described below.

[1483] Hardware and Software Used

[1484] The server includes hardware and software for collecting and analyzing the messaging history of users. Specifically, the server may have the following configuration:

[1485] Hardware: High-performance server computer

[1486] Software: Analytical software using natural language processing (NLP) techniques, generative AI models (e.g. Google's BERT, OpenAI's GPT-3)

[1487] The terminal is a device that allows the user to check and select reply suggestions. Specifically, the following configuration is possible.

[1488] Hardware: Smartphones, tablets, computers

[1489] Software: Messaging application

[1490] Data processing and calculation

[1491] The server collects the user's messaging history and analyzes it using natural language processing technology. As a result of the analysis, frequently used phrases and reply patterns are extracted from the user. Based on this, reply suggestions are generated using a generative AI model.

[1492] The specific processing flow is as follows:

[1493] 1. Message Data Collection:

[1494] The server retrieves past message data through the API of the user's messaging application, and stores the retrieved data in a database.

[1495] 2. Message data analysis:

[1496] The server inputs prompts into the generative AI model and analyzes the user's message data. As a result of the analysis, frequently used phrases and reply patterns are extracted.

[1497] 3. Generate reply suggestions:

[1498] The server generates reply suggestions based on the analysis results, using a generative AI model to generate reply suggestions that are suited to the user's reply habits and the flow of the conversation.

[1499] 4. Proposal of a response:

[1500] The device presents the generated reply suggestions to the user by displaying the reply suggestions below the message input field so that the user can select one.

[1501] Examples and prompts

[1502] If a user frequently uses the phrase "thank you" and has a pattern of expressing gratitude after being asked a question, the following prompt sentence can be input into the generative AI model:

[1503] Example prompt:

[1504] User's previous messages:

[1505] 1. "Thank you for answering my question."

[1506] 2. "Thanks, about that..."

[1507] 3. "Thanks, next time..."

[1508] The other person asks: "What do you think about this?"

[1509] Generate reply: "Thanks, about that..."

[1510] The server inputs this prompt sentence into a generative AI model to generate a reply proposal. The generated reply proposal is sent to the terminal and displayed at the bottom of the message input field. When the user taps on a reply proposal, the content is inserted into the message input field. This allows the user to significantly reduce the time spent thinking about a reply, and enables efficient communication. The flow of the specific process in the third embodiment will be described with reference to FIG. 15.

[1511] Step 1:

[1512] The server collects the user's messaging history.

[1513] Input: Historical messaging data obtained through the API of the user's messaging application.

[1514] Data processing: The acquired message data is stored in a database.

[1515] Output: The saved message data.

[1516] What happens: The server calls the messaging application's API to retrieve the user's past messages and save them in a database.

[1517] Step 2:

[1518] The server analyzes the collected message data.

[1519] Input: The saved message data.

[1520] Data calculation: Input the prompt sentence into the generative AI model and analyze the user's message data.

[1521] Output: A pattern of frequent phrases or replies used by users.

[1522] How it works: The server inputs prompts into the generative AI model and analyzes the user's message data. As a result of the analysis, it extracts frequently used phrases and reply patterns from the user.

[1523] Step 3:

[1524] The server generates a proposed reply based on the analysis.

[1525] Input: A pattern of frequently used phrases or replies by users.

[1526] Data calculation: Using a generative AI model, we generate reply suggestions that are suited to the user's reply habits and the flow of the conversation.

[1527] Output: The generated suggested replies.

[1528] Specific operation: Based on the analysis results, the server inputs the prompt sentence into the generative AI model to generate reply suggestions, which are then stored in a database.

[1529] Step 4:

[1530] The server transmits the generated reply proposal to the user's device.

[1531] Input: The generated suggested reply.

[1532] Data processing: Converting the proposed reply into a data format for sending to the user's device.

[1533] Output: The suggested reply sent to the user's device.

[1534] Specific behavior: The server converts the generated reply proposal into an appropriate data format and sends it to the user's device.

[1535] Step 5:

[1536] The terminal presents the generated reply suggestions to the user.

[1537] Input: The suggested reply sent to the user's device.

[1538] Data processing: Display reply suggestions at the bottom of the message input field.

[1539] Output: Suggested replies displayed in the message input field.

[1540] Specific operation: The device displays reply suggestions received from the server at the bottom of the message input field. When the user taps a reply suggestion, the reply is inserted into the message input field.

[1541] (Application example 3)

[1542] Next, a description will be given of Application Example 3 of Form Example 3. In the following description, the data processing device 12 will be referred to as a "server", and the headset type terminal 314 will be referred to as a "terminal".

[1543] In virtual stores, responding to customer inquiries quickly and appropriately is important to improve customer satisfaction. However, it takes time and effort for store clerks to respond to each customer individually, making efficient operation difficult. Furthermore, if the store clerk's responses are inconsistent, it may undermine customer trust. To solve these problems, a system is needed that can automatically generate replies from store clerks and provide efficient and consistent customer service.

[1544] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 3 is realized by the following means.

[1545] In this invention, the server includes means for reading a user's own reply habits from the user's messaging history, means for analyzing the flow of conversation between the user and the other party, means for generating reply suggestions that are characteristic of the user based on the analysis results, means for presenting the generated reply suggestions to the user, means for automatically generating replies for store clerks to streamline customer service in the virtual store, and means for displaying the generated reply suggestions on a smartphone application so that the user can select one. This makes it possible to automatically generate quick and appropriate replies to customer inquiries, reduce the burden on store clerks, and provide consistent customer service.

[1546] "User's messaging history" means a record of messages that a User has sent or received in the past.

[1547] "Reply habits" are patterns such as words and phrases that a user frequently uses when replying to messages, as well as the timing and order of replies.

[1548] "Chat flow" refers to the progress and context of a series of messages exchanged between a user and another person.

[1549] "Analysis Results" refers to information or data obtained after analyzing a User's messaging history and conversation flow.

[1550] A "suggested reply" is an automatically generated reply as a candidate for a message that the user may send.

[1551] A "virtual store" is a virtual store operated on the Internet where customers can purchase goods and services online.

[1552] "Customer service" refers to the task of responding to inquiries and requests from customers in the virtual store.

[1553] A "smartphone application" is a software program that runs on a smartphone.

[1554] "Automatic generation" means that the system generates suggested responses based on user input and past data, without manual intervention.

[1555] "Consistency" means always maintaining the same quality and standards in customer service.

[1556] To put this invention into practice, it is necessary to build a system that reads reply habits from a user's messaging history, analyzes the flow of the conversation, and generates reply suggestions that suit the user based on the analysis results. This system has the function of automatically generating replies for store clerks and displaying the generated reply suggestions on a smartphone application to streamline customer service in a virtual store.

[1557] Hardware and software used

[1558] Hardware: Smartphone

[1559] Software: OpenAI API, Python

[1560] Data processing and calculation

[1561] 1. Data Collection: The server collects the user's messaging history and identifies the user's reply habits, including the words and phrases the user uses most frequently, and the timing and sequence of replies.

[1562] 2. Conversation flow analysis: The server analyzes the conversation flow between the user and the other party to understand the context and progress.

[1563] 3. Reply generation: Based on the analysis results, the server uses a generative AI model to generate reply suggestions that are appropriate for the user. At this time, the prompt sentence is input to the generative AI model.

[1564] 4. Presentation of reply suggestions: The server displays the generated reply suggestions on the smartphone application, allowing the user to select one.

[1565] Examples

[1566] For example, if a customer asks "Do you have this item in stock?" in a virtual store, the server will generate a suggested reply such as "Thank you. We do have this item in stock..." based on the store clerk's past reply history and patterns.

[1567] Examples of prompt statements

[1568] User response history: Thank you. This item is in stock...

[1569] User response pattern: Question followed by a thank you

[1570] Generate the following reply:

[1571] In this way, the server can automatically generate prompt and appropriate replies to customer inquiries, reducing the burden on store clerks and achieving consistent customer service.

[1572] The flow of the specific process in the application example 3 will be described with reference to FIG.

[1573] Step 1:

[1574] The server collects the user's messaging history, i.e., records of messages the user has sent and received in the past, from a database. This input data includes the words and phrases the user frequently uses, the timing and sequence of replies, etc. As an output, a dataset is obtained that can be used to identify the user's reply habits.

[1575] Step 2:

[1576] The server analyzes the conversation flow between the user and the other party by analyzing the collected messaging history to understand the context and progress. The input data includes the sequence of messages exchanged between the user and the other party. The output is the analysis result of the conversation flow.

[1577] Step 3:

[1578] Based on the analysis results, the server uses a generative AI model to generate reply suggestions that are appropriate for the user. Specifically, the prompt sentence is input into the generative AI model, which generates appropriate reply suggestions. This input data includes the user's reply habits and the analysis results of the conversation flow. The output is a reply suggestion that is appropriate for the user.

[1579] Step 4:

[1580] The server displays the generated reply suggestions on the smartphone application. Specifically, the server transmits the generated reply suggestions to the user's device and displays them below the message input field of the application. The input data includes the generated reply suggestions. As an output, the smartphone screen displays the reply suggestions that the user can select.

[1581] Step 5:

[1582] The user can select a suggested reply or arrange it as they like. Specifically, the user can either insert the reply into the message input field by tapping on it, or edit the message based on the reply. The input data includes the reply selected by the user and the edited message. The output is the final reply message.

[1583] In addition, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[1584] "Example 1"

[1585] As one embodiment of the present invention, a system is provided that includes a means for reading a user's own reply habits from the user's messaging history, a means for analyzing the flow of conversation between the user and another person, a means for generating reply suggestions that are typical of the user based on the analysis results, a means for presenting the generated reply suggestions to the user, and an emotion engine for recognizing the user's emotions.

[1586] "Example 2"

[1587] The emotion engine extracts changes in a user's emotions from the user's messaging history, specifically from the context and wording of messages, and the use of emojis and stamps.

[1588] "Example 3"

[1589] The emotion engine generates reply suggestions that take into account the changes in the extracted emotions. Specifically, if the user is expressing emotions such as happiness or sadness, the engine generates reply suggestions that match that emotion. For example, if the user is expressing happiness, the engine generates reply suggestions that also express happiness.

[1590] "Example 4"

[1591] The emotion engine extracts changes in a user's emotions in real time and generates reply suggestions in real time that take those changes into account. Specifically, every time a user sends a message, the emotion is extracted from the message and a reply suggestion based on that emotion is instantly generated.

[1592] The process flow of each embodiment will be described below.

[1593] "Example 1"

[1594] Step 1: Read the user's own reply habits from their messaging history.

[1595] Step 2: Analyze the flow of conversation between the user and the other party.

[1596] Step 3: Generate appropriate replies based on the analysis results.

[1597] Step 4: Present the generated reply suggestions to the user.

[1598] Step 5: Activate the emotion engine to recognize the user's emotions.

[1599] "Example 2"

[1600] Step 1: Extract user sentiment changes from users' messaging history.

[1601] Step 2: Generate reply suggestions that take into account the extracted emotion changes.

[1602] "Example 3"

[1603] Step 1: Extract user sentiment changes from users' messaging history.

[1604] Step 2: Generate reply suggestions that take into account the extracted emotion changes.

[1605] "Example 4"

[1606] Step 1: Extract real-time changes in user sentiment from users’ messaging history.

[1607] Step 2: Generate reply suggestions in real time that take into account the extracted emotion changes.

[1608] Example 1

[1609] Next, a description will be given of Example 1 of the embodiment 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal".

[1610] In conventional messaging systems, it was difficult for users to respond consistently, and it was especially difficult to generate appropriate replies when they were busy or emotional. In addition, there was no system that provided reply suggestions that took the user's emotions into account, so it was not possible to respond in accordance with the user's intentions and emotions. This could lead to a decline in the quality of communication and the occurrence of misunderstandings and problems.

[1611] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[1612] In this invention, the server includes a means for reading a user's own reply habits from the user's messaging history, a means for analyzing the flow of conversation between the user and the other party, a means for generating reply suggestions that are appropriate for the user based on the analysis results, a means for presenting the generated reply suggestions to the user, and an emotion engine for recognizing the user's emotions. This allows the user to reply consistently, and further makes it possible to provide appropriate reply suggestions that take the user's emotions into consideration.

[1613] "User messaging history" refers to data such as the content of messages a user has sent in the past, the time they were sent, and the frequency of those messages.

[1614] "Reply habits" refers to characteristics such as the phrases and wording that a user frequently uses when replying to messages, as well as the timing and frequency of replies.

[1615] "Conversation flow" refers to information such as the continuity, context, and tone of messages exchanged between a user and the other person.

[1616] "Means for analyzing" refers to methods or devices that use natural language processing techniques and / or generative AI models to understand the content and context of a message and extract certain patterns or characteristics.

[1617] "Suggested replies" refers to sentences that are generated as candidates for messages that the user may send.

[1618] An "emotion engine" refers to technology or devices that analyze emotions contained in a user's messages and suggested replies and recognize the user's emotional state.

[1619] A "generative AI model" is a model for generating text using artificial intelligence technology, such as one that uses natural language processing technology.

[1620] "Presenting means" refers to a method or apparatus for displaying the generated suggested replies to a user.

[1621] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[1622] This invention is a system that analyzes a user's messaging history, extracts the user's reply habits, and further analyzes the flow of a conversation between the user and the other party to generate reply suggestions that are appropriate for the user and present them to the user. A specific embodiment of this system will be described below.

[1623] Collecting messaging history

[1624] The server retrieves message data from the user's messaging application (e.g., a general messaging application). The retrieved data is stored in a database. This data includes the content of messages the user has sent in the past, the time they were sent, and how frequently they were sent.

[1625] Analyzing messaging history

[1626] The server analyzes the stored message data. It uses natural language processing (NLP) techniques to extract frequently used phrases and vocabulary used by users. For example, it uses Python's NLTK library for text analysis. This reveals the user's reply habits.

[1627] Talk flow analysis

[1628] The server uses a generative AI model (e.g., a general generative AI model) to analyze the flow of the conversation between the user and the other party, thereby obtaining basic data for understanding the context and tone of the conversation and generating appropriate reply suggestions.

[1629] Generate user-friendly reply suggestions

[1630] The server generates reply suggestions that are appropriate for the user based on the analysis results. Using a generative AI model, it creates reply suggestions that take into account the user's reply habits and the flow of the conversation. For example, if a user often uses the phrase "thank you," a reply suggestion that includes that phrase will be generated.

[1631] Suggested replies

[1632] The server sends the reply proposal it has generated to the user's device. The user can review the proposed reply and make corrections or approvals as necessary. If the user approves the reply proposal, the reply is sent to the other party.

[1633] Emotion recognition by emotion engine

[1634] The server uses an emotion engine to recognize the user's emotions. The emotion engine analyzes the emotions contained in the user's messages and reply suggestions to understand the user's emotional state, which provides feedback to generate more appropriate reply suggestions.

[1635] Examples

[1636] For example, if a user frequently uses the phrase "Otsukaresama desu" (Thank you for your hard work), the server will extract this phrase as the user's reply habit. If the user receives a message saying "How was the meeting today?", the server will generate a reply suggestion of "Otsukaresama desu. The meeting went well" and present it to the user.

[1637] Examples of prompt statements

[1638] "Analyze the user's messaging history and extract the user's reply habits. Then, analyze the conversation flow between the user and the other party and generate reply suggestions that are appropriate for the user. Finally, present the generated reply suggestions to the user and provide feedback using an emotion engine that recognizes the user's emotions."

[1639] The system allows users to respond more naturally and consistently, and also provides appropriate reply suggestions that take the user's feelings into account.

[1640] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1641] Step 1:

[1642] The server retrieves message data from the user's messaging application. As input, it receives data such as the content of the user's past messages, the time of sending, and the frequency of sending. It then stores this data in a database. Specifically, it uses an API to collect data from the messaging application and stores it in a database.

[1643] Step 2:

[1644] The server analyzes the stored message data. As input, it uses the message data stored in the database. It uses natural language processing (NLP) techniques to extract phrases and wording frequently used by users. As output, it obtains data showing the user's reply habits. Specifically, it uses Python's NLTK library to perform text analysis.

[1645] Step 3:

[1646] The server uses a generative AI model to analyze the flow of the conversation between the user and the other party. As input, it uses message data between the user and the other party. It uses a generative AI model (e.g., a general generative AI model) to understand the context and tone of the conversation. As output, it obtains data showing the flow of the conversation. Specifically, it calls the generative AI model via an API and obtains the analysis results.

[1647] Step 4:

[1648] The server generates reply suggestions that are appropriate for the user based on the analysis results. As input, it uses data showing the user's reply habits and conversation flow. It uses a generative AI model to create reply suggestions that take into account the user's reply habits and conversation flow. As output, it obtains multiple reply suggestions. In concrete terms, it inputs a prompt sentence into the generative AI model and generates reply suggestions.

[1649] Step 5:

[1650] The server sends the generated reply suggestions to the user's terminal, using the generated reply suggestions as input. The user reviews the proposed reply suggestions and modifies or approves them if necessary. As output, the reply approved by the user is obtained. The specific operations include displaying the reply suggestions through a user interface and receiving the user's selection.

[1651] Step 6:

[1652] The server uses the emotion engine to recognize the user's emotions. It uses the user's message and suggested replies as input. The emotion engine analyzes the emotions contained in the user's message and suggested replies to understand the user's emotional state. As output, it obtains data indicating the user's emotional state. Specifically, it calls the emotion engine via an API and obtains the emotion analysis results.

[1653] (Application example 1)

[1654] Next, a description will be given of Application Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as a "server", and the headset type terminal 314 will be referred to as a "terminal".

[1655] Conventional messaging systems were unable to generate personalized reply suggestions that took into account the user's reply habits and emotions. They also lacked the ability to recommend appropriate content based on the user's interests and emotions. This made it difficult for users to quickly respond in a way that suited them, and it took a long time to find content that matched their interests.

[1656] The identification process by the identification pr...

Claims

1. A means to read a user's own reply habits from their messaging history; and A means for analyzing a flow of a conversation between the user and a person; means for generating a prompt sentence for instructing the user to generate a reply plan that is suitable for the user based on the analysis results of the reply habits and the talk flow; A generative AI model that receives the prompt sentence and generates a reply proposal that is likely to be received by the user; A means for displaying the reply suggestions generated by the generative AI model on the user's device and allowing the user to select one of the reply suggestions; A system including:

2. 2. The system of claim 1, wherein the generative AI model generates a plurality of suggested replies and displays the generated plurality of suggested replies on the user's device.

3. The system of claim 1 , wherein the suggested replies displayed on the user's device can be used as is or can be adapted by the user.

Images