System

Abstract

In a modern official account, a user can only passively communicate with a character or a famous person. For this reason, the interest in the official account is reduced, and as a result, the opening frequency of the account is reduced, and the account is often muted.SOLUTION: A system comprising: means for receiving a message from a user; means for preprocessing the received message; means for generating a response based on the preprocessed message using a generative AI model; means for applying a character or personality style to the generated response; and means for transmitting the response to the user.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 an instruction sentence related to a description of 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] Japanese Patent Publication No. 2022-180282 Summary of the Invention [Problem to be solved by the invention]

[0004] With today's official accounts, users can only passively communicate with characters or famous people. This has led to a decline in interest in official accounts, resulting in accounts being opened less frequently and sometimes muted. Furthermore, there is the issue of it being difficult for official account owners to effectively convey the information they want to convey to users. [Means for solving the problem]

[0005] To solve the above-mentioned problems, the present invention provides the following means. Specifically, a system including a means for receiving messages from users, a means for preprocessing the received messages, a means for generating responses using a generative AI model based on the preprocessed messages, a means for applying the style of a character or famous person to the generated responses, and a means for sending the responses to users allows users to proactively communicate with official accounts. This system allows the generative AI model to mimic the speaking style and writing style of the character or famous person, and has a function for adjusting the response content to the specific style of the character or famous person after generating a response, thereby providing a natural and engaging dialogue experience for users. The system also includes a means for tokenizing messages and removing unnecessary spaces and special characters during preprocessing, and a means for saving a dialogue history with the user and referencing the past dialogue when generating the next response, thereby achieving greater dialogue continuity and depth. Furthermore, the system includes a means for collecting user feedback after generating a response and updating the AI ​​model, aiming for continuous system improvement and increased user satisfaction.

[0006] "User" refers to an individual who uses the system to interact with the official account of a character or famous person.

[0007] "Message" refers to text data sent by a user to an official account.

[0008] "Server" refers to a computer system that receives a user's message, processes it, generates a response, and ultimately sends it to the user.

[0009] "Preprocessing" refers to the process of parsing incoming messages and performing tasks such as tokenizing, normalizing, and removing unnecessary spaces and special characters.

[0010] "Generative AI model" refers to an artificial intelligence model for generating appropriate responses based on a user's message.

[0011] "Response" refers to the text data generated by the generative AI model and sent to the user.

[0012] "Stylistically adapting" refers to the process of applying a particular writing style or speaking style of a character or famous figure to the generated responses.

[0013] "Tokenization" refers to parsing a message and converting it into a form suitable for natural language processing.

[0014] "Feedback" refers to the evaluations and opinions that users give about responses, and is information that can be used to improve AI models.

[0015] "Dialogue history" refers to data that records past message exchanges between a user and an official account.

[0016] "Normalization" refers to processing text data according to certain rules and converting it into a consistent format.

[0017] A "natural conversational experience" refers to an interaction that makes users feel as if they are having a direct conversation with a character or famous person. [Brief explanation of the drawings]

[0018] [Figure 1] 1 is a conceptual diagram showing an example of the configuration of a data processing system according to a first embodiment. [Figure 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. [Figure 3] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a second embodiment. [Figure 4] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and smart glasses according to a second embodiment. [Figure 5] FIG. 10 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. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a fourth embodiment. [Figure 8] FIG. 10 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. 3 is a sequence diagram showing a processing flow of the data processing system according to the first embodiment. [Figure 12] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 1. [Figure 13] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system according to the second embodiment when an emotion engine is combined. [Figure 14] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 2 when an emotion engine is combined. DETAILED DESCRIPTION OF THE INVENTION

[0019] An example of an embodiment of a system according to the technology of the present disclosure will be described below with reference to the accompanying drawings.

[0020] First, the terms used in the following description will be explained.

[0021] In the following embodiments, a coded processor (hereinafter simply referred to as a "processor") may be a single arithmetic device or a combination of multiple arithmetic devices. Furthermore, a processor may be a single 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), and an APU (Accelerated Processing Unit).

[0022] In the following embodiments, a coded RAM (Random Access Memory) is a memory in which information is temporarily stored and is used as a working memory by a processor.

[0023] 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.

[0024] In the following embodiments, a communication I / F (Interface) with a symbol is an interface including a communication processor, an antenna, etc. The communication I / F controls communication between multiple computers. Examples of communication standards applied to the communication I / F include wireless communication standards including 5G (5th Generation Mobile Communication System), Wi-Fi (registered trademark), Bluetooth (registered trademark), etc.

[0025] 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. Furthermore, in this specification, the same concept as "A and / or B" is also applied when three or more things are expressed connected by "and / or."

[0026] [First embodiment]

[0027] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0028] 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.

[0029] 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 WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0030] 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.

[0031] The reception device 38 includes a touch panel 38A, a microphone 38B, and the like, and receives user input. The touch panel 38A detects contact with an indicator (for example, a pen or a finger) to receive user input by the touch of the indicator. 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.

[0032] 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 (for example, audio and / or text). The display 40A displays visible information such as text and images in accordance with instructions from the processor 46. The speaker 40B outputs audio in accordance with instructions from the processor 46. The camera 42 is a compact digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0033] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 control the exchange of various information between the processor 46 and the processor 28 via the network 54.

[0034] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0035] 2, in the data processing device 12, a specific process is performed by the processor 28. A specific processing program 56 is stored in the storage 32. 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 process 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.

[0036] 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.

[0037] In the smart device 14, the processor 46 performs the reception output process. The storage 50 stores a reception output program 60. The reception output program 60 is used in conjunction with the specific processing program 56 by the data processing system 10. 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.

[0038] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. 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."

[0039] Regarding an embodiment of the present invention, the specific operation and processing flow of a dialogue system between official accounts and users using a generative AI model will be described.

[0040] overview

[0041] The present invention is a system that allows users to communicate naturally and proactively with the official accounts of characters and famous people. The system receives messages from users, generates responses using a generative AI model, and sends them in a style that matches the style of the specific character or famous person.

[0042] Program processing overview

[0043] 1. Receive a user's message

[0044] User: The user types and sends a message to the official account.

[0045] Terminal: Sends the entered message to the server.

[0046] 2. Message Reception and Preprocessing

[0047] Server: Receives messages sent from the terminal.

[0048] Server: Pre-processes the received message by tokenizing it and removing unnecessary spaces and special characters.

[0049] 3. Response Generation

[0050] Server: Generates a response to the user's message using a generative AI model based on the preprocessed message.

[0051] Server: Apply the style of a character or celebrity to the generated response, adjusting the writing style and speaking style.

[0052] 4. Sending the Response

[0053] Server: Sends the final generated response to the user's terminal.

[0054] Terminal: The user receives the response and the dialogue continues.

[0055] Specific examples

[0056] For example, consider the case where a user sends the following message:

[0057] User: "Tell me about your latest project."

[0058] When this message is sent, the following process occurs:

[0059] 1. Receive a user message

[0060] The server receives and stores this message.

[0061] 2. Message Preprocessing

[0062] The server parses and tokenizes the received message, removing unnecessary spaces and special characters.

[0063] 3. Response Generation

[0064] The server inputs the preprocessed messages into a generative AI model to generate a response.

[0065] For example, a generative AI model might generate a response like, "Of course, your recent projects have been very exciting. You're working on a new album."

[0066] 4. Applying Styles

[0067] The server then styles the generated response after a specific character or celebrity, adjusting it to, "Of course, my latest projects have been very exciting! I'm working on a new album."

[0068] 5. Sending the Response

[0069] The server sends this adjusted response to the user's terminal.

[0070] The terminal displays this response to the user.

[0071] This process allows users to continue natural and engaging interactions with characters and famous people.

[0072] The system also saves the user's interaction history and references it the next time it generates a response, providing a more consistent interaction. Furthermore, by collecting user feedback and periodically updating the generative AI model, the system's accuracy and user satisfaction can be improved.

[0073] The processing flow will be explained below.

[0074] Step 1:

[0075] User: Enter a message for the official account and tap the send button. For example, you could type, "Hello, tell me about your recent projects."

[0076] Step 2:

[0077] Terminal: Sends the message entered by the user to the server, where it is converted into an appropriate format (e.g., JSON).

[0078] Step 3:

[0079] Server: Receives the message sent from the device. For example, get the data user_message = "Hello, tell us about your recent projects."

[0080] Step 4:

[0081] Server: Stores the received message in a database. Message data is saved using an SQL INSERT query.

[0082] Step 5:

[0083] Server: Preprocesses the stored messages by tokenizing them and removing unnecessary spaces and special characters. For example, "Hello, tell me about your recent projects" is tokenized as "Hello, tell me about your recent projects".

[0084] Step 6:

[0085] Server: Feeds the pre-processed messages into a generative AI model, which is pre-trained to mimic the speaking and writing style of a specific character or famous figure.

[0086] Step 7:

[0087] Server: The generative AI model generates an appropriate response to the user's message. For example, the generative AI model generates a response like, "Of course, your recent projects are very exciting. I'm working on a new album."

[0088] Step 8:

[0089] Server: Apply the style of a character or famous person to the generated response. This style adjustment affects the writing style and speaking style. For example, "Of course, my latest projects have been very exciting! I'm working on a new album."

[0090] Step 9:

[0091] Server: Sends the final generated response to the user's terminal, as well as sending data according to the format.

[0092] Step 10:

[0093] Terminal: The user receives a response, for example, a message displayed on the screen saying, "Of course, my latest projects have been very exciting! I'm working on a new album."

[0094] Step 11:

[0095] User: Views the response received and types the next message. This process continues the conversation. For example, the user next types, "When is the album coming out?"

[0096] Step 12:

[0097] Server: Saves the dialogue history and references past dialogues when generating the next response, thereby maintaining a consistent conversation.

[0098] The above is the specific processing flow in the system.

[0099] Example 1

[0100] Next, a description will be given of 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."

[0101] In modern society, there is a growing demand for users to have natural-looking conversations with specific characters or famous people. However, existing dialogue systems have difficulty accurately imitating the speaking style and writing style of characters or famous people, which can lead to a poor user experience. Furthermore, there are technical challenges in maintaining consistency and naturalness in responses. Therefore, there is a need to develop a system that provides natural and engaging conversations based on the style of specific characters or famous people.

[0102] 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.

[0103] In this invention, the server includes means for receiving a message from a user, means for tokenizing the received message and preprocessing it to remove unnecessary spaces and special characters, means for generating a response using a generative AI model based on the preprocessed message, means for applying a style of a specific character or famous person to the generated response to adjust the writing style and speaking manner, and means for sending the adjusted response to the user, thereby enabling the user to have a natural and engaging conversation with the specific character or famous person.

[0104] "User" refers to a person who types and sends messages to interact with the system.

[0105] "Message" refers to text data that a user enters and sends to interact with the system.

[0106] "Tokenization" refers to the process of breaking down an incoming message into individual words or phrases.

[0107] "Preprocessing to remove unnecessary spaces and special characters" refers to the process of removing unnecessary spaces and specific symbols from the message to make it easier to analyze.

[0108] A "generative AI model" refers to an algorithm or software that generates natural language responses based on input text data.

[0109] "Response" refers to the text data of a reply generated by a generative AI model.

[0110] "Character or celebrity style" refers to the style or characteristics used to imitate the speech, writing style, or language of a particular character or celebrity.

[0111] "Tailoring" refers to the process of changing the writing style or speaking style to adapt the generated response to a particular style.

[0112] "Sending" refers to the act of sending the response generated by the server to the user's terminal as data.

[0113] This invention is a system that allows users to have natural conversations with the official accounts of specific characters or famous people. The system receives a message from the user, and a generative AI model generates a response based on the message. The generated response is adapted to the style of the specific character or famous person and returned to the user. To realize this process, the following steps are taken:

[0114] The server has a means for receiving messages from users. Users input and send messages to their official accounts. The terminal then sends these messages to the server. Specifically, the user input is passed to the server using a communication protocol such as an HTTP POST request.

[0115] The server parses the received message and performs preprocessing by tokenizing it and removing unnecessary spaces and special characters. The methods used here are, for example, Python's regular expression library or natural language processing library. For example, the message "Hello! Tell me about your recent projects." is split into tokens such as "Hello," "Recent," "Project," "About," "Tell me," and "Please."

[0116] Next, the server inputs the preprocessed message into a generative AI model to generate a response. For example, a large-scale language model such as GPT-3 is used as the generative AI model. The generative AI model receives the following input:

[0117] User: Hi! Tell us about your latest projects.

[0118] Character: Album Producer

[0119] Based on this input, the generative AI model generates an appropriate response, such as, "Of course, my recent projects have been very exciting. I'm working on a new album."

[0120] Apply the style of a particular character or celebrity to the generated response. The server does this by adjusting the writing style and word choice, for example, "Of course, my latest projects have been very exciting! I'm working on a new album."

[0121] The final adjusted response is sent to the user's device. The server packages the generated response in JSON format or similar and sends it to the user's device. The device receives it and displays it on the chat screen. Specifically, HTML and JavaScript are used to dynamically update the part that displays the message.

[0122] For example, if a user sends a message saying, "Tell me about your latest projects," the process will ultimately result in a response being returned to the user saying, "Of course, my latest projects are very exciting! I'm working on a new album."

[0123] The system also stores the user's interaction history and references it the next time it generates a response, allowing it to provide more consistent interactions. It also collects user feedback and periodically updates the generative AI model to improve system accuracy and user satisfaction.

[0124] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0125] Step 1:

[0126] The user enters and sends a message. Specifically, the user enters a text message in the chat box of the official account and clicks the send button. This operation causes the device to obtain the user's input and generate message data. The input is text data such as "Hello! Tell us about your recent projects." The output is that this text data is sent to the server via the device.

[0127] Step 2:

[0128] The device receives a message from the user and sends it to the server. The device packages the user's input as an HTTP POST request and sends it to the server's API endpoint. The input is the user's text message, for example, "Hello! Tell us about your latest project." The output is that this text message reaches the server.

[0129] Step 3:

[0130] The server receives messages sent from the terminal and stores them in a database. It then performs a process to insert the received messages into a table. Specifically, it stores the received data in the appropriate database table, along with metadata such as the message ID and timestamp. The input is the text data received from the terminal. The output is the state in which this data has been stored in the database.

[0131] Step 4:

[0132] The server performs preprocessing by tokenizing the received message and removing unnecessary spaces and special characters. The server uses regular expressions and natural language processing libraries to tokenize the message and remove unnecessary spaces and special characters. The input is the stored text message. The output is tokenized clean text data. For example, the message "Hello! Tell me about your recent projects." is split into "Hello", "Recent", "Project", "About", "Tell me", and "Please".

[0133] Step 5:

[0134] The server inputs the preprocessed message into a generative AI model to generate a response. A generative AI model (e.g., GPT-3) is used to generate natural-sounding responses based on the preprocessed message. The input is tokenized text data. The output is the generated response text. For example, suppose the input prompt is:

[0135] "User: Hi! Tell me about your latest project.

[0136] Character: Album Producer

[0137] The output from the generative AI model is, "Of course, your recent projects have been very exciting. You're working on a new album."

[0138] Step 6:

[0139] The server applies the style of a particular character or famous person to the generated response, adjusting the writing style and speaking style. The server adjusts the writing style and speaking style of the generated text based on the style of a particular character or famous person. For example, "Of course, my latest projects are very exciting! I'm working on a new album." The input is the generated response text. The output is the styled response text.

[0140] Step 7:

[0141] The server finally generates a response and sends it to the user's device. The server wraps the response text in JSON format or similar and sends it to the device as an HTTP response. The input is the response text after applying the style. The output is the response data sent to the device.

[0142] Step 8:

[0143] The device displays the response received from the server to the user. The device analyzes the received data and displays the response on the chat screen. Specifically, it updates the chat screen using HTML and JavaScript. The input is the response data received from the server. The output is the response text displayed on the user's chat screen. For example, the user's screen might display, "Of course, my recent projects have been very exciting! I'm working on a new album."

[0144] (Application example 1)

[0145] Next, a description will be given of Application 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."

[0146] The present invention relates to a system for improving engagement while maintaining consistency and naturalness of dialogue when users communicate naturally and proactively with the official accounts of characters and famous people. Conventional dialogue systems have difficulty faithfully reproducing the style of characters and famous people, and are unable to fully utilize dialogue history and feedback, limiting user satisfaction. For these reasons, there has been a demand for the development of a dialogue system that is more accurate and can be continuously improved.

[0147] 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.

[0148] In this invention, the server includes means for receiving messages from users, means for preprocessing the received messages, means for generating responses using a generative AI model based on the preprocessed messages, means for applying the style of a character or celebrity to the generated responses, means for sending the responses to the users, means for saving the response history and using it for the next dialogue, and means for collecting feedback from users and improving the generative AI model. This allows users to continue having natural and engaging dialogue with characters or celebrities, improving the consistency and accuracy of the dialogue. Furthermore, by utilizing user feedback, the generative AI model can be continuously improved to provide more accurate dialogue.

[0149] "Means for receiving a message from a user" refers to a device or software for receiving a text message entered by a user.

[0150] A "means for pre-processing received messages" is a device or software that tokenizes messages received from users and removes unnecessary spaces and special characters.

[0151] A "means for generating a response using a generative AI model based on a preprocessed message" is a device or software that uses a preprocessed message as input to generate a response using a generative AI model.

[0152] A "means for applying the style of a character or celebrity to generated responses" is a device or software that tailors the responses generated by a generative AI model to the speaking style or writing style of a particular character or celebrity.

[0153] The "means for sending a response to a user" is a device or software for sending the generated response to the user's terminal.

[0154] The "means for saving response history and using it in the next interaction" refers to a device or software that saves past interaction data between the user and the system and refers to that data during subsequent interactions.

[0155] A "means for collecting user feedback and improving a generative AI model" is a device or software that collects user-provided feedback and uses it as training data for a generative AI model to improve the model's performance.

[0156] "Template-based" refers to a method in which the corresponding response is tailored or generated based on a specific pattern or template.

[0157] A "prompt" is text entered to instruct a generative AI model, providing context for the response you want the model to generate.

[0158] In the embodiment for implementing the present invention, specific programs, processes, and examples of a system that realizes an interaction system between a user and the official account of a specific character or famous person will be described.

[0159] System Program

[0160] The system mainly consists of a client (user terminal) and a server. The client provides a UI for users to input messages, and the server analyzes and processes the received messages, then generates and sends a response. Specifically, it performs the following processes:

[0161] 1. User receives message

[0162] The user enters and sends a message to the official account.

[0163] The client (smartphone app) sends this message to the server.

[0164] 2. Message Preprocessing

[0165] The server tokenizes the received message, removing unnecessary spaces and special characters.

[0166] 3. Generating a Response

[0167] The server generates a response based on the preprocessed message using a generative AI model (e.g., OpenAI's GPT-4).

[0168] 4. Applying Styles

[0169] The generated responses are styled to suit specific characters or famous people, adjusting the writing style and speaking style, using a template-based approach.

[0170] 5. Sending the Response

[0171] The final generated response is sent to the user's terminal to continue the dialogue.

[0172] 6. Saving conversation history

[0173] The server stores the interaction history and references it the next time the interaction occurs, and also collects user feedback to continuously improve the generative AI model.

[0174] Hardware and software used

[0175] Hardware:

[0176] Server: Virtual server on Amazon Web Services (AWS)

[0177] Client: iOS or Android device (smartphone)

[0178] software:

[0179] Database: Uses Amazon RDS to store interaction history and feedback

[0180] Generative AI model: OpenAI's GPT-4

[0181] Preprocessing: A natural language processing library such as Python's NLTK or spaCy

[0182] Client app: Build the UI using Flutter

[0183] Examples of specific examples and prompts

[0184] For example, if a user sends the message "Tell me about your recent projects," the system generates a response as follows:

[0185] 1. Receive and preprocess the user's message:

[0186] The server receives the message "Tell us about your recent projects" and tokenizes and cleans it.

[0187] 2. Response generation using generative AI models:

[0188] The server generates a response using the following prompt sentence:

[0189] Prompt: "Respond in official character style: Tell me about your latest project."

[0190] The generative AI model (GPT-4) generates the response, "Of course, your recent projects have been very exciting. I'm working on a new album."

[0191] 3. Applying styles:

[0192] Tailor this response to the speech style of a particular character.

[0193] For example, "Of course, my latest projects have been very exciting! I'm working on a new album."

[0194] 4. Sending the response:

[0195] The server sends the tailored response to the user's device, which the client app displays.

[0196] This system not only allows users to enjoy natural and consistent interactions with characters and famous people, but it is also expected that the system itself will be continuously improved based on user feedback.

[0197] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0198] Step 1:

[0199] A user sends a message to the official account.

[0200] Input: The message the user types into the text box (e.g., "Tell me about your recent projects")

[0201] How it works: The user enters a message and presses the send button. The client (smartphone app) sends this message to the server.

[0202] Output: Message sent to the server

[0203] Step 2:

[0204] Preprocessing of messages received by the server.

[0205] Input: The message sent to the server

[0206] How it works: The server tokenizes the received message, removing unnecessary spaces and special characters, for example using Python's NLTK library or spaCy.

[0207] Output: Tokenized and cleaned message

[0208] Step 3:

[0209] The server generates a response based on the preprocessed message using a generative AI model.

[0210] Input: Preprocessed message

[0211] Operation: The server takes the preprocessed message as input to the generative AI model (GPT-4), generates a prompt sentence, and sends it to the generative AI model.

[0212] Sample prompt: "Respond in the style of your official character: Tell me about your latest project."

[0213] Output: The response generated by the generative AI model (e.g., "Of course, your recent projects have been very exciting. You're working on a new album.")

[0214] Step 4:

[0215] The server applies the style of a character or celebrity to the generated response.

[0216] Input: The response generated by the generative AI model

[0217] How it works: The server adapts the generated responses to suit the style of a particular character or celebrity, adjusting the writing style and speaking style. It uses a template-based approach to adapt the phrasing and writing style to suit the specific style.

[0218] Output: The styled response (e.g., "Of course, my latest projects have been very exciting! I'm working on a new album.")

[0219] Step 5:

[0220] The server finally sends the generated response to the user's terminal.

[0221] Input: The styled response

[0222] How it works: The server sends the styled response to the user's device, where the client app displays it.

[0223] Output: The response displayed on the user's terminal

[0224] Step 6:

[0225] The server saves the conversation history and uses it for the next conversation.

[0226] Input: The user's message and the generated response

[0227] How it works: The server saves the conversation history in a database such as Amazon RDS, and references it during subsequent conversations.

[0228] Output: Saved interaction history

[0229] Step 7:

[0230] The server collects user feedback and improves the generative AI model.

[0231] Input: User feedback

[0232] How it works: The server collects user feedback through the client app and uses it to improve the generative AI model, incorporating the feedback as training data for the model.

[0233] Output: An improved generative AI model

[0234] Furthermore, an emotion engine that estimates the user's emotion may be 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.

[0235] Regarding an embodiment of the present invention, the specific operation and processing flow of a dialogue system between official accounts and users using a generative AI model and an emotion engine will be described.

[0236] overview

[0237] The present invention is a system that allows users to communicate naturally and proactively with the official accounts of characters and celebrities, and also recognizes the user's emotions and provides appropriate responses. The system receives messages from users, generates responses using a generative AI model and an emotion engine, and sends them in a style that matches the specific character or celebrity's style. It also saves the user's emotion history and uses it for the next conversation.

[0238] Program processing overview

[0239] 1. Receive a user's message

[0240] User: The user types and sends a message to the official account. For example, they might type, "Hello, tell me about your recent projects."

[0241] Terminal: Sends the entered message to the server.

[0242] 2. Message Reception and Preprocessing

[0243] Server: Receives the message sent from the device. For example, get the data user_message = "Hello, tell us about your recent projects."

[0244] Server: Stores the received message in a database. Message data is saved using an SQL INSERT query.

[0245] Server: Preprocesses the stored messages by tokenizing them and removing unnecessary spaces and special characters. For example, "Hello, tell me about your recent projects" is tokenized as "Hello, tell me about your recent projects".

[0246] 3. Emotion recognition

[0247] Server: Inputs the preprocessed message into the emotion engine to recognize the user's emotion. The emotion engine extracts emotions such as positive, negative, and neutral from the message.

[0248] Server: Adjusts the input to the generative AI model based on the recognized emotion, for example, making the response more polite and supportive if the user has a negative emotion.

[0249] 4. Response Generation

[0250] Server: Using the trained data, the generative AI model generates a response to the user's message. For example, the generative AI model generates a response like, "Of course, your recent projects have been very exciting. I'm working on a new album."

[0251] Server: Style the generated response with a character or celebrity, transforming it into, "Of course, my latest projects have been very exciting! I'm working on a new album."

[0252] Server: Additionally, tailor the response based on the emotion engine's recognized emotion. For example, if the user is expressing negative emotion, provide additional encouragement such as, "Don't worry, the new album is going to be amazing!"

[0253] 5. Sending the Response

[0254] Server: Sends the final generated response to the user's terminal, as well as sending data according to the format.

[0255] Terminal: The user receives a response, for example, a message displayed on the screen saying, "Of course, your latest project is very exciting! Don't worry, the new album is going to be amazing!"

[0256] 6. Saving user emotion history

[0257] Server: The server stores the recognized user emotions and dialogue content in a database. This emotion history is referenced the next time a dialogue is generated.

[0258] 7. Refer to emotion history when generating the next response

[0259] Server: The next time the user sends a message, it references the user's past emotional history and provides a consistent dialogue.

[0260] Specific examples

[0261] For example, consider the case where a user sends the following message:

[0262] User: "I've been feeling a bit down lately. Could you tell me something to cheer me up?"

[0263] 1. Receive a user message

[0264] The server receives and stores this message.

[0265] 2. Message Preprocessing

[0266] The server parses and tokenizes the received message, removing unnecessary spaces and special characters.

[0267] 3. Emotion recognition

[0268] The server uses an emotion engine to recognize the user's emotion, in this case determining that the message indicates a negative emotion.

[0269] 4. Response Generation

[0270] The server then feeds the adjusted data into a generative AI model to generate a response, such as "Cheer up! I think your new project is very exciting."

[0271] The server applies the style of characters and famous people to the generated responses, and further adjusts them based on emotions recognized by the emotion engine: "Of course, cheer up! I think your new project will get you very excited! You have a great future ahead of you."

[0272] 5. Sending the Response

[0273] The server sends this adjusted response to the user's terminal.

[0274] The terminal displays this response to the user.

[0275] This process allows users to continue interacting with characters and celebrities in a natural, engaging and emotionally relevant way, and emotion history is stored and used to further personalize future responses.

[0276] The processing flow will be explained below.

[0277] Step 1:

[0278] User: Enter a message for the official account and tap the send button. For example, you could type, "Hello, tell me about your recent projects."

[0279] Step 2:

[0280] Terminal: Sends the message entered by the user to the server, where it is converted into an appropriate format (e.g., JSON).

[0281] Step 3:

[0282] Server: Receives the message sent from the device. For example, get the data user_message = "Hello, tell us about your recent projects."

[0283] Step 4:

[0284] Server: Stores the received message in a database. Message data is saved using an SQL INSERT query.

[0285] Step 5:

[0286] Server: Preprocesses the stored messages by tokenizing them and removing unnecessary spaces and special characters. For example, "Hello, tell me about your recent projects" is tokenized as "Hello, tell me about your recent projects".

[0287] Step 6:

[0288] Server: Inputs the preprocessed message into the emotion engine to recognize the user's emotion. The emotion engine extracts emotions such as positive, negative, and neutral from the message.

[0289] Step 7:

[0290] Server: The emotion engine receives the results of recognizing the user's emotion and stores them in the database. For example, if the emotion is recognized as "positive," it will be saved in the database.

[0291] Step 8:

[0292] Server: Adjusts the input to the generative AI model based on the recognized emotion. For example, if the user has a negative emotion, the response will be more polite and supportive. For example, if the emotion is recognized as "negative," the prompt to the generative AI model will be adjusted to "The user has a negative emotion. Please generate a polite and kind response."

[0293] Step 9:

[0294] Server: Using the trained data, the generative AI model generates a response to the user's message. For example, the generative AI model generates a response like, "Of course, your recent projects have been very exciting. I'm working on a new album."

[0295] Step 10:

[0296] Server: Apply the style of a character or celebrity to the generated response, for example, "Of course, my latest projects have been very exciting! I'm working on a new album."

[0297] Step 11:

[0298] Server: Additionally, tailor the response based on the emotion engine's perceived emotion. For example, if the user is expressing "negative" emotion, provide additional encouragement such as "Don't worry, the new album is going to be amazing!"

[0299] Step 12:

[0300] Server: Sends the final generated response to the user's terminal, as well as sending data according to the format.

[0301] Step 13:

[0302] Terminal: The user receives a response, for example, a message displayed on the screen saying, "Of course, your latest project is very exciting! Don't worry, the new album is going to be amazing!"

[0303] Step 14:

[0304] User: Views the response received and types the next message. This process continues the conversation. For example, the user next types, "When is the album coming out?"

[0305] Step 15:

[0306] Server: The server stores the recognized user emotions and dialogue content in a database. This emotion history is referenced the next time a dialogue is generated.

[0307] Step 16:

[0308] Server: The next time the user sends a message, the server refers to the previous emotion history and provides a consistent dialogue. For example, if the emotion was "negative," the server adjusts the response to continue being supportive.

[0309] The above is the specific processing flow in the system.

[0310] Example 2

[0311] Next, a description will be given of 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."

[0312] Conventional dialogue systems have suffered from mechanical dialogue with users, lack of consistency, and insufficient emotional response. It has also been difficult to naturally incorporate the style of a specific character or famous person into responses. Furthermore, the user's emotional history has not been utilized in the next dialogue, resulting in insufficient dialogue personalization. There is a need to solve these issues and realize more natural and emotionally responsive dialogue.

[0313] 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.

[0314] In this invention, the server includes means for receiving a message from a user, means for preprocessing the received message, means for recognizing the user's emotion based on the preprocessed message using an emotion engine, means for generating a response using a generative AI model based on the recognized emotion, means for applying a style of a character or famous person to the generated response and adjusting the response content based on the emotion, means for sending the response to the user, means for storing the response and the user's emotion history in a database, and means for generating a response by referring to the emotion history for the next interaction. This makes interactions with the user more natural and emotionally sensitive, and enables consistent and personalized responses incorporating the style of a specific character or famous person.

[0315] "User" refers to a person who interacts with the system.

[0316] "Server" refers to a computer system that receives and processes messages from users.

[0317] "Terminal" refers to the device that a user uses to enter and send messages to a server.

[0318] "Message" refers to text information that a user enters and sends to the system.

[0319] "Preprocessing" refers to analyzing the received message and performing processes such as tokenization and removing unnecessary spaces.

[0320] "Emotion engine" refers to a system that extracts emotions from users' messages.

[0321] A "generative AI model" refers to a system that uses artificial intelligence to automatically generate responses to user messages.

[0322] "Response" refers to a reply to a user's message created by a generative AI model.

[0323] "Character or celebrity style" refers to writing that imitates the speaking or writing style of a particular character or celebrity.

[0324] "Emotion history" refers to a record of emotional information from past interactions with a user.

[0325] "Database" refers to a collection of data that a system uses to store and manage information.

[0326] This invention is a system that uses a generative AI model and an emotion engine to enable dialogue between official accounts and users. The system is configured as follows, with specific operations and processing flows:

[0327] Basic system configuration

[0328] The system consists of a device used by the user, a server that processes messages, a database that stores data, a generative AI model, and an emotion engine.

[0329] Specific examples of hardware and software used

[0330] Devices: smartphones, computers, etc.

[0331] Server: Cloud server, dedicated server

[0332] Database: MySQL, PostgreSQL

[0333] Generative AI model: GPT-3 (OpenAI)

[0334] Emotion engine: VADER (Valence Aware Dictionary and sEntiment Reasoner)

[0335] Specific data processing and calculation

[0336] 1. User receives message

[0337] The user uses the device to send a message to the official account, for example, by typing "Hello, tell me about your recent projects." This message is then sent from the device to the server.

[0338] 2. Message Preprocessing

[0339] The server receives messages sent from the device, stores them in a database, and then pre-processes them by tokenizing them and removing unnecessary spaces and special characters.

[0340] 3. Emotion recognition

[0341] The server inputs the preprocessed messages into the emotion engine to recognize the user's emotions, which extracts emotions such as positive, negative, and neutral from the messages.

[0342] 4. Response Generation

[0343] The server then feeds data into a generative AI model based on the recognized emotions to generate a response to the user's message, adapting the style of a specific character or celebrity and further adjusting the response content based on the emotions recognized by the emotion engine.

[0344] 5. Sending the Response

[0345] The server finally sends the generated response to the user's terminal, where the user receives the response.

[0346] 6. Emotion history storage

[0347] The server stores the recognized emotions and dialogue content in a database, which allows the server to refer to the emotion history the next time a dialogue is generated, enabling a consistent dialogue.

[0348] Specific examples

[0349] For example, consider the case where a user sends the following message:

[0350] User: "I've been feeling a bit down lately. Could you tell me something to cheer me up?"

[0351] 1. User receives message

[0352] The server receives this message and stores it in a database.

[0353] 2. Message Preprocessing

[0354] The server parses and tokenizes the received message, removing unnecessary spaces and special characters.

[0355] 3. Emotion recognition

[0356] The server uses an emotion engine to recognize the user's emotion, in this case determining that the message indicates a negative emotion.

[0357] 4. Response Generation

[0358] The server then inputs the adjusted data into a generative AI model, generating a response such as, "Cheer up! I think your new project is very exciting."

[0359] You can also style this response as a character or celebrity, adjusting it to say, "Of course, cheer up! I think your new project is super exciting! You have a great future ahead of you."

[0360] 5. Sending the Response

[0361] The server sends this adjusted response to the user's terminal.

[0362] The terminal displays this response to the user.

[0363] In this way, users can continue to have natural, engaging, and emotionally relevant interactions with characters and celebrities, and emotion history is stored and used to further personalize future responses.

[0364] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0365] Step 1:

[0366] User receives message

[0367] The user types a message to the official account saying, "Hello, tell me about your recent projects" and sends it.

[0368] The terminal sends the message entered by the user to the server.

[0369] Input: The message entered by the user.

[0370] Output: The message sent to the server.

[0371] Step 2:

[0372] Receiving and saving messages

[0373] The server receives the message sent from the device, for example, user_message = "Hello, tell us about your recent projects."

[0374] The server stores this received message in a database using an SQL INSERT query:

[0375] sql

[0376] INSERT INTO messages (user_id, message) VALUES (1, "Hello, tell us about your recent projects");

[0377] Input: Message received from the terminal.

[0378] Output: The message stored in the database.

[0379] Step 3:

[0380] Message Preprocessing

[0381] The server preprocesses the stored messages by tokenizing them and removing unnecessary spaces and special characters. It uses Python's NLTK library to tokenize "Hello, tell me about your recent projects" into "Hello, tell me about your recent projects."

[0382] Input: The stored messages retrieved from the database.

[0383] Output: The tokenized message.

[0384] Step 4:

[0385] emotion recognition

[0386] The server inputs the preprocessed messages into the emotion engine to recognize the user's emotions. It uses VADER (Valence Aware Dictionary and sEntiment Reasoner) to extract emotions such as positive, negative, and neutral.

[0387] Input: The preprocessed tokenized message.

[0388] Output: User sentiment score.

[0389] Step 5:

[0390] Response Generation

[0391] Based on the recognized sentiment score, the server generates a response to the user's message using a generative AI model, for example, the GPT-3 model. The generative AI model generates a response such as, "Of course, your recent projects are very exciting. I'm working on a new album."

[0392] Input: Sentiment scores and preprocessed tokenized messages.

[0393] Output: The generated response.

[0394] Step 6:

[0395] Applying and adjusting styles

[0396] The server styles the generated response after a specific character or celebrity, such as "Of course, my latest projects have been very exciting! I'm working on a new album."

[0397] It also tailors its responses based on emotions recognized by its emotion engine: if the user is expressing negative emotions, it offers additional encouragement, such as, "Don't worry, the new album is going to be amazing!"

[0398] Input: Generated responses and sentiment scores.

[0399] Output: The styled and adjusted response.

[0400] Step 7:

[0401] Sending a Response

[0402] The server finally sends the adjusted response to the user's terminal.

[0403] The device will display this response to the user, for example, a message like "Of course, my latest projects are very exciting! Don't worry, the new album is going to be amazing!" will appear on the screen.

[0404] Input: The adjusted response.

[0405] Output: The response displayed on the terminal.

[0406] Step 8:

[0407] Emotion history storage

[0408] The server stores the recognized user emotions and the content of the conversation in a database, creating an emotion history that can be referenced the next time a conversation is generated.

[0409] For example, using a SQL query like this:

[0410] sql

[0411] INSERT INTO emotions (user_id, message_id, sentiment) VALUES (1, LAST_INSERT_ID(), "negative");

[0412] Input: User emotion data and dialogue content.

[0413] Output: Emotion history stored in a database.

[0414] Step 9:

[0415] Refer to emotion history when generating the next response

[0416] The next time the user sends a message, the server references the user's past emotion history and generates a new response based on that history to provide a consistent interaction.

[0417] For example, the next time a user sends a message, we get the sentiment history like this:

[0418] sql

[0419] SELECT sentiment FROM emotions WHERE user_id = 1 ORDER BY timestamp DESC LIMIT 1;

[0420] Input: Past emotional history.

[0421] Output: Consistent response.

[0422] (Application example 2)

[0423] Next, a description will be given of Application 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."

[0424] Conventional dialogue systems only have the ability to return a fixed response to messages sent by the user, and have difficulty recognizing the user's emotions and generating a response accordingly. Furthermore, they are unable to apply a specific context to the generated response or utilize the user's emotional history for the next dialogue, making it difficult to realize personalized communication.

[0425] The specific processing by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for receiving a message from a user, means for preprocessing the received message, means for generating a response using a generative AI model based on the preprocessed message, means for applying a specific context to the generated response, means for sending the response to the user, means for recognizing the user's emotion and generating a response according to the emotion, means for saving the preprocessed message and the recognized emotion in a database, and means for referring to the saved emotion history in the next dialogue. This makes it possible to generate a personalized response that takes the user's emotion into consideration and achieve continuous and consistent communication.

[0426] The "means for receiving a message from a user" is a mechanism for transferring a text message entered by a user to a server via a network.

[0427] The "means for preprocessing received messages" refers to functions for formatting messages to make them easier to parse and process, such as tokenizing messages and removing unnecessary spaces and special characters.

[0428] "Means for generating a response using a generative AI model based on a preprocessed message" refers to a mechanism that uses preprocessed message data as input and creates an appropriate response using a generative AI model.

[0429] "Means for applying specific context to generated responses" refers to the ability to add a specific character or contextual style to generated responses, making the responses presented to the user seem natural and familiar.

[0430] The "means for sending a response to the user" is a mechanism for transferring the generated response message to the user's terminal and displaying it on the screen.

[0431] The "means for recognizing the user's emotions and generating a response according to those emotions" is a function for detecting emotions from the user's message and generating a response with content and tone according to those emotions.

[0432] The "means for storing the preprocessed message and the recognized emotion in a database" is a mechanism for recording and storing the user's message and the emotion recognized from the message in a database.

[0433] The "means for referring to the emotion history stored in the next interaction" is a function for referring to the emotion history of the user stored in the past and utilizing it in the next interaction.

[0434] overview

[0435] The present invention is a dialogue system that generates personalized responses based on user messages. The system receives user messages, recognizes emotions, and generates responses based on those emotions using a generative AI model and an emotion engine. Furthermore, the system stores the user's emotion history and utilizes it in the next dialogue, providing consistent and personalized dialogue.

[0436] Hardware and Software Configuration

[0437] Hardware

[0438] The system hardware configuration is as follows:

[0439] 1. User terminal: A mobile device with communication capabilities, such as a smartphone or tablet.

[0440] 2. Server: A cloud server or dedicated server that processes data and generates responses.

[0441] software

[0442] The software used is as follows:

[0443] 1. Generative AI models: Natural language processing models such as the OpenAI API.

[0444] 2. Emotion engine: Emotion recognition algorithms (e.g., Sentiment Analysis API).

[0445] 3. Database management system: such as SQLite.

[0446] Program processing overview and specific examples

[0447] Message Preprocessing

[0448] The server pre-processes messages received from user devices by tokenizing them and removing unnecessary spaces and special characters. For example, a message such as "I'm very unhappy because my order hasn't arrived yet" is converted to "I'm very unhappy because my order hasn't arrived yet."

[0449] emotion recognition

[0450] The server inputs the preprocessed message into the emotion engine, which recognizes emotions such as positive, negative, and neutral. The emotion engine's algorithm extracts the user's emotion from the message content. For example, the above message is recognized as "negative."

[0451] Response Generation

[0452] The server inputs a prompt sentence into the generative AI model based on the emotion recognition results, and generates a response. For example, the following prompt sentence is input into the generative AI model:

[0453] User message: I'm very frustrated because my order hasn't arrived yet.

[0454] Emotion: Negative

[0455] Generate the corresponding response.

[0456] The generative AI model generates a natural response based on this prompt, such as, "We're sorry. We'll look into this immediately and get back to you as soon as possible."

[0457] Customizing and sending responses

[0458] The generated response is adapted to the specific context and style and sent to the user, where the server can amend or modify the generated response to present it in a user-friendly format.

[0459] Save user emotion history for future reference

[0460] The emotion recognition results and messages are stored in a database. The next time you interact with the system, the system will refer to the stored emotion history to provide a consistent and personalized response, allowing users to enjoy continuous and personalized communication.

[0461] In this way, the present invention enables natural dialogue according to the user's emotions, and realizes personalized support.

[0462] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0463] Step 1:

[0464] The server receives a message from the user terminal, for example, a user input such as "I'm very unhappy because my order hasn't arrived yet." The server receives this message and stores it in a database for further processing.

[0465] Step 2:

[0466] The server preprocesses the received message by tokenizing it and removing unnecessary spaces and special characters. For example, a message like "I'm very unhappy because my order hasn't arrived yet" is converted to "I'm very unhappy because my order hasn't arrived yet." This preprocessed message becomes the input data for the next emotion recognition.

[0467] Step 3:

[0468] The server inputs the preprocessed message into the emotion engine to recognize the user's emotion. The emotion engine extracts emotions such as positive, negative, and neutral from the input message. For example, for the preprocessed message "I'm very unhappy because my order hasn't arrived yet," the emotion engine outputs "negative."

[0469] Step 4:

[0470] The server inputs a prompt sentence into the generative AI model based on the emotion recognition results, and generates a response. The prompt sentence is set as follows:

[0471] User message: I'm very frustrated because my order hasn't arrived yet.

[0472] Emotion: Negative

[0473] Generate the corresponding response.

[0474] The generative AI model will then generate a natural response based on this prompt, such as "We apologize for the inconvenience. We will investigate this immediately and provide a prompt response."

[0475] Step 5:

[0476] The server applies specific context and style to the generated response, such as adding a more friendly "We're sorry, we'll look into this and get back to you as soon as possible" response, and customizes it as needed using templates and modules.

[0477] Step 6:

[0478] The server then sends the final response to the user's device. The user's device then displays the received response message on its screen. For example, the user may see a message such as, "We apologize for the inconvenience. We will investigate the issue immediately and respond as soon as possible."

[0479] Step 7:

[0480] The server stores the user's message and the emotion recognition results in a database. The stored data is referenced the next time a conversation occurs, allowing for consistent and personalized responses to be generated based on the user's past emotion history.

[0481] 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 audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio 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 audio data.

[0482] 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> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with 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.

[0483] In the above embodiment, an example in which the specific process is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific process may be performed by the smart device 14.

[0484] [Second embodiment]

[0485] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0486] 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.

[0487] 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 WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0488] 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.

[0489] 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 audio in accordance with instructions from the processor 46.

[0490] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and 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 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[0491] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 control the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0492] Fig. 4 shows an example of the 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.

[0493] 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.

[0494] 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.

[0495] 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.

[0496] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal."

[0497] Regarding an embodiment of the present invention, the specific operation and processing flow of a dialogue system between official accounts and users using a generative AI model will be described.

[0498] overview

[0499] The present invention is a system that allows users to communicate naturally and proactively with the official accounts of characters and famous people. The system receives messages from users, generates responses using a generative AI model, and sends them in a style that matches the style of the specific character or famous person.

[0500] Program processing overview

[0501] 1. Receive a user's message

[0502] User: The user types and sends a message to the official account.

[0503] Terminal: Sends the entered message to the server.

[0504] 2. Message Reception and Preprocessing

[0505] Server: Receives messages sent from the terminal.

[0506] Server: Pre-processes the received message by tokenizing it and removing unnecessary spaces and special characters.

[0507] 3. Response Generation

[0508] Server: Generates a response to the user's message using a generative AI model based on the preprocessed message.

[0509] Server: Apply the style of a character or celebrity to the generated response, adjusting the writing style and speaking style.

[0510] 4. Sending the Response

[0511] Server: Sends the final generated response to the user's terminal.

[0512] Terminal: The user receives the response and the dialogue continues.

[0513] Specific examples

[0514] For example, consider the case where a user sends the following message:

[0515] User: "Tell me about your latest project."

[0516] When this message is sent, the following process occurs:

[0517] 1. Receive a user message

[0518] The server receives and stores this message.

[0519] 2. Message Preprocessing

[0520] The server parses and tokenizes the received message, removing unnecessary spaces and special characters.

[0521] 3. Response Generation

[0522] The server inputs the preprocessed messages into a generative AI model to generate a response.

[0523] For example, a generative AI model might generate a response like, "Of course, your recent projects have been very exciting. You're working on a new album."

[0524] 4. Applying Styles

[0525] The server then styles the generated response after a specific character or celebrity, adjusting it to, "Of course, my latest projects have been very exciting! I'm working on a new album."

[0526] 5. Sending the Response

[0527] The server sends this adjusted response to the user's terminal.

[0528] The terminal displays this response to the user.

[0529] This process allows users to continue natural and engaging interactions with characters and famous people.

[0530] The system also saves the user's interaction history and references it the next time it generates a response, providing a more consistent interaction. Furthermore, by collecting user feedback and periodically updating the generative AI model, the system's accuracy and user satisfaction can be improved.

[0531] The processing flow will be explained below.

[0532] Step 1:

[0533] User: Enter a message for the official account and tap the send button. For example, you could type, "Hello, tell me about your recent projects."

[0534] Step 2:

[0535] Terminal: Sends the message entered by the user to the server, where it is converted into an appropriate format (e.g., JSON).

[0536] Step 3:

[0537] Server: Receives the message sent from the device. For example, get the data user_message = "Hello, tell us about your recent projects."

[0538] Step 4:

[0539] Server: Stores the received message in a database. Message data is saved using an SQL INSERT query.

[0540] Step 5:

[0541] Server: Preprocesses the stored messages by tokenizing them and removing unnecessary spaces and special characters. For example, "Hello, tell me about your recent projects" is tokenized as "Hello, tell me about your recent projects".

[0542] Step 6:

[0543] Server: Feeds the pre-processed messages into a generative AI model, which is pre-trained to mimic the speaking and writing style of a specific character or famous figure.

[0544] Step 7:

[0545] Server: The generative AI model generates an appropriate response to the user's message. For example, the generative AI model generates a response like, "Of course, your recent projects are very exciting. I'm working on a new album."

[0546] Step 8:

[0547] Server: Apply the style of a character or famous person to the generated response. This style adjustment affects the writing style and speaking style. For example, "Of course, my latest projects have been very exciting! I'm working on a new album."

[0548] Step 9:

[0549] Server: Sends the final generated response to the user's terminal, as well as sending data according to the format.

[0550] Step 10:

[0551] Terminal: The user receives a response, for example, a message displayed on the screen saying, "Of course, my latest projects have been very exciting! I'm working on a new album."

[0552] Step 11:

[0553] User: Views the response received and types the next message. This process continues the conversation. For example, the user next types, "When is the album coming out?"

[0554] Step 12:

[0555] Server: Saves the dialogue history and references past dialogues when generating the next response, thereby maintaining a consistent conversation.

[0556] The above is the specific processing flow in the system.

[0557] Example 1

[0558] Next, a description will be given of 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."

[0559] In modern society, there is a growing demand for users to have natural-looking conversations with specific characters or famous people. However, existing dialogue systems have difficulty accurately imitating the speaking style and writing style of characters or famous people, which can lead to a poor user experience. Furthermore, there are technical challenges in maintaining consistency and naturalness in responses. Therefore, there is a need to develop a system that provides natural and engaging conversations based on the style of specific characters or famous people.

[0560] 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.

[0561] In this invention, the server includes means for receiving a message from a user, means for tokenizing the received message and preprocessing it to remove unnecessary spaces and special characters, means for generating a response using a generative AI model based on the preprocessed message, means for applying a style of a specific character or famous person to the generated response to adjust the writing style and speaking manner, and means for sending the adjusted response to the user, thereby enabling the user to have a natural and engaging conversation with the specific character or famous person.

[0562] "User" refers to a person who types and sends messages to interact with the system.

[0563] "Message" refers to text data that a user enters and sends to interact with the system.

[0564] "Tokenization" refers to the process of breaking down an incoming message into individual words or phrases.

[0565] "Preprocessing to remove unnecessary spaces and special characters" refers to the process of removing unnecessary spaces and specific symbols from the message to make it easier to analyze.

[0566] A "generative AI model" refers to an algorithm or software that generates natural language responses based on input text data.

[0567] "Response" refers to the text data of a reply generated by a generative AI model.

[0568] "Character or celebrity style" refers to the style or characteristics used to imitate the speech, writing style, or language of a particular character or celebrity.

[0569] "Tailoring" refers to the process of changing the writing style or speaking style to adapt the generated response to a particular style.

[0570] "Sending" refers to the act of sending the response generated by the server to the user's terminal as data.

[0571] This invention is a system that allows users to have natural conversations with the official accounts of specific characters or famous people. The system receives a message from the user, and a generative AI model generates a response based on the message. The generated response is adapted to the style of the specific character or famous person and returned to the user. To realize this process, the following steps are taken:

[0572] The server has a means for receiving messages from users. Users input and send messages to their official accounts. The terminal then sends these messages to the server. Specifically, the user input is passed to the server using a communication protocol such as an HTTP POST request.

[0573] The server parses the received message and performs preprocessing by tokenizing it and removing unnecessary spaces and special characters. The methods used here are, for example, Python's regular expression library or natural language processing library. For example, the message "Hello! Tell me about your recent projects." is split into tokens such as "Hello," "Recent," "Project," "About," "Tell me," and "Please."

[0574] Next, the server inputs the preprocessed message into a generative AI model to generate a response. For example, a large-scale language model such as GPT-3 is used as the generative AI model. The generative AI model receives the following input:

[0575] User: Hi! Tell us about your latest projects.

[0576] Character: Album Producer

[0577] Based on this input, the generative AI model generates an appropriate response, such as, "Of course, my recent projects have been very exciting. I'm working on a new album."

[0578] Apply the style of a particular character or celebrity to the generated response. The server does this by adjusting the writing style and word choice, for example, "Of course, my latest projects have been very exciting! I'm working on a new album."

[0579] The final adjusted response is sent to the user's device. The server packages the generated response in JSON format or similar and sends it to the user's device. The device receives it and displays it on the chat screen. Specifically, HTML and JavaScript are used to dynamically update the part that displays the message.

[0580] For example, if a user sends a message saying, "Tell me about your latest projects," the process will ultimately result in a response being returned to the user saying, "Of course, my latest projects are very exciting! I'm working on a new album."

[0581] The system also stores the user's interaction history and references it the next time it generates a response, allowing it to provide more consistent interactions. It also collects user feedback and periodically updates the generative AI model to improve system accuracy and user satisfaction.

[0582] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0583] Step 1:

[0584] The user enters and sends a message. Specifically, the user enters a text message in the chat box of the official account and clicks the send button. This operation causes the device to obtain the user's input and generate message data. The input is text data such as "Hello! Tell us about your recent projects." The output is that this text data is sent to the server via the device.

[0585] Step 2:

[0586] The device receives a message from the user and sends it to the server. The device packages the user's input as an HTTP POST request and sends it to the server's API endpoint. The input is the user's text message, for example, "Hello! Tell us about your latest project." The output is that this text message reaches the server.

[0587] Step 3:

[0588] The server receives messages sent from the terminal and stores them in a database. It then performs a process to insert the received messages into a table. Specifically, it stores the received data in the appropriate database table, along with metadata such as the message ID and timestamp. The input is the text data received from the terminal. The output is the state in which this data has been stored in the database.

[0589] Step 4:

[0590] The server performs preprocessing by tokenizing the received message and removing unnecessary spaces and special characters. The server uses regular expressions and natural language processing libraries to tokenize the message and remove unnecessary spaces and special characters. The input is the stored text message. The output is tokenized clean text data. For example, the message "Hello! Tell me about your recent projects." is split into "Hello", "Recent", "Project", "About", "Tell me", and "Please".

[0591] Step 5:

[0592] The server inputs the preprocessed message into a generative AI model to generate a response. A generative AI model (e.g., GPT-3) is used to generate natural-sounding responses based on the preprocessed message. The input is tokenized text data. The output is the generated response text. For example, suppose the input prompt is:

[0593] "User: Hi! Tell me about your latest project.

[0594] Character: Album Producer

[0595] The output from the generative AI model is, "Of course, your recent projects have been very exciting. You're working on a new album."

[0596] Step 6:

[0597] The server applies the style of a particular character or famous person to the generated response, adjusting the writing style and speaking style. The server adjusts the writing style and speaking style of the generated text based on the style of a particular character or famous person. For example, "Of course, my latest projects are very exciting! I'm working on a new album." The input is the generated response text. The output is the styled response text.

[0598] Step 7:

[0599] The server finally generates a response and sends it to the user's device. The server wraps the response text in JSON format or similar and sends it to the device as an HTTP response. The input is the response text after applying the style. The output is the response data sent to the device.

[0600] Step 8:

[0601] The device displays the response received from the server to the user. The device analyzes the received data and displays the response on the chat screen. Specifically, it updates the chat screen using HTML and JavaScript. The input is the response data received from the server. The output is the response text displayed on the user's chat screen. For example, the user's screen might display, "Of course, my recent projects have been very exciting! I'm working on a new album."

[0602] (Application example 1)

[0603] Next, a description will be given of Application 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."

[0604] The present invention relates to a system for improving engagement while maintaining consistency and naturalness of dialogue when users communicate naturally and proactively with the official accounts of characters and famous people. Conventional dialogue systems have difficulty faithfully reproducing the style of characters and famous people, and are unable to fully utilize dialogue history and feedback, limiting user satisfaction. For these reasons, there has been a demand for the development of a dialogue system that is more accurate and can be continuously improved.

[0605] 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.

[0606] In this invention, the server includes means for receiving messages from users, means for preprocessing the received messages, means for generating responses using a generative AI model based on the preprocessed messages, means for applying the style of a character or celebrity to the generated responses, means for sending the responses to the users, means for saving the response history and using it for the next dialogue, and means for collecting feedback from users and improving the generative AI model. This allows users to continue having natural and engaging dialogue with characters or celebrities, improving the consistency and accuracy of the dialogue. Furthermore, by utilizing user feedback, the generative AI model can be continuously improved to provide more accurate dialogue.

[0607] "Means for receiving a message from a user" refers to a device or software for receiving a text message entered by a user.

[0608] A "means for pre-processing received messages" is a device or software that tokenizes messages received from users and removes unnecessary spaces and special characters.

[0609] A "means for generating a response using a generative AI model based on a preprocessed message" is a device or software that uses a preprocessed message as input to generate a response using a generative AI model.

[0610] A "means for applying the style of a character or celebrity to generated responses" is a device or software that tailors the responses generated by a generative AI model to the speaking style or writing style of a particular character or celebrity.

[0611] The "means for sending a response to a user" is a device or software for sending the generated response to the user's terminal.

[0612] The "means for saving response history and using it in the next interaction" refers to a device or software that saves past interaction data between the user and the system and refers to that data during subsequent interactions.

[0613] A "means for collecting user feedback and improving a generative AI model" is a device or software that collects user-provided feedback and uses it as training data for a generative AI model to improve the model's performance.

[0614] "Template-based" refers to a method in which the corresponding response is tailored or generated based on a specific pattern or template.

[0615] A "prompt" is text entered to instruct a generative AI model, providing context for the response you want the model to generate.

[0616] In the embodiment for implementing the present invention, specific programs, processes, and examples of a system that realizes an interaction system between a user and the official account of a specific character or famous person will be described.

[0617] System Program

[0618] The system mainly consists of a client (user terminal) and a server. The client provides a UI for users to input messages, and the server analyzes and processes the received messages, then generates and sends a response. Specifically, it performs the following processes:

[0619] 1. User receives message

[0620] The user enters and sends a message to the official account.

[0621] The client (smartphone app) sends this message to the server.

[0622] 2. Message Preprocessing

[0623] The server tokenizes the received message, removing unnecessary spaces and special characters.

[0624] 3. Generating a Response

[0625] The server generates a response based on the preprocessed message using a generative AI model (e.g., OpenAI's GPT-4).

[0626] 4. Applying Styles

[0627] The generated responses are styled to suit specific characters or famous people, adjusting the writing style and speaking style, using a template-based approach.

[0628] 5. Sending the Response

[0629] The final generated response is sent to the user's terminal to continue the dialogue.

[0630] 6. Saving conversation history

[0631] The server stores the interaction history and references it the next time the interaction occurs, and also collects user feedback to continuously improve the generative AI model.

[0632] Hardware and software used

[0633] Hardware:

[0634] Server: Virtual server on Amazon Web Services (AWS)

[0635] Client: iOS or Android device (smartphone)

[0636] software:

[0637] Database: Uses Amazon RDS to store interaction history and feedback

[0638] Generative AI model: OpenAI's GPT-4

[0639] Preprocessing: A natural language processing library such as Python's NLTK or spaCy

[0640] Client app: Build the UI using Flutter

[0641] Examples of specific examples and prompts

[0642] For example, if a user sends the message "Tell me about your recent projects," the system generates a response as follows:

[0643] 1. Receive and preprocess the user's message:

[0644] The server receives the message "Tell us about your recent projects" and tokenizes and cleans it.

[0645] 2. Response generation using generative AI models:

[0646] The server generates a response using the following prompt sentence:

[0647] Prompt: "Respond in official character style: Tell me about your latest project."

[0648] The generative AI model (GPT-4) generates the response, "Of course, your recent projects have been very exciting. I'm working on a new album."

[0649] 3. Applying styles:

[0650] Tailor this response to the speech style of a particular character.

[0651] For example, "Of course, my latest projects have been very exciting! I'm working on a new album."

[0652] 4. Sending the response:

[0653] The server sends the tailored response to the user's device, which the client app displays.

[0654] This system not only allows users to enjoy natural and consistent interactions with characters and famous people, but it is also expected that the system itself will be continuously improved based on user feedback.

[0655] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0656] Step 1:

[0657] A user sends a message to the official account.

[0658] Input: The message the user types into the text box (e.g., "Tell me about your recent projects")

[0659] How it works: The user enters a message and presses the send button. The client (smartphone app) sends this message to the server.

[0660] Output: Message sent to the server

[0661] Step 2:

[0662] Preprocessing of messages received by the server.

[0663] Input: The message sent to the server

[0664] How it works: The server tokenizes the received message, removing unnecessary spaces and special characters, for example using Python's NLTK library or spaCy.

[0665] Output: Tokenized and cleaned message

[0666] Step 3:

[0667] The server generates a response based on the preprocessed message using a generative AI model.

[0668] Input: Preprocessed message

[0669] Operation: The server takes the preprocessed message as input to the generative AI model (GPT-4), generates a prompt sentence, and sends it to the generative AI model.

[0670] Sample prompt: "Respond in the style of your official character: Tell me about your latest project."

[0671] Output: The response generated by the generative AI model (e.g., "Of course, your recent projects have been very exciting. You're working on a new album.")

[0672] Step 4:

[0673] The server applies the style of a character or celebrity to the generated response.

[0674] Input: The response generated by the generative AI model

[0675] How it works: The server adapts the generated responses to suit the style of a particular character or celebrity, adjusting the writing style and speaking style. It uses a template-based approach to adapt the phrasing and writing style to suit the specific style.

[0676] Output: The styled response (e.g., "Of course, my latest projects have been very exciting! I'm working on a new album.")

[0677] Step 5:

[0678] The server finally sends the generated response to the user's terminal.

[0679] Input: The styled response

[0680] How it works: The server sends the styled response to the user's device, where the client app displays it.

[0681] Output: The response displayed on the user's terminal

[0682] Step 6:

[0683] The server saves the conversation history and uses it for the next conversation.

[0684] Input: The user's message and the generated response

[0685] How it works: The server saves the conversation history in a database such as Amazon RDS, and references it during subsequent conversations.

[0686] Output: Saved interaction history

[0687] Step 7:

[0688] The server collects user feedback and improves the generative AI model.

[0689] Input: User feedback

[0690] How it works: The server collects user feedback through the client app and uses it to improve the generative AI model, incorporating the feedback as training data for the model.

[0691] Output: An improved generative AI model

[0692] Furthermore, 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.

[0693] Regarding an embodiment of the present invention, the specific operation and processing flow of a dialogue system between official accounts and users using a generative AI model and an emotion engine will be described.

[0694] overview

[0695] The present invention is a system that allows users to communicate naturally and proactively with the official accounts of characters and celebrities, and also recognizes the user's emotions and provides appropriate responses. The system receives messages from users, generates responses using a generative AI model and an emotion engine, and sends them in a style that matches the specific character or celebrity's style. It also saves the user's emotion history and uses it for the next conversation.

[0696] Program processing overview

[0697] 1. Receive a user's message

[0698] User: The user types and sends a message to the official account. For example, they might type, "Hello, tell me about your recent projects."

[0699] Terminal: Sends the entered message to the server.

[0700] 2. Message Reception and Preprocessing

[0701] Server: Receives the message sent from the device. For example, get the data user_message = "Hello, tell us about your recent projects."

[0702] Server: Stores the received message in a database. Message data is saved using an SQL INSERT query.

[0703] Server: Preprocesses the stored messages by tokenizing them and removing unnecessary spaces and special characters. For example, "Hello, tell me about your recent projects" is tokenized as "Hello, tell me about your recent projects".

[0704] 3. Emotion recognition

[0705] Server: Inputs the preprocessed message into the emotion engine to recognize the user's emotion. The emotion engine extracts emotions such as positive, negative, and neutral from the message.

[0706] Server: Adjusts the input to the generative AI model based on the recognized emotion, for example, making the response more polite and supportive if the user has a negative emotion.

[0707] 4. Response Generation

[0708] Server: Using the trained data, the generative AI model generates a response to the user's message. For example, the generative AI model generates a response like, "Of course, your recent projects have been very exciting. I'm working on a new album."

[0709] Server: Style the generated response with a character or celebrity, transforming it into, "Of course, my latest projects have been very exciting! I'm working on a new album."

[0710] Server: Additionally, tailor the response based on the emotion engine's recognized emotion. For example, if the user is expressing negative emotion, provide additional encouragement such as, "Don't worry, the new album is going to be amazing!"

[0711] 5. Sending the Response

[0712] Server: Sends the final generated response to the user's terminal, as well as sending data according to the format.

[0713] Terminal: The user receives a response, for example, a message displayed on the screen saying, "Of course, your latest project is very exciting! Don't worry, the new album is going to be amazing!"

[0714] 6. Saving user emotion history

[0715] Server: The server stores the recognized user emotions and dialogue content in a database. This emotion history is referenced the next time a dialogue is generated.

[0716] 7. Refer to emotion history when generating the next response

[0717] Server: The next time the user sends a message, it references the user's past emotional history and provides a consistent dialogue.

[0718] Specific examples

[0719] For example, consider the case where a user sends the following message:

[0720] User: "I've been feeling a bit down lately. Could you tell me something to cheer me up?"

[0721] 1. Receive a user message

[0722] The server receives and stores this message.

[0723] 2. Message Preprocessing

[0724] The server parses and tokenizes the received message, removing unnecessary spaces and special characters.

[0725] 3. Emotion recognition

[0726] The server uses an emotion engine to recognize the user's emotion, in this case determining that the message indicates a negative emotion.

[0727] 4. Response Generation

[0728] The server then feeds the adjusted data into a generative AI model to generate a response, such as "Cheer up! I think your new project is very exciting."

[0729] The server applies the style of characters and famous people to the generated responses, and further adjusts them based on emotions recognized by the emotion engine: "Of course, cheer up! I think your new project will get you very excited! You have a great future ahead of you."

[0730] 5. Sending the Response

[0731] The server sends this adjusted response to the user's terminal.

[0732] The terminal displays this response to the user.

[0733] This process allows users to continue interacting with characters and celebrities in a natural, engaging and emotionally relevant way, and emotion history is stored and used to further personalize future responses.

[0734] The processing flow will be explained below.

[0735] Step 1:

[0736] User: Enter a message for the official account and tap the send button. For example, you could type, "Hello, tell me about your recent projects."

[0737] Step 2:

[0738] Terminal: Sends the message entered by the user to the server, where it is converted into an appropriate format (e.g., JSON).

[0739] Step 3:

[0740] Server: Receives the message sent from the device. For example, get the data user_message = "Hello, tell us about your recent projects."

[0741] Step 4:

[0742] Server: Stores the received message in a database. Message data is saved using an SQL INSERT query.

[0743] Step 5:

[0744] Server: Preprocesses the stored messages by tokenizing them and removing unnecessary spaces and special characters. For example, "Hello, tell me about your recent projects" is tokenized as "Hello, tell me about your recent projects".

[0745] Step 6:

[0746] Server: Inputs the preprocessed message into the emotion engine to recognize the user's emotion. The emotion engine extracts emotions such as positive, negative, and neutral from the message.

[0747] Step 7:

[0748] Server: The emotion engine receives the results of recognizing the user's emotion and stores them in the database. For example, if the emotion is recognized as "positive," it will be saved in the database.

[0749] Step 8:

[0750] Server: Adjusts the input to the generative AI model based on the recognized emotion. For example, if the user has a negative emotion, the response will be more polite and supportive. For example, if the emotion is recognized as "negative," the prompt to the generative AI model will be adjusted to "The user has a negative emotion. Please generate a polite and kind response."

[0751] Step 9:

[0752] Server: Using the trained data, the generative AI model generates a response to the user's message. For example, the generative AI model generates a response like, "Of course, your recent projects have been very exciting. I'm working on a new album."

[0753] Step 10:

[0754] Server: Apply the style of a character or celebrity to the generated response, for example, "Of course, my latest projects have been very exciting! I'm working on a new album."

[0755] Step 11:

[0756] Server: Additionally, tailor the response based on the emotion engine's perceived emotion. For example, if the user is expressing "negative" emotion, provide additional encouragement such as "Don't worry, the new album is going to be amazing!"

[0757] Step 12:

[0758] Server: Sends the final generated response to the user's terminal, as well as sending data according to the format.

[0759] Step 13:

[0760] Terminal: The user receives a response, for example, a message displayed on the screen saying, "Of course, your latest project is very exciting! Don't worry, the new album is going to be amazing!"

[0761] Step 14:

[0762] User: Views the response received and types the next message. This process continues the conversation. For example, the user next types, "When is the album coming out?"

[0763] Step 15:

[0764] Server: The server stores the recognized user emotions and dialogue content in a database. This emotion history is referenced the next time a dialogue is generated.

[0765] Step 16:

[0766] Server: The next time the user sends a message, the server refers to the previous emotion history and provides a consistent dialogue. For example, if the emotion was "negative," the server adjusts the response to continue being supportive.

[0767] The above is the specific processing flow in the system.

[0768] Example 2

[0769] Next, a description will be given of 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."

[0770] Conventional dialogue systems have suffered from mechanical dialogue with users, lack of consistency, and insufficient emotional response. It has also been difficult to naturally incorporate the style of a specific character or famous person into responses. Furthermore, the user's emotional history has not been utilized in the next dialogue, resulting in insufficient dialogue personalization. There is a need to solve these issues and realize more natural and emotionally responsive dialogue.

[0771] 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.

[0772] In this invention, the server includes means for receiving a message from a user, means for preprocessing the received message, means for recognizing the user's emotion based on the preprocessed message using an emotion engine, means for generating a response using a generative AI model based on the recognized emotion, means for applying a style of a character or famous person to the generated response and adjusting the response content based on the emotion, means for sending the response to the user, means for storing the response and the user's emotion history in a database, and means for generating a response by referring to the emotion history for the next interaction. This makes interactions with the user more natural and emotionally sensitive, and enables consistent and personalized responses incorporating the style of a specific character or famous person.

[0773] "User" refers to a person who interacts with the system.

[0774] "Server" refers to a computer system that receives and processes messages from users.

[0775] "Terminal" refers to the device that a user uses to enter and send messages to a server.

[0776] "Message" refers to text information that a user enters and sends to the system.

[0777] "Preprocessing" refers to analyzing the received message and performing processes such as tokenization and removing unnecessary spaces.

[0778] "Emotion engine" refers to a system that extracts emotions from users' messages.

[0779] A "generative AI model" refers to a system that uses artificial intelligence to automatically generate responses to user messages.

[0780] "Response" refers to a reply to a user's message created by a generative AI model.

[0781] "Character or celebrity style" refers to writing that imitates the speaking or writing style of a particular character or celebrity.

[0782] "Emotion history" refers to a record of emotional information from past interactions with a user.

[0783] "Database" refers to a collection of data that a system uses to store and manage information.

[0784] This invention is a system that uses a generative AI model and an emotion engine to enable dialogue between official accounts and users. The system is configured as follows, with specific operations and processing flows:

[0785] Basic system configuration

[0786] The system consists of a device used by the user, a server that processes messages, a database that stores data, a generative AI model, and an emotion engine.

[0787] Specific examples of hardware and software used

[0788] Devices: smartphones, computers, etc.

[0789] Server: Cloud server, dedicated server

[0790] Database: MySQL, PostgreSQL

[0791] Generative AI model: GPT-3 (OpenAI)

[0792] Emotion engine: VADER (Valence Aware Dictionary and sEntiment Reasoner)

[0793] Specific data processing and calculation

[0794] 1. User receives message

[0795] The user uses the device to send a message to the official account, for example, by typing "Hello, tell me about your recent projects." This message is then sent from the device to the server.

[0796] 2. Message Preprocessing

[0797] The server receives messages sent from the device, stores them in a database, and then pre-processes them by tokenizing them and removing unnecessary spaces and special characters.

[0798] 3. Emotion recognition

[0799] The server inputs the preprocessed messages into the emotion engine to recognize the user's emotions, which extracts emotions such as positive, negative, and neutral from the messages.

[0800] 4. Response Generation

[0801] The server then feeds data into a generative AI model based on the recognized emotions to generate a response to the user's message, adapting the style of a specific character or celebrity and further adjusting the response content based on the emotions recognized by the emotion engine.

[0802] 5. Sending the Response

[0803] The server finally sends the generated response to the user's terminal, where the user receives the response.

[0804] 6. Emotion history storage

[0805] The server stores the recognized emotions and dialogue content in a database, which allows the server to refer to the emotion history the next time a dialogue is generated, enabling a consistent dialogue.

[0806] Specific examples

[0807] For example, consider the case where a user sends the following message:

[0808] User: "I've been feeling a bit down lately. Could you tell me something to cheer me up?"

[0809] 1. User receives message

[0810] The server receives this message and stores it in a database.

[0811] 2. Message Preprocessing

[0812] The server parses and tokenizes the received message, removing unnecessary spaces and special characters.

[0813] 3. Emotion recognition

[0814] The server uses an emotion engine to recognize the user's emotion, in this case determining that the message indicates a negative emotion.

[0815] 4. Response Generation

[0816] The server then inputs the adjusted data into a generative AI model, generating a response such as, "Cheer up! I think your new project is very exciting."

[0817] You can also style this response as a character or celebrity, adjusting it to say, "Of course, cheer up! I think your new project is super exciting! You have a great future ahead of you."

[0818] 5. Sending the Response

[0819] The server sends this adjusted response to the user's terminal.

[0820] The terminal displays this response to the user.

[0821] In this way, users can continue to have natural, engaging, and emotionally relevant interactions with characters and celebrities, and emotion history is stored and used to further personalize future responses.

[0822] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0823] Step 1:

[0824] User receives message

[0825] The user types a message to the official account saying, "Hello, tell me about your recent projects" and sends it.

[0826] The terminal sends the message entered by the user to the server.

[0827] Input: The message entered by the user.

[0828] Output: The message sent to the server.

[0829] Step 2:

[0830] Receiving and saving messages

[0831] The server receives the message sent from the device, for example, user_message = "Hello, tell us about your recent projects."

[0832] The server stores this received message in a database using an SQL INSERT query:

[0833] sql

[0834] INSERT INTO messages (user_id, message) VALUES (1, "Hello, tell us about your recent projects");

[0835] Input: Message received from the terminal.

[0836] Output: The message stored in the database.

[0837] Step 3:

[0838] Message Preprocessing

[0839] The server preprocesses the stored messages by tokenizing them and removing unnecessary spaces and special characters. It uses Python's NLTK library to tokenize "Hello, tell me about your recent projects" into "Hello, tell me about your recent projects."

[0840] Input: The stored messages retrieved from the database.

[0841] Output: The tokenized message.

[0842] Step 4:

[0843] emotion recognition

[0844] The server inputs the preprocessed messages into the emotion engine to recognize the user's emotions. It uses VADER (Valence Aware Dictionary and sEntiment Reasoner) to extract emotions such as positive, negative, and neutral.

[0845] Input: The preprocessed tokenized message.

[0846] Output: User sentiment score.

[0847] Step 5:

[0848] Response Generation

[0849] Based on the recognized sentiment score, the server generates a response to the user's message using a generative AI model, for example, the GPT-3 model. The generative AI model generates a response such as, "Of course, your recent projects are very exciting. I'm working on a new album."

[0850] Input: Sentiment scores and preprocessed tokenized messages.

[0851] Output: The generated response.

[0852] Step 6:

[0853] Applying and adjusting styles

[0854] The server styles the generated response after a specific character or celebrity, such as "Of course, my latest projects have been very exciting! I'm working on a new album."

[0855] It also tailors its responses based on emotions recognized by its emotion engine: if the user is expressing negative emotions, it offers additional encouragement, such as, "Don't worry, the new album is going to be amazing!"

[0856] Input: Generated responses and sentiment scores.

[0857] Output: The styled and adjusted response.

[0858] Step 7:

[0859] Sending a Response

[0860] The server finally sends the adjusted response to the user's terminal.

[0861] The device will display this response to the user, for example, a message like "Of course, my latest projects are very exciting! Don't worry, the new album is going to be amazing!" will appear on the screen.

[0862] Input: The adjusted response.

[0863] Output: The response displayed on the terminal.

[0864] Step 8:

[0865] Emotion history storage

[0866] The server stores the recognized user emotions and the content of the conversation in a database, creating an emotion history that can be referenced the next time a conversation is generated.

[0867] For example, using a SQL query like this:

[0868] sql

[0869] INSERT INTO emotions (user_id, message_id, sentiment) VALUES (1, LAST_INSERT_ID(), "negative");

[0870] Input: User emotion data and dialogue content.

[0871] Output: Emotion history stored in a database.

[0872] Step 9:

[0873] Refer to emotion history when generating the next response

[0874] The next time the user sends a message, the server references the user's past emotion history and generates a new response based on that history to provide a consistent interaction.

[0875] For example, the next time a user sends a message, we get the sentiment history like this:

[0876] sql

[0877] SELECT sentiment FROM emotions WHERE user_id = 1 ORDER BY timestamp DESC LIMIT 1;

[0878] Input: Past emotional history.

[0879] Output: Consistent response.

[0880] (Application example 2)

[0881] Next, a description will be given of Application 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."

[0882] Conventional dialogue systems only have the ability to return a fixed response to messages sent by the user, and have difficulty recognizing the user's emotions and generating a response accordingly. Furthermore, they are unable to apply a specific context to the generated response or utilize the user's emotional history for the next dialogue, making it difficult to realize personalized communication.

[0883] The specific processing by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for receiving a message from a user, means for preprocessing the received message, means for generating a response using a generative AI model based on the preprocessed message, means for applying a specific context to the generated response, means for sending the response to the user, means for recognizing the user's emotion and generating a response according to the emotion, means for saving the preprocessed message and the recognized emotion in a database, and means for referring to the saved emotion history in the next dialogue. This makes it possible to generate a personalized response that takes the user's emotion into consideration and achieve continuous and consistent communication.

[0884] The "means for receiving a message from a user" is a mechanism for transferring a text message entered by a user to a server via a network.

[0885] The "means for preprocessing received messages" refers to functions for formatting messages to make them easier to parse and process, such as tokenizing messages and removing unnecessary spaces and special characters.

[0886] "Means for generating a response using a generative AI model based on a preprocessed message" refers to a mechanism that uses preprocessed message data as input and creates an appropriate response using a generative AI model.

[0887] "Means for applying specific context to generated responses" refers to the ability to add a specific character or contextual style to generated responses, making the responses presented to the user seem natural and familiar.

[0888] The "means for sending a response to the user" is a mechanism for transferring the generated response message to the user's terminal and displaying it on the screen.

[0889] The "means for recognizing the user's emotions and generating a response according to those emotions" is a function for detecting emotions from the user's message and generating a response with content and tone according to those emotions.

[0890] The "means for storing the preprocessed message and the recognized emotion in a database" is a mechanism for recording and storing the user's message and the emotion recognized from the message in a database.

[0891] The "means for referring to the emotion history stored in the next interaction" is a function for referring to the emotion history of the user stored in the past and utilizing it in the next interaction.

[0892] overview

[0893] The present invention is a dialogue system that generates personalized responses based on user messages. The system receives user messages, recognizes emotions, and generates responses based on those emotions using a generative AI model and an emotion engine. Furthermore, the system stores the user's emotion history and utilizes it in the next dialogue, providing consistent and personalized dialogue.

[0894] Hardware and Software Configuration

[0895] Hardware

[0896] The system hardware configuration is as follows:

[0897] 1. User terminal: A mobile device with communication capabilities, such as a smartphone or tablet.

[0898] 2. Server: A cloud server or dedicated server that processes data and generates responses.

[0899] software

[0900] The software used is as follows:

[0901] 1. Generative AI models: Natural language processing models such as the OpenAI API.

[0902] 2. Emotion engine: Emotion recognition algorithms (e.g., Sentiment Analysis API).

[0903] 3. Database management system: such as SQLite.

[0904] Program processing overview and specific examples

[0905] Message Preprocessing

[0906] The server pre-processes messages received from user devices by tokenizing them and removing unnecessary spaces and special characters. For example, a message such as "I'm very unhappy because my order hasn't arrived yet" is converted to "I'm very unhappy because my order hasn't arrived yet."

[0907] emotion recognition

[0908] The server inputs the preprocessed message into the emotion engine, which recognizes emotions such as positive, negative, and neutral. The emotion engine's algorithm extracts the user's emotion from the message content. For example, the above message is recognized as "negative."

[0909] Response Generation

[0910] The server inputs a prompt sentence into the generative AI model based on the emotion recognition results, and generates a response. For example, the following prompt sentence is input into the generative AI model:

[0911] User message: I'm very frustrated because my order hasn't arrived yet.

[0912] Emotion: Negative

[0913] Generate the corresponding response.

[0914] The generative AI model generates a natural response based on this prompt, such as, "We're sorry. We'll look into this immediately and get back to you as soon as possible."

[0915] Customizing and sending responses

[0916] The generated response is adapted to the specific context and style and sent to the user, where the server can amend or modify the generated response to present it in a user-friendly format.

[0917] Save user emotion history for future reference

[0918] The emotion recognition results and messages are stored in a database. The next time you interact with the system, the system will refer to the stored emotion history to provide a consistent and personalized response, allowing users to enjoy continuous and personalized communication.

[0919] In this way, the present invention enables natural dialogue according to the user's emotions, and realizes personalized support.

[0920] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0921] Step 1:

[0922] The server receives a message from the user terminal, for example, a user input such as "I'm very unhappy because my order hasn't arrived yet." The server receives this message and stores it in a database for further processing.

[0923] Step 2:

[0924] The server preprocesses the received message by tokenizing it and removing unnecessary spaces and special characters. For example, a message like "I'm very unhappy because my order hasn't arrived yet" is converted to "I'm very unhappy because my order hasn't arrived yet." This preprocessed message becomes the input data for the next emotion recognition.

[0925] Step 3:

[0926] The server inputs the preprocessed message into the emotion engine to recognize the user's emotion. The emotion engine extracts emotions such as positive, negative, and neutral from the input message. For example, for the preprocessed message "I'm very unhappy because my order hasn't arrived yet," the emotion engine outputs "negative."

[0927] Step 4:

[0928] The server inputs a prompt sentence into the generative AI model based on the emotion recognition results, and generates a response. The prompt sentence is set as follows:

[0929] User message: I'm very frustrated because my order hasn't arrived yet.

[0930] Emotion: Negative

[0931] Generate the corresponding response.

[0932] The generative AI model will then generate a natural response based on this prompt, such as "We apologize for the inconvenience. We will investigate this immediately and provide a prompt response."

[0933] Step 5:

[0934] The server applies specific context and style to the generated response, such as adding a more friendly "We're sorry, we'll look into this and get back to you as soon as possible" response, and customizes it as needed using templates and modules.

[0935] Step 6:

[0936] The server then sends the final response to the user's device. The user's device then displays the received response message on its screen. For example, the user may see a message such as, "We apologize for the inconvenience. We will investigate the issue immediately and respond as soon as possible."

[0937] Step 7:

[0938] The server stores the user's message and the emotion recognition results in a database. The stored data is referenced the next time a conversation occurs, allowing for consistent and personalized responses to be generated based on the user's past emotion history.

[0939] 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 audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio 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 audio data.

[0940] 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> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with 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.

[0941] In the above embodiment, an example in which the specific processing is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the smart glasses 214.

[0942] [Third embodiment]

[0943] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[0944] 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.

[0945] 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 WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0946] 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.

[0947] 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 audio in accordance with instructions from the processor 46.

[0948] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and 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 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[0949] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 control the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0950] Fig. 6 shows an example of the 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.

[0951] 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.

[0952] 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.

[0953] In the headset type terminal 314, a 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.

[0954] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as the "server" and the headset type terminal 314 will be referred to as the "terminal."

[0955] Regarding an embodiment of the present invention, the specific operation and processing flow of a dialogue system between official accounts and users using a generative AI model will be described.

[0956] overview

[0957] The present invention is a system that allows users to communicate naturally and proactively with the official accounts of characters and famous people. The system receives messages from users, generates responses using a generative AI model, and sends them in a style that matches the style of the specific character or famous person.

[0958] Program processing overview

[0959] 1. Receive a user's message

[0960] User: The user types and sends a message to the official account.

[0961] Terminal: Sends the entered message to the server.

[0962] 2. Message Reception and Preprocessing

[0963] Server: Receives messages sent from the terminal.

[0964] Server: Pre-processes the received message by tokenizing it and removing unnecessary spaces and special characters.

[0965] 3. Response Generation

[0966] Server: Generates a response to the user's message using a generative AI model based on the preprocessed message.

[0967] Server: Apply the style of a character or celebrity to the generated response, adjusting the writing style and speaking style.

[0968] 4. Sending the Response

[0969] Server: Sends the final generated response to the user's terminal.

[0970] Terminal: The user receives the response and the dialogue continues.

[0971] Specific examples

[0972] For example, consider the case where a user sends the following message:

[0973] User: "Tell me about your latest project."

[0974] When this message is sent, the following process occurs:

[0975] 1. Receive a user message

[0976] The server receives and stores this message.

[0977] 2. Message Preprocessing

[0978] The server parses and tokenizes the received message, removing unnecessary spaces and special characters.

[0979] 3. Response Generation

[0980] The server inputs the preprocessed messages into a generative AI model to generate a response.

[0981] For example, a generative AI model might generate a response like, "Of course, your recent projects have been very exciting. You're working on a new album."

[0982] 4. Applying Styles

[0983] The server then styles the generated response after a specific character or celebrity, adjusting it to, "Of course, my latest projects have been very exciting! I'm working on a new album."

[0984] 5. Sending the Response

[0985] The server sends this adjusted response to the user's terminal.

[0986] The terminal displays this response to the user.

[0987] This process allows users to continue natural and engaging interactions with characters and famous people.

[0988] The system also saves the user's interaction history and references it the next time it generates a response, providing a more consistent interaction. Furthermore, by collecting user feedback and periodically updating the generative AI model, the system's accuracy and user satisfaction can be improved.

[0989] The processing flow will be explained below.

[0990] Step 1:

[0991] User: Enter a message for the official account and tap the send button. For example, you could type, "Hello, tell me about your recent projects."

[0992] Step 2:

[0993] Terminal: Sends the message entered by the user to the server, where it is converted into an appropriate format (e.g., JSON).

[0994] Step 3:

[0995] Server: Receives the message sent from the device. For example, get the data user_message = "Hello, tell us about your recent projects."

[0996] Step 4:

[0997] Server: Stores the received message in a database. Message data is saved using an SQL INSERT query.

[0998] Step 5:

[0999] Server: Preprocesses the stored messages by tokenizing them and removing unnecessary spaces and special characters. For example, "Hello, tell me about your recent projects" is tokenized as "Hello, tell me about your recent projects".

[1000] Step 6:

[1001] Server: Feeds the pre-processed messages into a generative AI model, which is pre-trained to mimic the speaking and writing style of a specific character or famous figure.

[1002] Step 7:

[1003] Server: The generative AI model generates an appropriate response to the user's message. For example, the generative AI model generates a response like, "Of course, your recent projects are very exciting. I'm working on a new album."

[1004] Step 8:

[1005] Server: Apply the style of a character or famous person to the generated response. This style adjustment affects the writing style and speaking style. For example, "Of course, my latest projects have been very exciting! I'm working on a new album."

[1006] Step 9:

[1007] Server: Sends the final generated response to the user's terminal, as well as sending data according to the format.

[1008] Step 10:

[1009] Terminal: The user receives a response, for example, a message displayed on the screen saying, "Of course, my latest projects have been very exciting! I'm working on a new album."

[1010] Step 11:

[1011] User: Views the response received and types the next message. This process continues the conversation. For example, the user next types, "When is the album coming out?"

[1012] Step 12:

[1013] Server: Saves the dialogue history and references past dialogues when generating the next response, thereby maintaining a consistent conversation.

[1014] The above is the specific processing flow in the system.

[1015] Example 1

[1016] Next, a description will be given of 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."

[1017] In modern society, there is a growing demand for users to have natural-looking conversations with specific characters or famous people. However, existing dialogue systems have difficulty accurately imitating the speaking style and writing style of characters or famous people, which can lead to a poor user experience. Furthermore, there are technical challenges in maintaining consistency and naturalness in responses. Therefore, there is a need to develop a system that provides natural and engaging conversations based on the style of specific characters or famous people.

[1018] 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.

[1019] In this invention, the server includes means for receiving a message from a user, means for tokenizing the received message and preprocessing it to remove unnecessary spaces and special characters, means for generating a response using a generative AI model based on the preprocessed message, means for applying a style of a specific character or famous person to the generated response to adjust the writing style and speaking manner, and means for sending the adjusted response to the user, thereby enabling the user to have a natural and engaging conversation with the specific character or famous person.

[1020] "User" refers to a person who types and sends messages to interact with the system.

[1021] "Message" refers to text data that a user enters and sends to interact with the system.

[1022] "Tokenization" refers to the process of breaking down an incoming message into individual words or phrases.

[1023] "Preprocessing to remove unnecessary spaces and special characters" refers to the process of removing unnecessary spaces and specific symbols from the message to make it easier to analyze.

[1024] A "generative AI model" refers to an algorithm or software that generates natural language responses based on input text data.

[1025] "Response" refers to the text data of a reply generated by a generative AI model.

[1026] "Character or celebrity style" refers to the style or characteristics used to imitate the speech, writing style, or language of a particular character or celebrity.

[1027] "Tailoring" refers to the process of changing the writing style or speaking style to adapt the generated response to a particular style.

[1028] "Sending" refers to the act of sending the response generated by the server to the user's terminal as data.

[1029] This invention is a system that allows users to have natural conversations with the official accounts of specific characters or famous people. The system receives a message from the user, and a generative AI model generates a response based on the message. The generated response is adapted to the style of the specific character or famous person and returned to the user. To realize this process, the following steps are taken:

[1030] The server has a means for receiving messages from users. Users input and send messages to their official accounts. The terminal then sends these messages to the server. Specifically, the user input is passed to the server using a communication protocol such as an HTTP POST request.

[1031] The server parses the received message and performs preprocessing by tokenizing it and removing unnecessary spaces and special characters. The methods used here are, for example, Python's regular expression library or natural language processing library. For example, the message "Hello! Tell me about your recent projects." is split into tokens such as "Hello," "Recent," "Project," "About," "Tell me," and "Please."

[1032] Next, the server inputs the preprocessed message into a generative AI model to generate a response. For example, a large-scale language model such as GPT-3 is used as the generative AI model. The generative AI model receives the following input:

[1033] User: Hi! Tell us about your latest projects.

[1034] Character: Album Producer

[1035] Based on this input, the generative AI model generates an appropriate response, such as, "Of course, my recent projects have been very exciting. I'm working on a new album."

[1036] Apply the style of a particular character or celebrity to the generated response. The server does this by adjusting the writing style and word choice, for example, "Of course, my latest projects have been very exciting! I'm working on a new album."

[1037] The final adjusted response is sent to the user's device. The server packages the generated response in JSON format or similar and sends it to the user's device. The device receives it and displays it on the chat screen. Specifically, HTML and JavaScript are used to dynamically update the part that displays the message.

[1038] For example, if a user sends a message saying, "Tell me about your latest projects," the process will ultimately result in a response being returned to the user saying, "Of course, my latest projects are very exciting! I'm working on a new album."

[1039] The system also stores the user's interaction history and references it the next time it generates a response, allowing it to provide more consistent interactions. It also collects user feedback and periodically updates the generative AI model to improve system accuracy and user satisfaction.

[1040] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1041] Step 1:

[1042] The user enters and sends a message. Specifically, the user enters a text message in the chat box of the official account and clicks the send button. This operation causes the device to obtain the user's input and generate message data. The input is text data such as "Hello! Tell us about your recent projects." The output is that this text data is sent to the server via the device.

[1043] Step 2:

[1044] The device receives a message from the user and sends it to the server. The device packages the user's input as an HTTP POST request and sends it to the server's API endpoint. The input is the user's text message, for example, "Hello! Tell us about your latest project." The output is that this text message reaches the server.

[1045] Step 3:

[1046] The server receives messages sent from the terminal and stores them in a database. It then performs a process to insert the received messages into a table. Specifically, it stores the received data in the appropriate database table, along with metadata such as the message ID and timestamp. The input is the text data received from the terminal. The output is the state in which this data has been stored in the database.

[1047] Step 4:

[1048] The server performs preprocessing by tokenizing the received message and removing unnecessary spaces and special characters. The server uses regular expressions and natural language processing libraries to tokenize the message and remove unnecessary spaces and special characters. The input is the stored text message. The output is tokenized clean text data. For example, the message "Hello! Tell me about your recent projects." is split into "Hello", "Recent", "Project", "About", "Tell me", and "Please".

[1049] Step 5:

[1050] The server inputs the preprocessed message into a generative AI model to generate a response. A generative AI model (e.g., GPT-3) is used to generate natural-sounding responses based on the preprocessed message. The input is tokenized text data. The output is the generated response text. For example, suppose the input prompt is:

[1051] "User: Hi! Tell me about your latest project.

[1052] Character: Album Producer

[1053] The output from the generative AI model is, "Of course, your recent projects have been very exciting. You're working on a new album."

[1054] Step 6:

[1055] The server applies the style of a particular character or famous person to the generated response, adjusting the writing style and speaking style. The server adjusts the writing style and speaking style of the generated text based on the style of a particular character or famous person. For example, "Of course, my latest projects are very exciting! I'm working on a new album." The input is the generated response text. The output is the styled response text.

[1056] Step 7:

[1057] The server finally generates a response and sends it to the user's device. The server wraps the response text in JSON format or similar and sends it to the device as an HTTP response. The input is the response text after applying the style. The output is the response data sent to the device.

[1058] Step 8:

[1059] The device displays the response received from the server to the user. The device analyzes the received data and displays the response on the chat screen. Specifically, it updates the chat screen using HTML and JavaScript. The input is the response data received from the server. The output is the response text displayed on the user's chat screen. For example, the user's screen might display, "Of course, my recent projects have been very exciting! I'm working on a new album."

[1060] (Application example 1)

[1061] Next, a description will be given of Application 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."

[1062] The present invention relates to a system for improving engagement while maintaining consistency and naturalness of dialogue when users communicate naturally and proactively with the official accounts of characters and famous people. Conventional dialogue systems have difficulty faithfully reproducing the style of characters and famous people, and are unable to fully utilize dialogue history and feedback, limiting user satisfaction. For these reasons, there has been a demand for the development of a dialogue system that is more accurate and can be continuously improved.

[1063] 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.

[1064] In this invention, the server includes means for receiving messages from users, means for preprocessing the received messages, means for generating responses using a generative AI model based on the preprocessed messages, means for applying the style of a character or celebrity to the generated responses, means for sending the responses to the users, means for saving the response history and using it for the next dialogue, and means for collecting feedback from users and improving the generative AI model. This allows users to continue having natural and engaging dialogue with characters or celebrities, improving the consistency and accuracy of the dialogue. Furthermore, by utilizing user feedback, the generative AI model can be continuously improved to provide more accurate dialogue.

[1065] "Means for receiving a message from a user" refers to a device or software for receiving a text message entered by a user.

[1066] A "means for pre-processing received messages" is a device or software that tokenizes messages received from users and removes unnecessary spaces and special characters.

[1067] A "means for generating a response using a generative AI model based on a preprocessed message" is a device or software that uses a preprocessed message as input to generate a response using a generative AI model.

[1068] A "means for applying the style of a character or celebrity to generated responses" is a device or software that tailors the responses generated by a generative AI model to the speaking style or writing style of a particular character or celebrity.

[1069] The "means for sending a response to a user" is a device or software for sending the generated response to the user's terminal.

[1070] The "means for saving response history and using it in the next interaction" refers to a device or software that saves past interaction data between the user and the system and refers to that data during subsequent interactions.

[1071] A "means for collecting user feedback and improving a generative AI model" is a device or software that collects user-provided feedback and uses it as training data for a generative AI model to improve the model's performance.

[1072] "Template-based" refers to a method in which the corresponding response is tailored or generated based on a specific pattern or template.

[1073] A "prompt" is text entered to instruct a generative AI model, providing context for the response you want the model to generate.

[1074] In the embodiment for implementing the present invention, specific programs, processes, and examples of a system that realizes an interaction system between a user and the official account of a specific character or famous person will be described.

[1075] System Program

[1076] The system mainly consists of a client (user terminal) and a server. The client provides a UI for users to input messages, and the server analyzes and processes the received messages, then generates and sends a response. Specifically, it performs the following processes:

[1077] 1. User receives message

[1078] The user enters and sends a message to the official account.

[1079] The client (smartphone app) sends this message to the server.

[1080] 2. Message Preprocessing

[1081] The server tokenizes the received message, removing unnecessary spaces and special characters.

[1082] 3. Generating a Response

[1083] The server generates a response based on the preprocessed message using a generative AI model (e.g., OpenAI's GPT-4).

[1084] 4. Applying Styles

[1085] The generated responses are styled to suit specific characters or famous people, adjusting the writing style and speaking style, using a template-based approach.

[1086] 5. Sending the Response

[1087] The final generated response is sent to the user's terminal to continue the dialogue.

[1088] 6. Saving conversation history

[1089] The server stores the interaction history and references it the next time the interaction occurs, and also collects user feedback to continuously improve the generative AI model.

[1090] Hardware and software used

[1091] Hardware:

[1092] Server: Virtual server on Amazon Web Services (AWS)

[1093] Client: iOS or Android device (smartphone)

[1094] software:

[1095] Database: Uses Amazon RDS to store interaction history and feedback

[1096] Generative AI model: OpenAI's GPT-4

[1097] Preprocessing: A natural language processing library such as Python's NLTK or spaCy

[1098] Client app: Build the UI using Flutter

[1099] Examples of specific examples and prompts

[1100] For example, if a user sends the message "Tell me about your recent projects," the system generates a response as follows:

[1101] 1. Receive and preprocess the user's message:

[1102] The server receives the message "Tell us about your recent projects" and tokenizes and cleans it.

[1103] 2. Response generation using generative AI models:

[1104] The server generates a response using the following prompt sentence:

[1105] Prompt: "Respond in official character style: Tell me about your latest project."

[1106] The generative AI model (GPT-4) generates the response, "Of course, your recent projects have been very exciting. I'm working on a new album."

[1107] 3. Applying styles:

[1108] Tailor this response to the speech style of a particular character.

[1109] For example, "Of course, my latest projects have been very exciting! I'm working on a new album."

[1110] 4. Sending the response:

[1111] The server sends the tailored response to the user's device, which the client app displays.

[1112] This system not only allows users to enjoy natural and consistent interactions with characters and famous people, but it is also expected that the system itself will be continuously improved based on user feedback.

[1113] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1114] Step 1:

[1115] A user sends a message to the official account.

[1116] Input: The message the user types into the text box (e.g., "Tell me about your recent projects")

[1117] How it works: The user enters a message and presses the send button. The client (smartphone app) sends this message to the server.

[1118] Output: Message sent to the server

[1119] Step 2:

[1120] Preprocessing of messages received by the server.

[1121] Input: The message sent to the server

[1122] How it works: The server tokenizes the received message, removing unnecessary spaces and special characters, for example using Python's NLTK library or spaCy.

[1123] Output: Tokenized and cleaned message

[1124] Step 3:

[1125] The server generates a response based on the preprocessed message using a generative AI model.

[1126] Input: Preprocessed message

[1127] Operation: The server takes the preprocessed message as input to the generative AI model (GPT-4), generates a prompt sentence, and sends it to the generative AI model.

[1128] Sample prompt: "Respond in the style of your official character: Tell me about your latest project."

[1129] Output: The response generated by the generative AI model (e.g., "Of course, your recent projects have been very exciting. You're working on a new album.")

[1130] Step 4:

[1131] The server applies the style of a character or celebrity to the generated response.

[1132] Input: The response generated by the generative AI model

[1133] How it works: The server adapts the generated responses to suit the style of a particular character or celebrity, adjusting the writing style and speaking style. It uses a template-based approach to adapt the phrasing and writing style to suit the specific style.

[1134] Output: The styled response (e.g., "Of course, my latest projects have been very exciting! I'm working on a new album.")

[1135] Step 5:

[1136] The server finally sends the generated response to the user's terminal.

[1137] Input: The styled response

[1138] How it works: The server sends the styled response to the user's device, where the client app displays it.

[1139] Output: The response displayed on the user's terminal

[1140] Step 6:

[1141] The server saves the conversation history and uses it for the next conversation.

[1142] Input: The user's message and the generated response

[1143] How it works: The server saves the conversation history in a database such as Amazon RDS, and references it during subsequent conversations.

[1144] Output: Saved interaction history

[1145] Step 7:

[1146] The server collects user feedback and improves the generative AI model.

[1147] Input: User feedback

[1148] How it works: The server collects user feedback through the client app and uses it to improve the generative AI model, incorporating the feedback as training data for the model.

[1149] Output: An improved generative AI model

[1150] Furthermore, 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.

[1151] Regarding an embodiment of the present invention, the specific operation and processing flow of a dialogue system between official accounts and users using a generative AI model and an emotion engine will be described.

[1152] overview

[1153] The present invention is a system that allows users to communicate naturally and proactively with the official accounts of characters and celebrities, and also recognizes the user's emotions and provides appropriate responses. The system receives messages from users, generates responses using a generative AI model and an emotion engine, and sends them in a style that matches the specific character or celebrity's style. It also saves the user's emotion history and uses it for the next conversation.

[1154] Program processing overview

[1155] 1. Receive a user's message

[1156] User: The user types and sends a message to the official account. For example, they might type, "Hello, tell me about your recent projects."

[1157] Terminal: Sends the entered message to the server.

[1158] 2. Message Reception and Preprocessing

[1159] Server: Receives the message sent from the device. For example, get the data user_message = "Hello, tell us about your recent projects."

[1160] Server: Stores the received message in a database. Message data is saved using an SQL INSERT query.

[1161] Server: Preprocesses the stored messages by tokenizing them and removing unnecessary spaces and special characters. For example, "Hello, tell me about your recent projects" is tokenized as "Hello, tell me about your recent projects".

[1162] 3. Emotion recognition

[1163] Server: Inputs the preprocessed message into the emotion engine to recognize the user's emotion. The emotion engine extracts emotions such as positive, negative, and neutral from the message.

[1164] Server: Adjusts the input to the generative AI model based on the recognized emotion, for example, making the response more polite and supportive if the user has a negative emotion.

[1165] 4. Response Generation

[1166] Server: Using the trained data, the generative AI model generates a response to the user's message. For example, the generative AI model generates a response like, "Of course, your recent projects have been very exciting. I'm working on a new album."

[1167] Server: Style the generated response with a character or celebrity, transforming it into, "Of course, my latest projects have been very exciting! I'm working on a new album."

[1168] Server: Additionally, tailor the response based on the emotion engine's recognized emotion. For example, if the user is expressing negative emotion, provide additional encouragement such as, "Don't worry, the new album is going to be amazing!"

[1169] 5. Sending the Response

[1170] Server: Sends the final generated response to the user's terminal, as well as sending data according to the format.

[1171] Terminal: The user receives a response, for example, a message displayed on the screen saying, "Of course, your latest project is very exciting! Don't worry, the new album is going to be amazing!"

[1172] 6. Saving user emotion history

[1173] Server: The server stores the recognized user emotions and dialogue content in a database. This emotion history is referenced the next time a dialogue is generated.

[1174] 7. Refer to emotion history when generating the next response

[1175] Server: The next time the user sends a message, it references the user's past emotional history and provides a consistent dialogue.

[1176] Specific examples

[1177] For example, consider the case where a user sends the following message:

[1178] User: "I've been feeling a bit down lately. Could you tell me something to cheer me up?"

[1179] 1. Receive a user message

[1180] The server receives and stores this message.

[1181] 2. Message Preprocessing

[1182] The server parses and tokenizes the received message, removing unnecessary spaces and special characters.

[1183] 3. Emotion recognition

[1184] The server uses an emotion engine to recognize the user's emotion, in this case determining that the message indicates a negative emotion.

[1185] 4. Response Generation

[1186] The server then feeds the adjusted data into a generative AI model to generate a response, such as "Cheer up! I think your new project is very exciting."

[1187] The server applies the style of characters and famous people to the generated responses, and further adjusts them based on emotions recognized by the emotion engine: "Of course, cheer up! I think your new project will get you very excited! You have a great future ahead of you."

[1188] 5. Sending the Response

[1189] The server sends this adjusted response to the user's terminal.

[1190] The terminal displays this response to the user.

[1191] This process allows users to continue interacting with characters and celebrities in a natural, engaging and emotionally relevant way, and emotion history is stored and used to further personalize future responses.

[1192] The processing flow will be explained below.

[1193] Step 1:

[1194] User: Enter a message for the official account and tap the send button. For example, you could type, "Hello, tell me about your recent projects."

[1195] Step 2:

[1196] Terminal: Sends the message entered by the user to the server, where it is converted into an appropriate format (e.g., JSON).

[1197] Step 3:

[1198] Server: Receives the message sent from the device. For example, get the data user_message = "Hello, tell us about your recent projects."

[1199] Step 4:

[1200] Server: Stores the received message in a database. Message data is saved using an SQL INSERT query.

[1201] Step 5:

[1202] Server: Preprocesses the stored messages by tokenizing them and removing unnecessary spaces and special characters. For example, "Hello, tell me about your recent projects" is tokenized as "Hello, tell me about your recent projects".

[1203] Step 6:

[1204] Server: Inputs the preprocessed message into the emotion engine to recognize the user's emotion. The emotion engine extracts emotions such as positive, negative, and neutral from the message.

[1205] Step 7:

[1206] Server: The emotion engine receives the results of recognizing the user's emotion and stores them in the database. For example, if the emotion is recognized as "positive," it will be saved in the database.

[1207] Step 8:

[1208] Server: Adjusts the input to the generative AI model based on the recognized emotion. For example, if the user has a negative emotion, the response will be more polite and supportive. For example, if the emotion is recognized as "negative," the prompt to the generative AI model will be adjusted to "The user has a negative emotion. Please generate a polite and kind response."

[1209] Step 9:

[1210] Server: Using the trained data, the generative AI model generates a response to the user's message. For example, the generative AI model generates a response like, "Of course, your recent projects have been very exciting. I'm working on a new album."

[1211] Step 10:

[1212] Server: Apply the style of a character or celebrity to the generated response, for example, "Of course, my latest projects have been very exciting! I'm working on a new album."

[1213] Step 11:

[1214] Server: Additionally, tailor the response based on the emotion engine's perceived emotion. For example, if the user is expressing "negative" emotion, provide additional encouragement such as "Don't worry, the new album is going to be amazing!"

[1215] Step 12:

[1216] Server: Sends the final generated response to the user's terminal, as well as sending data according to the format.

[1217] Step 13:

[1218] Terminal: The user receives a response, for example, a message displayed on the screen saying, "Of course, your latest project is very exciting! Don't worry, the new album is going to be amazing!"

[1219] Step 14:

[1220] User: Views the response received and types the next message. This process continues the conversation. For example, the user next types, "When is the album coming out?"

[1221] Step 15:

[1222] Server: The server stores the recognized user emotions and dialogue content in a database. This emotion history is referenced the next time a dialogue is generated.

[1223] Step 16:

[1224] Server: The next time the user sends a message, the server refers to the previous emotion history and provides a consistent dialogue. For example, if the emotion was "negative," the server adjusts the response to continue being supportive.

[1225] The above is the specific processing flow in the system.

[1226] Example 2

[1227] Next, a description will be given of 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."

[1228] Conventional dialogue systems have suffered from mechanical dialogue with users, lack of consistency, and insufficient emotional response. It has also been difficult to naturally incorporate the style of a specific character or famous person into responses. Furthermore, the user's emotional history has not been utilized in the next dialogue, resulting in insufficient dialogue personalization. There is a need to solve these issues and realize more natural and emotionally responsive dialogue.

[1229] 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.

[1230] In this invention, the server includes means for receiving a message from a user, means for preprocessing the received message, means for recognizing the user's emotion based on the preprocessed message using an emotion engine, means for generating a response using a generative AI model based on the recognized emotion, means for applying a style of a character or famous person to the generated response and adjusting the response content based on the emotion, means for sending the response to the user, means for storing the response and the user's emotion history in a database, and means for generating a response by referring to the emotion history for the next interaction. This makes interactions with the user more natural and emotionally sensitive, and enables consistent and personalized responses incorporating the style of a specific character or famous person.

[1231] "User" refers to a person who interacts with the system.

[1232] "Server" refers to a computer system that receives and processes messages from users.

[1233] "Terminal" refers to the device that a user uses to enter and send messages to a server.

[1234] "Message" refers to text information that a user enters and sends to the system.

[1235] "Preprocessing" refers to analyzing the received message and performing processes such as tokenization and removing unnecessary spaces.

[1236] "Emotion engine" refers to a system that extracts emotions from users' messages.

[1237] A "generative AI model" refers to a system that uses artificial intelligence to automatically generate responses to user messages.

[1238] "Response" refers to a reply to a user's message created by a generative AI model.

[1239] "Character or celebrity style" refers to writing that imitates the speaking or writing style of a particular character or celebrity.

[1240] "Emotion history" refers to a record of emotional information from past interactions with a user.

[1241] "Database" refers to a collection of data that a system uses to store and manage information.

[1242] This invention is a system that uses a generative AI model and an emotion engine to enable dialogue between official accounts and users. The system is configured as follows, with specific operations and processing flows:

[1243] Basic system configuration

[1244] The system consists of a device used by the user, a server that processes messages, a database that stores data, a generative AI model, and an emotion engine.

[1245] Specific examples of hardware and software used

[1246] Devices: smartphones, computers, etc.

[1247] Server: Cloud server, dedicated server

[1248] Database: MySQL, PostgreSQL

[1249] Generative AI model: GPT-3 (OpenAI)

[1250] Emotion engine: VADER (Valence Aware Dictionary and sEntiment Reasoner)

[1251] Specific data processing and calculation

[1252] 1. User receives message

[1253] The user uses the device to send a message to the official account, for example, by typing "Hello, tell me about your recent projects." This message is then sent from the device to the server.

[1254] 2. Message Preprocessing

[1255] The server receives messages sent from the device, stores them in a database, and then pre-processes them by tokenizing them and removing unnecessary spaces and special characters.

[1256] 3. Emotion recognition

[1257] The server inputs the preprocessed messages into the emotion engine to recognize the user's emotions, which extracts emotions such as positive, negative, and neutral from the messages.

[1258] 4. Response Generation

[1259] The server then feeds data into a generative AI model based on the recognized emotions to generate a response to the user's message, adapting the style of a specific character or celebrity and further adjusting the response content based on the emotions recognized by the emotion engine.

[1260] 5. Sending the Response

[1261] The server finally sends the generated response to the user's terminal, where the user receives the response.

[1262] 6. Emotion history storage

[1263] The server stores the recognized emotions and dialogue content in a database, which allows the server to refer to the emotion history the next time a dialogue is generated, enabling a consistent dialogue.

[1264] Specific examples

[1265] For example, consider the case where a user sends the following message:

[1266] User: "I've been feeling a bit down lately. Could you tell me something to cheer me up?"

[1267] 1. User receives message

[1268] The server receives this message and stores it in a database.

[1269] 2. Message Preprocessing

[1270] The server parses and tokenizes the received message, removing unnecessary spaces and special characters.

[1271] 3. Emotion recognition

[1272] The server uses an emotion engine to recognize the user's emotion, in this case determining that the message indicates a negative emotion.

[1273] 4. Response Generation

[1274] The server then inputs the adjusted data into a generative AI model, generating a response such as, "Cheer up! I think your new project is very exciting."

[1275] You can also style this response as a character or celebrity, adjusting it to say, "Of course, cheer up! I think your new project is super exciting! You have a great future ahead of you."

[1276] 5. Sending the Response

[1277] The server sends this adjusted response to the user's terminal.

[1278] The terminal displays this response to the user.

[1279] In this way, users can continue to have natural, engaging, and emotionally relevant interactions with characters and celebrities, and emotion history is stored and used to further personalize future responses.

[1280] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1281] Step 1:

[1282] User receives message

[1283] The user types a message to the official account saying, "Hello, tell me about your recent projects" and sends it.

[1284] The terminal sends the message entered by the user to the server.

[1285] Input: The message entered by the user.

[1286] Output: The message sent to the server.

[1287] Step 2:

[1288] Receiving and saving messages

[1289] The server receives the message sent from the device, for example, user_message = "Hello, tell us about your recent projects."

[1290] The server stores this received message in a database using an SQL INSERT query:

[1291] sql

[1292] INSERT INTO messages (user_id, message) VALUES (1, "Hello, tell us about your recent projects");

[1293] Input: Message received from the terminal.

[1294] Output: The message stored in the database.

[1295] Step 3:

[1296] Message Preprocessing

[1297] The server preprocesses the stored messages by tokenizing them and removing unnecessary spaces and special characters. It uses Python's NLTK library to tokenize "Hello, tell me about your recent projects" into "Hello, tell me about your recent projects."

[1298] Input: The stored messages retrieved from the database.

[1299] Output: The tokenized message.

[1300] Step 4:

[1301] emotion recognition

[1302] The server inputs the preprocessed messages into the emotion engine to recognize the user's emotions. It uses VADER (Valence Aware Dictionary and sEntiment Reasoner) to extract emotions such as positive, negative, and neutral.

[1303] Input: The preprocessed tokenized message.

[1304] Output: User sentiment score.

[1305] Step 5:

[1306] Response Generation

[1307] Based on the recognized sentiment score, the server generates a response to the user's message using a generative AI model, for example, the GPT-3 model. The generative AI model generates a response such as, "Of course, your recent projects are very exciting. I'm working on a new album."

[1308] Input: Sentiment scores and preprocessed tokenized messages.

[1309] Output: The generated response.

[1310] Step 6:

[1311] Applying and adjusting styles

[1312] The server styles the generated response after a specific character or celebrity, such as "Of course, my latest projects have been very exciting! I'm working on a new album."

[1313] It also tailors its responses based on emotions recognized by its emotion engine: if the user is expressing negative emotions, it offers additional encouragement, such as, "Don't worry, the new album is going to be amazing!"

[1314] Input: Generated responses and sentiment scores.

[1315] Output: The styled and adjusted response.

[1316] Step 7:

[1317] Sending a Response

[1318] The server finally sends the adjusted response to the user's terminal.

[1319] The device will display this response to the user, for example, a message like "Of course, my latest projects are very exciting! Don't worry, the new album is going to be amazing!" will appear on the screen.

[1320] Input: The adjusted response.

[1321] Output: The response displayed on the terminal.

[1322] Step 8:

[1323] Emotion history storage

[1324] The server stores the recognized user emotions and the content of the conversation in a database, creating an emotion history that can be referenced the next time a conversation is generated.

[1325] For example, using a SQL query like this:

[1326] sql

[1327] INSERT INTO emotions (user_id, message_id, sentiment) VALUES (1, LAST_INSERT_ID(), "negative");

[1328] Input: User emotion data and dialogue content.

[1329] Output: Emotion history stored in a database.

[1330] Step 9:

[1331] Refer to emotion history when generating the next response

[1332] The next time the user sends a message, the server references the user's past emotion history and generates a new response based on that history to provide a consistent interaction.

[1333] For example, the next time a user sends a message, we get the sentiment history like this:

[1334] sql

[1335] SELECT sentiment FROM emotions WHERE user_id = 1 ORDER BY timestamp DESC LIMIT 1;

[1336] Input: Past emotional history.

[1337] Output: Consistent response.

[1338] (Application example 2)

[1339] Next, a description will be given of Application 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."

[1340] Conventional dialogue systems only have the ability to return a fixed response to messages sent by the user, and have difficulty recognizing the user's emotions and generating a response accordingly. Furthermore, they are unable to apply a specific context to the generated response or utilize the user's emotional history for the next dialogue, making it difficult to realize personalized communication.

[1341] The specific processing by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for receiving a message from a user, means for preprocessing the received message, means for generating a response using a generative AI model based on the preprocessed message, means for applying a specific context to the generated response, means for sending the response to the user, means for recognizing the user's emotion and generating a response according to the emotion, means for saving the preprocessed message and the recognized emotion in a database, and means for referring to the saved emotion history in the next dialogue. This makes it possible to generate a personalized response that takes the user's emotion into consideration and achieve continuous and consistent communication.

[1342] The "means for receiving a message from a user" is a mechanism for transferring a text message entered by a user to a server via a network.

[1343] The "means for preprocessing received messages" refers to functions for formatting messages to make them easier to parse and process, such as tokenizing messages and removing unnecessary spaces and special characters.

[1344] "Means for generating a response using a generative AI model based on a preprocessed message" refers to a mechanism that uses preprocessed message data as input and creates an appropriate response using a generative AI model.

[1345] "Means for applying specific context to generated responses" refers to the ability to add a specific character or contextual style to generated responses, making the responses presented to the user seem natural and familiar.

[1346] The "means for sending a response to the user" is a mechanism for transferring the generated response message to the user's terminal and displaying it on the screen.

[1347] The "means for recognizing the user's emotions and generating a response according to those emotions" is a function for detecting emotions from the user's message and generating a response with content and tone according to those emotions.

[1348] The "means for storing the preprocessed message and the recognized emotion in a database" is a mechanism for recording and storing the user's message and the emotion recognized from the message in a database.

[1349] The "means for referring to the emotion history stored in the next interaction" is a function for referring to the emotion history of the user stored in the past and utilizing it in the next interaction.

[1350] overview

[1351] The present invention is a dialogue system that generates personalized responses based on user messages. The system receives user messages, recognizes emotions, and generates responses based on those emotions using a generative AI model and an emotion engine. Furthermore, the system stores the user's emotion history and utilizes it in the next dialogue, providing consistent and personalized dialogue.

[1352] Hardware and Software Configuration

[1353] Hardware

[1354] The system hardware configuration is as follows:

[1355] 1. User terminal: A mobile device with communication capabilities, such as a smartphone or tablet.

[1356] 2. Server: A cloud server or dedicated server that processes data and generates responses.

[1357] software

[1358] The software used is as follows:

[1359] 1. Generative AI models: Natural language processing models such as the OpenAI API.

[1360] 2. Emotion engine: Emotion recognition algorithms (e.g., Sentiment Analysis API).

[1361] 3. Database management system: such as SQLite.

[1362] Program processing overview and specific examples

[1363] Message Preprocessing

[1364] The server pre-processes messages received from user devices by tokenizing them and removing unnecessary spaces and special characters. For example, a message such as "I'm very unhappy because my order hasn't arrived yet" is converted to "I'm very unhappy because my order hasn't arrived yet."

[1365] emotion recognition

[1366] The server inputs the preprocessed message into the emotion engine, which recognizes emotions such as positive, negative, and neutral. The emotion engine's algorithm extracts the user's emotion from the message content. For example, the above message is recognized as "negative."

[1367] Response Generation

[1368] The server inputs a prompt sentence into the generative AI model based on the emotion recognition results, and generates a response. For example, the following prompt sentence is input into the generative AI model:

[1369] User message: I'm very frustrated because my order hasn't arrived yet.

[1370] Emotion: Negative

[1371] Generate the corresponding response.

[1372] The generative AI model generates a natural response based on this prompt, such as, "We're sorry. We'll look into this immediately and get back to you as soon as possible."

[1373] Customizing and sending responses

[1374] The generated response is adapted to the specific context and style and sent to the user, where the server can amend or modify the generated response to present it in a user-friendly format.

[1375] Save user emotion history for future reference

[1376] The emotion recognition results and messages are stored in a database. The next time you interact with the system, the system will refer to the stored emotion history to provide a consistent and personalized response, allowing users to enjoy continuous and personalized communication.

[1377] In this way, the present invention enables natural dialogue according to the user's emotions, and realizes personalized support.

[1378] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1379] Step 1:

[1380] The server receives a message from the user terminal, for example, a user input such as "I'm very unhappy because my order hasn't arrived yet." The server receives this message and stores it in a database for further processing.

[1381] Step 2:

[1382] The server preprocesses the received message by tokenizing it and removing unnecessary spaces and special characters. For example, a message like "I'm very unhappy because my order hasn't arrived yet" is converted to "I'm very unhappy because my order hasn't arrived yet." This preprocessed message becomes the input data for the next emotion recognition.

[1383] Step 3:

[1384] The server inputs the preprocessed message into the emotion engine to recognize the user's emotion. The emotion engine extracts emotions such as positive, negative, and neutral from the input message. For example, for the preprocessed message "I'm very unhappy because my order hasn't arrived yet," the emotion engine outputs "negative."

[1385] Step 4:

[1386] The server inputs a prompt sentence into the generative AI model based on the emotion recognition results, and generates a response. The prompt sentence is set as follows:

[1387] User message: I'm very frustrated because my order hasn't arrived yet.

[1388] Emotion: Negative

[1389] Generate the corresponding response.

[1390] The generative AI model will then generate a natural response based on this prompt, such as "We apologize for the inconvenience. We will investigate this immediately and provide a prompt response."

[1391] Step 5:

[1392] The server applies specific context and style to the generated response, such as adding a more friendly "We're sorry, we'll look into this and get back to you as soon as possible" response, and customizes it as needed using templates and modules.

[1393] Step 6:

[1394] The server then sends the final response to the user's device. The user's device then displays the received response message on its screen. For example, the user may see a message such as, "We apologize for the inconvenience. We will investigate the issue immediately and respond as soon as possible."

[1395] Step 7:

[1396] The server stores the user's message and the emotion recognition results in a database. The stored data is referenced the next time a conversation occurs, allowing for consistent and personalized responses to be generated based on the user's past emotion history.

[1397] The specific processing unit 290 transmits the result of the specific processing to the headset type terminal 314. In the headset type terminal 314, the control unit 46A causes the speaker 240 and the display 343 to output the result of the specific processing. The microphone 238 acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio 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 audio data.

[1398] 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> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with 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.

[1399] 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 headset type terminal 314.

[1400] [Fourth embodiment]

[1401] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1402] 7, a data processing system 410 includes a data processing device 12 and a robot 414. An example of the data processing device 12 is a server.

[1403] 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 WAN (Wide Area Network) and / or a LAN (Local Area Network).

[1404] The robot 414 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a control target 443. 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 control target 443 are also connected to the bus 52.

[1405] 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 audio in accordance with instructions from the processor 46.

[1406] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and 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 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[1407] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 control the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[1408] The control object 443 includes a display device, LEDs in the eyes, and motors for driving the arms, hands, and feet. The posture and gestures of the robot 414 are controlled by controlling the motors of the arms, hands, and feet. Some of the emotions of the robot 414 can be expressed by controlling these motors. In addition, the facial expressions of the robot 414 can also be expressed by controlling the light emission state of the LEDs in the eyes of the robot 414.

[1409] Fig. 8 shows an example of the main functions of the data processing device 12 and the robot 414. As shown in Fig. 8, 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.

[1410] 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.

[1411] 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.

[1412] In the robot 414, the processor 46 performs the reception output process. 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.

[1413] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1414] Regarding an embodiment of the present invention, the specific operation and processing flow of a dialogue system between official accounts and users using a generative AI model will be described.

[1415] overview

[1416] The present invention is a system that allows users to communicate naturally and proactively with the official accounts of characters and famous people. The system receives messages from users, generates responses using a generative AI model, and sends them in a style that matches the style of the specific character or famous person.

[1417] Program processing overview

[1418] 1. Receive a user's message

[1419] User: The user types and sends a message to the official account.

[1420] Terminal: Sends the entered message to the server.

[1421] 2. Message Reception and Preprocessing

[1422] Server: Receives messages sent from the terminal.

[1423] Server: Pre-processes the received message by tokenizing it and removing unnecessary spaces and special characters.

[1424] 3. Response Generation

[1425] Server: Generates a response to the user's message using a generative AI model based on the preprocessed message.

[1426] Server: Apply the style of a character or celebrity to the generated response, adjusting the writing style and speaking style.

[1427] 4. Sending the Response

[1428] Server: Sends the final generated response to the user's terminal.

[1429] Terminal: The user receives the response and the dialogue continues.

[1430] Specific examples

[1431] For example, consider the case where a user sends the following message:

[1432] User: "Tell me about your latest project."

[1433] When this message is sent, the following process occurs:

[1434] 1. Receive a user message

[1435] The server receives and stores this message.

[1436] 2. Message Preprocessing

[1437] The server parses and tokenizes the received message, removing unnecessary spaces and special characters.

[1438] 3. Response Generation

[1439] The server inputs the preprocessed messages into a generative AI model to generate a response.

[1440] For example, a generative AI model might generate a response like, "Of course, your recent projects have been very exciting. You're working on a new album."

[1441] 4. Applying Styles

[1442] The server then styles the generated response after a specific character or celebrity, adjusting it to, "Of course, my latest projects have been very exciting! I'm working on a new album."

[1443] 5. Sending the Response

[1444] The server sends this adjusted response to the user's terminal.

[1445] The terminal displays this response to the user.

[1446] This process allows users to continue natural and engaging interactions with characters and famous people.

[1447] The system also saves the user's interaction history and references it the next time it generates a response, providing a more consistent interaction. Furthermore, by collecting user feedback and periodically updating the generative AI model, the system's accuracy and user satisfaction can be improved.

[1448] The processing flow will be explained below.

[1449] Step 1:

[1450] User: Enter a message for the official account and tap the send button. For example, you could type, "Hello, tell me about your recent projects."

[1451] Step 2:

[1452] Terminal: Sends the message entered by the user to the server, where it is converted into an appropriate format (e.g., JSON).

[1453] Step 3:

[1454] Server: Receives the message sent from the device. For example, get the data user_message = "Hello, tell us about your recent projects."

[1455] Step 4:

[1456] Server: Stores the received message in a database. Message data is saved using an SQL INSERT query.

[1457] Step 5:

[1458] Server: Preprocesses the stored messages by tokenizing them and removing unnecessary spaces and special characters. For example, "Hello, tell me about your recent projects" is tokenized as "Hello, tell me about your recent projects".

[1459] Step 6:

[1460] Server: Feeds the pre-processed messages into a generative AI model, which is pre-trained to mimic the speaking and writing style of a specific character or famous figure.

[1461] Step 7:

[1462] Server: The generative AI model generates an appropriate response to the user's message. For example, the generative AI model generates a response like, "Of course, your recent projects are very exciting. I'm working on a new album."

[1463] Step 8:

[1464] Server: Apply the style of a character or famous person to the generated response. This style adjustment affects the writing style and speaking style. For example, "Of course, my latest projects have been very exciting! I'm working on a new album."

[1465] Step 9:

[1466] Server: Sends the final generated response to the user's terminal, as well as sending data according to the format.

[1467] Step 10:

[1468] Terminal: The user receives a response, for example, a message displayed on the screen saying, "Of course, my latest projects have been very exciting! I'm working on a new album."

[1469] Step 11:

[1470] User: Views the response received and types the next message. This process continues the conversation. For example, the user next types, "When is the album coming out?"

[1471] Step 12:

[1472] Server: Saves the dialogue history and references past dialogues when generating the next response, thereby maintaining a consistent conversation.

[1473] The above is the specific processing flow in the system.

[1474] Example 1

[1475] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1476] In modern society, there is a growing demand for users to have natural-looking conversations with specific characters or famous people. However, existing dialogue systems have difficulty accurately imitating the speaking style and writing style of characters or famous people, which can lead to a poor user experience. Furthermore, there are technical challenges in maintaining consistency and naturalness in responses. Therefore, there is a need to develop a system that provides natural and engaging conversations based on the style of specific characters or famous people.

[1477] 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.

[1478] In this invention, the server includes means for receiving a message from a user, means for tokenizing the received message and preprocessing it to remove unnecessary spaces and special characters, means for generating a response using a generative AI model based on the preprocessed message, means for applying a style of a specific character or famous person to the generated response to adjust the writing style and speaking manner, and means for sending the adjusted response to the user, thereby enabling the user to have a natural and engaging conversation with the specific character or famous person.

[1479] "User" refers to a person who types and sends messages to interact with the system.

[1480] "Message" refers to text data that a user enters and sends to interact with the system.

[1481] "Tokenization" refers to the process of breaking down an incoming message into individual words or phrases.

[1482] "Preprocessing to remove unnecessary spaces and special characters" refers to the process of removing unnecessary spaces and specific symbols from the message to make it easier to analyze.

[1483] A "generative AI model" refers to an algorithm or software that generates natural language responses based on input text data.

[1484] "Response" refers to the text data of a reply generated by a generative AI model.

[1485] "Character or celebrity style" refers to the style or characteristics used to imitate the speech, writing style, or language of a particular character or celebrity.

[1486] "Tailoring" refers to the process of changing the writing style or speaking style to adapt the generated response to a particular style.

[1487] "Sending" refers to the act of sending the response generated by the server to the user's terminal as data.

[1488] This invention is a system that allows users to have natural conversations with the official accounts of specific characters or famous people. The system receives a message from the user, and a generative AI model generates a response based on the message. The generated response is adapted to the style of the specific character or famous person and returned to the user. To realize this process, the following steps are taken:

[1489] The server has a means for receiving messages from users. Users input and send messages to their official accounts. The terminal then sends these messages to the server. Specifically, the user input is passed to the server using a communication protocol such as an HTTP POST request.

[1490] The server parses the received message and performs preprocessing by tokenizing it and removing unnecessary spaces and special characters. The methods used here are, for example, Python's regular expression library or natural language processing library. For example, the message "Hello! Tell me about your recent projects." is split into tokens such as "Hello," "Recent," "Project," "About," "Tell me," and "Please."

[1491] Next, the server inputs the preprocessed message into a generative AI model to generate a response. For example, a large-scale language model such as GPT-3 is used as the generative AI model. The generative AI model receives the following input:

[1492] User: Hi! Tell us about your latest projects.

[1493] Character: Album Producer

[1494] Based on this input, the generative AI model generates an appropriate response, such as, "Of course, my recent projects have been very exciting. I'm working on a new album."

[1495] Apply the style of a particular character or celebrity to the generated response. The server does this by adjusting the writing style and word choice, for example, "Of course, my latest projects have been very exciting! I'm working on a new album."

[1496] The final adjusted response is sent to the user's device. The server packages the generated response in JSON format or similar and sends it to the user's device. The device receives it and displays it on the chat screen. Specifically, HTML and JavaScript are used to dynamically update the part that displays the message.

[1497] For example, if a user sends a message saying, "Tell me about your latest projects," the process will ultimately result in a response being returned to the user saying, "Of course, my latest projects are very exciting! I'm working on a new album."

[1498] The system also stores the user's interaction history and references it the next time it generates a response, allowing it to provide more consistent interactions. It also collects user feedback and periodically updates the generative AI model to improve system accuracy and user satisfaction.

[1499] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1500] Step 1:

[1501] The user enters and sends a message. Specifically, the user enters a text message in the chat box of the official account and clicks the send button. This operation causes the device to obtain the user's input and generate message data. The input is text data such as "Hello! Tell us about your recent projects." The output is that this text data is sent to the server via the device.

[1502] Step 2:

[1503] The device receives a message from the user and sends it to the server. The device packages the user's input as an HTTP POST request and sends it to the server's API endpoint. The input is the user's text message, for example, "Hello! Tell us about your latest project." The output is that this text message reaches the server.

[1504] Step 3:

[1505] The server receives messages sent from the terminal and stores them in a database. It then performs a process to insert the received messages into a table. Specifically, it stores the received data in the appropriate database table, along with metadata such as the message ID and timestamp. The input is the text data received from the terminal. The output is the state in which this data has been stored in the database.

[1506] Step 4:

[1507] The server performs preprocessing by tokenizing the received message and removing unnecessary spaces and special characters. The server uses regular expressions and natural language processing libraries to tokenize the message and remove unnecessary spaces and special characters. The input is the stored text message. The output is tokenized clean text data. For example, the message "Hello! Tell me about your recent projects." is split into "Hello", "Recent", "Project", "About", "Tell me", and "Please".

[1508] Step 5:

[1509] The server inputs the preprocessed message into a generative AI model to generate a response. A generative AI model (e.g., GPT-3) is used to generate natural-sounding responses based on the preprocessed message. The input is tokenized text data. The output is the generated response text. For example, suppose the input prompt is:

[1510] "User: Hi! Tell me about your latest project.

[1511] Character: Album Producer

[1512] The output from the generative AI model is, "Of course, your recent projects have been very exciting. You're working on a new album."

[1513] Step 6:

[1514] The server applies the style of a particular character or famous person to the generated response, adjusting the writing style and speaking style. The server adjusts the writing style and speaking style of the generated text based on the style of a particular character or famous person. For example, "Of course, my latest projects are very exciting! I'm working on a new album." The input is the generated response text. The output is the styled response text.

[1515] Step 7:

[1516] The server finally generates a response and sends it to the user's device. The server wraps the response text in JSON format or similar and sends it to the device as an HTTP response. The input is the response text after applying the style. The output is the response data sent to the device.

[1517] Step 8:

[1518] The device displays the response received from the server to the user. The device analyzes the received data and displays the response on the chat screen. Specifically, it updates the chat screen using HTML and JavaScript. The input is the response data received from the server. The output is the response text displayed on the user's chat screen. For example, the user's screen might display, "Of course, my recent projects have been very exciting! I'm working on a new album."

[1519] (Application example 1)

[1520] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1521] The present invention relates to a system for improving engagement while maintaining consistency and naturalness of dialogue when users communicate naturally and proactively with the official accounts of characters and famous people. Conventional dialogue systems have difficulty faithfully reproducing the style of characters and famous people, and are unable to fully utilize dialogue history and feedback, limiting user satisfaction. For these reasons, there has been a demand for the development of a dialogue system that is more accurate and can be continuously improved.

[1522] 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.

[1523] In this invention, the server includes means for receiving messages from users, means for preprocessing the received messages, means for generating responses using a generative AI model based on the preprocessed messages, means for applying the style of a character or celebrity to the generated responses, means for sending the responses to the users, means for saving the response history and using it for the next dialogue, and means for collecting feedback from users and improving the generative AI model. This allows users to continue having natural and engaging dialogue with characters or celebrities, improving the consistency and accuracy of the dialogue. Furthermore, by utilizing user feedback, the generative AI model can be continuously improved to provide more accurate dialogue.

[1524] "Means for receiving a message from a user" refers to a device or software for receiving a text message entered by a user.

[1525] A "means for pre-processing received messages" is a device or software that tokenizes messages received from users and removes unnecessary spaces and special characters.

[1526] A "means for generating a response using a generative AI model based on a preprocessed message" is a device or software that uses a preprocessed message as input to generate a response using a generative AI model.

[1527] A "means for applying the style of a character or celebrity to generated responses" is a device or software that tailors the responses generated by a generative AI model to the speaking style or writing style of a particular character or celebrity.

[1528] The "means for sending a response to a user" is a device or software for sending the generated response to the user's terminal.

[1529] The "means for saving response history and using it in the next interaction" refers to a device or software that saves past interaction data between the user and the system and refers to that data during subsequent interactions.

[1530] A "means for collecting user feedback and improving a generative AI model" is a device or software that collects user-provided feedback and uses it as training data for a generative AI model to improve the model's performance.

[1531] "Template-based" refers to a method in which the corresponding response is tailored or generated based on a specific pattern or template.

[1532] A "prompt" is text entered to instruct a generative AI model, providing context for the response you want the model to generate.

[1533] In the embodiment for implementing the present invention, specific programs, processes, and examples of a system that realizes an interaction system between a user and the official account of a specific character or famous person will be described.

[1534] System Program

[1535] The system mainly consists of a client (user terminal) and a server. The client provides a UI for users to input messages, and the server analyzes and processes the received messages, then generates and sends a response. Specifically, it performs the following processes:

[1536] 1. User receives message

[1537] The user enters and sends a message to the official account.

[1538] The client (smartphone app) sends this message to the server.

[1539] 2. Message Preprocessing

[1540] The server tokenizes the received message, removing unnecessary spaces and special characters.

[1541] 3. Generating a Response

[1542] The server generates a response based on the preprocessed message using a generative AI model (e.g., OpenAI's GPT-4).

[1543] 4. Applying Styles

[1544] The generated responses are styled to suit specific characters or famous people, adjusting the writing style and speaking style, using a template-based approach.

[1545] 5. Sending the Response

[1546] The final generated response is sent to the user's terminal to continue the dialogue.

[1547] 6. Saving conversation history

[1548] The server stores the interaction history and references it the next time the interaction occurs, and also collects user feedback to continuously improve the generative AI model.

[1549] Hardware and software used

[1550] Hardware:

[1551] Server: Virtual server on Amazon Web Services (AWS)

[1552] Client: iOS or Android device (smartphone)

[1553] software:

[1554] Database: Uses Amazon RDS to store interaction history and feedback

[1555] Generative AI model: OpenAI's GPT-4

[1556] Preprocessing: A natural language processing library such as Python's NLTK or spaCy

[1557] Client app: Build the UI using Flutter

[1558] Examples of specific examples and prompts

[1559] For example, if a user sends the message "Tell me about your recent projects," the system generates a response as follows:

[1560] 1. Receive and preprocess the user's message:

[1561] The server receives the message "Tell us about your recent projects" and tokenizes and cleans it.

[1562] 2. Response generation using generative AI models:

[1563] The server generates a response using the following prompt sentence:

[1564] Prompt: "Respond in official character style: Tell me about your latest project."

[1565] The generative AI model (GPT-4) generates the response, "Of course, your recent projects have been very exciting. I'm working on a new album."

[1566] 3. Applying styles:

[1567] Tailor this response to the speech style of a particular character.

[1568] For example, "Of course, my latest projects have been very exciting! I'm working on a new album."

[1569] 4. Sending the response:

[1570] The server sends the tailored response to the user's device, which the client app displays.

[1571] This system not only allows users to enjoy natural and consistent interactions with characters and famous people, but it is also expected that the system itself will be continuously improved based on user feedback.

[1572] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1573] Step 1:

[1574] A user sends a message to the official account.

[1575] Input: The message the user types into the text box (e.g., "Tell me about your recent projects")

[1576] How it works: The user enters a message and presses the send button. The client (smartphone app) sends this message to the server.

[1577] Output: Message sent to the server

[1578] Step 2:

[1579] Preprocessing of messages received by the server.

[1580] Input: The message sent to the server

[1581] How it works: The server tokenizes the received message, removing unnecessary spaces and special characters, for example using Python's NLTK library or spaCy.

[1582] Output: Tokenized and cleaned message

[1583] Step 3:

[1584] The server generates a response based on the preprocessed message using a generative AI model.

[1585] Input: Preprocessed message

[1586] Operation: The server takes the preprocessed message as input to the generative AI model (GPT-4), generates a prompt sentence, and sends it to the generative AI model.

[1587] Sample prompt: "Respond in the style of your official character: Tell me about your latest project."

[1588] Output: The response generated by the generative AI model (e.g., "Of course, your recent projects have been very exciting. You're working on a new album.")

[1589] Step 4:

[1590] The server applies the style of a character or celebrity to the generated response.

[1591] Input: The response generated by the generative AI model

[1592] How it works: The server adapts the generated responses to suit the style of a particular character or celebrity, adjusting the writing style and speaking style. It uses a template-based approach to adapt the phrasing and writing style to suit the specific style.

[1593] Output: The styled response (e.g., "Of course, my latest projects have been very exciting! I'm working on a new album.")

[1594] Step 5:

[1595] The server finally sends the generated response to the user's terminal.

[1596] Input: The styled response

[1597] How it works: The server sends the styled response to the user's device, where the client app displays it.

[1598] Output: The response displayed on the user's terminal

[1599] Step 6:

[1600] The server saves the conversation history and uses it for the next conversation.

[1601] Input: The user's message and the generated response

[1602] How it works: The server saves the conversation history in a database such as Amazon RDS, and references it during subsequent conversations.

[1603] Output: Saved interaction history

[1604] Step 7:

[1605] The server collects user feedback and improves the generative AI model.

[1606] Input: User feedback

[1607] How it works: The server collects user feedback through the client app and uses it to improve the generative AI model, incorporating the feedback as training data for the model.

[1608] Output: An improved generative AI model

[1609] Furthermore, 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.

[1610] Regarding an embodiment of the present invention, the specific operation and processing flow of a dialogue system between official accounts and users using a generative AI model and an emotion engine will be described.

[1611] overview

[1612] The present invention is a system that allows users to communicate naturally and proactively with the official accounts of characters and celebrities, and also recognizes the user's emotions and provides appropriate responses. The system receives messages from users, generates responses using a generative AI model and an emotion engine, and sends them in a style that matches the specific character or celebrity's style. It also saves the user's emotion history and uses it for the next conversation.

[1613] Program processing overview

[1614] 1. Receive a user's message

[1615] User: The user types and sends a message to the official account. For example, they might type, "Hello, tell me about your recent projects."

[1616] Terminal: Sends the entered message to the server.

[1617] 2. Message Reception and Preprocessing

[1618] Server: Receives the message sent from the device. For example, get the data user_message = "Hello, tell us about your recent projects."

[1619] Server: Stores the received message in a database. Message data is saved using an SQL INSERT query.

[1620] Server: Preprocesses the stored messages by tokenizing them and removing unnecessary spaces and special characters. For example, "Hello, tell me about your recent projects" is tokenized as "Hello, tell me about your recent projects".

[1621] 3. Emotion recognition

[1622] Server: Inputs the preprocessed message into the emotion engine to recognize the user's emotion. The emotion engine extracts emotions such as positive, negative, and neutral from the message.

[1623] Server: Adjusts the input to the generative AI model based on the recognized emotion, for example, making the response more polite and supportive if the user has a negative emotion.

[1624] 4. Response Generation

[1625] Server: Using the trained data, the generative AI model generates a response to the user's message. For example, the generative AI model generates a response like, "Of course, your recent projects have been very exciting. I'm working on a new album."

[1626] Server: Style the generated response with a character or celebrity, transforming it into, "Of course, my latest projects have been very exciting! I'm working on a new album."

[1627] Server: Additionally, tailor the response based on the emotion engine's recognized emotion. For example, if the user is expressing negative emotion, provide additional encouragement such as, "Don't worry, the new album is going to be amazing!"

[1628] 5. Sending the Response

[1629] Server: Sends the final generated response to the user's terminal, as well as sending data according to the format.

[1630] Terminal: The user receives a response, for example, a message displayed on the screen saying, "Of course, your latest project is very exciting! Don't worry, the new album is going to be amazing!"

[1631] 6. Saving user emotion history

[1632] Server: The server stores the recognized user emotions and dialogue content in a database. This emotion history is referenced the next time a dialogue is generated.

[1633] 7. Refer to emotion history when generating the next response

[1634] Server: The next time the user sends a message, it references the user's past emotional history and provides a consistent dialogue.

[1635] Specific examples

[1636] For example, consider the case where a user sends the following message:

[1637] User: "I've been feeling a bit down lately. Could you tell me something to cheer me up?"

[1638] 1. Receive a user message

[1639] The server receives and stores this message.

[1640] 2. Message Preprocessing

[1641] The server parses and tokenizes the received message, removing unnecessary spaces and special characters.

[1642] 3. Emotion recognition

[1643] The server uses an emotion engine to recognize the user's emotion, in this case determining that the message indicates a negative emotion.

[1644] 4. Response Generation

[1645] The server then feeds the adjusted data into a generative AI model to generate a response, such as "Cheer up! I think your new project is very exciting."

[1646] The server applies the style of characters and famous people to the generated responses, and further adjusts them based on emotions recognized by the emotion engine: "Of course, cheer up! I think your new project will get you very excited! You have a great future ahead of you."

[1647] 5. Sending the Response

[1648] The server sends this adjusted response to the user's terminal.

[1649] The terminal displays this response to the user.

[1650] This process allows users to continue interacting with characters and celebrities in a natural, engaging and emotionally relevant way, and emotion history is stored and used to further personalize future responses.

[1651] The processing flow will be explained below.

[1652] Step 1:

[1653] User: Enter a message for the official account and tap the send button. For example, you could type, "Hello, tell me about your recent projects."

[1654] Step 2:

[1655] Terminal: Sends the message entered by the user to the server, where it is converted into an appropriate format (e.g., JSON).

[1656] Step 3:

[1657] Server: Receives the message sent from the device. For example, get the data user_message = "Hello, tell us about your recent projects."

[1658] Step 4:

[1659] Server: Stores the received message in a database. Message data is saved using an SQL INSERT query.

[1660] Step 5:

[1661] Server: Preprocesses the stored messages by tokenizing them and removing unnecessary spaces and special characters. For example, "Hello, tell me about your recent projects" is tokenized as "Hello, tell me about your recent projects".

[1662] Step 6:

[1663] Server: Inputs the preprocessed message into the emotion engine to recognize the user's emotion. The emotion engine extracts emotions such as positive, negative, and neutral from the message.

[1664] Step 7:

[1665] Server: The emotion engine receives the results of recognizing the user's emotion and stores them in the database. For example, if the emotion is recognized as "positive," it will be saved in the database.

[1666] Step 8:

[1667] Server: Adjusts the input to the generative AI model based on the recognized emotion. For example, if the user has a negative emotion, the response will be more polite and supportive. For example, if the emotion is recognized as "negative," the prompt to the generative AI model will be adjusted to "The user has a negative emotion. Please generate a polite and kind response."

[1668] Step 9:

[1669] Server: Using the trained data, the generative AI model generates a response to the user's message. For example, the generative AI model generates a response like, "Of course, your recent projects have been very exciting. I'm working on a new album."

[1670] Step 10:

[1671] Server: Apply the style of a character or celebrity to the generated response, for example, "Of course, my latest projects have been very exciting! I'm working on a new album."

[1672] Step 11:

[1673] Server: Additionally, tailor the response based on the emotion engine's perceived emotion. For example, if the user is expressing "negative" emotion, provide additional encouragement such as "Don't worry, the new album is going to be amazing!"

[1674] Step 12:

[1675] Server: Sends the final generated response to the user's terminal, as well as sending data according to the format.

[1676] Step 13:

[1677] Terminal: The user receives a response, for example, a message displayed on the screen saying, "Of course, your latest project is very exciting! Don't worry, the new album is going to be amazing!"

[1678] Step 14:

[1679] User: Views the response received and types the next message. This process continues the conversation. For example, the user next types, "When is the album coming out?"

[1680] Step 15:

[1681] Server: The server stores the recognized user emotions and dialogue content in a database. This emotion history is referenced the next time a dialogue is generated.

[1682] Step 16:

[1683] Server: The next time the user sends a message, the server refers to the previous emotion history and provides a consistent dialogue. For example, if the emotion was "negative," the server adjusts the response to continue being supportive.

[1684] The above is the specific processing flow in the system.

[1685] Example 2

[1686] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1687] Conventional dialogue systems have suffered from mechanical dialogue with users, lack of consistency, and insufficient emotional response. It has also been difficult to naturally incorporate the style of a specific character or famous person into responses. Furthermore, the user's emotional history has not been utilized in the next dialogue, resulting in insufficient dialogue personalization. There is a need to solve these issues and realize more natural and emotionally responsive dialogue.

[1688] 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.

[1689] In this invention, the server includes means for receiving a message from a user, means for preprocessing the received message, means for recognizing the user's emotion based on the preprocessed message using an emotion engine, means for generating a response using a generative AI model based on the recognized emotion, means for applying a style of a character or famous person to the generated response and adjusting the response content based on the emotion, means for sending the response to the user, means for storing the response and the user's emotion history in a database, and means for generating a response by referring to the emotion history for the next interaction. This makes interactions with the user more natural and emotionally sensitive, and enables consistent and personalized responses incorporating the style of a specific character or famous person.

[1690] "User" refers to a person who interacts with the system.

[1691] "Server" refers to a computer system that receives and processes messages from users.

[1692] "Terminal" refers to the device that a user uses to enter and send messages to a server.

[1693] "Message" refers to text information that a user enters and sends to the system.

[1694] "Preprocessing" refers to analyzing the received message and performing processes such as tokenization and removing unnecessary spaces.

[1695] "Emotion engine" refers to a system that extracts emotions from users' messages.

[1696] A "generative AI model" refers to a system that uses artificial intelligence to automatically generate responses to user messages.

[1697] "Response" refers to a reply to a user's message created by a generative AI model.

[1698] "Character or celebrity style" refers to writing that imitates the speaking or writing style of a particular character or celebrity.

[1699] "Emotion history" refers to a record of emotional information from past interactions with a user.

[1700] "Database" refers to a collection of data that a system uses to store and manage information.

[1701] This invention is a system that uses a generative AI model and an emotion engine to enable dialogue between official accounts and users. The system is configured as follows, with specific operations and processing flows:

[1702] Basic system configuration

[1703] The system consists of a device used by the user, a server that processes messages, a database that stores data, a generative AI model, and an emotion engine.

[1704] Specific examples of hardware and software used

[1705] Devices: smartphones, computers, etc.

[1706] Server: Cloud server, dedicated server

[1707] Database: MySQL, PostgreSQL

[1708] Generative AI model: GPT-3 (OpenAI)

[1709] Emotion engine: VADER (Valence Aware Dictionary and sEntiment Reasoner)

[1710] Specific data processing and calculation

[1711] 1. User receives message

[1712] The user uses the device to send a message to the official account, for example, by typing "Hello, tell me about your recent projects." This message is then sent from the device to the server.

[1713] 2. Message Preprocessing

[1714] The server receives messages sent from the device, stores them in a database, and then pre-processes them by tokenizing them and removing unnecessary spaces and special characters.

[1715] 3. Emotion recognition

[1716] The server inputs the preprocessed messages into the emotion engine to recognize the user's emotions, which extracts emotions such as positive, negative, and neutral from the messages.

[1717] 4. Response Generation

[1718] The server then feeds data into a generative AI model based on the recognized emotions to generate a response to the user's message, adapting the style of a specific character or celebrity and further adjusting the response content based on the emotions recognized by the emotion engine.

[1719] 5. Sending the Response

[1720] The server finally sends the generated response to the user's terminal, where the user receives the response.

[1721] 6. Emotion history storage

[1722] The server stores the recognized emotions and dialogue content in a database, which allows the server to refer to the emotion history the next time a dialogue is generated, enabling a consistent dialogue.

[1723] Specific examples

[1724] For example, consider the case where a user sends the following message:

[1725] User: "I've been feeling a bit down lately. Could you tell me something to cheer me up?"

[1726] 1. User receives message

[1727] The server receives this message and stores it in a database.

[1728] 2. Message Preprocessing

[1729] The server parses and tokenizes the received message, removing unnecessary spaces and special characters.

[1730] 3. Emotion recognition

[1731] The server uses an emotion engine to recognize the user's emotion, in this case determining that the message indicates a negative emotion.

[1732] 4. Response Generation

[1733] The server then inputs the adjusted data into a generative AI model, generating a response such as, "Cheer up! I think your new project is very exciting."

[1734] You can also style this response as a character or celebrity, adjusting it to say, "Of course, cheer up! I think your new project is super exciting! You have a great future ahead of you."

[1735] 5. Sending the Response

[1736] The server sends this adjusted response to the user's terminal.

[1737] The terminal displays this response to the user.

[1738] In this way, users can continue to have natural, engaging, and emotionally relevant interactions with characters and celebrities, and emotion history is stored and used to further personalize future responses.

[1739] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1740] Step 1:

[1741] User receives message

[1742] The user types a message to the official account saying, "Hello, tell me about your recent projects" and sends it.

[1743] The terminal sends the message entered by the user to the server.

[1744] Input: The message entered by the user.

[1745] Output: The message sent to the server.

[1746] Step 2:

[1747] Receiving and saving messages

[1748] The server receives the message sent from the device, for example, user_message = "Hello, tell us about your recent projects."

[1749] The server stores this received message in a database using an SQL INSERT query:

[1750] sql

[1751] INSERT INTO messages (user_id, message) VALUES (1, "Hello, tell us about your recent projects");

[1752] Input: Message received from the terminal.

[1753] Output: The message stored in the database.

[1754] Step 3:

[1755] Message Preprocessing

[1756] The server preprocesses the stored messages by tokenizing them and removing unnecessary spaces and special characters. It uses Python's NLTK library to tokenize "Hello, tell me about your recent projects" into "Hello, tell me about your recent projects."

[1757] Input: The stored messages retrieved from the database.

[1758] Output: The tokenized message.

[1759] Step 4:

[1760] emotion recognition

[1761] The server inputs the preprocessed messages into the emotion engine to recognize the user's emotions. It uses VADER (Valence Aware Dictionary and sEntiment Reasoner) to extract emotions such as positive, negative, and neutral.

[1762] Input: The preprocessed tokenized message.

[1763] Output: User sentiment score.

[1764] Step 5:

[1765] Response Generation

[1766] Based on the recognized sentiment score, the server generates a response to the user's message using a generative AI model, for example, the GPT-3 model. The generative AI model generates a response such as, "Of course, your recent projects are very exciting. I'm working on a new album."

[1767] Input: Sentiment scores and preprocessed tokenized messages.

[1768] Output: The generated response.

[1769] Step 6:

[1770] Applying and adjusting styles

[1771] The server styles the generated response after a specific character or celebrity, such as "Of course, my latest projects have been very exciting! I'm working on a new album."

[1772] It also tailors its responses based on emotions recognized by its emotion engine: if the user is expressing negative emotions, it offers additional encouragement, such as, "Don't worry, the new album is going to be amazing!"

[1773] Input: Generated responses and sentiment scores.

[1774] Output: The styled and adjusted response.

[1775] Step 7:

[1776] Sending a Response

[1777] The server finally sends the adjusted response to the user's terminal.

[1778] The device will display this response to the user, for example, a message like "Of course, my latest projects are very exciting! Don't worry, the new album is going to be amazing!" will appear on the screen.

[1779] Input: The adjusted response.

[1780] Output: The response displayed on the terminal.

[1781] Step 8:

[1782] Emotion history storage

[1783] The server stores the recognized user emotions and the content of the conversation in a database, creating an emotion history that can be referenced the next time a conversation is generated.

[1784] For example, using a SQL query like this:

[1785] sql

[1786] INSERT INTO emotions (user_id, message_id, sentiment) VALUES (1, LAST_INSERT_ID(), "negative");

[1787] Input: User emotion data and dialogue content.

[1788] Output: Emotion history stored in a database.

[1789] Step 9:

[1790] Refer to emotion history when generating the next response

[1791] The next time the user sends a message, the server references the user's past emotion history and generates a new response based on that history to provide a consistent interaction.

[1792] For example, the next time a user sends a message, we get the sentiment history like this:

[1793] sql

[1794] SELECT sentiment FROM emotions WHERE user_id = 1 ORDER BY timestamp DESC LIMIT 1;

[1795] Input: Past emotional history.

[1796] Output: Consistent response.

[1797] (Application example 2)

[1798] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1799] Conventional dialogue systems only have the ability to return a fixed response to messages sent by the user, and have difficulty recognizing the user's emotions and generating a response accordingly. Furthermore, they are unable to apply a specific context to the generated response or utilize the user's emotional history for the next dialogue, making it difficult to realize personalized communication.

[1800] The specific processing by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for receiving a message from a user, means for preprocessing the received message, means for generating a response using a generative AI model based on the preprocessed message, means for applying a specific context to the generated response, means for sending the response to the user, means for recognizing the user's emotion and generating a response according to the emotion, means for saving the preprocessed message and the recognized emotion in a database, and means for referring to the saved emotion history in the next dialogue. This makes it possible to generate a personalized response that takes the user's emotion into consideration and achieve continuous and consistent communication.

[1801] The "means for receiving a message from a user" is a mechanism for transferring a text message entered by a user to a server via a network.

[1802] The "means for preprocessing received messages" refers to functions for formatting messages to make them easier to parse and process, such as tokenizing messages and removing unnecessary spaces and special characters.

[1803] "Means for generating a response using a generative AI model based on a preprocessed message" refers to a mechanism that uses preprocessed message data as input and creates an appropriate response using a generative AI model.

[1804] "Means for applying specific context to generated responses" refers to the ability to add a specific character or contextual style to generated responses, making the responses presented to the user seem natural and familiar.

[1805] The "means for sending a response to the user" is a mechanism for transferring the generated response message to the user's terminal and displaying it on the screen.

[1806] The "means for recognizing the user's emotions and generating a response according to those emotions" is a function for detecting emotions from the user's message and generating a response with content and tone according to those emotions.

[1807] The "means for storing the preprocessed message and the recognized emotion in a database" is a mechanism for recording and storing the user's message and the emotion recognized from the message in a database.

[1808] The "means for referring to the emotion history stored in the next interaction" is a function for referring to the emotion history of the user stored in the past and utilizing it in the next interaction.

[1809] overview

[1810] The present invention is a dialogue system that generates personalized responses based on user messages. The system receives user messages, recognizes emotions, and generates responses based on those emotions using a generative AI model and an emotion engine. Furthermore, the system stores the user's emotion history and utilizes it in the next dialogue, providing consistent and personalized dialogue.

[1811] Hardware and Software Configuration

[1812] Hardware

[1813] The system hardware configuration is as follows:

[1814] 1. User terminal: A mobile device with communication capabilities, such as a smartphone or tablet.

[1815] 2. Server: A cloud server or dedicated server that processes data and generates responses.

[1816] software

[1817] The software used is as follows:

[1818] 1. Generative AI models: Natural language processing models such as the OpenAI API.

[1819] 2. Emotion engine: Emotion recognition algorithms (e.g., Sentiment Analysis API).

[1820] 3. Database management system: such as SQLite.

[1821] Program processing overview and specific examples

[1822] Message Preprocessing

[1823] The server pre-processes messages received from user devices by tokenizing them and removing unnecessary spaces and special characters. For example, a message such as "I'm very unhappy because my order hasn't arrived yet" is converted to "I'm very unhappy because my order hasn't arrived yet."

[1824] emotion recognition

[1825] The server inputs the preprocessed message into the emotion engine, which recognizes emotions such as positive, negative, and neutral. The emotion engine's algorithm extracts the user's emotion from the message content. For example, the above message is recognized as "negative."

[1826] Response Generation

[1827] The server inputs a prompt sentence into the generative AI model based on the emotion recognition results, and generates a response. For example, the following prompt sentence is input into the generative AI model:

[1828] User message: I'm very frustrated because my order hasn't arrived yet.

[1829] Emotion: Negative

[1830] Generate the corresponding response.

[1831] The generative AI model generates a natural response based on this prompt, such as, "We're sorry. We'll look into this immediately and get back to you as soon as possible."

[1832] Customizing and sending responses

[1833] The generated response is adapted to the specific context and style and sent to the user, where the server can amend or modify the generated response to present it in a user-friendly format.

[1834] Save user emotion history for future reference

[1835] The emotion recognition results and messages are stored in a database. The next time you interact with the system, the system will refer to the stored emotion history to provide a consistent and personalized response, allowing users to enjoy continuous and personalized communication.

[1836] In this way, the present invention enables natural dialogue according to the user's emotions, and realizes personalized support.

[1837] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1838] Step 1:

[1839] The server receives a message from the user terminal, for example, a user input such as "I'm very unhappy because my order hasn't arrived yet." The server receives this message and stores it in a database for further processing.

[1840] Step 2:

[1841] The server preprocesses the received message by tokenizing it and removing unnecessary spaces and special characters. For example, a message like "I'm very unhappy because my order hasn't arrived yet" is converted to "I'm very unhappy because my order hasn't arrived yet." This preprocessed message becomes the input data for the next emotion recognition.

[1842] Step 3:

[1843] The server inputs the preprocessed message into the emotion engine to recognize the user's emotion. The emotion engine extracts emotions such as positive, negative, and neutral from the input message. For example, for the preprocessed message "I'm very unhappy because my order hasn't arrived yet," the emotion engine outputs "negative."

[1844] Step 4:

[1845] The server inputs a prompt sentence into the generative AI model based on the emotion recognition results, and generates a response. The prompt sentence is set as follows:

[1846] User message: I'm very frustrated because my order hasn't arrived yet.

[1847] Emotion: Negative

[1848] Generate the corresponding response.

[1849] The generative AI model will then generate a natural response based on this prompt, such as "We apologize for the inconvenience. We will investigate this immediately and provide a prompt response."

[1850] Step 5:

[1851] The server applies specific context and style to the generated response, such as adding a more friendly "We're sorry, we'll look into this and get back to you as soon as possible" response, and customizes it as needed using templates and modules.

[1852] Step 6:

[1853] The server then sends the final response to the user's device. The user's device then displays the received response message on its screen. For example, the user may see a message such as, "We apologize for the inconvenience. We will investigate the issue immediately and respond as soon as possible."

[1854] Step 7:

[1855] The server stores the user's message and the emotion recognition results in a database. The stored data is referenced the next time a conversation occurs, allowing for consistent and personalized responses to be generated based on the user's past emotion history.

[1856] The specific processing unit 290 transmits the result of the specific processing to the robot 414. In the robot 414, the control unit 46A causes the speaker 240 and the control target 443 to output the result of the specific processing. The microphone 238 acquires voice indicating a user input regarding the result of the specific processing. The control unit 46A transmits 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.

[1857] 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> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with 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.

[1858] 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 robot 414.

[1859] The emotion identification model 59 as an emotion engine may determine the user's emotion according to a specific mapping. Specifically, the emotion identification model 59 may determine the user's emotion according to an emotion map (see FIG. 9), which is a specific mapping. Similarly, the emotion identification model 59 may determine the robot's emotion, and the identification processing unit 290 may perform identification processing using the robot's emotion.

[1860] FIG. 9 is a diagram illustrating an emotion map 400 on which multiple emotions are mapped. In the emotion map 400, emotions are arranged in concentric circles radiating from the center. Emotions closer to the center of the concentric circles are more primitive. Emotions representing states and actions arising from a state of mind are arranged on the outer edges of the concentric circles. The concept of emotion includes both affect and mental states. Emotions generally generated from reactions occurring in the brain are arranged on the left side of the concentric circles. Emotions generally induced by situational judgment are arranged on the right side of the concentric circles. Emotions generally generated from reactions occurring in the brain and induced by situational judgment are arranged on the upper and lower sides of the concentric circles. Furthermore, the emotion of "pleasure" is arranged on the upper side of the concentric circles, and the emotion of "discomfort" is arranged on the lower side. In this way, in the emotion map 400, multiple emotions are mapped based on the structure by which emotions are generated, and emotions that tend to occur simultaneously are mapped close to each other.

[1861] These emotions are distributed in the 3 o'clock direction on emotion map 400, and typically fluctuate between relief and anxiety. In the right half of emotion map 400, situational awareness dominates over internal sensations, resulting in a sense of calm.

[1862] The inside of emotion map 400 represents what is going on in the mind, and the outside of emotion map 400 represents behavior, so the further you go outside emotion map 400, the more visible the emotions become (the more they are expressed in behavior).

[1863] Human emotions are based on various balances, such as posture and blood sugar levels. When these balances deviate from the ideal, a state of discomfort is indicated, and when they approach the ideal, a state of pleasure is indicated. Emotions can also be created for robots, automobiles, and motorcycles, based on various balances, such as posture and remaining battery life. When these balances deviate from the ideal, a state of discomfort is indicated, and when they approach the ideal, a state of pleasure is indicated. An emotion map can be generated, for example, based on Dr. Mitsuyoshi's emotion map (Research on Voice Emotion Recognition and Emotional Brain Physiological Signal Analysis Systems, Tokushima University, Doctoral Dissertation: https: / / ci.nii.ac.jp / naid / 500000375379). The left half of the emotion map lists emotions belonging to the "reaction" domain, where sensation is dominant. The right half of the emotion map lists emotions belonging to the "situation" domain, where situational awareness is dominant.

[1864] The emotion map defines two emotions that promote learning. One is a negative emotion on the situation side, around the middle of "repentance" or "reflection." In other words, this occurs when the robot experiences negative emotions such as "I never want to feel this way again" or "I don't want to be scolded again." The other is a positive emotion on the response side, around "desire." In other words, this occurs when the robot experiences positive feelings such as "I want more" or "I want to know more."

[1865] The emotion identification model 59 inputs user input into a pre-trained neural network, obtains emotion values ​​indicating each emotion shown in the emotion map 400, and determines the user's emotion. This neural network is pre-trained based on multiple pieces of training data that are combinations of user input and emotion values ​​indicating each emotion shown in the emotion map 400. Furthermore, this neural network is trained so that emotions that are located close to each other have similar values, as in the emotion map 900 shown in FIG. 10. FIG. 10 shows an example in which multiple emotions, "relieved," "calm," and "reassuring," have similar emotion values.

[1866] The system according to the present disclosure has been described above mainly with respect to the functions of the data processing device 12, but the system according to the present disclosure is not necessarily implemented on a server. The system according to the present disclosure may be implemented as a general information processing system. The present disclosure may be implemented, for example, as a software program running on a personal computer or an application running on a smartphone, etc. The method according to the present disclosure may be provided to users in the form of SaaS (Software as a Service).

[1867] In the above embodiment, an example was given in which the specific processing is performed by one computer 22, but the technology of the present disclosure is not limited to this, and the specific processing may be distributed and performed by a plurality of computers including the computer 22. For example, the data generation model 58 may be provided in an external device of the data processing device 12, and data may be generated in the external device in accordance with input data.

[1868] In the above embodiment, an example in which the specific processing program 56 is stored in the storage 32 has been described, but the technology of the present disclosure is not limited to this. For example, the specific processing program 56 may be stored in a portable, computer-readable, non-transitory storage medium such as a USB (Universal Serial Bus) memory. The specific processing program 56 stored in the non-transitory storage medium is installed in the computer 22 of the data processing device 12. The processor 28 executes the specific processing in accordance with the specific processing program 56.

[1869] Alternatively, the specific processing program 56 may be stored in a storage device such as a server connected to the data processing device 12 via the network 54, and the specific processing program 56 may be downloaded and installed on the computer 22 in response to a request from the data processing device 12.

[1870] It is not necessary to store all of the specific processing program 56 in a storage device such as a server connected to the data processing device 12 via the network 54, or to store all of the specific processing program 56 in the storage 32; only a portion of the specific processing program 56 may be stored.

[1871] The hardware resource for executing a specific process can be any of the following processors: An example of a processor is a CPU, which is a general-purpose processor that functions as a hardware resource for executing a specific process by executing software, i.e., a program. Another example of a processor is a dedicated electrical circuit, such as an FPGA (Field-Programmable Gate Array), a PLD (Programmable Logic Device), or an ASIC (Application Specific Integrated Circuit), which is a processor with a circuit configuration designed specifically for executing a specific process. Each processor has built-in or connected memory, and each processor uses the memory to execute the specific process.

[1872] The hardware resource that executes the specific processing may be configured with one of these various processors, or may be configured with a combination of two or more processors of the same or different types (for example, a combination of multiple FPGAs, or a combination of a CPU and an FPGA). Also, the hardware resource that executes the specific processing may be a single processor.

[1873] As an example of a system configured with a single processor, first, one processor is configured by combining one or more CPUs and software, and this processor functions as a hardware resource that executes a specific process. Second, there is a system that uses a processor that realizes the functions of an entire system including multiple hardware resources that execute a specific process on a single IC chip, as typified by SoC (System-on-a-chip). In this way, a specific process is realized using one or more of the above-mentioned various processors as hardware resources.

[1874] Furthermore, the hardware structure of these various processors can be, more specifically, an electric circuit that combines circuit elements such as semiconductor devices. The specific processing described above is merely an example. Therefore, it goes without saying that unnecessary steps may be deleted, new steps may be added, or the processing order may be rearranged, without departing from the spirit of the invention.

[1875] The above-described description and illustrations are a detailed explanation of the parts related to the technology of the present disclosure and are merely an example of the technology of the present disclosure. For example, the above description of the configuration, functions, actions, and effects is an explanation of an example of the configuration, functions, actions, and effects of the parts related to the technology of the present disclosure. Therefore, it goes without saying that unnecessary parts may be deleted, new elements may be added, or replacements may be made to the above-described description and illustrations within the scope of the gist of the technology of the present disclosure. Furthermore, to avoid confusion and facilitate understanding of the parts related to the technology of the present disclosure, the above-described description and illustrations omit explanations of common technical knowledge that do not require particular explanation to enable the implementation of the technology of the present disclosure.

[1876] All publications, patent applications, and technical standards mentioned in this specification are herein incorporated by reference to the same extent as if each individual publication, patent application, or technical standard was specifically and individually indicated to be incorporated by reference.

[1877] The following is further disclosed regarding the above embodiment.

[1878] (Claim 1)

[1879] means for receiving a message from a user;

[1880] means for pre-processing received messages;

[1881] a means for generating a response using a generative AI model based on the preprocessed message;

[1882] A means to style generated responses with characters or famous people;

[1883] means for transmitting a response to a user;

[1884] A system including:

[1885] (Claim 2)

[1886] 10. The system of claim 1, wherein the generative AI model has settings that mimic the speaking style and writing style of a character or famous person.

[1887] (Claim 3)

[1888] 10. The system of claim 1, further comprising means for tailoring the content of the response to a particular style of a character or celebrity after the response is generated.

[1889] (Claim 4)

[1890] 10. The system of claim 1, further comprising means for tokenizing the message and removing unnecessary spaces and special characters during preprocessing.

[1891] (Claim 5)

[1892] 2. The system according to claim 1, further comprising means for saving a history of interactions with the user and for referring to the past interactions when generating a next response.

[1893] (Claim 6)

[1894] 2. The system of claim 1, further comprising normalization means for appropriately transforming the preprocessed messages.

[1895] (Claim 7)

[1896] 10. The system of claim 1, further comprising means for collecting feedback from a user and updating the AI ​​model after generating the response.

[1897] "Example 1"

[1898] (Claim 1)

[1899] means for receiving a message from a user;

[1900] A means of pre-processing received messages to tokenize them and remove unnecessary spaces and special characters;

[1901] a means for generating a response using a generative AI model based on the preprocessed message;

[1902] A means to adapt the generated responses to the style of a particular character or celebrity, adjusting the writing style and speaking style;

[1903] means for transmitting the tailored response to the user;

[1904] A system including:

[1905] (Claim 2)

[1906] 10. The system of claim 1, wherein the generative AI model has settings that mimic the speaking style and writing style of a particular character or famous person.

[1907] (Claim 3)

[1908] 10. The system of claim 1, further comprising means for tailoring the content of the response to a particular style of a character or celebrity after the response is generated.

[1909] "Application Example 1"

[1910] (Claim 1)

[1911] means for receiving a message from a user;

[1912] means for pre-processing received messages;

[1913] a means for generating a response using a generative AI model based on the preprocessed message;

[1914] A means to style generated responses with characters or famous people;

[1915] means for transmitting a response to a user;

[1916] A means to save response history and use it for the next conversation;

[1917] A means of collecting user feedback and improving the generative AI model; and

[1918] A system including:

[1919] (Claim 2)

[1920] 2. The system of claim 1, wherein the generative AI model has settings that mimic the speaking style and writing style of a character or famous person, and includes means for template-based adjustment of the style.

[1921] (Claim 3)

[1922] 10. The system of claim 1, further comprising means for tailoring the response content to a particular style of a character or celebrity after the response is generated, and further comprising including a dialogue history in the prompt.

[1923] "Example 2: Combining Emotion Engines"

[1924] (Claim 1)

[1925] means for receiving a message from a user;

[1926] means for pre-processing received messages;

[1927] means for recognizing a user's emotion based on the preprocessed message using an emotion engine;

[1928] a means for generating a response using a generative AI model based on the recognized emotion;

[1929] a means for applying a style of character or celebrity to the generated responses and tailoring the response content based on emotion;

[1930] means for transmitting a response to a user;

[1931] a means for storing the responses and the user's emotion history in a database;

[1932] A means for generating a response by referring to the emotion history for the next interaction;

[1933] A system including:

[1934] (Claim 2)

[1935] 10. The system of claim 1, wherein the generative AI model has settings that mimic the speaking style and writing style of a character or famous person.

[1936] (Claim 3)

[1937] 10. The system of claim 1, further comprising means for tailoring the content of the response to a particular style of a character or celebrity after the response is generated.

[1938] "Application example 2 when combining emotion engines"

[1939] (Claim 1)

[1940] means for receiving a message from a user;

[1941] means for pre-processing received messages;

[1942] a means for generating a response using a generative AI model based on the preprocessed message;

[1943] a means of applying a specific context to the generated response;

[1944] means for transmitting a response to a user;

[1945] means for recognizing a user's emotion and generating a response according to the emotion;

[1946] means for storing the preprocessed messages and the recognized emotions in a database;

[1947] A means for referencing the stored emotion history in the next interaction;

[1948] A system including:

[1949] (Claim 2)

[1950] 2. The system of claim 1, wherein the generative AI model has settings that mimic a particular speaking style or writing style.

[1951] (Claim 3)

[1952] 10. The system of claim 1, further comprising means for tailoring the content of the response to a particular style after the response is generated. [Explanation of symbols]

[1953] 10, 210, 310, 410 Data Processing Systems 12 Data Processing Device 14 Smart Devices 214 Smart Glasses 314 Headset-type terminal 414 Robot< / url:> < / url:> < / url:> < / url:>

Claims

1. means for receiving a message from a user; means for pre-processing received messages; a means for generating a response using a generative AI model based on the preprocessed message; A means to style generated responses with characters or famous people; means for transmitting a response to a user; A system including:

2. 2. The system of claim 1, wherein the generative AI model has settings that mimic the speaking style and writing style of a character or famous person.

3. 10. The system of claim 1, further comprising means for tailoring the content of the response to a particular style of a character or celebrity after the response is generated.

4. 2. The system of claim 1, further comprising means for tokenizing the message and removing unnecessary spaces and special characters during preprocessing.

5. 2. The system according to claim 1, further comprising means for saving a history of interactions with the user and for referring to the past interactions when generating a next response.

6. 2. The system of claim 1, further comprising normalization means for appropriately transforming the preprocessed messages.

7. 10. The system of claim 1, further comprising means for collecting feedback from a user and updating the AI ​​model after generating the response.

Images