System

Abstract

To provide a system.SOLUTION: A system includes: means for accumulating talk flows that have been deployed in the past; means for loading specifications and features of a new service into an artificial intelligence; means for generating a proposed talk flow that matches the new service; means for collecting log data of past sales conversations; means for analyzing the collected log data by inputting a specific prompt sentence into a generative AI model, to extract a pattern of successful sales conversations; means for analyzing the tone of a user's voice, the volume of the user's voice and the speaking speed of the user's voice, to recognize the user's emotions; means for selecting a talk flow based on the emotions; and means for displaying the selected talk flow on the user's terminal.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] When developing a new service, it is important to efficiently generate talk flows that are optimized for that service. [Means for solving the problem]

[0005] As a means for solving this problem, the system disclosed herein includes a means for accumulating talk flows that have been developed in the past, a means for loading the specifications and features of a new service into artificial intelligence, a means for generating a proposed talk flow tailored to the new service, a means for collecting log data of past sales conversations, a means for analyzing the collected log data by inputting specific prompt sentences into a generative AI model to extract patterns of successful sales conversations, a means for analyzing the tone of the user's voice, the volume of the user's voice, and the speaking rate of the user's voice to recognize the user's emotions, a means for selecting a talk flow based on the emotions, and a means for displaying the selected talk flow on the user's terminal. [Brief explanation of the drawings]

[0006] [Figure 1] 1 is a conceptual diagram showing an example of the configuration of a data processing system according to a first embodiment. [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. 2 is a sequence diagram showing a flow of processing in the data processing system according to the first embodiment of the first form example. [Figure 12] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 1 of Embodiment 1. [Figure 13] FIG. 10 is a sequence diagram showing a processing flow of a data processing system in a second embodiment of the second form example. [Figure 14] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 2 of Embodiment Example 2. [Figure 15] FIG. 10 is a sequence diagram showing the flow of processing in a data processing system according to a third embodiment of the third embodiment. [Figure 16] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 3 of Embodiment 3. [Figure 17] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in the first embodiment of the first form example when an emotion engine is combined. [Figure 18] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 1 of Form Example 1 when an emotion engine is combined. [Figure 19] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in the second embodiment of the second form example when an emotion engine is combined. [Figure 20] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 2 of Form Example 2 when an emotion engine is combined. [Figure 21] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in the third embodiment of the third form example when an emotion engine is combined. [Figure 22] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 3 of Form Example 3 when an emotion engine is combined. DETAILED DESCRIPTION OF THE INVENTION

[0007] 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.

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

[0009] 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, the 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), an APU (Accelerated Processing Unit), or a TPU (TENSOR PROCESSING UNIT (registered trademark)).

[0010] 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.

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

[0012] In the following embodiments, a communication I / F (Interface) with a 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.

[0013] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B." In other words, "A and / or B" means that it may be only A, only B, or a combination of A and B. 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."

[0014] [First embodiment]

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

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

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

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

[0019] The reception device 38 includes a touch panel 38A, 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.

[0020] The output device 40 includes a display 40A and a speaker 40B, and presents data to the user 20 by outputting the data in a form of expression that the user 20 can perceive (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.

[0021] The camera 42 is a small digital camera equipped with an optical system including a lens, an aperture, a shutter, and the like, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0022] 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.

[0023] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0024] 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.

[0025] 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.

[0026] 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.

[0027] Next, the specific processing by the specific processing unit 290 of the data processing device 12 will be described.

[0028] "Example 1"

[0029] One embodiment of the present invention is a system that stores excellent talk flows that have been developed in the past in a database. This system collects data on past talk flows, analyzes it to extract excellent talk flows, and stores them in a database. Specifically, it extracts successful sales conversation patterns from logs of past sales conversations and stores them as talk flows.

[0030] "Example 2"

[0031] Another embodiment of the present invention is a means for loading the specifications and features of a new service into AI. Specifically, the specifications and manuals of the new service are loaded into AI as text data, and the AI ​​understands the content and generates a talk flow based on that. For example, the manual for the new service is analyzed using a text analysis tool, and the AI ​​learns from the results.

[0032] "Example 3"

[0033] Furthermore, one embodiment of the present invention also includes a means for generating the optimal proposed talk flow tailored to the service. Specifically, the AI ​​understands the specifications and features of the new service, selects the optimal one from the accumulated talk flows, and generates a new talk flow based on that. For example, if the new service is a product that emphasizes "ecology," the AI ​​selects a talk flow related to "ecology" and generates a new talk flow based on that.

[0034] The processing flow of each embodiment will be described below.

[0035] "Example 1"

[0036] Step 1: Collect logs of past sales conversations.

[0037] Step 2: Extract patterns of successful sales conversations from the collected logs.

[0038] Step 3: The extracted patterns are stored in a database as talk flows.

[0039] "Example 2"

[0040] Step 1: Obtain the specifications and manuals for the new service as text data.

[0041] Step 2: Analyze the acquired text data using a text analysis tool.

[0042] Step 3: Have the AI ​​learn the analysis results.

[0043] "Example 3"

[0044] Step 1: AI understands the specifications and features of the new service.

[0045] Step 2: Based on what you understand, select the most appropriate talk flow from the accumulated list.

[0046] Step 3: Generate a new talk flow based on the selected talk flow.

[0047] Example 1

[0048] Next, a description will be given of Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0049] Conventional talk flow generation systems were unable to effectively utilize past conversation data, making it difficult to generate optimal talk flows suited to new services. Furthermore, the process of extracting successful conversation patterns and storing them in a database was not automated, making it difficult to generate efficient talk flows.

[0050] The identification process by the identification processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means. In this invention, the server includes means for accumulating talk flows developed in the past, means for collecting past conversation log data, means for preprocessing the collected log data, means for analyzing the preprocessed data, means for extracting successful conversation patterns, means for saving the extracted patterns in a database, means for loading the specifications and features of a new service into an artificial intelligence, and means for generating an optimal proposed talk flow tailored to the service. This makes it possible to effectively utilize past conversation data and automatically generate an optimal talk flow suited to a new service.

[0051] "Talk flow" is a series of steps or patterns that indicate the progression and flow of a conversation.

[0052] "Log data" refers to records of operation history and events generated by systems and applications.

[0053] "Preprocessing" is the process of cleaning and shaping data before data analysis or machine learning.

[0054] "Data analysis" is the process of analyzing collected data and extracting useful information and patterns.

[0055] A "success pattern" is the characteristics and flow of a conversation that achieves a specific purpose (e.g., purchasing a product).

[0056] A "database" is a system for efficiently storing, managing, and retrieving data.

[0057] "Artificial intelligence" refers to computer systems that mimic human intelligence and can learn and reason.

[0058] "Natural language processing" is a technology that allows computers to understand, interpret, and generate human language.

[0059] The "optimal proposed talk flow" is a talk flow that indicates the progress and flow of a conversation that is most suitable for a specific purpose.

[0060] MODE FOR CARRYING OUT THE INVENTION

[0061] This invention is a system that effectively utilizes past conversation data to automatically generate an optimal talk flow suitable for a new service. Specific embodiments of this system will be described below.

[0062] Data collection

[0063] The server collects log data of past sales conversations. This log data is a record of conversations between salespeople and customers and is stored in text format. The server reads the log file from a specified directory (e.g., / data / sales_logs / ) and parses the log file line by line to identify the start and end of the conversation.

[0064] Data Preprocessing

[0065] The server preprocesses the collected log data. This preprocessing includes text cleaning (removing unnecessary characters and symbols), tokenization (dividing the data into words), and removal of stop words. Specifically, it uses the Python library NLTK to tokenize the text and remove stop words.

[0066] Data analysis

[0067] The server uses natural language processing (NLP) techniques to analyze the preprocessed data. Specifically, it uses spaCy to extract conversation topics, sentiment, keywords, etc. The server identifies the topic of each conversation, performs sentiment analysis, and classifies the conversation's sentiment as positive, negative, or neutral.

[0068] Extracting success patterns

[0069] Based on the analysis results, the server extracts patterns of successful sales conversations. A successful conversation is one in which the customer purchases a product or receives positive feedback. The server uses a machine learning algorithm (e.g., random forest) to identify successful patterns and organizes them as a conversation flow.

[0070] Saving to a database

[0071] The server saves the extracted successful patterns as talk flows in a database. A relational database such as MySQL (registered trademark) or PostgreSQL is used as the database. The server connects to the MySQL database, saves the talk flows using INSERT statements, and reads and verifies the data from the database to confirm that the save was successful.

[0072] Loading the specifications of the new service

[0073] The server loads the specifications and features of the new service into the AI. The specifications and features of the new service are provided as text data, which the server analyzes and inputs into the AI.

[0074] Generation of optimal proposed talk flow

[0075] The server uses artificial intelligence (AI)-based natural language processing to generate optimal proposed talk flows tailored to new services. The generated talk flows are based on past success patterns and the specifications of the new services.

[0076] Specific examples

[0077] For example, consider the following sales conversation log:

[0078] Salesperson: Hello, what item are you looking for today?

[0079] Customer: I'm looking for a new smartphone.

[0080] Salesperson: That's great. What features are important to you?

[0081] Customer: I want something with a good camera.

[0082] Salesperson: Then this model is recommended. It has a high-performance camera.

[0083] Customer: Thank you, I'll buy it.

[0084] This conversation log is analyzed, and successful patterns (customers purchasing products) are extracted and saved in a database as talk flows.

[0085] Prompt Sentence Examples

[0086] Examples of prompts to input to a generative AI model might include:

[0087] "Analyze past sales conversation logs and extract patterns of successful sales conversations. A successful conversation is one in which the customer purchases a product or leaves positive feedback. Store the extracted patterns in a database as conversation flows."

[0088] In this way, the server collects and analyzes logs of past sales conversations and stores successful talk flows in a database, creating a system.

[0089] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0090] Step 1: Data collection

[0091] The server collects log data of past sales conversations. As input, it reads the log files from a specified directory (e.g., / data / sales_logs / ). The server parses the log files line by line to identify the start and end of a conversation. As output, it receives the parsed conversation log data.

[0092] Specific behavior:

[0093] The server will read the log files from the specified directory.

[0094] The server parses the log file line by line to identify the beginning and end of a conversation.

[0095] Step 2: Data Preprocessing

[0096] The server preprocesses the collected log data. As input, it receives the collected conversation log data. Preprocessing includes text cleaning (removing unnecessary characters and symbols), tokenization (splitting into words), and removal of stop words. As output, it obtains the preprocessed, clean data.

[0097] Specific behavior:

[0098] The server uses the Python library NLTK to tokenize the text.

[0099] The server removes stop words (e.g., a, the, is, etc.) and leaves only the important words.

[0100] Step 3: Data analysis

[0101] The server analyzes the preprocessed data. As input, it receives the preprocessed, clean data. It uses natural language processing (NLP) techniques to extract conversation topics, sentiment, keywords, etc. As output, it provides the analysis results.

[0102] Specific behavior:

[0103] The server uses spaCy to identify the topic of each conversation.

[0104] The server performs sentiment analysis and classifies the conversation into positive, negative, or neutral sentiment.

[0105] Step 4: Extracting success patterns

[0106] The server extracts patterns of successful sales conversations based on the analysis results. It receives the analysis results as input. A successful conversation is when the customer purchases a product or receives positive feedback. The server obtains a success pattern as output.

[0107] Specific behavior:

[0108] The server uses machine learning algorithms (e.g., random forests) to identify successful patterns.

[0109] The server organizes the successful patterns as talk flows.

[0110] Step 5: Saving to the Database

[0111] The server saves the extracted successful patterns as talk flows in a database. The successful patterns are received as input, and the talk flows saved in the database are obtained as output.

[0112] Specific behavior:

[0113] The server connects to a MySQL database and saves the talk flow using an INSERT statement.

[0114] The server reads and verifies the data from the database to ensure the save was successful.

[0115] Step 6: Load the specifications for the new service

[0116] The server loads the specifications and features of the new service into the AI. As input, it receives the specifications and features of the new service as text data. As output, it obtains the specification data loaded into the AI.

[0117] Specific behavior:

[0118] The server analyzes the specifications and features of the new service as text data.

[0119] The server inputs the analysis results into the artificial intelligence.

[0120] Step 7: Generate optimal proposed talk flow

[0121] The server uses natural language processing with artificial intelligence to generate the optimal proposed talk flow for the new service. As input, it receives past success patterns and new service specification data. As output, it obtains the optimal proposed talk flow.

[0122] Specific behavior:

[0123] The server generates the optimal proposed talk flow based on past success patterns and specification data for new services.

[0124] The server stores the generated talk flow in a database.

[0125] In this way, the server collects and analyzes logs of past sales conversations, stores successful talk flows in a database, and realizes a system that automatically generates optimal talk flows suitable for new services.

[0126] (Application example 1)

[0127] Next, a description will be given of Application Example 1 of Embodiment 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."

[0128] In traditional customer service work, a store clerk needs a wealth of experience and knowledge to make optimal suggestions to customers, which is difficult for new or inexperienced store clerks. In addition, there is a lack of means to provide real-time conversation flows that allow for quick and accurate responses to customer needs. This has led to issues such as inconsistent customer service quality and a decline in customer satisfaction.

[0129] 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.

[0130] In this invention, the server includes means for storing previously developed talk flows, means for loading specifications and features of new services into an artificial intelligence, means for generating an optimal proposed talk flow tailored to the service, means for displaying the optimal proposed talk flow in real time on a visual device worn by a store clerk, and means for generating the next talk flow based on a conversation with a customer, thereby enabling store clerks to provide customers with consistently high-quality customer service.

[0131] "Past talk flow" refers to the flow and patterns of conversations with customers that have taken place in the past, and by accumulating and analyzing this data, patterns of successful sales conversations can be extracted.

[0132] "Specifications and features of new services" refers to the detailed content and characteristics of the newly provided services, and by inputting this into artificial intelligence, it becomes the basic data for generating optimal proposals for the services.

[0133] "Artificial intelligence" refers to the technology that enables computer systems to mimic human intelligence, learn, reason, and self-correct, generating optimal conversation flow through natural language processing and data analysis.

[0134] "Optimal proposal talk flow" refers to the most effective flow of proposals and responses in conversations with customers, and is generated by artificial intelligence.

[0135] A "visual device" refers to a device worn by a store clerk that visually displays information, such as smart glasses.

[0136] "Real-time" refers to the fact that data processing and information provision are immediate, meaning that the optimal conversation flow is displayed instantly during a conversation with a customer.

[0137] "Means for generating the next talk flow based on a conversation with a customer" refers to technology that analyzes a customer's comments and reactions in real time and uses artificial intelligence to generate the next conversation flow accordingly.

[0138] A system for implementing this invention includes a means for storing previously developed talk flows, a means for loading specifications and features of new services into artificial intelligence, a means for generating an optimal proposed talk flow tailored to the service, a means for displaying the optimal proposed talk flow in real time on a visual device worn by a store clerk, and a means for generating the next talk flow based on a conversation with a customer.

[0139] The server first stores past talk flows in a database. This process involves collecting logs of past sales conversations and extracting patterns of successful sales conversations. As a result, the best talk flows are stored in the database.

[0140] Next, the specifications and features of the new service are input as text data into the AI, which then learns information about the new service and obtains the basic data needed to generate optimal proposals.

[0141] The AI ​​uses natural language processing technology to generate optimal proposed talk flows based on accumulated talk flows and the specifications and features of new services. This generated talk flow is displayed in real time on a visual device (e.g., smart glasses) worn by the store clerk.

[0142] When a store clerk starts a conversation with a customer, they serve them while referring to the optimal proposed talk flow displayed on the visual device. Based on the customer's comments and reactions, the AI ​​generates the next talk flow in real time and displays it on the visual device.

[0143] As a concrete example, the following prompt sentence is input to the generative AI model:

[0144] "Hello, what products are you looking for today?"

[0145] "Customer: I'm looking for a new smartphone."

[0146] Salesperson: Thank you. What features are most important to you?

[0147] Based on this prompt, the generative AI model generates the next conversation flow, which store clerks can refer to in real time through the smart glasses, enabling them to provide consistent, high-quality customer service to customers.

[0148] The hardware used includes smart glasses (e.g., Google® Glass®), and the software includes SQLite (database management) and OpenAI® API (generative AI model), which allows the system to instantly generate and display the optimal talk flow based on the conversation with the customer.

[0149] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0150] Step 1:

[0151] The server stores past talk flows in a database. Specifically, it collects logs of past sales conversations and analyzes them to extract patterns of successful sales conversations. The input is the log data of past sales conversations, and the output is the patterns of successful talk flows that are saved in the database.

[0152] Step 2:

[0153] The server loads the specifications and features of the new service into the AI ​​as text data. Specifically, the detailed content and characteristics of the new service are input in text format, and the AI ​​learns from them. The input is the text data of the specifications and features of the new service, and the output is the AI ​​in a state where it has learned information about the new service.

[0154] Step 3:

[0155] The server uses artificial intelligence to generate the optimal proposed talk flow. Specifically, it uses natural language processing technology to generate the optimal talk flow based on the accumulated talk flows and the specifications and features of the new service. The inputs are past successful talk flows and the specifications and features of the new service, and the output is the optimal proposed talk flow.

[0156] Step 4:

[0157] The terminal (a visual device worn by the store clerk) displays the generated optimal proposed talk flow in real time. Specifically, it receives the optimal proposed talk flow sent from the server and displays it on the display of the visual device. The input is the optimal proposed talk flow sent from the server, and the output is the talk flow displayed on the visual device.

[0158] Step 5:

[0159] The user (store clerk) converses with the customer while referring to the optimal proposed talk flow displayed through a visual device. Specifically, based on the customer's comments and reactions, the user sends information to the server to generate the next talk flow. The input is the customer's comments and reactions, and the output is the information to generate the next talk flow sent to the server.

[0160] Step 6:

[0161] The server generates the next talk flow based on the conversation with the customer. Specifically, it analyzes the customer's remarks and reactions, and artificial intelligence generates the next talk flow accordingly. The input is data on the customer's remarks and reactions, and the output is the next talk flow generated and sent to the visual device.

[0162] Step 7:

[0163] The terminal (visual device) displays the generated next talk flow in real time. Specifically, it receives the next talk flow sent from the server and displays it on the display of the visual device. The input is the next talk flow sent from the server, and the output is the next talk flow displayed on the visual device.

[0164] Example 2

[0165] Next, a description will be given of Example 2 of Form Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0166] Conventional talk flow generation systems have had difficulty generating talk flows that quickly and accurately reflect the specifications and features of new services. Furthermore, the accumulation of past talk flows and the use of analysis results were not fully utilized, making it difficult to generate talk flows efficiently. This resulted in the issue of creating talk flows when introducing new services taking a great deal of time and effort.

[0167] 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.

[0168] In this invention, the server includes means for storing talk flows that have been developed in the past, means for loading specifications and features of new services into an artificial intelligence, means for receiving specifications and manuals of new services as text data and analyzing them using a text analysis tool, means for inputting the analysis results into a generative AI model and training the AI ​​to understand the content, means for generating talk flows based on the content understood by the AI, and means for providing the generated talk flows. This makes it possible to generate talk flows that quickly and accurately reflect the specifications and features of new services.

[0169] "Talk flows developed in the past" refer to the flow and patterns of dialogue that have been previously executed, and are data that can be accumulated and reused.

[0170] "Specifications and features of new services" refers to information that provides a detailed description and characteristics of the newly provided services, and is the basic data for generating talk flows based on this information.

[0171] "Artificial intelligence" refers to computer systems that mimic human intelligence using techniques such as machine learning and natural language processing.

[0172] "Text data" refers to information expressed as a string of characters, such as specifications, manuals, and other documents expressed in digital format.

[0173] "Text analysis tools" refer to software and algorithms used to analyze text data and extract meaning and structure.

[0174] A "generative AI model" refers to an artificial intelligence model that generates new information or responses based on input data.

[0175] "Talk flow" indicates the flow and pattern of a dialogue, and is a blueprint for smoothly advancing interaction with users.

[0176] "Analysis results" refers to information about the meaning and structure of data obtained by text analysis tools.

[0177] "Learning" refers to the process by which artificial intelligence finds patterns and rules based on input data to improve future predictions and responses.

[0178] "Providing" refers to the act of making the generated talk flow available to users or other systems.

[0179] The present invention relates to a system for generating talk flows that quickly and accurately reflect the specifications and features of new services. A specific embodiment of this system will be described below.

[0180] System configuration

[0181] Subject: Server

[0182] The server receives specifications and manuals for new services as text data. Specifically, it receives HTTP requests from users or devices and saves the text data. Next, the server analyzes the text data using a text analysis tool (e.g., NLTK or SpaCy). Based on the analysis results, the server inputs the data into a generative AI model (e.g., GPT-4 (registered trademark)) and trains the AI ​​to understand the content. Finally, the server generates a talk flow based on the content understood by the AI ​​and provides the generated talk flow to the user or device.

[0183] Subject: Terminal

[0184] The device inputs the specifications and manuals for the new service as text data. The user selects a file using the device's file selection dialog, and the device sends this data to the server. When the server returns the analysis results, the device inputs them into the generative AI model, training the AI ​​to understand the content. Finally, the device displays the talk flow generated by the generative AI model to the user.

[0185] Subject: User

[0186] The user inputs the specifications and manuals for the new service as text data into their device. The user then sends this data to the server via their device, which then analyzes it using a text analysis tool. Once the analysis results are returned from the server, the user inputs the data into the generative AI model via their device, training the AI ​​to understand the content. Finally, the user checks the conversation flow generated by the AI.

[0187] Specific examples

[0188] Example 1: Loading the specifications of a new service into AI

[0189] The user has a PDF specification for the new service. This PDF is converted to text data and entered into the device. The device sends this data to the server, which performs text analysis using NLTK. Based on the analysis results, the server inputs the data into GPT-4, training the AI ​​to understand the content. Finally, the server provides the user with the AI-generated conversation flow.

[0190] Example 2: Loading a new service manual into AI

[0191] The user has a manual for the new service in Word format. The user converts this Word file into text data and inputs it into the device. The device then sends this data to the server, which performs text analysis using SpaCy. Based on the analysis results, the server inputs the data into GPT-4, training the AI ​​to understand the content. Finally, the server provides the user with the AI-generated conversation flow.

[0192] Prompt Sentence Examples

[0193] "Please input the specifications of the new service as text data and analyze it using NLTK. Input the analysis results into GPT-4 and train the AI ​​to understand the content. Finally, please provide the AI-generated conversation flow."

[0194] In this way, the present invention enables the creation of talk flows that quickly and accurately reflect the specifications and features of new services.

[0195] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0196] Step 1:

[0197] The server receives text data such as specifications or manuals for new services from users or terminals. Specifically, the server receives an HTTP request and saves the text data. The input is the text data for specifications or manuals for new services, and the output is the saved text data.

[0198] Step 2:

[0199] The server analyzes the received text data using a text analysis tool (e.g., NLTK or SpaCy). Specifically, the server calls the text analysis tool to perform tokenization, part-of-speech tagging, dependency analysis, etc. on the text data. The input is the saved text data, and the output is the analysis results.

[0200] Step 3:

[0201] The server converts the results of text analysis into a format for input to a generative AI model (e.g., GPT-4). Specifically, the server converts the analysis results into an appropriate data format, such as JSON. The input is the result of text analysis, and the output is the converted data format.

[0202] Step 4:

[0203] The server inputs the converted data into the generative AI model and trains the AI ​​to understand the content. Specifically, the server calls the API of the generative AI model and sends the data. The input is the converted data format, and the output is the content understood by the AI.

[0204] Step 5:

[0205] The server generates a talk flow based on what the generative AI model understands. Specifically, the server receives a response from the AI ​​and executes the logic to construct the talk flow. The input is what the AI ​​understands, and the output is the generated talk flow.

[0206] Step 6:

[0207] The server provides the generated talk flow to the user or terminal. Specifically, the server generates an HTTP response and sends data including the talk flow. The input is the generated talk flow, and the output is the talk flow provided to the user or terminal.

[0208] (Application example 2)

[0209] Next, a description will be given of Application Example 2 of Form Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0210] Whenever a new service is introduced, it is difficult to help users understand its specifications and how to use it. In particular, for complex services and multi-functional platforms, users need a guide to use the service properly. However, with conventional methods, simply referring to manuals or FAQs is insufficient, and there is an issue of making it difficult for users to quickly access the information they are looking for.

[0211] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[0212] In this invention, the server includes means for storing previously deployed talk flows, means for loading specifications and features of new services into an AI, means for generating an optimal proposed talk flow tailored to the service, means for generating answers to user questions, and means for answering user questions based on the generated talk flows, thereby enabling users to smoothly understand and use new services.

[0213] "Talk flow developed in the past" refers to the flow and patterns of conversations that have been carried out in the past, and by storing these, they can be used to help with future conversations.

[0214] "Means of having AI read the specifications and features of new services" refers to a method of inputting detailed descriptions and characteristics of new services into AI as text data, allowing the AI ​​to understand their content.

[0215] "Means for generating optimal proposed talk flow" refers to a method in which AI automatically creates the most suitable dialogue flow for users based on the specifications and features of a new service.

[0216] "Means for generating answers to user questions" refers to the method by which AI receives questions from users and generates appropriate answers to them.

[0217] "Means of answering user questions based on the generated talk flow" refers to a method of providing specific answers to user questions based on the dialogue flow generated by AI.

[0218] A system for implementing this invention is configured as follows: First, the server has a means for storing talk flows that have been developed in the past. This means allows the flow and patterns of previously executed conversations to be saved and used for future conversations.

[0219] The next step involves having the AI ​​learn the specifications and features of the new service. This involves inputting a detailed description and characteristics of the new service into the AI ​​as text data, allowing the AI ​​to understand the content. Specifically, a text analysis tool is used to analyze the manual for the new service, and the AI ​​is then trained on the results.

[0220] Furthermore, the server has a means for generating optimal proposed talk flows, in which AI automatically creates the most suitable dialogue flow for users based on the specifications and features of the new service. It also includes a means for generating answers to users' questions using AI-based natural language processing technology.

[0221] The method for generating answers to user questions involves receiving questions from users and having AI generate appropriate answers to them. It also includes a method for answering user questions based on the generated talk flow, providing specific answers to user questions based on the dialogue flow generated by the AI.

[0222] Hardware and software used

[0223] Hardware: Smartphone

[0224] Software: OpenAI API, Python

[0225] Data processing and calculation

[0226] 1. Reading text data: Read the manual for the new service as a text file.

[0227] 2. Training the AI: Use OpenAI's API to train the AI ​​to learn the contents of the manual.

[0228] 3. Generate talk flow: The AI ​​generates a talk flow based on the manual and saves it in JSON format.

[0229] 4. Generating answers to user questions: Upon receiving a user question, the AI ​​generates an appropriate answer based on the conversation flow.

[0230] Specific examples

[0231] If a user asks, "How do I search for new content?", the AI ​​will answer, based on the contents of the manual, "Enter keywords in the search bar and press the search button."

[0232] Prompt Sentence Examples

[0233] Please answer the user's questions based on the following conversation flow.

[0234] Talk flow:

[0235] 1. How to find new content

[0236] 2. How to view content

[0237] 3. How to set up your account

[0238] User Question: How do I search for new content?

[0239] In this way, users can smoothly understand and use new services.

[0240] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0241] Step 1:

[0242] The server reads the manual for the new service as a text file.

[0243] Input: New service manual (text file)

[0244] Output: Text data

[0245] Specific operation: The server reads the manual for the new service from the specified file path and stores the contents in memory as text data.

[0246] Step 2:

[0247] The server uses a text analysis tool to analyze the read text data.

[0248] Input: Text data

[0249] Output: Analysis results (structured text data)

[0250] Specific operation: The server uses a text analysis tool (for example, a natural language processing library) to analyze the text data and extract sentence structures and keywords.

[0251] Step 3:

[0252] The server trains the AI ​​model based on the analysis results.

[0253] Input: Analysis results (structured text data)

[0254] Output: Trained AI model

[0255] Specific operation: The server uses the OpenAI API to input the analysis results into the AI ​​model, allowing the AI ​​to learn to understand the specifications and features of the new service.

[0256] Step 4:

[0257] The server generates the optimal proposed talk flow using the trained AI model.

[0258] Input: Trained AI model

[0259] Output: Optimal proposal talk flow (JSON format)

[0260] Specific operation: The server inputs the usage scenario of the new service into the AI ​​model, and the AI ​​generates the optimal dialogue flow. The generated dialogue flow is saved in JSON format.

[0261] Step 5:

[0262] The user enters a question using a smartphone.

[0263] Input: User question (text)

[0264] Output: None (passed on to next step)

[0265] Specific operation: The user opens the application on their smartphone and enters a question in text format.

[0266] Step 6:

[0267] The server receives the user's question and generates an answer based on the generated talk flow.

[0268] Input: User question (text), optimal suggestion talk flow (JSON format)

[0269] Output: AI response (text)

[0270] Specific operation: The server inputs the user's question into the AI ​​model and generates an appropriate answer while referring to the generated talk flow.

[0271] Step 7:

[0272] The server sends the generated answer to the user's smartphone.

[0273] Input: AI response (text)

[0274] Output: The answer displayed on the user's smartphone

[0275] Specific operation: The server sends the generated answer to the user's smartphone and displays it in the user's application.

[0276] In this way, users can smoothly understand and use new services.

[0277] Example 3

[0278] Next, a description will be given of a third embodiment of the third embodiment. 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."

[0279] Conventional talk flow generation systems have had difficulty quickly and effectively generating talk flows that are optimal for the specifications and features of new services. Furthermore, there were problems with the time it took for users to input information about new services, and with variations in the quality of the generated talk flows. This could lead to ineffective service introductions and proposals, and could result in lower customer satisfaction.

[0280] The specific processing by the specific processing unit 290 of the data processing device 12 in the third embodiment is realized by the following means.

[0281] In this invention, the server includes means for storing previously developed talk flows, means for loading specifications and features of a new service into an AI, means for generating an optimal proposed talk flow tailored to the service, means for a user to input the specifications and features of the new service, means for the server to receive the input information and send a prompt to the generative AI model, means for the server to select a relevant talk flow from a database of stored talk flows, means for the server to generate a new talk flow using the generative AI model, and means for the server to provide the generated talk flow to the user. This makes it possible to quickly and effectively generate a talk flow optimal for the specifications and features of a new service, reducing the user's effort and providing a high-quality talk flow.

[0282] "Past-developed talk flows" refer to previously used dialogue flows and patterns, which can be stored in a database to make them reusable.

[0283] "Specifications and features of new services" refers to detailed descriptions and characteristics of newly provided services, and is information used to generate optimal talk flows.

[0284] "Artificial intelligence" refers to the technology that allows computer systems to mimic human intelligence to learn, reason, and self-correct, and is used in this case to generate talk flow.

[0285] "Optimal proposal talk flow" refers to the dialogue flow that is most suited to the specifications and features of a new service, and is intended to effectively introduce the service to customers.

[0286] "Means for users to input specifications and features of new services" refers to an interface or method by which users can provide information about new services to the system.

[0287] "Means for the server to receive input information and send prompts to the generative AI model" refers to a method for receiving information provided by a user and issuing instructions to the generative AI model based on that information.

[0288] "Means for the server to select relevant talk flows from a database of accumulated talk flows" refers to a method for selecting past talk flows stored in a database that are relevant to the new service.

[0289] "Means for the server to generate a new talk flow using a generative AI model" refers to a method for creating a new talk flow using a generative AI model based on a selected talk flow.

[0290] The term "means by which the server provides the generated talk flow to the user" refers to a method for displaying or transmitting the generated talk flow to the user.

[0291] MODE FOR CARRYING OUT THE INVENTION

[0292] The present invention relates to a system for generating talk flows that are optimal for the specifications and features of new services. Specific embodiments of this system will be described below.

[0293] System configuration

[0294] This system consists of three main components: a server, a terminal, and a user. The server contains a database of previously deployed talk flows, a generative AI model (e.g., OpenAI's GPT-4), and a talk flow generation algorithm. The terminal provides an interface for users to input specifications and features of new services.

[0295] Program processing

[0296] The user uses the terminal to input the specifications and features of the new service. For example, the user may input information such as "This is an eco-friendly product made from renewable materials." The terminal then sends this information to the server.

[0297] The server receives the information sent by the user and creates a prompt to send to the generative AI model based on that information. For example, it generates a prompt such as, "Please generate a talk flow to introduce a new eco-friendly product. This product is made from environmentally friendly materials and is highly energy efficient."

[0298] Next, the server selects a relevant talk flow from a database of stored talk flows based on the prompt sentence. This selection is done using natural language processing technology. For example, a talk flow that emphasizes the benefits of eco-friendly products or a talk flow that suggests an environmentally friendly lifestyle may be selected.

[0299] The server generates a new talk flow based on the selected talk flow using a generative AI model. The generative AI model understands the context and flow of the talk flow and reconstructs it in a way that is optimal for the new service.

[0300] The generated talk flow might look something like this:

[0301] 1. "Hello, today I'd like to introduce you to a new eco-friendly product."

[0302] 2. "This product is made from renewable materials and is manufactured with maximum energy efficiency."

[0303] 3. "Furthermore, using this product will help reduce energy consumption in daily life."

[0304] Finally, the server sends the generated talk flow to the user's device, where the user can view the talk flow on the device screen and make corrections or adjustments as necessary.

[0305] Specific examples

[0306] For example, if the new service is a product that emphasizes "ecology," the user can input the following prompt into the generative AI model:

[0307] Example prompt sentence:

[0308] "Generate a talk flow to introduce a new eco-friendly product. This product is made from environmentally friendly materials and is highly energy efficient."

[0309] Based on this prompt, the server generates an optimal talk flow and provides it to the user. In this way, the user can quickly and effectively generate a talk flow that is optimal for the specifications and features of the new service. The flow of the specification process in the third embodiment will be explained with reference to Figure 15.

[0310] Program processing flow

[0311] Step 1:

[0312] Users input specifications and features of new services

[0313] The user uses the terminal to input the specifications and features of the new service. For example, the user might input "It's an eco-friendly product made from renewable materials." This input data is sent to the server in text format.

[0314] input:

[0315] Specifications and features of the new service (e.g., "It is an ecological product and uses renewable materials")

[0316] output:

[0317] Text data sent to the server

[0318] Step 2:

[0319] The server receives the input information and sends a prompt to the generative AI model.

[0320] The server receives the information sent by the user and creates a prompt to send to the generative AI model based on that information. For example, it generates a prompt such as, "Please generate a talk flow to introduce a new eco-friendly product. This product is made from environmentally friendly materials and is highly energy efficient."

[0321] input:

[0322] Text data sent by the user

[0323] Data processing:

[0324] Generate prompt sentences (e.g., "Generate a talk flow to introduce a new eco-friendly product. This product is made from environmentally friendly materials and is highly energy efficient.")

[0325] output:

[0326] Generated prompt statement

[0327] Step 3:

[0328] The server selects relevant talk flows from its database of accumulated talk flows.

[0329] The server uses natural language processing technology to select a relevant talk flow from a database of stored talk flows based on the generated prompt. For example, a talk flow emphasizing the benefits of eco-friendly products or a talk flow proposing an environmentally friendly lifestyle may be selected.

[0330] input:

[0331] Generated prompt statement

[0332] Data processing:

[0333] Select relevant talk flows using natural language processing technology

[0334] output:

[0335] Selected Talk Flow

[0336] Step 4:

[0337] The server generates a new talk flow using the generative AI model.

[0338] The server generates a new talk flow based on the selected talk flow using a generative AI model. The generative AI model understands the context and flow of the talk flow and reconstructs it in a way that is optimal for the new service.

[0339] input:

[0340] Selected Talk Flow

[0341] Data processing:

[0342] Generate new conversation flows using generative AI models

[0343] output:

[0344] The new talk flow generated

[0345] Step 5:

[0346] The server provides the generated talk flow to the user.

[0347] The server sends the generated talk flow to the user's device, where the user can check the talk flow on the device screen and make corrections or adjustments as necessary.

[0348] input:

[0349] The new talk flow generated

[0350] output:

[0351] Dialogue flow sent to the user's device

[0352] Specific examples of operation

[0353] For example, if the new service is a product that emphasizes "ecology," the user can input the following prompt into the generative AI model:

[0354] Example prompt sentence:

[0355] "Generate a talk flow to introduce a new eco-friendly product. This product is made from environmentally friendly materials and is highly energy efficient."

[0356] Based on this prompt, the server generates an optimal talk flow and provides it to the user. In this way, the user can quickly and effectively generate a talk flow that is optimal for the specifications and features of a new service.

[0357] (Application example 3)

[0358] Next, a description will be given of Application Example 3 of Form Example 3. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0359] Conventional talk flow generation systems have had difficulty quickly and accurately generating optimal talk flows that match the specifications and features of new services. Furthermore, they were unable to provide users with the information they wanted in real time, making improving the user experience a challenge.

[0360] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 3 is realized by the following means. In this invention, the server includes means for storing talk flows developed in the past, means for loading specifications and features of new services into AI, means for generating an optimal proposed talk flow tailored to the service, means for inputting features of products searched by the user, and means for displaying the generated talk flow to the user. This makes it possible to provide the information desired by the user in real time and improve the user experience.

[0361] "Past talk flows" refers to previously used dialogue flows and patterns, and are data that can be accumulated and reused.

[0362] "New service specifications" refers to information that indicates the detailed design, functions, and characteristics of a newly provided service.

[0363] "Features" is information that indicates the characteristics and features of a service or product.

[0364] "AI" is an abbreviation for artificial intelligence, a system that analyzes data and makes decisions using technologies such as machine learning and natural language processing.

[0365] "Optimal proposal talk flow" is a process that generates the most appropriate dialogue flow based on the specifications and features of a specific service or product.

[0366] "Characteristics of the product searched by the user" refers to information that indicates the characteristics or features of the product that the user searches for on the online platform.

[0367] "Generated talk flow" refers to the flow and patterns of dialogue generated by AI based on the specifications and features of the new service.

[0368] The "means for displaying to the user" refers to a method or technology for visually presenting the generated talk flow to the user.

[0369] A system for implementing this invention includes a server, a user terminal, and an AI model. The server stores previously deployed talk flows and loads the specifications and features of new services into the AI. The user terminal inputs the features of the product searched by the user and displays the generated talk flows.

[0370] Hardware and software used

[0371] Hardware: Servers, smartphones

[0372] Software: Python, OpenAI API

[0373] Data processing and calculation

[0374] 1. Data entry: The user uses a smartphone to enter the characteristics of the product they are searching for, for example, "products made with renewable energy, recyclable packaging."

[0375] 2. Prompt generation: Based on the input product characteristics, the server generates a prompt sentence to be input into the generative AI model.

[0376] 3. AI model invocation: The server uses OpenAI's API to send prompts to a generative AI model (e.g., GPT-3 (registered trademark)) to generate an optimal talk flow.

[0377] 4. Displaying the results: The server sends the generated talk flow to the user's terminal and displays it to the user.

[0378] Specific examples

[0379] If a user searches for eco-friendly products that have the characteristics of "products that use renewable energy and recyclable packaging," the following prompt sentence will be generated:

[0380] Prompt Sentence Examples

[0381] Generate a talk flow about a new eco-friendly product. Product features: Renewable energy-based product, recyclable packaging

[0382] The following is an example of a generated talk flow:

[0383] This product is made using renewable energy, making it an eco-friendly choice, and the packaging is recyclable, reducing waste. Perfect for those who want to live an eco-conscious life.

[0384] In this way, the information desired by the user can be provided in real time, improving the user experience.

[0385] The flow of the specific processing in Application Example 3 will be described with reference to FIG.

[0386] Step 1:

[0387] The user uses a smartphone to input the characteristics of the product they are searching for, such as "products that use renewable energy, recyclable packaging." The input data is sent to the server in text format.

[0388] Step 2:

[0389] The server receives the product features sent by the user. Based on the received text data, it generates a prompt sentence to be input into the generative AI model. Specifically, it creates a prompt sentence that includes the product features.

[0390] Step 3:

[0391] The server calls the OpenAI API using the generated prompt sentence. The prompt sentence is sent to the API, and the optimal talk flow is generated by a generative AI model (e.g., GPT-3). The input is the prompt sentence, and the output is the generated talk flow.

[0392] Step 4:

[0393] The server receives the generated talk flow and converts it into an appropriate format for display to the user. Specifically, it converts text data into a format that is easy for the user to understand.

[0394] Step 5:

[0395] The server sends the formatted talk flow to the user's smartphone. The user's device displays the received talk flow on its screen. The user can check the displayed talk flow and obtain detailed product information.

[0396] This series of processes allows users to receive the optimal talk flow in real time based on the characteristics of the product they searched for.

[0397] 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.

[0398] "Example 1"

[0399] As one embodiment of the present invention, a system is provided that includes an emotion engine in which AI recognizes a user's emotions. In this system, the emotion engine recognizes the user's emotions in real time during a dialogue with the user and generates an optimal talk flow according to the emotion. For example, if the user expresses joy, the emotion engine conveys that information to the AI, and the AI ​​generates a talk flow corresponding to joy.

[0400] "Example 2"

[0401] In another embodiment of the present invention, a system is provided in which an emotion engine includes means for recognizing emotions from the tone of a user's voice. This system analyzes audio information such as the tone, volume, and speaking rate of the user's voice, and estimates the user's emotional state from the results. For example, if the user's voice tone is high and the speaking rate is fast, it is determined that the user is likely excited.

[0402] "Example 3"

[0403] In a further embodiment of the present invention, a system is provided that includes a means for adjusting a talk flow according to a recognized user's emotion. In this system, an AI recognizes the user's emotion and generates a talk flow that is optimal for that emotion. For example, if the user expresses anger, the AI ​​uses that information to generate a talk flow that calms the user.

[0404] The processing flow of each embodiment will be described below.

[0405] "Example 1"

[0406] Step 1: A user interaction is initiated.

[0407] Step 2: The emotion engine recognizes the user's emotions in real time.

[0408] Step 3: The recognized emotional information is conveyed to the AI.

[0409] Step 4: The AI ​​generates the optimal conversation flow based on the emotion.

[0410] "Example 2"

[0411] Step 1: The user's voice information (tone of voice, volume, speaking rate, etc.) is analyzed by the emotion engine.

[0412] Step 2: The user's emotional state is inferred from the analysis results.

[0413] Step 3: The estimated emotional information is conveyed to the AI.

[0414] Step 4: The AI ​​generates the optimal conversation flow based on the emotion.

[0415] "Example 3"

[0416] Step 1: A user interaction is initiated.

[0417] Step 2: The emotion engine recognizes the user's emotions in real time.

[0418] Step 3: The recognized emotional information is conveyed to the AI.

[0419] Step 4: The AI ​​generates the optimal conversation flow based on the emotion.

[0420] Step 5: The generated talk flow is presented to the user.

[0421] Example 1

[0422] Next, a description will be given of Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0423] Conventional talk flow generation systems were unable to effectively utilize past conversation data, making it difficult to generate optimal talk flows in real time that matched the user's emotions. Furthermore, there was a lack of a way to quickly generate talk flows that reflected the specifications and features of new services. This meant that improvements in sales efficiency and customer satisfaction were not fully achieved.

[0424] 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.

[0425] In this invention, the server includes means for storing talk flows developed in the past, means for inputting specifications and features of new services into an artificial intelligence, means for generating an optimal proposed talk flow tailored to the service, means for collecting log data of past sales conversations, means for analyzing the collected log data to extract patterns of successful sales conversations, means for recognizing user emotions in real time, and means for generating an optimal talk flow based on the recognized emotion information. This makes it possible to effectively utilize patterns of successful sales conversations in the past and generate an optimal talk flow in real time that corresponds to the user's emotions.

[0426] "Talk flow" refers to a series of statements and responses in a dialogue with a user.

[0427] "Artificial intelligence" refers to the technology that enables computer systems to mimic human intelligence to learn, reason, perceive, and make decisions.

[0428] "Log Data" refers to a series of events and activities that a system or application records during its operation.

[0429] "Natural language processing" refers to the technology that enables computers to understand, generate, and manipulate human language.

[0430] "Emotion recognition" refers to the technology of identifying and classifying emotions from a user's voice or text.

[0431] "Optimal proposal talk flow" refers to the dialogue flow that is most suitable for a specific situation or condition.

[0432] A "database" refers to a system for efficiently storing, managing, and retrieving data.

[0433] "Pattern extraction" refers to the process of finding specific patterns or characteristics in data.

[0434] "Real-time" refers to data processing and response occurring immediately.

[0435] This invention is a system that collects log data of past sales conversations, analyzes it to extract patterns of successful sales conversations, recognizes user emotions in real time, and generates optimal talk flows. This system operates in cooperation with a server, a terminal, and a user.

[0436] Hardware and software used

[0437] Hardware: Servers, terminals (PCs, smartphones)

[0438] Software: database management systems (e.g., MySQL), emotion recognition engines (e.g., IBM Watson®), generative AI models (e.g., GPT-4)

[0439] Data collection

[0440] The server collects log data of past sales conversations. Specifically, it obtains log data from sales systems and customer relationship management systems (CRM). The collected data is saved in text and audio formats.

[0441] Data analysis

[0442] The server analyzes the collected sales conversation log data, analyzes the content of the conversations using natural language processing (NLP) techniques, and extracts patterns of successful sales conversations using NLP libraries (e.g., spaCy).

[0443] Saving to a database

[0444] The server stores the extracted talk flows in a database. The server uses a database management system (e.g., MySQL) to store the talk flows as structured data.

[0445] User Emotion Recognition

[0446] The device activates an emotion engine during interaction with the user. It uses an emotion recognition engine (e.g., IBM Watson) to recognize emotions from the user's voice and text in real time.

[0447] Generate talk flow

[0448] The server generates an optimal talk flow using a generative AI model based on the emotion information. Specifically, it inputs prompt sentences into the generative AI model (e.g., GPT-4) and generates an appropriate talk flow.

[0449] Specific examples

[0450] Examples of data collection

[0451] The server collects sales conversation logs from the past year, including text and audio data of conversations.

[0452] Specific examples of emotion recognition

[0453] If a user says, "This product is absolutely amazing!", the emotion engine recognizes the emotion "delight."

[0454] Example of talk flow generation

[0455] If the emotion engine recognizes "joy," the server inputs the following prompt sentence into the generative AI model:

[0456] Generate the optimal conversation flow when the user is expressing the emotion of joy.

[0457] The generative AI model generates the following conversation flow:

[0458] "Thank you! I'm glad you're a satisfied customer. Please let me know if there are any other products you're interested in."

[0459] This system effectively utilizes patterns of past successful sales conversations and can generate optimal conversation flows in real time according to the user's emotions.

[0460] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0461] Step 1:

[0462] Data collection

[0463] The server collects log data of past sales conversations. Specifically, it obtains log data from the sales system and customer relationship management system (CRM). The input is log data from the sales system and CRM, and the output is the collected log data stored in temporary storage within the server. The server accesses the CRM system via an API and obtains sales conversation logs from the past year.

[0464] Step 2:

[0465] Data analysis

[0466] The server analyzes the collected sales conversation log data. Specifically, it uses natural language processing (NLP) technology to analyze the content of the conversation. The input is the collected log data, and the output is the extraction of patterns of successful sales conversations. The server tokenizes the log data using an NLP library (e.g., spaCy) and extracts important keywords and phrases. The server identifies the characteristics of successful sales conversations and classifies them as excellent talk flows.

[0467] Step 3:

[0468] Saving to a database

[0469] The server saves the extracted good talk flows in a database. The input is the extracted talk flows, and the output is the talk flows saved in the database. The server connects to the database and executes SQL queries to insert the talk flow data. The saved talk flows are stored in the database for later retrieval and reference.

[0470] Step 4:

[0471] User Emotion Recognition

[0472] The device activates an emotion engine during a dialogue with the user. Specifically, it uses an emotion recognition engine (e.g., IBM Watson) to recognize emotions from the user's voice or text in real time. The input is the user's speech, and the output is the recognized emotion information. The device receives the user's speech as voice input or text input, sends the input data to the emotion recognition engine, and receives the results of the emotion analysis.

[0473] Step 5:

[0474] Generate talk flow

[0475] The server generates an optimal talk flow using a generative AI model based on the emotion information. Specifically, it inputs a prompt into the generative AI model (e.g., GPT-4) to generate an appropriate talk flow. The inputs are the recognized emotion information and the prompt, and the output is the generated talk flow. If the emotion engine recognizes "joy," the server inputs the following prompt into the generative AI model:

[0476] Generate the optimal conversation flow when the user is expressing the emotion of joy.

[0477] The generative AI model generates the following conversation flow:

[0478] "Thank you! I'm glad you're a satisfied customer. Please let me know if there are any other products you're interested in."

[0479] The server transmits the generated talk flow to the terminal, which uses the talk flow for interaction with the user.

[0480] (Application example 1)

[0481] Next, a description will be given of Application Example 1 of Embodiment 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."

[0482] Conventional customer service systems have difficulty recognizing customer emotions in real time and providing optimal conversation flows accordingly. As a result, improvements in customer satisfaction and sales efficiency have not been fully achieved. In addition, there are limited ways to effectively utilize successful conversation flows from the past.

[0483] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means. In this invention, the server includes means for storing previously developed talk flows, means for loading specifications and features of new services into an AI, means for generating an optimal proposed talk flow tailored to the service, means for recognizing a user's emotions in real time, means for selecting an optimal talk flow based on the recognized emotions, and means for displaying the selected talk flow. This makes it possible to provide an optimal talk flow based on a customer's emotions in real time, thereby improving customer satisfaction and sales efficiency.

[0484] "Past talk flow" refers to a series of conversation patterns used in past sales conversations and customer service.

[0485] "Specifications and features of the new service" refers to the detailed content and characteristics of the newly provided service.

[0486] "Means of loading data into AI" refers to a method of inputting data into artificial intelligence and having it learn and analyze based on that data.

[0487] "Means for generating optimal proposed talk flow" refers to a method for creating the most effective conversation pattern according to specific conditions and situations.

[0488] "Means for recognizing a user's emotions in real time" refers to a method for instantly determining a user's emotions at that time from their facial expressions, tone of voice, etc.

[0489] "Means for selecting an optimal talk flow based on recognized emotions" refers to a method for selecting the most appropriate conversation pattern depending on the user's emotional state.

[0490] "Means for displaying the selected talk flow" refers to a method for visually presenting the selected conversation pattern.

[0491] A system for implementing this invention includes a means for storing previously deployed talk flows, a means for loading the specifications and features of a new service into an AI, a means for generating an optimal proposed talk flow tailored to the service, a means for recognizing a user's emotions in real time, a means for selecting an optimal talk flow based on the recognized emotions, and a means for displaying the selected talk flow.

[0492] System Program

[0493] This system is realized using the following hardware and software.

[0494] Hardware

[0495] Smart glasses: equipped with a camera and microphone

[0496] Server: A server to host the database and AI models

[0497] software

[0498] OpenCV: Image processing library

[0499] Keras: A deep learning library

[0500] SpeechRecognition: A speech recognition library

[0501] Data processing and calculation

[0502] Image Processing

[0503] The images captured by the smart glasses camera are processed using OpenCV to extract facial features, enabling real-time analysis of the user's facial expressions.

[0504] emotion recognition

[0505] The extracted facial image is input into a Keras model to recognize emotions, allowing us to instantly grasp the user's emotional state.

[0506] Talk flow generation

[0507] Based on the recognized emotion, the system selects the optimal talk flow from the database, thereby providing the optimal conversation pattern according to the user's emotion.

[0508] display

[0509] The selected talk flow is displayed on the smart glasses' display, allowing store staff to provide service while checking the optimal talk flow in real time.

[0510] Specific examples

[0511] For example, suppose a customer is looking for a new smartphone. The smart glasses recognize the customer's "interest" from their facial expression and select "explanations for customers who are interested" from past successful conversation flows. The store clerk then follows that conversation flow to recommend the best smartphone for the customer.

[0512] Prompt Sentence Examples

[0513] When a customer is looking for a new smartphone, the smart glasses recognize "interest" from the customer's facial expression. Select an "explanation for an interested customer" from a past successful conversation flow and present it to the store associate.

[0514] In this way, it is possible to provide the optimal conversation flow in real time according to the customer's emotions, thereby improving customer satisfaction and sales efficiency.

[0515] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0516] Step 1:

[0517] The camera in the smart glasses captures the customer's video. The input is real-time video data, and the output is the captured video frame. The smart glasses then transmits the video frame to the server.

[0518] Step 2:

[0519] The server extracts facial features from video frames using OpenCV. The input is the captured video frame and the output is the extracted face image. The server runs a face detection algorithm to identify the face location.

[0520] Step 3:

[0521] The server recognizes emotions from extracted facial images using a Keras model. The input is the extracted facial image, and the output is the recognized emotion label (e.g., joy, sadness, interest, etc.). The server preprocesses the facial image and inputs it to the emotion recognition model.

[0522] Step 4:

[0523] The server selects the optimal talk flow from the database based on the recognized emotion. The input is the recognized emotion label, and the output is the optimal talk flow. The server searches the database for the talk flow corresponding to the emotion label.

[0524] Step 5:

[0525] The server displays the selected talk flow on the display of the smart glasses. The input is the optimal talk flow, and the output is the talk flow displayed on the display of the smart glasses. The server sends the talk flow in text format to the smart glasses.

[0526] Step 6:

[0527] The salesperson responds to the customer according to the talk flow displayed on the smart glasses display. The input is the displayed talk flow, and the output is the conversation with the customer. The salesperson makes the best suggestions to the customer based on the talk flow.

[0528] Step 7:

[0529] After serving the customer, the salesperson inputs feedback into the system. The input is the feedback data, and the output is an updated talk flow database. The server receives the feedback and updates the talk flow database.

[0530] Example 2

[0531] Next, a description will be given of Example 2 of Form Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0532] With conventional systems, it was difficult to efficiently understand the specifications and features of new services and generate optimal conversation flows based on them. It was also difficult to analyze the user's emotional state in real time and respond appropriately. This hindered improvements to the user experience.

[0533] 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.

[0534] In this invention, the server includes means for storing previously developed talk flows, means for loading specifications and features of a new service into an artificial intelligence, means for generating an optimal proposed talk flow tailored to the service, means for analyzing specifications and manuals for the new service as text data, means for collecting user voice data and converting it into text data using voice recognition software, and means for transmitting the converted text data to an emotion analysis engine and analyzing the user's emotional state. This makes it possible to efficiently understand the specifications and features of the new service, generate an optimal talk flow, and further analyze the user's emotional state in real time to take appropriate action.

[0535] The "means for storing talk flows that have been developed in the past" is a function for saving data on talk flows that have been executed in the past so that they can be referenced later.

[0536] "Means for loading the specifications and features of new services into artificial intelligence" refers to a function that inputs information such as specifications and manuals for new services into artificial intelligence and allows it to understand their contents.

[0537] The "means for generating optimal proposed talk flow" is a function that enables artificial intelligence to automatically create optimal talk flow based on the specifications and features of a new service.

[0538] "Means for analyzing specifications and manuals for new services as text data" refers to a function for importing documents related to new services as text data and analyzing the contents using an analysis tool.

[0539] "Means for collecting user voice data and converting it into text data using voice recognition software" refers to a function for collecting user voice and converting it into text data using voice recognition technology.

[0540] The "means for sending the converted text data to an emotion analysis engine and analyzing the user's emotional state" is a function for inputting the text data obtained by speech recognition into an emotion analysis engine and estimating the user's emotional state.

[0541] MODE FOR CARRYING OUT THE INVENTION

[0542] This invention is a system that efficiently understands the specifications and features of new services and generates optimal talk flows based on them. It is also a system that can analyze the user's emotional state in real time and respond appropriately.

[0543] Server Operation

[0544] The server first receives the specifications and manuals for the new service as text data from the user. This text data is saved in the server's storage. The server then analyzes the saved text data using a text analysis tool (e.g., NLTK or SpaCy). The analysis results are input into a generative AI model (e.g., OpenAI's GPT-4). The generative AI model generates an optimal talk flow based on the input data. The generated talk flow is then provided to the user by the server.

[0545] As a concrete example, when analyzing the manual for a new cloud service, the server first obtains the text data of the manual and performs grammatical analysis using NLTK. The analysis results are then input into GPT-4 to generate a talk flow that explains how to use the service and its features.

[0546] Example prompt sentence:

[0547] "Please analyze the manual for the following cloud service and generate a conversation flow for users to use the service."

[0548] Device behavior

[0549] The device collects the user's voice data in real time. It captures the voice through a microphone and sends it to speech recognition software (e.g., Google Speech-to-Text API). The speech recognition software converts the voice data into text data. The converted text data is sent to an emotion analysis engine (e.g., IBM Watson Tone Analyzer) to analyze the user's emotional state. The analysis results are displayed to the user on the device.

[0550] For example, when a user calls customer support, the device collects the user's voice in real time and converts it into text using the Google Speech-to-Text API.Then, it uses IBM Watson Tone Analyzer to analyze the user's emotions from the text data and determine whether the user is excited or not.

[0551] Example prompt sentence:

[0552] "Analyze the user's voice data and estimate their emotional state based on their tone of voice, volume, and speaking rate."

[0553] User Actions

[0554] Users provide the system with specifications and manuals for new services. The text data provided by the user is analyzed by the server and used as data for AI learning. The user can then check the talk flow generated based on the analysis results and make any necessary corrections.

[0555] As a concrete example, when a user uploads a manual for new software to the system, the user first prepares the text data of the manual and uploads it to the system. The server then analyzes it and provides the generated talk flow to the user. The user can then check the talk flow and make any necessary corrections.

[0556] Example prompt sentence:

[0557] "Upload the new software manual and generate the talk flow."

[0558] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0559] Server Processing Steps

[0560] Step 1:

[0561] The server receives specifications and manuals for new services as text data from users, and stores the text data uploaded by users in the server's storage.

[0562] Input: New service specifications and manuals (text data)

[0563] Output: Text data saved in storage

[0564] Specific operation: The user uploads "New Cloud Service Manual.txt" and the server saves it in storage.

[0565] Step 2:

[0566] The server analyzes the stored text data using text analysis tools (e.g., NLTK or SpaCy), performing grammatical analysis and keyword extraction.

[0567] Input: Text data saved in storage

[0568] Output: Analysis results (grammar analysis data, keywords)

[0569] Specific operation: The server analyzes "New Cloud Service Manual.txt" using NLTK and extracts important keywords and phrases.

[0570] Step 3:

[0571] The server inputs the analysis results into a generative AI model (e.g., OpenAI's GPT-4), which generates a talk flow based on the input data.

[0572] Input: Analysis results (grammar analysis data, keywords)

[0573] Output: Generated talk flow

[0574] Specific operation: The server inputs the extracted data into GPT-4 and generates a talk flow.

[0575] Step 4:

[0576] The server obtains the talk flow output from the generative AI model and provides it to the user.

[0577] Input: Generated talk flow

[0578] Output: The talk flow provided to the user

[0579] Specific operation: The server provides the generated talk flow to the user so that the user can check it.

[0580] Terminal processing steps

[0581] Step 1:

[0582] The device collects the user's voice data in real time by capturing speech through a microphone and sending it to speech recognition software (e.g., Google Speech-to-Text API).

[0583] Input: User's voice data

[0584] Output: Voice data sent to speech recognition software

[0585] Specific operation: The device uses a microphone to collect audio when the user calls customer support.

[0586] Step 2:

[0587] The device converts the voice data into text data using voice recognition software and temporarily stores the converted text data.

[0588] Input: Voice data sent to speech recognition software

[0589] Output: Text data

[0590] Specific operation: The voice is converted into text using the Google Speech-to-Text API, and information such as "the user's voice is high-pitched and they speak quickly" is extracted.

[0591] Step 3:

[0592] The device sends the converted text data to an emotion analysis engine (e.g., IBM Watson Tone Analyzer) to analyze the user's emotional state.

[0593] Input: Text data

[0594] Output: Emotion analysis results

[0595] Specific operation: Using IBM Watson Tone Analyzer, the system analyzes the user's emotions from text data and infers that the user is excited.

[0596] Step 4:

[0597] The device displays the results obtained from the emotion analysis engine to the user, providing feedback according to the user's emotional state.

[0598] Input: Sentiment analysis results

[0599] Output: Sentiment analysis results displayed to the user

[0600] Specific operation: Display the estimation results to the user and prompt them to take appropriate action.

[0601] User processing steps

[0602] Step 1:

[0603] Users provide specifications and manuals for new services to the system, and the uploaded data is sent to the server.

[0604] Input: New service specifications and manuals (text data)

[0605] Output: Text data sent to the server

[0606] Specific operation: The user uploads "New Software Manual.pdf" to the system.

[0607] Step 2:

[0608] The user checks the analysis results and generated talk flow provided by the server and makes corrections as necessary.

[0609] Input: Analysis results and talk flow provided by the server

[0610] Output: Modified talk flow

[0611] Specific operation: The user checks the talk flow and makes corrections such as "I would like this part to be explained in more detail."

[0612] Step 3:

[0613] Users can provide feedback on the generated talk flow and sentiment analysis results, which will be used to improve the system.

[0614] Input: Generated talk flow and sentiment analysis results

[0615] Output: Feedback

[0616] Specific operation: The user provides feedback on the sentiment analysis results, such as "This analysis was accurate."

[0617] (Application example 2)

[0618] Next, a description will be given of Application Example 2 of Form Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0619] Conventional talk flow generation systems had difficulty quickly reflecting the specifications and features of new services, and were unable to properly recognize and respond to users' emotional states. Furthermore, they were unable to issue appropriate warnings even in emergency situations where a quick response was required.

[0620] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[0621] In this invention, the server includes means for storing previously deployed talk flows, means for loading the specifications and features of a new service into an AI, means for generating an optimal proposed talk flow tailored to the service, means for analyzing a user's voice and converting it into text data, means for analyzing the user's tone, volume, and speaking rate to estimate their emotional state, means for responding to the user based on the generated talk flow, and means for issuing a warning when an emergency is detected. This allows the specifications and features of the new service to be quickly reflected, making it possible to appropriately recognize and respond to the user's emotional state. It also enables a rapid response in an emergency.

[0622] A "talk flow" is a series of steps and responses that show the flow of a conversation with a user.

[0623] A "New Service Specification" is a document that describes the detailed design, functions, and features of a newly provided service.

[0624] "AI" is an abbreviation for artificial intelligence, a technology that allows machines to learn and reason by imitating human intelligence.

[0625] "Text data" refers to data that represents character information in a digital format.

[0626] "Natural language processing" is a technology that allows computers to understand, interpret, and generate human language.

[0627] "Voice analysis" is a technology that analyzes voice data and understands its content.

[0628] "Emotional state" refers to the user's emotional or psychological state.

[0629] An "emergency" is a situation in which a danger or problem exists that requires immediate action.

[0630] A "warning" is a notice or alert intended to notify you of a danger or problem.

[0631] A system for implementing this invention has the following configuration: First, the server has a means for storing talk flows that have been developed in the past. This means allows past dialogue data to be stored and used to generate future talk flows.

[0632] Next, we have a method for loading the specifications and features of the new service into the AI. This method involves loading the specifications and manuals of the new service into the AI ​​as text data, allowing the AI ​​to understand their contents. This makes it possible to generate the optimal conversation flow for the new service.

[0633] Furthermore, it has a means to analyze the user's voice and convert it into text data. This means uses speech recognition technology to convert the user's voice into text data and analyzes the content. Specifically, it collects voice using the smartphone's microphone and converts it into text data using the speech_recognition library.

[0634] It also has a means to estimate the user's emotional state by analyzing the tone, volume, and speaking rate of the user's voice. This means analyzes the audio data and estimates the user's emotional state. For example, if the volume is loud and the speaking rate is fast, it is determined that the user is excited. This analysis uses the numpy library.

[0635] It also has a means to respond to users based on the generated talk flow. This means uses a generative AI model (e.g., GPT-3) to generate appropriate answers to users' questions. Specifically, it uses the transformers library to generate talk flows and respond to users.

[0636] Furthermore, it has a means to issue a warning when an emergency situation is detected. This means analyzes the user's emotional state and issues a warning when it is judged to be an emergency. For example, if the user says "help me," it will immediately issue a warning.

[0637] For example, if a user says, "My home security system isn't working," the following prompt sentence can be fed into the generative AI model:

[0638] User: My home security system isn't working.

[0639] AI: Is your security system turned on and does your internet connection work?

[0640] In this way, by generating an appropriate talk flow for a user's question and performing emotion recognition, it is possible to provide faster and more appropriate support.

[0641] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0642] Step 1:

[0643] The user speaks their question or problem.

[0644] Input: User's voice data

[0645] Output: Audio data

[0646] Specific action: The user speaks a question or problem into the smartphone microphone.

[0647] Step 2:

[0648] The terminal converts the voice data into text data.

[0649] Input: Audio data

[0650] Output: Text data

[0651] Specific operation: The device uses the speech_recognition library to convert voice data into text data.

[0652] Step 3:

[0653] The server analyzes the text data and generates an appropriate talk flow.

[0654] Input: Text data

[0655] Output: Talkflow

[0656] Specific operation: The server uses a generative AI model (e.g., GPT-3) to generate a talk flow based on text data. It uses the transformers library.

[0657] Step 4:

[0658] The server analyzes the user's tone of voice, volume, and speaking rate to estimate their emotional state.

[0659] Input: Audio data

[0660] Output: Emotional state

[0661] Specific operation: The server uses the numpy library to analyze the tone, volume, and speaking rate of the audio data and estimate the emotional state.

[0662] Step 5:

[0663] The server responds to the user based on the generated talk flow.

[0664] Input: Talk flow, emotional state

[0665] Output:Answer

[0666] Specific operation: The server uses the generated talk flow to generate an appropriate response for the user and sends it to the terminal.

[0667] Step 6:

[0668] If the server detects an emergency, it will issue an alert.

[0669] Input: Emotional state

[0670] Output: warning

[0671] Specific behavior: The server analyzes the user's emotional state and issues a warning if it determines that the user is in an emergency. For example, if the user says "help me," it will immediately issue a warning.

[0672] Example 3

[0673] Next, a description will be given of a third embodiment of the third embodiment. 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."

[0674] Conventional talk flow generation systems have had issues with generating talk flows that are optimal for the specifications and features of new services, and are unable to adjust the talk flow according to the user's emotions. This has resulted in a poor user experience and made it difficult to effectively propose services.

[0675] The specific processing by the specific processing unit 290 of the data processing device 12 in the third embodiment is realized by the following means.

[0676] In this invention, the server includes means for storing previously developed talk flows, means for inputting specifications and features of a new service into an artificial intelligence, means for generating an optimal proposed talk flow tailored to the service, means for recognizing a user's emotion, and means for adjusting the talk flow in accordance with the recognized user's emotion. This makes it possible to generate an optimal talk flow for a new service and adjust the talk flow in accordance with the user's emotion.

[0677] "Talk flows developed in the past" refers to previously used dialogue flows and patterns, which are stored in a database.

[0678] "Specifications and features of the new service" refers to the detailed content and characteristics of the newly provided service, and is information that will be read into the artificial intelligence.

[0679] "Artificial intelligence" refers to the technology that enables computer systems to mimic human intelligence to learn, reason, and self-correct, as used here to generate and moderate talk flow.

[0680] The "optimal proposed talk flow" generates the most effective dialogue flow based on the specifications and features of the new service.

[0681] "User's emotion" refers to the emotional state shown by the user, and is recognized from voice, facial expressions, etc.

[0682] "Means for recognizing emotions" refers to technology that analyzes the user's voice and facial expressions to identify emotions, thereby understanding the user's emotional state.

[0683] The "means for adjusting the talk flow" refers to a technology for changing the flow of a conversation in accordance with the user's recognized emotions, thereby providing a conversation that is optimal for the user's emotions.

[0684] This invention is a system that uses AI to generate optimal proposed talk flows for new services and further adjusts the talk flows according to the user's emotions. Specific embodiments of this system are described below.

[0685] Hardware and software used

[0686] Hardware: Servers, terminals

[0687] Software: Generative AI models, emotion recognition software

[0688] System Overview

[0689] 1. Input specifications and features of new services

[0690] The user inputs the specifications and features of the new service into the terminal.

[0691] The terminal transmits the input data to the server.

[0692] The server passes the received data to the generative AI model.

[0693] 2. Selecting the optimal conversation flow

[0694] The server has access to a database of stored talk flows.

[0695] The generative AI model selects the optimal talk flow from a database based on the characteristics of the new service.

[0696] 3. Creating a new talk flow

[0697] The generative AI model generates a new talk flow based on the selected talk flow.

[0698] The server transmits the generated talk flow to the terminal.

[0699] 4. User Emotion Recognition

[0700] The device analyzes the user's voice and facial expressions using emotion recognition software.

[0701] The emotion recognition software recognizes the user's emotions and sends this information to the server.

[0702] 5. Adjusting conversation flow according to emotions

[0703] Based on the received emotional data, the server instructs the generative AI model to adjust the talk flow.

[0704] The generative AI model generates a conversation flow that best suits the user's emotions.

[0705] The server sends the adjusted talk flow to the terminal.

[0706] Specific examples

[0707] New service: eco-friendly detergent

[0708] User Emotion: Anger

[0709] Example prompt sentence:

[0710] Generate a talk flow for a new eco-friendly cleaner. If the user is showing signs of anger, include a talk flow to calm the user.

[0711] An example of a generated talk flow:

[0712] Hello. This detergent is made with eco-friendly ingredients, making it a better choice for the planet. If you have any complaints, please let us know. We value your feedback.

[0713] In this way, the system can generate optimal talk flows for new services and adjust them according to the user's emotions. This improves the user experience and enables effective proposal of services. The flow of the identification process in the third embodiment will be explained with reference to FIG. 21.

[0714] Step 1:

[0715] Entering specifications and features of new services

[0716] The user inputs the specifications and features of the new service into the terminal.

[0717] Input: Specifications and features of the new service (e.g., properties of an eco-friendly detergent)

[0718] The terminal transmits the input data to the server.

[0719] Output: New service specifications and features data sent to the server

[0720] Specific operation: The user enters "ecological detergent" and its characteristics into the input form on the device and presses the send button.

[0721] Step 2:

[0722] Choosing the best talk flow

[0723] The server passes the received data to the generative AI model.

[0724] Input: Data on the specifications and features of the new service

[0725] The server has access to a database of stored talk flows.

[0726] The generative AI model selects the optimal talk flow from a database based on the characteristics of the new service.

[0727] Output: Optimal talk flow data

[0728] Specific operation: The server searches the database for talk flows related to "ecology" and passes them to the generative AI model.

[0729] Step 3:

[0730] Creating a new talk flow

[0731] The generative AI model generates a new talk flow based on the selected talk flow.

[0732] Input: Data for optimal talk flow

[0733] The server transmits the generated talk flow to the terminal.

[0734] Output: Newly generated talk flow data

[0735] Specific operation: The generative AI model generates a new talk flow that reflects the characteristics of the "ecological detergent," and the server sends it to the device.

[0736] Step 4:

[0737] User Emotion Recognition

[0738] The device analyzes the user's voice and facial expressions using emotion recognition software.

[0739] Input: User voice and facial expression data

[0740] The emotion recognition software recognizes the user's emotions and sends this information to the server.

[0741] Output: Recognized user emotion data

[0742] How it works: When a user speaks into the device's camera, emotion recognition software analyzes the user's facial expressions and tone of voice to identify anger.

[0743] Step 5:

[0744] Adjusting conversation flow according to emotions

[0745] Based on the received emotional data, the server instructs the generative AI model to adjust the talk flow.

[0746] Input: Recognized user emotion data

[0747] The generative AI model generates a conversation flow that best suits the user's emotions.

[0748] The server sends the adjusted talk flow to the terminal.

[0749] Output: Adjusted talk flow data

[0750] Specific operation: The server passes data that "the user is expressing anger" to the generative AI model, the generative AI model generates a conversation flow that calms the user, and the server sends it to the device.

[0751] (Application example 3)

[0752] Next, a description will be given of Application Example 3 of Form Example 3. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0753] Conventional talk flow generation systems have had issues with generating optimal talk flows based on the specifications and features of new services, and with being unable to adjust the talk flow in response to user emotions. This makes it difficult to improve customer satisfaction and provide effective customer service. Furthermore, it has not been possible to display talk flows in real time on smart devices.

[0754] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 3 is realized by the following means.

[0755] In this invention, the server includes a means for storing previously deployed talk flows, a means for loading the specifications and features of a new service into an AI, a means for generating an optimal proposed talk flow tailored to the service, a means for recognizing a user's emotion, a means for adjusting the talk flow according to the recognized user's emotion, and a means for displaying the talk flow on a smart device. This enables the generation of an optimal talk flow based on the specifications and features of a new service, and further enables the adjustment of the talk flow according to the user's emotion. It also enables the display of talk flows in real time using a smart device.

[0756] "Means for storing talk flows that have been developed in the past" refers to a means for saving talk flows that have been used in the past in a database or the like so that they can be referenced later.

[0757] "Means of loading the specifications and features of new services into AI" refers to the means of inputting the details and characteristics of new services into AI as text data so that the AI ​​can understand that information.

[0758] "Means for generating optimal proposed talk flow" refers to a means by which AI automatically creates the most appropriate talk flow based on the specifications and features of a new service.

[0759] "Means for recognizing user emotions" refers to means for analyzing the user's facial expressions and tone of voice using cameras, sensors, etc., to identify their emotions.

[0760] "Means for adjusting the talk flow according to the recognized user emotions" refers to a means by which AI dynamically changes the talk flow based on the user's emotional data and responds optimally.

[0761] The "means for displaying a talk flow on a smart device" refers to a means for displaying the generated talk flow in real time on a device such as smart glasses or a smartphone.

[0762] The system for implementing this invention is configured as follows: First, the server has a database that stores talk flows that have been used in the past. This database stores talk flows that have been used in the past and allows them to be referenced later.

[0763] Next, a method is needed to have the AI ​​learn the specifications and features of the new service. This method involves inputting the details and characteristics of the new service into the AI ​​as text data so that the AI ​​can understand that information. Specifically, natural language processing technology is used to analyze the text data and enable the AI ​​to understand its content.

[0764] Furthermore, it includes a means for generating an optimal proposed talk flow, in which the AI ​​automatically creates the most suitable talk flow based on the specifications and features of the new service.The generative AI model is used to generate the talk flow based on the prompt sentence.

[0765] To recognize a user's emotions, hardware such as cameras and sensors is required. These devices are used to analyze the user's facial expressions and tone of voice to identify their emotions. Specifically, image processing libraries such as OpenCV are used to recognize emotions in real time.

[0766] To adjust the talk flow according to the recognized user emotion, the AI ​​dynamically changes the talk flow based on the emotion data. In this process, the generative AI model is used again to generate the optimal talk flow according to the emotion.

[0767] Finally, a means for displaying the talk flow on a smart device is included. The generated talk flow is displayed in real time on a device such as smart glasses or a smartphone, allowing a store clerk to provide an optimal talk flow to a customer.

[0768] For example, if a customer is interested in ecology, here is a prompt:

[0769] Service Features: Ecology

[0770] Customer Sentiment: Interest

[0771] Optimal conversation flow:

[0772] By inputting this prompt into a generative AI model, an optimal conversation flow about ecology can be generated. For example, a conversation flow such as "This product is made from renewable materials and is environmentally friendly" can be generated.

[0773] The hardware used includes smart glasses and smartphones, and the software uses OpenCV (image processing) and Transformers (AI model for generating Hugging Face), which enables real-time emotion recognition and the generation and display of talk flow.

[0774] The flow of the specific processing in Application Example 3 will be described with reference to FIG.

[0775] Step 1:

[0776] The server accumulates previously deployed talk flows in a database. As input, it receives data on previously used talk flows and stores them in the database. As output, it obtains a database of accumulated talk flows. This database is structured so that it can be referenced later.

[0777] Step 2:

[0778] The server loads the specifications and features of the new service as text data into the AI. As input, text data containing the details and characteristics of the new service is provided, which is analyzed using natural language processing technology. As output, data in a format that the AI ​​can understand is obtained. This data is then used to generate the talk flow.

[0779] Step 3:

[0780] The server uses a generative AI model to generate the optimal proposed talk flow. As input, a prompt sentence based on the specifications and features of the new service is given, and the generative AI model generates the talk flow based on this. The optimal talk flow is obtained as output. Specifically, the prompt sentence is input into the generative AI model, and the generated talk flow is obtained.

[0781] Step 4:

[0782] The device uses a camera and sensors to recognize the user's emotions. As input, real-time data including the user's facial expressions and tone of voice is provided, which is analyzed using an image processing library (e.g., OpenCV). As output, the user's emotional data is obtained. Specifically, the device analyzes the images captured by the camera and identifies the emotion.

[0783] Step 5:

[0784] The server adjusts the talk flow according to the recognized user emotion. As input, the user's emotion data and the generated talk flow are given, and the generative AI model dynamically changes the talk flow based on this. The adjusted talk flow is obtained as output. Specifically, the emotion data is added to the prompt sentence and re-input into the generative AI model.

[0785] Step 6:

[0786] The terminal displays the talk flow on the smart device. As input, the adjusted talk flow is given and displayed in real time on smart glasses or a smartphone. As output, the optimal talk flow is provided to the user. As a specific operation, the generated talk flow is displayed on the device's display.

[0787] 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.

[0788] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (registered trademark) (Internet search engine).<URL: https: / / openai.com / blog / chatgpt> ) 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.

[0789] Another example of generative AI is Gemini (registered trademark) (Internet search engine). <url: https: gemini.google.com ?hl="ja">) are mentioned.

[0790] 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.

[0791] [Second embodiment]

[0792] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0793] 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.

[0794] 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.

[0795] The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and 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).

[0796] 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.

[0797] 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.

[0798] 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).

[0799] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for 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.

[0800] 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.

[0801] 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.

[0802] 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.

[0803] 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.

[0804] Next, the specific processing by the specific processing unit 290 of the data processing device 12 will be described.

[0805] "Example 1"

[0806] One embodiment of the present invention is a system that stores excellent talk flows that have been developed in the past in a database. This system collects data on past talk flows, analyzes it to extract excellent talk flows, and stores them in a database. Specifically, it extracts successful sales conversation patterns from logs of past sales conversations and stores them as talk flows.

[0807] "Example 2"

[0808] Another embodiment of the present invention is a means for loading the specifications and features of a new service into AI. Specifically, the specifications and manuals of the new service are loaded into AI as text data, and the AI ​​understands the content and generates a talk flow based on that. For example, the manual for the new service is analyzed using a text analysis tool, and the AI ​​learns from the results.

[0809] "Example 3"

[0810] Furthermore, one embodiment of the present invention also includes a means for generating the optimal proposed talk flow tailored to the service. Specifically, the AI ​​understands the specifications and features of the new service, selects the optimal one from the accumulated talk flows, and generates a new talk flow based on that. For example, if the new service is a product that emphasizes "ecology," the AI ​​selects a talk flow related to "ecology" and generates a new talk flow based on that.

[0811] The processing flow of each embodiment will be described below.

[0812] "Example 1"

[0813] Step 1: Collect logs of past sales conversations.

[0814] Step 2: Extract patterns of successful sales conversations from the collected logs.

[0815] Step 3: The extracted patterns are stored in a database as talk flows.

[0816] "Example 2"

[0817] Step 1: Obtain the specifications and manuals for the new service as text data.

[0818] Step 2: Analyze the acquired text data using a text analysis tool.

[0819] Step 3: Have the AI ​​learn the analysis results.

[0820] "Example 3"

[0821] Step 1: AI understands the specifications and features of the new service.

[0822] Step 2: Based on what you understand, select the most appropriate talk flow from the accumulated list.

[0823] Step 3: Generate a new talk flow based on the selected talk flow.

[0824] Example 1

[0825] Next, a description will be given of Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0826] Conventional talk flow generation systems were unable to effectively utilize past conversation data, making it difficult to generate optimal talk flows suited to new services. Furthermore, the process of extracting successful conversation patterns and storing them in a database was not automated, making it difficult to generate efficient talk flows.

[0827] The identification process by the identification processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means. In this invention, the server includes means for accumulating talk flows developed in the past, means for collecting past conversation log data, means for preprocessing the collected log data, means for analyzing the preprocessed data, means for extracting successful conversation patterns, means for saving the extracted patterns in a database, means for loading the specifications and features of a new service into an artificial intelligence, and means for generating an optimal proposed talk flow tailored to the service. This makes it possible to effectively utilize past conversation data and automatically generate an optimal talk flow suited to a new service.

[0828] "Talk flow" is a series of steps or patterns that indicate the progression and flow of a conversation.

[0829] "Log data" refers to records of operation history and events generated by systems and applications.

[0830] "Preprocessing" is the process of cleaning and shaping data before data analysis or machine learning.

[0831] "Data analysis" is the process of analyzing collected data and extracting useful information and patterns.

[0832] A "success pattern" is the characteristics and flow of a conversation that achieves a specific purpose (e.g., purchasing a product).

[0833] A "database" is a system for efficiently storing, managing, and retrieving data.

[0834] "Artificial intelligence" refers to computer systems that mimic human intelligence and can learn and reason.

[0835] "Natural language processing" is a technology that allows computers to understand, interpret, and generate human language.

[0836] The "optimal proposed talk flow" is a talk flow that indicates the progress and flow of a conversation that is most suitable for a specific purpose.

[0837] MODE FOR CARRYING OUT THE INVENTION

[0838] This invention is a system that effectively utilizes past conversation data to automatically generate an optimal talk flow suitable for a new service. Specific embodiments of this system will be described below.

[0839] Data collection

[0840] The server collects log data of past sales conversations. This log data is a record of conversations between salespeople and customers and is stored in text format. The server reads the log file from a specified directory (e.g., / data / sales_logs / ) and parses the log file line by line to identify the start and end of the conversation.

[0841] Data Preprocessing

[0842] The server preprocesses the collected log data. This preprocessing includes text cleaning (removing unnecessary characters and symbols), tokenization (dividing the data into words), and removal of stop words. Specifically, it uses the Python library NLTK to tokenize the text and remove stop words.

[0843] Data analysis

[0844] The server uses natural language processing (NLP) techniques to analyze the preprocessed data. Specifically, it uses spaCy to extract conversation topics, sentiment, keywords, etc. The server identifies the topic of each conversation, performs sentiment analysis, and classifies the conversation's sentiment as positive, negative, or neutral.

[0845] Extracting success patterns

[0846] Based on the analysis results, the server extracts patterns of successful sales conversations. A successful conversation is one in which the customer purchases a product or receives positive feedback. The server uses a machine learning algorithm (e.g., random forest) to identify successful patterns and organizes them as a conversation flow.

[0847] Saving to a database

[0848] The server saves the extracted successful patterns as talk flows in a database. A relational database such as MySQL or PostgreSQL is used for the database. The server connects to the MySQL database, saves the talk flows using INSERT statements, and reads and verifies the data from the database to confirm that the save was successful.

[0849] Loading the specifications of the new service

[0850] The server loads the specifications and features of the new service into the AI. The specifications and features of the new service are provided as text data, which the server analyzes and inputs into the AI.

[0851] Generation of optimal proposed talk flow

[0852] The server uses artificial intelligence (AI)-based natural language processing to generate optimal proposed talk flows tailored to new services. The generated talk flows are based on past success patterns and the specifications of the new services.

[0853] Specific examples

[0854] For example, consider the following sales conversation log:

[0855] Salesperson: Hello, what item are you looking for today?

[0856] Customer: I'm looking for a new smartphone.

[0857] Salesperson: That's great. What features are important to you?

[0858] Customer: I want something with a good camera.

[0859] Salesperson: Then this model is recommended. It has a high-performance camera.

[0860] Customer: Thank you, I'll buy it.

[0861] This conversation log is analyzed, and successful patterns (customers purchasing products) are extracted and saved in a database as talk flows.

[0862] Prompt Sentence Examples

[0863] Examples of prompts to input to a generative AI model might include:

[0864] "Analyze past sales conversation logs and extract patterns of successful sales conversations. A successful conversation is one in which the customer purchases a product or leaves positive feedback. Store the extracted patterns in a database as conversation flows."

[0865] In this way, the server collects and analyzes logs of past sales conversations and stores successful talk flows in a database, creating a system.

[0866] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0867] Step 1: Data collection

[0868] The server collects log data of past sales conversations. As input, it reads the log files from a specified directory (e.g., / data / sales_logs / ). The server parses the log files line by line to identify the start and end of a conversation. As output, it receives the parsed conversation log data.

[0869] Specific behavior:

[0870] The server will read the log files from the specified directory.

[0871] The server parses the log file line by line to identify the beginning and end of a conversation.

[0872] Step 2: Data Preprocessing

[0873] The server preprocesses the collected log data. As input, it receives the collected conversation log data. Preprocessing includes text cleaning (removing unnecessary characters and symbols), tokenization (splitting into words), and removal of stop words. As output, it obtains the preprocessed, clean data.

[0874] Specific behavior:

[0875] The server uses the Python library NLTK to tokenize the text.

[0876] The server removes stop words (e.g., a, the, is, etc.) and leaves only the important words.

[0877] Step 3: Data analysis

[0878] The server analyzes the preprocessed data. As input, it receives the preprocessed, clean data. It uses natural language processing (NLP) techniques to extract conversation topics, sentiment, keywords, etc. As output, it provides the analysis results.

[0879] Specific behavior:

[0880] The server uses spaCy to identify the topic of each conversation.

[0881] The server performs sentiment analysis and classifies the conversation into positive, negative, or neutral sentiment.

[0882] Step 4: Extracting success patterns

[0883] The server extracts patterns of successful sales conversations based on the analysis results. It receives the analysis results as input. A successful conversation is when the customer purchases a product or receives positive feedback. The server obtains a success pattern as output.

[0884] Specific behavior:

[0885] The server uses machine learning algorithms (e.g., random forests) to identify successful patterns.

[0886] The server organizes the successful patterns as talk flows.

[0887] Step 5: Saving to the Database

[0888] The server saves the extracted successful patterns as talk flows in a database. The successful patterns are received as input, and the talk flows saved in the database are obtained as output.

[0889] Specific behavior:

[0890] The server connects to a MySQL database and saves the talk flow using an INSERT statement.

[0891] The server reads and verifies the data from the database to ensure the save was successful.

[0892] Step 6: Load the specifications for the new service

[0893] The server loads the specifications and features of the new service into the AI. As input, it receives the specifications and features of the new service as text data. As output, it obtains the specification data loaded into the AI.

[0894] Specific behavior:

[0895] The server analyzes the specifications and features of the new service as text data.

[0896] The server inputs the analysis results into the artificial intelligence.

[0897] Step 7: Generate optimal proposed talk flow

[0898] The server uses natural language processing with artificial intelligence to generate the optimal proposed talk flow for the new service. As input, it receives past success patterns and new service specification data. As output, it obtains the optimal proposed talk flow.

[0899] Specific behavior:

[0900] The server generates the optimal proposed talk flow based on past success patterns and specification data for new services.

[0901] The server stores the generated talk flow in a database.

[0902] In this way, the server collects and analyzes logs of past sales conversations, stores successful talk flows in a database, and realizes a system that automatically generates optimal talk flows suitable for new services.

[0903] (Application example 1)

[0904] Next, a description will be given of Application Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0905] In traditional customer service work, a store clerk needs a wealth of experience and knowledge to make optimal suggestions to customers, which is difficult for new or inexperienced store clerks. In addition, there is a lack of means to provide real-time conversation flows that allow for quick and accurate responses to customer needs. This has led to issues such as inconsistent customer service quality and a decline in customer satisfaction.

[0906] 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.

[0907] In this invention, the server includes means for storing previously developed talk flows, means for loading specifications and features of new services into an artificial intelligence, means for generating an optimal proposed talk flow tailored to the service, means for displaying the optimal proposed talk flow in real time on a visual device worn by a store clerk, and means for generating the next talk flow based on a conversation with a customer, thereby enabling store clerks to provide customers with consistently high-quality customer service.

[0908] "Past talk flow" refers to the flow and patterns of conversations with customers that have taken place in the past, and by accumulating and analyzing this data, patterns of successful sales conversations can be extracted.

[0909] "Specifications and features of new services" refers to the detailed content and characteristics of the newly provided services, and by inputting this into artificial intelligence, it becomes the basic data for generating optimal proposals for the services.

[0910] "Artificial intelligence" refers to the technology that enables computer systems to mimic human intelligence, learn, reason, and self-correct, generating optimal conversation flow through natural language processing and data analysis.

[0911] "Optimal proposal talk flow" refers to the most effective flow of proposals and responses in conversations with customers, and is generated by artificial intelligence.

[0912] A "visual device" refers to a device worn by a store clerk that visually displays information, such as smart glasses.

[0913] "Real-time" refers to the fact that data processing and information provision are immediate, meaning that the optimal conversation flow is displayed instantly during a conversation with a customer.

[0914] "Means for generating the next talk flow based on a conversation with a customer" refers to technology that analyzes a customer's comments and reactions in real time and uses artificial intelligence to generate the next conversation flow accordingly.

[0915] A system for implementing this invention includes a means for storing previously developed talk flows, a means for loading specifications and features of new services into artificial intelligence, a means for generating an optimal proposed talk flow tailored to the service, a means for displaying the optimal proposed talk flow in real time on a visual device worn by a store clerk, and a means for generating the next talk flow based on a conversation with a customer.

[0916] The server first stores past talk flows in a database. This process involves collecting logs of past sales conversations and extracting patterns of successful sales conversations. As a result, the best talk flows are stored in the database.

[0917] Next, the specifications and features of the new service are input as text data into the AI, which then learns information about the new service and obtains the basic data needed to generate optimal proposals.

[0918] The AI ​​uses natural language processing technology to generate optimal proposed talk flows based on accumulated talk flows and the specifications and features of new services. This generated talk flow is displayed in real time on a visual device (e.g., smart glasses) worn by the store clerk.

[0919] When a store clerk starts a conversation with a customer, they serve them while referring to the optimal proposed talk flow displayed on the visual device. Based on the customer's comments and reactions, the AI ​​generates the next talk flow in real time and displays it on the visual device.

[0920] As a concrete example, the following prompt sentence is input to the generative AI model:

[0921] "Hello, what products are you looking for today?"

[0922] "Customer: I'm looking for a new smartphone."

[0923] Salesperson: Thank you. What features are most important to you?

[0924] Based on this prompt, the generative AI model generates the next conversation flow, which store clerks can refer to in real time through the smart glasses, enabling them to provide consistent, high-quality customer service to customers.

[0925] The hardware used includes smart glasses (e.g., Google Glass), and the software includes SQLite (database management) and OpenAI API (generative AI model), which allows the system to instantly generate and display the optimal conversation flow based on the conversation with the customer.

[0926] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0927] Step 1:

[0928] The server stores past talk flows in a database. Specifically, it collects logs of past sales conversations and analyzes them to extract patterns of successful sales conversations. The input is the log data of past sales conversations, and the output is the patterns of successful talk flows that are saved in the database.

[0929] Step 2:

[0930] The server loads the specifications and features of the new service into the AI ​​as text data. Specifically, the detailed content and characteristics of the new service are input in text format, and the AI ​​learns from them. The input is the text data of the specifications and features of the new service, and the output is the AI ​​in a state where it has learned information about the new service.

[0931] Step 3:

[0932] The server uses artificial intelligence to generate the optimal proposed talk flow. Specifically, it uses natural language processing technology to generate the optimal talk flow based on the accumulated talk flows and the specifications and features of the new service. The inputs are past successful talk flows and the specifications and features of the new service, and the output is the optimal proposed talk flow.

[0933] Step 4:

[0934] The terminal (a visual device worn by the store clerk) displays the generated optimal proposed talk flow in real time. Specifically, it receives the optimal proposed talk flow sent from the server and displays it on the display of the visual device. The input is the optimal proposed talk flow sent from the server, and the output is the talk flow displayed on the visual device.

[0935] Step 5:

[0936] The user (store clerk) converses with the customer while referring to the optimal proposed talk flow displayed through a visual device. Specifically, based on the customer's comments and reactions, the user sends information to the server to generate the next talk flow. The input is the customer's comments and reactions, and the output is the information to generate the next talk flow sent to the server.

[0937] Step 6:

[0938] The server generates the next talk flow based on the conversation with the customer. Specifically, it analyzes the customer's remarks and reactions, and artificial intelligence generates the next talk flow accordingly. The input is data on the customer's remarks and reactions, and the output is the next talk flow generated and sent to the visual device.

[0939] Step 7:

[0940] The terminal (visual device) displays the generated next talk flow in real time. Specifically, it receives the next talk flow sent from the server and displays it on the display of the visual device. The input is the next talk flow sent from the server, and the output is the next talk flow displayed on the visual device.

[0941] Example 2

[0942] Next, a description will be given of Example 2 of Form Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0943] Conventional talk flow generation systems have had difficulty generating talk flows that quickly and accurately reflect the specifications and features of new services. Furthermore, the accumulation of past talk flows and the use of analysis results were not fully utilized, making it difficult to generate talk flows efficiently. This resulted in the issue of creating talk flows when introducing new services taking a great deal of time and effort.

[0944] 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.

[0945] In this invention, the server includes means for storing talk flows that have been developed in the past, means for loading specifications and features of new services into an artificial intelligence, means for receiving specifications and manuals of new services as text data and analyzing them using a text analysis tool, means for inputting the analysis results into a generative AI model and training the AI ​​to understand the content, means for generating talk flows based on the content understood by the AI, and means for providing the generated talk flows. This makes it possible to generate talk flows that quickly and accurately reflect the specifications and features of new services.

[0946] "Talk flows developed in the past" refer to the flow and patterns of dialogue that have been previously executed, and are data that can be accumulated and reused.

[0947] "Specifications and features of new services" refers to information that provides a detailed description and characteristics of the newly provided services, and is the basic data for generating talk flows based on this information.

[0948] "Artificial intelligence" refers to computer systems that mimic human intelligence using techniques such as machine learning and natural language processing.

[0949] "Text data" refers to information expressed as a string of characters, such as specifications, manuals, and other documents expressed in digital format.

[0950] "Text analysis tools" refer to software and algorithms used to analyze text data and extract meaning and structure.

[0951] A "generative AI model" refers to an artificial intelligence model that generates new information or responses based on input data.

[0952] "Talk flow" indicates the flow and pattern of a dialogue, and is a blueprint for smoothly advancing interaction with users.

[0953] "Analysis results" refers to information about the meaning and structure of data obtained by text analysis tools.

[0954] "Learning" refers to the process by which artificial intelligence finds patterns and rules based on input data to improve future predictions and responses.

[0955] "Providing" refers to the act of making the generated talk flow available to users or other systems.

[0956] The present invention relates to a system for generating talk flows that quickly and accurately reflect the specifications and features of new services. A specific embodiment of this system will be described below.

[0957] System configuration

[0958] Subject: Server

[0959] The server receives the specifications and manuals for the new service as text data. Specifically, it receives an HTTP request from the user or device and saves the text data. Next, the server analyzes this text data using a text analysis tool (e.g., NLTK or SpaCy). Based on the analysis results, the server inputs the data into a generative AI model (e.g., GPT-4) and trains the AI ​​to understand the content. Finally, the server generates a talk flow based on the content understood by the AI ​​and provides the generated talk flow to the user or device.

[0960] Subject: Terminal

[0961] The device inputs the specifications and manuals for the new service as text data. The user selects a file using the device's file selection dialog, and the device sends this data to the server. When the server returns the analysis results, the device inputs them into the generative AI model, training the AI ​​to understand the content. Finally, the device displays the talk flow generated by the generative AI model to the user.

[0962] Subject: User

[0963] The user inputs the specifications and manuals for the new service as text data into their device. The user then sends this data to the server via their device, which then analyzes it using a text analysis tool. Once the analysis results are returned from the server, the user inputs the data into the generative AI model via their device, training the AI ​​to understand the content. Finally, the user checks the conversation flow generated by the AI.

[0964] Specific examples

[0965] Example 1: Loading the specifications of a new service into AI

[0966] The user has a PDF specification for the new service. This PDF is converted to text data and entered into the device. The device sends this data to the server, which performs text analysis using NLTK. Based on the analysis results, the server inputs the data into GPT-4, training the AI ​​to understand the content. Finally, the server provides the user with the AI-generated conversation flow.

[0967] Example 2: Loading a new service manual into AI

[0968] The user has a manual for the new service in Word format. The user converts this Word file into text data and inputs it into the device. The device then sends this data to the server, which performs text analysis using SpaCy. Based on the analysis results, the server inputs the data into GPT-4, training the AI ​​to understand the content. Finally, the server provides the user with the AI-generated conversation flow.

[0969] Prompt Sentence Examples

[0970] "Please input the specifications of the new service as text data and analyze it using NLTK. Input the analysis results into GPT-4 and train the AI ​​to understand the content. Finally, please provide the AI-generated conversation flow."

[0971] In this way, the present invention enables the creation of talk flows that quickly and accurately reflect the specifications and features of new services.

[0972] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0973] Step 1:

[0974] The server receives text data such as specifications or manuals for new services from users or terminals. Specifically, the server receives an HTTP request and saves the text data. The input is the text data for specifications or manuals for new services, and the output is the saved text data.

[0975] Step 2:

[0976] The server analyzes the received text data using a text analysis tool (e.g., NLTK or SpaCy). Specifically, the server calls the text analysis tool to perform tokenization, part-of-speech tagging, dependency analysis, etc. on the text data. The input is the saved text data, and the output is the analysis results.

[0977] Step 3:

[0978] The server converts the results of text analysis into a format for input to a generative AI model (e.g., GPT-4). Specifically, the server converts the analysis results into an appropriate data format, such as JSON. The input is the result of text analysis, and the output is the converted data format.

[0979] Step 4:

[0980] The server inputs the converted data into the generative AI model and trains the AI ​​to understand the content. Specifically, the server calls the API of the generative AI model and sends the data. The input is the converted data format, and the output is the content understood by the AI.

[0981] Step 5:

[0982] The server generates a talk flow based on what the generative AI model understands. Specifically, the server receives a response from the AI ​​and executes the logic to construct the talk flow. The input is what the AI ​​understands, and the output is the generated talk flow.

[0983] Step 6:

[0984] The server provides the generated talk flow to the user or terminal. Specifically, the server generates an HTTP response and sends data including the talk flow. The input is the generated talk flow, and the output is the talk flow provided to the user or terminal.

[0985] (Application example 2)

[0986] Next, a description will be given of Application Example 2 of Form Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0987] Whenever a new service is introduced, it is difficult to help users understand its specifications and how to use it. In particular, for complex services and multi-functional platforms, users need a guide to use the service properly. However, with conventional methods, simply referring to manuals or FAQs is insufficient, and there is an issue of making it difficult for users to quickly access the information they are looking for.

[0988] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[0989] In this invention, the server includes means for storing previously deployed talk flows, means for loading specifications and features of new services into an AI, means for generating an optimal proposed talk flow tailored to the service, means for generating answers to user questions, and means for answering user questions based on the generated talk flows, thereby enabling users to smoothly understand and use new services.

[0990] "Talk flow developed in the past" refers to the flow and patterns of conversations that have been carried out in the past, and by storing these, they can be used to help with future conversations.

[0991] "Means of having AI read the specifications and features of new services" refers to a method of inputting detailed descriptions and characteristics of new services into AI as text data, allowing the AI ​​to understand their content.

[0992] "Means for generating optimal proposed talk flow" refers to a method in which AI automatically creates the most suitable dialogue flow for users based on the specifications and features of a new service.

[0993] "Means for generating answers to user questions" refers to the method by which AI receives questions from users and generates appropriate answers to them.

[0994] "Means of answering user questions based on the generated talk flow" refers to a method of providing specific answers to user questions based on the dialogue flow generated by AI.

[0995] A system for implementing this invention is configured as follows: First, the server has a means for storing talk flows that have been developed in the past. This means allows the flow and patterns of previously executed conversations to be saved and used for future conversations.

[0996] The next step involves having the AI ​​learn the specifications and features of the new service. This involves inputting a detailed description and characteristics of the new service into the AI ​​as text data, allowing the AI ​​to understand the content. Specifically, a text analysis tool is used to analyze the manual for the new service, and the AI ​​is then trained on the results.

[0997] Furthermore, the server has a means for generating optimal proposed talk flows, in which AI automatically creates the most suitable dialogue flow for users based on the specifications and features of the new service. It also includes a means for generating answers to users' questions using AI-based natural language processing technology.

[0998] The method for generating answers to user questions involves receiving questions from users and having AI generate appropriate answers to them. It also includes a method for answering user questions based on the generated talk flow, providing specific answers to user questions based on the dialogue flow generated by the AI.

[0999] Hardware and software used

[1000] Hardware: Smartphone

[1001] Software: OpenAI API, Python

[1002] Data processing and calculation

[1003] 1. Reading text data: Read the manual for the new service as a text file.

[1004] 2. Training the AI: Use OpenAI's API to train the AI ​​to learn the contents of the manual.

[1005] 3. Generate talk flow: The AI ​​generates a talk flow based on the manual and saves it in JSON format.

[1006] 4. Generating answers to user questions: Upon receiving a user question, the AI ​​generates an appropriate answer based on the conversation flow.

[1007] Specific examples

[1008] If a user asks, "How do I search for new content?", the AI ​​will answer, based on the contents of the manual, "Enter keywords in the search bar and press the search button."

[1009] Prompt Sentence Examples

[1010] Please answer the user's questions based on the following conversation flow.

[1011] Talk flow:

[1012] 1. How to find new content

[1013] 2. How to view content

[1014] 3. How to set up your account

[1015] User Question: How do I search for new content?

[1016] In this way, users can smoothly understand and use new services.

[1017] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1018] Step 1:

[1019] The server reads the manual for the new service as a text file.

[1020] Input: New service manual (text file)

[1021] Output: Text data

[1022] Specific operation: The server reads the manual for the new service from the specified file path and stores the contents in memory as text data.

[1023] Step 2:

[1024] The server uses a text analysis tool to analyze the read text data.

[1025] Input: Text data

[1026] Output: Analysis results (structured text data)

[1027] Specific operation: The server uses a text analysis tool (for example, a natural language processing library) to analyze the text data and extract sentence structures and keywords.

[1028] Step 3:

[1029] The server trains the AI ​​model based on the analysis results.

[1030] Input: Analysis results (structured text data)

[1031] Output: Trained AI model

[1032] Specific operation: The server uses the OpenAI API to input the analysis results into the AI ​​model, allowing the AI ​​to learn to understand the specifications and features of the new service.

[1033] Step 4:

[1034] The server generates the optimal proposed talk flow using the trained AI model.

[1035] Input: Trained AI model

[1036] Output: Optimal proposal talk flow (JSON format)

[1037] Specific operation: The server inputs the usage scenario of the new service into the AI ​​model, and the AI ​​generates the optimal dialogue flow. The generated dialogue flow is saved in JSON format.

[1038] Step 5:

[1039] The user enters a question using a smartphone.

[1040] Input: User question (text)

[1041] Output: None (passed on to next step)

[1042] Specific operation: The user opens the application on their smartphone and enters a question in text format.

[1043] Step 6:

[1044] The server receives the user's question and generates an answer based on the generated talk flow.

[1045] Input: User question (text), optimal suggestion talk flow (JSON format)

[1046] Output: AI response (text)

[1047] Specific operation: The server inputs the user's question into the AI ​​model and generates an appropriate answer while referring to the generated talk flow.

[1048] Step 7:

[1049] The server sends the generated answer to the user's smartphone.

[1050] Input: AI response (text)

[1051] Output: The answer displayed on the user's smartphone

[1052] Specific operation: The server sends the generated answer to the user's smartphone and displays it in the user's application.

[1053] In this way, users can smoothly understand and use new services.

[1054] Example 3

[1055] Next, a description will be given of Example 3 of Form Example 3. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[1056] Conventional talk flow generation systems have had difficulty quickly and effectively generating talk flows that are optimal for the specifications and features of new services. Furthermore, there were problems with the time it took for users to input information about new services, and with variations in the quality of the generated talk flows. This could lead to ineffective service introductions and proposals, and could result in lower customer satisfaction.

[1057] The specific processing by the specific processing unit 290 of the data processing device 12 in the third embodiment is realized by the following means.

[1058] In this invention, the server includes means for storing previously developed talk flows, means for loading specifications and features of a new service into an AI, means for generating an optimal proposed talk flow tailored to the service, means for a user to input the specifications and features of the new service, means for the server to receive the input information and send a prompt to the generative AI model, means for the server to select a relevant talk flow from a database of stored talk flows, means for the server to generate a new talk flow using the generative AI model, and means for the server to provide the generated talk flow to the user. This makes it possible to quickly and effectively generate a talk flow optimal for the specifications and features of a new service, reducing the user's effort and providing a high-quality talk flow.

[1059] "Past-developed talk flows" refer to previously used dialogue flows and patterns, which can be stored in a database to make them reusable.

[1060] "Specifications and features of new services" refers to detailed descriptions and characteristics of newly provided services, and is information used to generate optimal talk flows.

[1061] "Artificial intelligence" refers to the technology that allows computer systems to mimic human intelligence to learn, reason, and self-correct, and is used in this case to generate talk flow.

[1062] "Optimal proposal talk flow" refers to the dialogue flow that is most suited to the specifications and features of a new service, and is intended to effectively introduce the service to customers.

[1063] "Means for users to input specifications and features of new services" refers to an interface or method by which users can provide information about new services to the system.

[1064] "Means for the server to receive input information and send prompts to the generative AI model" refers to a method for receiving information provided by a user and issuing instructions to the generative AI model based on that information.

[1065] "Means for the server to select relevant talk flows from a database of accumulated talk flows" refers to a method for selecting past talk flows stored in a database that are relevant to the new service.

[1066] "Means for the server to generate a new talk flow using a generative AI model" refers to a method for creating a new talk flow using a generative AI model based on a selected talk flow.

[1067] The term "means by which the server provides the generated talk flow to the user" refers to a method for displaying or transmitting the generated talk flow to the user.

[1068] MODE FOR CARRYING OUT THE INVENTION

[1069] The present invention relates to a system for generating talk flows that are optimal for the specifications and features of new services. Specific embodiments of this system will be described below.

[1070] System configuration

[1071] This system consists of three main components: a server, a terminal, and a user. The server contains a database of previously deployed talk flows, a generative AI model (e.g., OpenAI's GPT-4), and a talk flow generation algorithm. The terminal provides an interface for users to input specifications and features of new services.

[1072] Program processing

[1073] The user uses the terminal to input the specifications and features of the new service. For example, the user may input information such as "This is an eco-friendly product made from renewable materials." The terminal then sends this information to the server.

[1074] The server receives the information sent by the user and creates a prompt to send to the generative AI model based on that information. For example, it generates a prompt such as, "Please generate a talk flow to introduce a new eco-friendly product. This product is made from environmentally friendly materials and is highly energy efficient."

[1075] Next, the server selects a relevant talk flow from a database of stored talk flows based on the prompt sentence. This selection is done using natural language processing technology. For example, a talk flow that emphasizes the benefits of eco-friendly products or a talk flow that suggests an environmentally friendly lifestyle may be selected.

[1076] The server generates a new talk flow based on the selected talk flow using a generative AI model. The generative AI model understands the context and flow of the talk flow and reconstructs it in a way that is optimal for the new service.

[1077] The generated talk flow might look something like this:

[1078] 1. "Hello, today I'd like to introduce you to a new eco-friendly product."

[1079] 2. "This product is made from renewable materials and is manufactured with maximum energy efficiency."

[1080] 3. "Furthermore, using this product will help reduce energy consumption in daily life."

[1081] Finally, the server sends the generated talk flow to the user's device, where the user can view the talk flow on the device screen and make corrections or adjustments as necessary.

[1082] Specific examples

[1083] For example, if the new service is a product that emphasizes "ecology," the user can input the following prompt into the generative AI model:

[1084] Example prompt sentence:

[1085] "Generate a talk flow to introduce a new eco-friendly product. This product is made from environmentally friendly materials and is highly energy efficient."

[1086] Based on this prompt, the server generates an optimal talk flow and provides it to the user. In this way, the user can quickly and effectively generate a talk flow that is optimal for the specifications and features of the new service. The flow of the specification process in the third embodiment will be explained with reference to Figure 15.

[1087] Program processing flow

[1088] Step 1:

[1089] Users input specifications and features of new services

[1090] The user uses the terminal to input the specifications and features of the new service. For example, the user might input "It's an eco-friendly product made from renewable materials." This input data is sent to the server in text format.

[1091] input:

[1092] Specifications and features of the new service (e.g., "It is an ecological product and uses renewable materials")

[1093] output:

[1094] Text data sent to the server

[1095] Step 2:

[1096] The server receives the input information and sends a prompt to the generative AI model.

[1097] The server receives the information sent by the user and creates a prompt to send to the generative AI model based on that information. For example, it generates a prompt such as, "Please generate a talk flow to introduce a new eco-friendly product. This product is made from environmentally friendly materials and is highly energy efficient."

[1098] input:

[1099] Text data sent by the user

[1100] Data processing:

[1101] Generate prompt sentences (e.g., "Generate a talk flow to introduce a new eco-friendly product. This product is made from environmentally friendly materials and is highly energy efficient.")

[1102] output:

[1103] Generated prompt statement

[1104] Step 3:

[1105] The server selects relevant talk flows from its database of accumulated talk flows.

[1106] The server uses natural language processing technology to select a relevant talk flow from a database of stored talk flows based on the generated prompt. For example, a talk flow emphasizing the benefits of eco-friendly products or a talk flow proposing an environmentally friendly lifestyle may be selected.

[1107] input:

[1108] Generated prompt statement

[1109] Data processing:

[1110] Select relevant talk flows using natural language processing technology

[1111] output:

[1112] Selected Talk Flow

[1113] Step 4:

[1114] The server generates a new talk flow using the generative AI model.

[1115] The server generates a new talk flow based on the selected talk flow using a generative AI model. The generative AI model understands the context and flow of the talk flow and reconstructs it in a way that is optimal for the new service.

[1116] input:

[1117] Selected Talk Flow

[1118] Data processing:

[1119] Generate new conversation flows using generative AI models

[1120] output:

[1121] The new talk flow generated

[1122] Step 5:

[1123] The server provides the generated talk flow to the user.

[1124] The server sends the generated talk flow to the user's device, where the user can check the talk flow on the device screen and make corrections or adjustments as necessary.

[1125] input:

[1126] The new talk flow generated

[1127] output:

[1128] Dialogue flow sent to the user's device

[1129] Specific examples of operation

[1130] For example, if the new service is a product that emphasizes "ecology," the user can input the following prompt into the generative AI model:

[1131] Example prompt sentence:

[1132] "Generate a talk flow to introduce a new eco-friendly product. This product is made from environmentally friendly materials and is highly energy efficient."

[1133] Based on this prompt, the server generates an optimal talk flow and provides it to the user. In this way, the user can quickly and effectively generate a talk flow that is optimal for the specifications and features of a new service.

[1134] (Application example 3)

[1135] Next, a description will be given of Application Example 3 of Form Example 3. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[1136] Conventional talk flow generation systems have had difficulty quickly and accurately generating optimal talk flows that match the specifications and features of new services. Furthermore, they were unable to provide users with the information they wanted in real time, making improving the user experience a challenge.

[1137] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 3 is realized by the following means. In this invention, the server includes means for storing talk flows developed in the past, means for loading specifications and features of new services into AI, means for generating an optimal proposed talk flow tailored to the service, means for inputting features of products searched by the user, and means for displaying the generated talk flow to the user. This makes it possible to provide the information desired by the user in real time and improve the user experience.

[1138] "Past talk flows" refers to previously used dialogue flows and patterns, and are data that can be accumulated and reused.

[1139] "New service specifications" refers to information that indicates the detailed design, functions, and characteristics of a newly provided service.

[1140] "Features" is information that indicates the characteristics and features of a service or product.

[1141] "AI" is an abbreviation for artificial intelligence, a system that analyzes data and makes decisions using technologies such as machine learning and natural language processing.

[1142] "Optimal proposal talk flow" is a process that generates the most appropriate dialogue flow based on the specifications and features of a specific service or product.

[1143] "Characteristics of the product searched by the user" refers to information that indicates the characteristics or features of the product that the user searches for on the online platform.

[1144] "Generated talk flow" refers to the flow and patterns of dialogue generated by AI based on the specifications and features of the new service.

[1145] The "means for displaying to the user" refers to a method or technology for visually presenting the generated talk flow to the user.

[1146] A system for implementing this invention includes a server, a user terminal, and an AI model. The server stores previously deployed talk flows and loads the specifications and features of new services into the AI. The user terminal inputs the features of the product searched by the user and displays the generated talk flows.

[1147] Hardware and software used

[1148] Hardware: Servers, smartphones

[1149] Software: Python, OpenAI API

[1150] Data processing and calculation

[1151] 1. Data entry: The user uses a smartphone to enter the characteristics of the product they are searching for, for example, "products made with renewable energy, recyclable packaging."

[1152] 2. Prompt generation: Based on the input product characteristics, the server generates a prompt sentence to be input into the generative AI model.

[1153] 3. AI model call: The server uses OpenAI's API to send prompts to a generative AI model (e.g., GPT-3) to generate the optimal conversation flow.

[1154] 4. Displaying the results: The server sends the generated talk flow to the user's terminal and displays it to the user.

[1155] Specific examples

[1156] If a user searches for eco-friendly products that have the characteristics of "products that use renewable energy and recyclable packaging," the following prompt sentence will be generated:

[1157] Prompt Sentence Examples

[1158] Generate a talk flow about a new eco-friendly product. Product features: Renewable energy-based product, recyclable packaging

[1159] The following is an example of a generated talk flow:

[1160] This product is made using renewable energy, making it an eco-friendly choice, and the packaging is recyclable, reducing waste. Perfect for those who want to live an eco-conscious life.

[1161] In this way, the information desired by the user can be provided in real time, improving the user experience.

[1162] The flow of the specific processing in Application Example 3 will be described with reference to FIG.

[1163] Step 1:

[1164] The user uses a smartphone to input the characteristics of the product they are searching for, such as "products that use renewable energy, recyclable packaging." The input data is sent to the server in text format.

[1165] Step 2:

[1166] The server receives the product features sent by the user. Based on the received text data, it generates a prompt sentence to be input into the generative AI model. Specifically, it creates a prompt sentence that includes the product features.

[1167] Step 3:

[1168] The server calls the OpenAI API using the generated prompt sentence. The prompt sentence is sent to the API, and the optimal talk flow is generated by a generative AI model (e.g., GPT-3). The input is the prompt sentence, and the output is the generated talk flow.

[1169] Step 4:

[1170] The server receives the generated talk flow and converts it into an appropriate format for display to the user. Specifically, it converts text data into a format that is easy for the user to understand.

[1171] Step 5:

[1172] The server sends the formatted talk flow to the user's smartphone. The user's device displays the received talk flow on its screen. The user can check the displayed talk flow and obtain detailed product information.

[1173] This series of processes allows users to receive the optimal talk flow in real time based on the characteristics of the product they searched for.

[1174] 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.

[1175] "Example 1"

[1176] As one embodiment of the present invention, a system is provided that includes an emotion engine in which AI recognizes a user's emotions. In this system, the emotion engine recognizes the user's emotions in real time during a dialogue with the user and generates an optimal talk flow according to the emotion. For example, if the user expresses joy, the emotion engine conveys that information to the AI, and the AI ​​generates a talk flow corresponding to joy.

[1177] "Example 2"

[1178] In another embodiment of the present invention, a system is provided in which an emotion engine includes means for recognizing emotions from the tone of a user's voice. This system analyzes audio information such as the tone, volume, and speaking rate of the user's voice, and estimates the user's emotional state from the results. For example, if the user's voice tone is high and the speaking rate is fast, it is determined that the user is likely excited.

[1179] "Example 3"

[1180] In a further embodiment of the present invention, the talk flow is adjusted according to the recognized user's emotions.

[1181] In this system, AI recognizes the user's emotions and generates a talk flow that is optimal for those emotions. For example, if the user is expressing anger, the AI ​​uses that information to generate a talk flow that calms the user.

[1182] The processing flow of each embodiment will be described below.

[1183] "Example 1"

[1184] Step 1: A user interaction is initiated.

[1185] Step 2: The emotion engine recognizes the user's emotions in real time.

[1186] Step 3: The recognized emotional information is conveyed to the AI.

[1187] Step 4: The AI ​​generates the optimal conversation flow based on the emotion.

[1188] "Example 2"

[1189] Step 1: The user's voice information (tone of voice, volume, speaking rate, etc.) is analyzed by the emotion engine.

[1190] Step 2: The user's emotional state is inferred from the analysis results.

[1191] Step 3: The estimated emotional information is conveyed to the AI.

[1192] Step 4: The AI ​​generates the optimal conversation flow based on the emotion.

[1193] "Example 3"

[1194] Step 1: A user interaction is initiated.

[1195] Step 2: The emotion engine recognizes the user's emotions in real time.

[1196] Step 3: The recognized emotional information is conveyed to the AI.

[1197] Step 4: The AI ​​generates the optimal conversation flow based on the emotion.

[1198] Step 5: The generated talk flow is presented to the user.

[1199] Example 1

[1200] Next, a description will be given of Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[1201] Conventional talk flow generation systems were unable to effectively utilize past conversation data, making it difficult to generate optimal talk flows in real time that matched the user's emotions. Furthermore, there was a lack of a way to quickly generate talk flows that reflected the specifications and features of new services. This meant that improvements in sales efficiency and customer satisfaction were not fully achieved.

[1202] 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.

[1203] In this invention, the server includes means for storing talk flows developed in the past, means for inputting specifications and features of new services into an artificial intelligence, means for generating an optimal proposed talk flow tailored to the service, means for collecting log data of past sales conversations, means for analyzing the collected log data to extract patterns of successful sales conversations, means for recognizing user emotions in real time, and means for generating an optimal talk flow based on the recognized emotion information. This makes it possible to effectively utilize patterns of successful sales conversations in the past and generate an optimal talk flow in real time that corresponds to the user's emotions.

[1204] "Talk flow" refers to a series of statements and responses in a dialogue with a user.

[1205] "Artificial intelligence" refers to the technology that enables computer systems to mimic human intelligence to learn, reason, perceive, and make decisions.

[1206] "Log Data" refers to a series of events and activities that a system or application records during its operation.

[1207] "Natural language processing" refers to the technology that enables computers to understand, generate, and manipulate human language.

[1208] "Emotion recognition" refers to the technology of identifying and classifying emotions from a user's voice or text.

[1209] "Optimal proposal talk flow" refers to the dialogue flow that is most suitable for a specific situation or condition.

[1210] A "database" refers to a system for efficiently storing, managing, and retrieving data.

[1211] "Pattern extraction" refers to the process of finding specific patterns or characteristics in data.

[1212] "Real-time" refers to data processing and response occurring immediately.

[1213] This invention is a system that collects log data of past sales conversations, analyzes it to extract patterns of successful sales conversations, recognizes user emotions in real time, and generates optimal talk flows. This system operates in cooperation with a server, a terminal, and a user.

[1214] Hardware and software used

[1215] Hardware: Servers, terminals (PCs, smartphones)

[1216] Software: Database management systems (e.g., MySQL), emotion recognition engines (e.g., IBM Watson), generative AI models (e.g., GPT-4)

[1217] Data collection

[1218] The server collects log data of past sales conversations. Specifically, it obtains log data from sales systems and customer relationship management systems (CRM). The collected data is saved in text and audio formats.

[1219] Data analysis

[1220] The server analyzes the collected sales conversation log data, analyzes the content of the conversations using natural language processing (NLP) techniques, and extracts patterns of successful sales conversations using NLP libraries (e.g., spaCy).

[1221] Saving to a database

[1222] The server stores the extracted talk flows in a database. The server uses a database management system (e.g., MySQL) to store the talk flows as structured data.

[1223] User Emotion Recognition

[1224] The device activates an emotion engine during interaction with the user. It uses an emotion recognition engine (e.g., IBM Watson) to recognize emotions from the user's voice and text in real time.

[1225] Generate talk flow

[1226] The server generates an optimal talk flow using a generative AI model based on the emotion information. Specifically, it inputs prompt sentences into the generative AI model (e.g., GPT-4) and generates an appropriate talk flow.

[1227] Specific examples

[1228] Examples of data collection

[1229] The server collects sales conversation logs from the past year, including text and audio data of conversations.

[1230] Specific examples of emotion recognition

[1231] If a user says, "This product is absolutely amazing!", the emotion engine recognizes the emotion "delight."

[1232] Example of talk flow generation

[1233] If the emotion engine recognizes "joy," the server inputs the following prompt sentence into the generative AI model:

[1234] Generate the optimal conversation flow when the user is expressing the emotion of joy.

[1235] The generative AI model generates the following conversation flow:

[1236] "Thank you! I'm glad you're a satisfied customer. Please let me know if there are any other products you're interested in."

[1237] This system effectively utilizes patterns of past successful sales conversations and can generate optimal conversation flows in real time according to the user's emotions.

[1238] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1239] Step 1:

[1240] Data collection

[1241] The server collects log data of past sales conversations. Specifically, it obtains log data from the sales system and customer relationship management system (CRM). The input is log data from the sales system and CRM, and the output is the collected log data stored in temporary storage within the server. The server accesses the CRM system via an API and obtains sales conversation logs from the past year.

[1242] Step 2:

[1243] Data analysis

[1244] The server analyzes the collected sales conversation log data. Specifically, it uses natural language processing (NLP) technology to analyze the content of the conversation. The input is the collected log data, and the output is the extraction of patterns of successful sales conversations. The server tokenizes the log data using an NLP library (e.g., spaCy) and extracts important keywords and phrases. The server identifies the characteristics of successful sales conversations and classifies them as excellent talk flows.

[1245] Step 3:

[1246] Saving to a database

[1247] The server saves the extracted good talk flows in a database. The input is the extracted talk flows, and the output is the talk flows saved in the database. The server connects to the database and executes SQL queries to insert the talk flow data. The saved talk flows are stored in the database for later retrieval and reference.

[1248] Step 4:

[1249] User Emotion Recognition

[1250] The device activates an emotion engine during a dialogue with the user. Specifically, it uses an emotion recognition engine (e.g., IBM Watson) to recognize emotions from the user's voice or text in real time. The input is the user's speech, and the output is the recognized emotion information. The device receives the user's speech as voice input or text input, sends the input data to the emotion recognition engine, and receives the results of the emotion analysis.

[1251] Step 5:

[1252] Generate talk flow

[1253] The server generates an optimal talk flow using a generative AI model based on the emotion information. Specifically, it inputs a prompt into the generative AI model (e.g., GPT-4) to generate an appropriate talk flow. The inputs are the recognized emotion information and the prompt, and the output is the generated talk flow. If the emotion engine recognizes "joy," the server inputs the following prompt into the generative AI model:

[1254] Generate the optimal conversation flow when the user is expressing the emotion of joy.

[1255] The generative AI model generates the following conversation flow:

[1256] "Thank you! I'm glad you're a satisfied customer. Please let me know if there are any other products you're interested in."

[1257] The server transmits the generated talk flow to the terminal, which uses the talk flow for interaction with the user.

[1258] (Application example 1)

[1259] Next, a description will be given of Application Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[1260] Conventional customer service systems have difficulty recognizing customer emotions in real time and providing optimal conversation flows accordingly. As a result, improvements in customer satisfaction and sales efficiency have not been fully achieved. In addition, there are limited ways to effectively utilize successful conversation flows from the past.

[1261] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means. In this invention, the server includes means for storing previously developed talk flows, means for loading specifications and features of new services into an AI, means for generating an optimal proposed talk flow tailored to the service, means for recognizing a user's emotions in real time, means for selecting an optimal talk flow based on the recognized emotions, and means for displaying the selected talk flow. This makes it possible to provide an optimal talk flow based on a customer's emotions in real time, thereby improving customer satisfaction and sales efficiency.

[1262] "Past talk flow" refers to a series of conversation patterns used in past sales conversations and customer service.

[1263] "Specifications and features of the new service" refers to the detailed content and characteristics of the newly provided service.

[1264] "Means of loading data into AI" refers to a method of inputting data into artificial intelligence and having it learn and analyze based on that data.

[1265] "Means for generating optimal proposed talk flow" refers to a method for creating the most effective conversation pattern according to specific conditions and situations.

[1266] "Means for recognizing a user's emotions in real time" refers to a method for instantly determining a user's emotions at that time from their facial expressions, tone of voice, etc.

[1267] "Means for selecting an optimal talk flow based on recognized emotions" refers to a method for selecting the most appropriate conversation pattern depending on the user's emotional state.

[1268] "Means for displaying the selected talk flow" refers to a method for visually presenting the selected conversation pattern.

[1269] A system for implementing this invention includes a means for storing previously deployed talk flows, a means for loading the specifications and features of a new service into an AI, a means for generating an optimal proposed talk flow tailored to the service, a means for recognizing a user's emotions in real time, a means for selecting an optimal talk flow based on the recognized emotions, and a means for displaying the selected talk flow.

[1270] System Program

[1271] This system is realized using the following hardware and software.

[1272] Hardware

[1273] Smart glasses: equipped with a camera and microphone

[1274] Server: A server to host the database and AI models

[1275] software

[1276] OpenCV: Image processing library

[1277] Keras: A deep learning library

[1278] SpeechRecognition: A speech recognition library

[1279] Data processing and calculation

[1280] Image Processing

[1281] The images captured by the smart glasses camera are processed using OpenCV to extract facial features, enabling real-time analysis of the user's facial expressions.

[1282] emotion recognition

[1283] The extracted facial image is input into a Keras model to recognize emotions, allowing us to instantly grasp the user's emotional state.

[1284] Talk flow generation

[1285] Based on the recognized emotion, the system selects the optimal talk flow from the database, thereby providing the optimal conversation pattern according to the user's emotion.

[1286] display

[1287] The selected talk flow is displayed on the smart glasses' display, allowing store staff to provide service while checking the optimal talk flow in real time.

[1288] Specific examples

[1289] For example, suppose a customer is looking for a new smartphone. The smart glasses recognize the customer's "interest" from their facial expression and select "explanations for customers who are interested" from past successful conversation flows. The store clerk then follows that conversation flow to recommend the best smartphone for the customer.

[1290] Prompt Sentence Examples

[1291] When a customer is looking for a new smartphone, the smart glasses recognize "interest" from the customer's facial expression. Select an "explanation for an interested customer" from a past successful conversation flow and present it to the store associate.

[1292] In this way, it is possible to provide the optimal conversation flow in real time according to the customer's emotions, thereby improving customer satisfaction and sales efficiency.

[1293] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1294] Step 1:

[1295] The camera in the smart glasses captures the customer's video. The input is real-time video data, and the output is the captured video frame. The smart glasses then transmits the video frame to the server.

[1296] Step 2:

[1297] The server extracts facial features from video frames using OpenCV. The input is the captured video frame and the output is the extracted face image. The server runs a face detection algorithm to identify the face location.

[1298] Step 3:

[1299] The server recognizes emotions from extracted facial images using a Keras model. The input is the extracted facial image, and the output is the recognized emotion label (e.g., joy, sadness, interest, etc.). The server preprocesses the facial image and inputs it to the emotion recognition model.

[1300] Step 4:

[1301] The server selects the optimal talk flow from the database based on the recognized emotion. The input is the recognized emotion label, and the output is the optimal talk flow. The server searches the database for the talk flow corresponding to the emotion label.

[1302] Step 5:

[1303] The server displays the selected talk flow on the display of the smart glasses. The input is the optimal talk flow, and the output is the talk flow displayed on the display of the smart glasses. The server sends the talk flow in text format to the smart glasses.

[1304] Step 6:

[1305] The salesperson responds to the customer according to the talk flow displayed on the smart glasses display. The input is the displayed talk flow, and the output is the conversation with the customer. The salesperson makes the best suggestions to the customer based on the talk flow.

[1306] Step 7:

[1307] After serving the customer, the salesperson inputs feedback into the system. The input is the feedback data, and the output is an updated talk flow database. The server receives the feedback and updates the talk flow database.

[1308] Example 2

[1309] Next, a description will be given of Example 2 of Form Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[1310] With conventional systems, it was difficult to efficiently understand the specifications and features of new services and generate optimal conversation flows based on them. It was also difficult to analyze the user's emotional state in real time and respond appropriately. This hindered improvements to the user experience.

[1311] 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.

[1312] In this invention, the server includes means for storing previously developed talk flows, means for loading specifications and features of a new service into an artificial intelligence, means for generating an optimal proposed talk flow tailored to the service, means for analyzing specifications and manuals for the new service as text data, means for collecting user voice data and converting it into text data using voice recognition software, and means for transmitting the converted text data to an emotion analysis engine and analyzing the user's emotional state. This makes it possible to efficiently understand the specifications and features of the new service, generate an optimal talk flow, and further analyze the user's emotional state in real time to take appropriate action.

[1313] The "means for storing talk flows that have been developed in the past" is a function for saving data on talk flows that have been executed in the past so that they can be referenced later.

[1314] "Means for loading the specifications and features of new services into artificial intelligence" refers to a function that inputs information such as specifications and manuals for new services into artificial intelligence and allows it to understand their contents.

[1315] The "means for generating optimal proposed talk flow" is a function that enables artificial intelligence to automatically create optimal talk flow based on the specifications and features of a new service.

[1316] "Means for analyzing specifications and manuals for new services as text data" refers to a function for importing documents related to new services as text data and analyzing the contents using an analysis tool.

[1317] "Means for collecting user voice data and converting it into text data using voice recognition software" refers to a function for collecting user voice and converting it into text data using voice recognition technology.

[1318] The "means for sending the converted text data to an emotion analysis engine and analyzing the user's emotional state" is a function for inputting the text data obtained by speech recognition into an emotion analysis engine and estimating the user's emotional state.

[1319] MODE FOR CARRYING OUT THE INVENTION

[1320] This invention is a system that efficiently understands the specifications and features of new services and generates optimal talk flows based on them. It is also a system that can analyze the user's emotional state in real time and respond appropriately.

[1321] Server Operation

[1322] The server first receives the specifications and manuals for the new service as text data from the user. This text data is saved in the server's storage. The server then analyzes the saved text data using a text analysis tool (e.g., NLTK or SpaCy). The analysis results are input into a generative AI model (e.g., OpenAI's GPT-4). The generative AI model generates an optimal talk flow based on the input data. The generated talk flow is then provided to the user by the server.

[1323] As a concrete example, when analyzing the manual for a new cloud service, the server first obtains the text data of the manual and performs grammatical analysis using NLTK. The analysis results are then input into GPT-4 to generate a talk flow that explains how to use the service and its features.

[1324] Example prompt sentence:

[1325] "Please analyze the manual for the following cloud service and generate a conversation flow for users to use the service."

[1326] Device behavior

[1327] The device collects the user's voice data in real time. It captures the voice through a microphone and sends it to speech recognition software (e.g., Google Speech-to-Text API). The speech recognition software converts the voice data into text data. The converted text data is sent to an emotion analysis engine (e.g., IBM Watson Tone Analyzer) to analyze the user's emotional state. The analysis results are displayed to the user on the device.

[1328] For example, when a user calls customer support, the device collects the user's voice in real time and converts it into text using the Google Speech-to-Text API.Then, it uses IBM Watson Tone Analyzer to analyze the user's emotions from the text data and determine whether the user is excited or not.

[1329] Example prompt sentence:

[1330] "Analyze the user's voice data and estimate their emotional state based on their tone of voice, volume, and speaking rate."

[1331] User Actions

[1332] Users provide the system with specifications and manuals for new services. The text data provided by the user is analyzed by the server and used as data for AI learning. The user can then check the talk flow generated based on the analysis results and make any necessary corrections.

[1333] As a concrete example, when a user uploads a manual for new software to the system, the user first prepares the text data of the manual and uploads it to the system. The server then analyzes it and provides the generated talk flow to the user. The user can then check the talk flow and make any necessary corrections.

[1334] Example prompt sentence:

[1335] "Upload the new software manual and generate the talk flow."

[1336] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1337] Server Processing Steps

[1338] Step 1:

[1339] The server receives specifications and manuals for new services as text data from users, and stores the text data uploaded by users in the server's storage.

[1340] Input: New service specifications and manuals (text data)

[1341] Output: Text data saved in storage

[1342] Specific operation: The user uploads "New Cloud Service Manual.txt" and the server saves it in storage.

[1343] Step 2:

[1344] The server analyzes the stored text data using text analysis tools (e.g., NLTK or SpaCy), performing grammatical analysis and keyword extraction.

[1345] Input: Text data saved in storage

[1346] Output: Analysis results (grammar analysis data, keywords)

[1347] Specific operation: The server analyzes "New Cloud Service Manual.txt" using NLTK and extracts important keywords and phrases.

[1348] Step 3:

[1349] The server inputs the analysis results into a generative AI model (e.g., OpenAI's GPT-4), which generates a talk flow based on the input data.

[1350] Input: Analysis results (grammar analysis data, keywords)

[1351] Output: Generated talk flow

[1352] Specific operation: The server inputs the extracted data into GPT-4 and generates a talk flow.

[1353] Step 4:

[1354] The server obtains the talk flow output from the generative AI model and provides it to the user.

[1355] Input: Generated talk flow

[1356] Output: The talk flow provided to the user

[1357] Specific operation: The server provides the generated talk flow to the user so that the user can check it.

[1358] Terminal processing steps

[1359] Step 1:

[1360] The device collects the user's voice data in real time by capturing speech through a microphone and sending it to speech recognition software (e.g., Google Speech-to-Text API).

[1361] Input: User's voice data

[1362] Output: Voice data sent to speech recognition software

[1363] Specific operation: The device uses a microphone to collect audio when the user calls customer support.

[1364] Step 2:

[1365] The device converts the voice data into text data using voice recognition software and temporarily stores the converted text data.

[1366] Input: Voice data sent to speech recognition software

[1367] Output: Text data

[1368] Specific operation: The voice is converted into text using the Google Speech-to-Text API, and information such as "the user's voice is high-pitched and they speak quickly" is extracted.

[1369] Step 3:

[1370] The device sends the converted text data to an emotion analysis engine (e.g., IBM Watson Tone Analyzer) to analyze the user's emotional state.

[1371] Input: Text data

[1372] Output: Emotion analysis results

[1373] Specific operation: Using IBM Watson Tone Analyzer, the system analyzes the user's emotions from text data and infers that the user is excited.

[1374] Step 4:

[1375] The device displays the results obtained from the emotion analysis engine to the user, providing feedback according to the user's emotional state.

[1376] Input: Sentiment analysis results

[1377] Output: Sentiment analysis results displayed to the user

[1378] Specific operation: Display the estimation results to the user and prompt them to take appropriate action.

[1379] User processing steps

[1380] Step 1:

[1381] Users provide specifications and manuals for new services to the system, and the uploaded data is sent to the server.

[1382] Input: New service specifications and manuals (text data)

[1383] Output: Text data sent to the server

[1384] Specific operation: The user uploads "New Software Manual.pdf" to the system.

[1385] Step 2:

[1386] The user checks the analysis results and generated talk flow provided by the server and makes corrections as necessary.

[1387] Input: Analysis results and talk flow provided by the server

[1388] Output: Modified talk flow

[1389] Specific operation: The user checks the talk flow and makes corrections such as "I would like this part to be explained in more detail."

[1390] Step 3:

[1391] Users can provide feedback on the generated talk flow and sentiment analysis results, which will be used to improve the system.

[1392] Input: Generated talk flow and sentiment analysis results

[1393] Output: Feedback

[1394] Specific operation: The user provides feedback on the sentiment analysis results, such as "This analysis was accurate."

[1395] (Application example 2)

[1396] Next, a description will be given of Application Example 2 of Form Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[1397] Conventional talk flow generation systems had difficulty quickly reflecting the specifications and features of new services, and were unable to properly recognize and respond to users' emotional states. Furthermore, they were unable to issue appropriate warnings even in emergency situations where a quick response was required.

[1398] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[1399] In this invention, the server includes means for storing previously deployed talk flows, means for loading the specifications and features of a new service into an AI, means for generating an optimal proposed talk flow tailored to the service, means for analyzing a user's voice and converting it into text data, means for analyzing the user's tone, volume, and speaking rate to estimate their emotional state, means for responding to the user based on the generated talk flow, and means for issuing a warning when an emergency is detected. This allows the specifications and features of the new service to be quickly reflected, making it possible to appropriately recognize and respond to the user's emotional state. It also enables a rapid response in an emergency.

[1400] A "talk flow" is a series of steps and responses that show the flow of a conversation with a user.

[1401] A "New Service Specification" is a document that describes the detailed design, functions, and features of a newly provided service.

[1402] "AI" is an abbreviation for artificial intelligence, a technology that allows machines to learn and reason by imitating human intelligence.

[1403] "Text data" refers to data that represents character information in a digital format.

[1404] "Natural language processing" is a technology that allows computers to understand, interpret, and generate human language.

[1405] "Voice analysis" is a technology that analyzes voice data and understands its content.

[1406] "Emotional state" refers to the user's emotional or psychological state.

[1407] An "emergency" is a situation in which a danger or problem exists that requires immediate action.

[1408] A "warning" is a notice or alert intended to notify you of a danger or problem.

[1409] A system for implementing this invention has the following configuration: First, the server has a means for storing talk flows that have been developed in the past. This means allows past dialogue data to be stored and used to generate future talk flows.

[1410] Next, we have a method for loading the specifications and features of the new service into the AI. This method involves loading the specifications and manuals of the new service into the AI ​​as text data, allowing the AI ​​to understand their contents. This makes it possible to generate the optimal conversation flow for the new service.

[1411] Furthermore, it has a means to analyze the user's voice and convert it into text data. This means uses speech recognition technology to convert the user's voice into text data and analyzes the content. Specifically, it collects voice using the smartphone's microphone and converts it into text data using the speech_recognition library.

[1412] It also has a means to estimate the user's emotional state by analyzing the tone, volume, and speaking rate of the user's voice. This means analyzes the audio data and estimates the user's emotional state. For example, if the volume is loud and the speaking rate is fast, it is determined that the user is excited. This analysis uses the numpy library.

[1413] It also has a means to respond to users based on the generated talk flow. This means uses a generative AI model (e.g., GPT-3) to generate appropriate answers to users' questions. Specifically, it uses the transformers library to generate talk flows and respond to users.

[1414] Furthermore, it has a means to issue a warning when an emergency situation is detected. This means analyzes the user's emotional state and issues a warning when it is judged to be an emergency. For example, if the user says "help me," it will immediately issue a warning.

[1415] For example, if a user says, "My home security system isn't working," the following prompt sentence can be fed into the generative AI model:

[1416] User: My home security system isn't working.

[1417] AI: Is your security system turned on and does your internet connection work?

[1418] In this way, by generating an appropriate talk flow for a user's question and performing emotion recognition, it is possible to provide faster and more appropriate support.

[1419] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1420] Step 1:

[1421] The user speaks their question or problem.

[1422] Input: User's voice data

[1423] Output: Audio data

[1424] Specific action: The user speaks a question or problem into the smartphone microphone.

[1425] Step 2:

[1426] The terminal converts the voice data into text data.

[1427] Input: Audio data

[1428] Output: Text data

[1429] Specific operation: The device uses the speech_recognition library to convert voice data into text data.

[1430] Step 3:

[1431] The server analyzes the text data and generates an appropriate talk flow.

[1432] Input: Text data

[1433] Output: Talkflow

[1434] Specific operation: The server uses a generative AI model (e.g., GPT-3) to generate a talk flow based on text data. It uses the transformers library.

[1435] Step 4:

[1436] The server analyzes the user's tone of voice, volume, and speaking rate to estimate their emotional state.

[1437] Input: Audio data

[1438] Output: Emotional state

[1439] Specific operation: The server uses the numpy library to analyze the tone, volume, and speaking rate of the audio data and estimate the emotional state.

[1440] Step 5:

[1441] The server responds to the user based on the generated talk flow.

[1442] Input: Talk flow, emotional state

[1443] Output:Answer

[1444] Specific operation: The server uses the generated talk flow to generate an appropriate response for the user and sends it to the terminal.

[1445] Step 6:

[1446] If the server detects an emergency, it will issue an alert.

[1447] Input: Emotional state

[1448] Output: warning

[1449] Specific behavior: The server analyzes the user's emotional state and issues a warning if it determines that the user is in an emergency. For example, if the user says "help me," it will immediately issue a warning.

[1450] Example 3

[1451] Next, a description will be given of Example 3 of Form Example 3. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[1452] Conventional talk flow generation systems have had issues with generating talk flows that are optimal for the specifications and features of new services, and are unable to adjust the talk flow according to the user's emotions. This has resulted in a poor user experience and made it difficult to effectively propose services.

[1453] The specific processing by the specific processing unit 290 of the data processing device 12 in the third embodiment is realized by the following means.

[1454] In this invention, the server includes means for storing previously developed talk flows, means for inputting specifications and features of a new service into an artificial intelligence, means for generating an optimal proposed talk flow tailored to the service, means for recognizing a user's emotion, and means for adjusting the talk flow in accordance with the recognized user's emotion. This makes it possible to generate an optimal talk flow for a new service and adjust the talk flow in accordance with the user's emotion.

[1455] "Talk flows developed in the past" refers to previously used dialogue flows and patterns, which are stored in a database.

[1456] "Specifications and features of the new service" refers to the detailed content and characteristics of the newly provided service, and is information that will be read into the artificial intelligence.

[1457] "Artificial intelligence" refers to the technology that enables computer systems to mimic human intelligence to learn, reason, and self-correct, as used here to generate and moderate talk flow.

[1458] The "optimal proposed talk flow" generates the most effective dialogue flow based on the specifications and features of the new service.

[1459] "User's emotion" refers to the emotional state shown by the user, and is recognized from voice, facial expressions, etc.

[1460] "Means for recognizing emotions" refers to technology that analyzes the user's voice and facial expressions to identify emotions, thereby understanding the user's emotional state.

[1461] The "means for adjusting the talk flow" refers to a technology for changing the flow of a conversation in accordance with the user's recognized emotions, thereby providing a conversation that is optimal for the user's emotions.

[1462] This invention is a system that uses AI to generate optimal proposed talk flows for new services and further adjusts the talk flows according to the user's emotions. Specific embodiments of this system are described below.

[1463] Hardware and software used

[1464] Hardware: Servers, terminals

[1465] Software: Generative AI models, emotion recognition software

[1466] System Overview

[1467] 1. Input specifications and features of new services

[1468] The user inputs the specifications and features of the new service into the terminal.

[1469] The terminal transmits the input data to the server.

[1470] The server passes the received data to the generative AI model.

[1471] 2. Selecting the optimal conversation flow

[1472] The server has access to a database of stored talk flows.

[1473] The generative AI model selects the optimal talk flow from a database based on the characteristics of the new service.

[1474] 3. Creating a new talk flow

[1475] The generative AI model generates a new talk flow based on the selected talk flow.

[1476] The server transmits the generated talk flow to the terminal.

[1477] 4. User Emotion Recognition

[1478] The device analyzes the user's voice and facial expressions using emotion recognition software.

[1479] The emotion recognition software recognizes the user's emotions and sends this information to the server.

[1480] 5. Adjusting conversation flow according to emotions

[1481] Based on the received emotional data, the server instructs the generative AI model to adjust the talk flow.

[1482] The generative AI model generates a conversation flow that best suits the user's emotions.

[1483] The server sends the adjusted talk flow to the terminal.

[1484] Specific examples

[1485] New service: eco-friendly detergent

[1486] User Emotion: Anger

[1487] Example prompt sentence:

[1488] Generate a talk flow for a new eco-friendly cleaner. If the user is showing signs of anger, include a talk flow to calm the user.

[1489] An example of a generated talk flow:

[1490] Hello. This detergent is made with eco-friendly ingredients, making it a better choice for the planet. If you have any complaints, please let us know. We value your feedback.

[1491] In this way, the system can generate optimal talk flows for new services and adjust them according to the user's emotions. This improves the user experience and enables effective proposal of services. The flow of the identification process in the third embodiment will be explained with reference to FIG. 21.

[1492] Step 1:

[1493] Entering specifications and features of new services

[1494] The user inputs the specifications and features of the new service into the terminal.

[1495] Input: Specifications and features of the new service (e.g., properties of an eco-friendly detergent)

[1496] The terminal transmits the input data to the server.

[1497] Output: New service specifications and features data sent to the server

[1498] Specific operation: The user enters "ecological detergent" and its characteristics into the input form on the device and presses the send button.

[1499] Step 2:

[1500] Choosing the best talk flow

[1501] The server passes the received data to the generative AI model.

[1502] Input: Data on the specifications and features of the new service

[1503] The server has access to a database of stored talk flows.

[1504] The generative AI model selects the optimal talk flow from a database based on the characteristics of the new service.

[1505] Output: Optimal talk flow data

[1506] Specific operation: The server searches the database for talk flows related to "ecology" and passes them to the generative AI model.

[1507] Step 3:

[1508] Creating a new talk flow

[1509] The generative AI model generates a new talk flow based on the selected talk flow.

[1510] Input: Data for optimal talk flow

[1511] The server transmits the generated talk flow to the terminal.

[1512] Output: Newly generated talk flow data

[1513] Specific operation: The generative AI model generates a new talk flow that reflects the characteristics of the "ecological detergent," and the server sends it to the device.

[1514] Step 4:

[1515] User Emotion Recognition

[1516] The device analyzes the user's voice and facial expressions using emotion recognition software.

[1517] Input: User voice and facial expression data

[1518] The emotion recognition software recognizes the user's emotions and sends this information to the server.

[1519] Output: Recognized user emotion data

[1520] How it works: When a user speaks into the device's camera, emotion recognition software analyzes the user's facial expressions and tone of voice to identify anger.

[1521] Step 5:

[1522] Adjusting conversation flow according to emotions

[1523] Based on the received emotional data, the server instructs the generative AI model to adjust the talk flow.

[1524] Input: Recognized user emotion data

[1525] The generative AI model generates a conversation flow that best suits the user's emotions.

[1526] The server sends the adjusted talk flow to the terminal.

[1527] Output: Adjusted talk flow data

[1528] Specific operation: The server passes data that "the user is expressing anger" to the generative AI model, the generative AI model generates a conversation flow that calms the user, and the server sends it to the device.

[1529] (Application example 3)

[1530] Next, a description will be given of Application Example 3 of Form Example 3. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[1531] Conventional talk flow generation systems have had issues with generating optimal talk flows based on the specifications and features of new services, and with being unable to adjust the talk flow in response to user emotions. This makes it difficult to improve customer satisfaction and provide effective customer service. Furthermore, it has not been possible to display talk flows in real time on smart devices.

[1532] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 3 is realized by the following means.

[1533] In this invention, the server includes a means for storing previously deployed talk flows, a means for loading the specifications and features of a new service into an AI, a means for generating an optimal proposed talk flow tailored to the service, a means for recognizing a user's emotion, a means for adjusting the talk flow according to the recognized user's emotion, and a means for displaying the talk flow on a smart device. This enables the generation of an optimal talk flow based on the specifications and features of a new service, and further enables the adjustment of the talk flow according to the user's emotion. It also enables the display of talk flows in real time using a smart device.

[1534] "Means for storing talk flows that have been developed in the past" refers to a means for saving talk flows that have been used in the past in a database or the like so that they can be referenced later.

[1535] "Means of loading the specifications and features of new services into AI" refers to the means of inputting the details and characteristics of new services into AI as text data so that the AI ​​can understand that information.

[1536] "Means for generating optimal proposed talk flow" refers to a means by which AI automatically creates the most appropriate talk flow based on the specifications and features of a new service.

[1537] "Means for recognizing user emotions" refers to means for analyzing the user's facial expressions and tone of voice using cameras, sensors, etc., to identify their emotions.

[1538] "Means for adjusting the talk flow according to the recognized user emotions" refers to a means by which AI dynamically changes the talk flow based on the user's emotional data and responds optimally.

[1539] The "means for displaying a talk flow on a smart device" refers to a means for displaying the generated talk flow in real time on a device such as smart glasses or a smartphone.

[1540] The system for implementing this invention is configured as follows: First, the server has a database that stores talk flows that have been used in the past. This database stores talk flows that have been used in the past and allows them to be referenced later.

[1541] Next, a method is needed to have the AI ​​learn the specifications and features of the new service. This method involves inputting the details and characteristics of the new service into the AI ​​as text data so that the AI ​​can understand that information. Specifically, natural language processing technology is used to analyze the text data and enable the AI ​​to understand its content.

[1542] Furthermore, it includes a means for generating an optimal proposed talk flow, in which the AI ​​automatically creates the most suitable talk flow based on the specifications and features of the new service.The generative AI model is used to generate the talk flow based on the prompt sentence.

[1543] To recognize a user's emotions, hardware such as cameras and sensors is required. These devices are used to analyze the user's facial expressions and tone of voice to identify their emotions. Specifically, image processing libraries such as OpenCV are used to recognize emotions in real time.

[1544] To adjust the talk flow according to the recognized user emotion, the AI ​​dynamically changes the talk flow based on the emotion data. In this process, the generative AI model is used again to generate the optimal talk flow according to the emotion.

[1545] Finally, a means for displaying the talk flow on a smart device is included. The generated talk flow is displayed in real time on a device such as smart glasses or a smartphone, allowing a store clerk to provide an optimal talk flow to a customer.

[1546] For example, if a customer is interested in ecology, here is a prompt:

[1547] Service Features: Ecology

[1548] Customer Sentiment: Interest

[1549] Optimal conversation flow:

[1550] By inputting this prompt into a generative AI model, an optimal conversation flow about ecology can be generated. For example, a conversation flow such as "This product is made from renewable materials and is environmentally friendly" can be generated.

[1551] The hardware used includes smart glasses and smartphones, and the software uses OpenCV (image processing) and Transformers (AI model for generating Hugging Face), which enables real-time emotion recognition and the generation and display of talk flow.

[1552] The flow of the specific processing in Application Example 3 will be described with reference to FIG.

[1553] Step 1:

[1554] The server accumulates previously deployed talk flows in a database. As input, it receives data on previously used talk flows and stores them in the database. As output, it obtains a database of accumulated talk flows. This database is structured so that it can be referenced later.

[1555] Step 2:

[1556] The server loads the specifications and features of the new service as text data into the AI. As input, text data containing the details and characteristics of the new service is provided, which is analyzed using natural language processing technology. As output, data in a format that the AI ​​can understand is obtained. This data is then used to generate the talk flow.

[1557] Step 3:

[1558] The server uses a generative AI model to generate the optimal proposed talk flow. As input, a prompt sentence based on the specifications and features of the new service is given, and the generative AI model generates the talk flow based on this. The optimal talk flow is obtained as output. Specifically, the prompt sentence is input into the generative AI model, and the generated talk flow is obtained.

[1559] Step 4:

[1560] The device uses a camera and sensors to recognize the user's emotions. As input, real-time data including the user's facial expressions and tone of voice is provided, which is analyzed using an image processing library (e.g., OpenCV). As output, the user's emotional data is obtained. Specifically, the device analyzes the images captured by the camera and identifies the emotion.

[1561] Step 5:

[1562] The server adjusts the talk flow according to the recognized user emotion. As input, the user's emotion data and the generated talk flow are given, and the generative AI model dynamically changes the talk flow based on this. The adjusted talk flow is obtained as output. Specifically, the emotion data is added to the prompt sentence and re-input into the generative AI model.

[1563] Step 6:

[1564] The terminal displays the talk flow on the smart device. As input, the adjusted talk flow is given and displayed in real time on smart glasses or a smartphone. As output, the optimal talk flow is provided to the user. As a specific operation, the generated talk flow is displayed on the device's display.

[1565] 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.

[1566] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ) and other generation AIs. The data generation model 58 is obtained by 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.

[1567] Another example of generative AI is Gemini (internet search engine). <url: https: gemini.google.com ?hl="ja">) are mentioned.

[1568] 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.

[1569] [Third embodiment]

[1570] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[1571] 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.

[1572] 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).

[1573] 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.

[1574] 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.

[1575] 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).

[1576] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for 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.

[1577] 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.

[1578] 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.

[1579] 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.

[1580] In the headset type terminal 314, the processor 46 performs the reception output process.

[1581] 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.

[1582] Next, the specific processing by the specific processing unit 290 of the data processing device 12 will be described.

[1583] "Example 1"

[1584] One embodiment of the present invention is a system that stores excellent talk flows that have been developed in the past in a database. This system collects data on past talk flows, analyzes it to extract excellent talk flows, and stores them in a database. Specifically, it extracts successful sales conversation patterns from logs of past sales conversations and stores them as talk flows.

[1585] "Example 2"

[1586] Another embodiment of the present invention is a means for loading the specifications and features of a new service into AI. Specifically, the specifications and manuals of the new service are loaded into AI as text data, and the AI ​​understands the content and generates a talk flow based on that. For example, the manual for the new service is analyzed using a text analysis tool, and the AI ​​learns from the results.

[1587] "Example 3"

[1588] Furthermore, one embodiment of the present invention also includes a means for generating the optimal proposed talk flow tailored to the service. Specifically, the AI ​​understands the specifications and features of the new service, selects the optimal one from the accumulated talk flows, and generates a new talk flow based on that. For example, if the new service is a product that emphasizes "ecology," the AI ​​selects a talk flow related to "ecology" and generates a new talk flow based on that.

[1589] The processing flow of each embodiment will be described below.

[1590] "Example 1"

[1591] Step 1: Collect logs of past sales conversations.

[1592] Step 2: Extract patterns of successful sales conversations from the collected logs.

[1593] Step 3: The extracted patterns are stored in a database as talk flows.

[1594] "Example 2"

[1595] Step 1: Obtain the specifications and manuals for the new service as text data.

[1596] Step 2: Analyze the acquired text data using a text analysis tool.

[1597] Step 3: Have the AI ​​learn the analysis results.

[1598] "Example 3"

[1599] Step 1: AI understands the specifications and features of the new service.

[1600] Step 2: Based on what you understand, select the most appropriate talk flow from the accumulated list.

[1601] Step 3: Generate a new talk flow based on the selected talk flow.

[1602] Example 1

[1603] Next, a description will be given of Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1604] Conventional talk flow generation systems were unable to effectively utilize past conversation data, making it difficult to generate optimal talk flows suited to new services. Furthermore, the process of extracting successful conversation patterns and storing them in a database was not automated, making it difficult to generate efficient talk flows.

[1605] The identification process by the identification processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means. In this invention, the server includes means for accumulating talk flows developed in the past, means for collecting past conversation log data, means for preprocessing the collected log data, means for analyzing the preprocessed data, means for extracting successful conversation patterns, means for saving the extracted patterns in a database, means for loading the specifications and features of a new service into an artificial intelligence, and means for generating an optimal proposed talk flow tailored to the service. This makes it possible to effectively utilize past conversation data and automatically generate an optimal talk flow suited to a new service.

[1606] "Talk flow" is a series of steps or patterns that indicate the progression and flow of a conversation.

[1607] "Log data" refers to records of operation history and events generated by systems and applications.

[1608] "Preprocessing" is the process of cleaning and shaping data before data analysis or machine learning.

[1609] "Data analysis" is the process of analyzing collected data and extracting useful information and patterns.

[1610] A "success pattern" is the characteristics and flow of a conversation that achieves a specific purpose (e.g., purchasing a product).

[1611] A "database" is a system for efficiently storing, managing, and retrieving data.

[1612] "Artificial intelligence" refers to computer systems that mimic human intelligence and can learn and reason.

[1613] "Natural language processing" is a technology that allows computers to understand, interpret, and generate human language.

[1614] The "optimal proposed talk flow" is a talk flow that indicates the progress and flow of a conversation that is most suitable for a specific purpose.

[1615] MODE FOR CARRYING OUT THE INVENTION

[1616] This invention is a system that effectively utilizes past conversation data to automatically generate an optimal talk flow suitable for a new service. Specific embodiments of this system will be described below.

[1617] Data collection

[1618] The server collects log data of past sales conversations. This log data is a record of conversations between salespeople and customers and is stored in text format. The server reads the log file from a specified directory (e.g., / data / sales_logs / ) and parses the log file line by line to identify the start and end of the conversation.

[1619] Data Preprocessing

[1620] The server preprocesses the collected log data. This preprocessing includes text cleaning (removing unnecessary characters and symbols), tokenization (dividing the data into words), and removal of stop words. Specifically, it uses the Python library NLTK to tokenize the text and remove stop words.

[1621] Data analysis

[1622] The server uses natural language processing (NLP) techniques to analyze the preprocessed data. Specifically, it uses spaCy to extract conversation topics, sentiment, keywords, etc. The server identifies the topic of each conversation, performs sentiment analysis, and classifies the conversation's sentiment as positive, negative, or neutral.

[1623] Extracting success patterns

[1624] Based on the analysis results, the server extracts patterns of successful sales conversations. A successful conversation is one in which the customer purchases a product or receives positive feedback. The server uses a machine learning algorithm (e.g., random forest) to identify successful patterns and organizes them as a conversation flow.

[1625] Saving to a database

[1626] The server saves the extracted successful patterns as talk flows in a database. A relational database such as MySQL or PostgreSQL is used for the database. The server connects to the MySQL database, saves the talk flows using INSERT statements, and reads and verifies the data from the database to confirm that the save was successful.

[1627] Loading the specifications of the new service

[1628] The server loads the specifications and features of the new service into the AI. The specifications and features of the new service are provided as text data, which the server analyzes and inputs into the AI.

[1629] Generation of optimal proposed talk flow

[1630] The server uses artificial intelligence (AI)-based natural language processing to generate optimal proposed talk flows tailored to new services. The generated talk flows are based on past success patterns and the specifications of the new services.

[1631] Specific examples

[1632] For example, consider the following sales conversation log:

[1633] Salesperson: Hello, what item are you looking for today?

[1634] Customer: I'm looking for a new smartphone.

[1635] Salesperson: That's great. What features are important to you?

[1636] Customer: I want something with a good camera.

[1637] Salesperson: Then this model is recommended. It has a high-performance camera.

[1638] Customer: Thank you, I'll buy it.

[1639] This conversation log is analyzed, and successful patterns (customers purchasing products) are extracted and saved in a database as talk flows.

[1640] Prompt Sentence Examples

[1641] Examples of prompts to input to a generative AI model might include:

[1642] "Analyze past sales conversation logs and extract patterns of successful sales conversations. A successful conversation is one in which the customer purchases a product or leaves positive feedback. Store the extracted patterns in a database as conversation flows."

[1643] In this way, the server collects and analyzes logs of past sales conversations and stores successful talk flows in a database, creating a system.

[1644] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1645] Step 1: Data collection

[1646] The server collects log data of past sales conversations. As input, it reads the log files from a specified directory (e.g., / data / sales_logs / ). The server parses the log files line by line to identify the start and end of a conversation. As output, it receives the parsed conversation log data.

[1647] Specific behavior:

[1648] The server will read the log files from the specified directory.

[1649] The server parses the log file line by line to identify the beginning and end of a conversation.

[1650] Step 2: Data Preprocessing

[1651] The server preprocesses the collected log data. As input, it receives the collected conversation log data. Preprocessing includes text cleaning (removing unnecessary characters and symbols), tokenization (splitting into words), and removal of stop words. As output, it obtains the preprocessed, clean data.

[1652] Specific behavior:

[1653] The server uses the Python library NLTK to tokenize the text.

[1654] The server removes stop words (e.g., a, the, is, etc.) and leaves only the important words.

[1655] Step 3: Data analysis

[1656] The server analyzes the preprocessed data. As input, it receives the preprocessed, clean data. It uses natural language processing (NLP) techniques to extract conversation topics, sentiment, keywords, etc. As output, it provides the analysis results.

[1657] Specific behavior:

[1658] The server uses spaCy to identify the topic of each conversation.

[1659] The server performs sentiment analysis and classifies the conversation into positive, negative, or neutral sentiment.

[1660] Step 4: Extracting success patterns

[1661] The server extracts patterns of successful sales conversations based on the analysis results. It receives the analysis results as input. A successful conversation is when the customer purchases a product or receives positive feedback. The server obtains a success pattern as output.

[1662] Specific behavior:

[1663] The server uses machine learning algorithms (e.g., random forests) to identify successful patterns.

[1664] The server organizes the successful patterns as talk flows.

[1665] Step 5: Saving to the Database

[1666] The server saves the extracted successful patterns as talk flows in a database. The successful patterns are received as input, and the talk flows saved in the database are obtained as output.

[1667] Specific behavior:

[1668] The server connects to a MySQL database and saves the talk flow using an INSERT statement.

[1669] The server reads and verifies the data from the database to ensure the save was successful.

[1670] Step 6: Load the specifications for the new service

[1671] The server loads the specifications and features of the new service into the AI. As input, it receives the specifications and features of the new service as text data. As output, it obtains the specification data loaded into the AI.

[1672] Specific behavior:

[1673] The server analyzes the specifications and features of the new service as text data.

[1674] The server inputs the analysis results into the artificial intelligence.

[1675] Step 7: Generate optimal proposed talk flow

[1676] The server uses natural language processing with artificial intelligence to generate the optimal proposed talk flow for the new service. As input, it receives past success patterns and new service specification data. As output, it obtains the optimal proposed talk flow.

[1677] Specific behavior:

[1678] The server generates the optimal proposed talk flow based on past success patterns and specification data for new services.

[1679] The server stores the generated talk flow in a database.

[1680] In this way, the server collects and analyzes logs of past sales conversations, stores successful talk flows in a database, and realizes a system that automatically generates optimal talk flows suitable for new services.

[1681] (Application example 1)

[1682] Next, a description will be given of Application Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1683] In traditional customer service work, a store clerk needs a wealth of experience and knowledge to make optimal suggestions to customers, which is difficult for new or inexperienced store clerks. In addition, there is a lack of means to provide real-time conversation flows that allow for quick and accurate responses to customer needs. This has led to issues such as inconsistent customer service quality and a decline in customer satisfaction.

[1684] 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.

[1685] In this invention, the server includes means for storing previously developed talk flows, means for loading specifications and features of new services into an artificial intelligence, means for generating an optimal proposed talk flow tailored to the service, means for displaying the optimal proposed talk flow in real time on a visual device worn by a store clerk, and means for generating the next talk flow based on a conversation with a customer, thereby enabling store clerks to provide customers with consistently high-quality customer service.

[1686] "Past talk flow" refers to the flow and patterns of conversations with customers that have taken place in the past, and by accumulating and analyzing this data, patterns of successful sales conversations can be extracted.

[1687] "Specifications and features of new services" refers to the detailed content and characteristics of the newly provided services, and by inputting this into artificial intelligence, it becomes the basic data for generating optimal proposals for the services.

[1688] "Artificial intelligence" refers to the technology that enables computer systems to mimic human intelligence, learn, reason, and self-correct, generating optimal conversation flow through natural language processing and data analysis.

[1689] "Optimal proposal talk flow" refers to the most effective flow of proposals and responses in conversations with customers, and is generated by artificial intelligence.

[1690] A "visual device" refers to a device worn by a store clerk that visually displays information, such as smart glasses.

[1691] "Real-time" refers to the fact that data processing and information provision are immediate, meaning that the optimal conversation flow is displayed instantly during a conversation with a customer.

[1692] "Means for generating the next talk flow based on a conversation with a customer" refers to technology that analyzes a customer's comments and reactions in real time and uses artificial intelligence to generate the next conversation flow accordingly.

[1693] A system for implementing this invention includes a means for storing previously developed talk flows, a means for loading specifications and features of new services into artificial intelligence, a means for generating an optimal proposed talk flow tailored to the service, a means for displaying the optimal proposed talk flow in real time on a visual device worn by a store clerk, and a means for generating the next talk flow based on a conversation with a customer.

[1694] The server first stores past talk flows in a database. This process involves collecting logs of past sales conversations and extracting patterns of successful sales conversations. As a result, the best talk flows are stored in the database.

[1695] Next, the specifications and features of the new service are input as text data into the AI, which then learns information about the new service and obtains the basic data needed to generate optimal proposals.

[1696] The AI ​​uses natural language processing technology to generate optimal proposed talk flows based on accumulated talk flows and the specifications and features of new services. This generated talk flow is displayed in real time on a visual device (e.g., smart glasses) worn by the store clerk.

[1697] When a store clerk starts a conversation with a customer, they serve them while referring to the optimal proposed talk flow displayed on the visual device. Based on the customer's comments and reactions, the AI ​​generates the next talk flow in real time and displays it on the visual device.

[1698] As a concrete example, the following prompt sentence is input to the generative AI model:

[1699] "Hello, what products are you looking for today?"

[1700] "Customer: I'm looking for a new smartphone."

[1701] Salesperson: Thank you. What features are most important to you?

[1702] Based on this prompt, the generative AI model generates the next conversation flow, which store clerks can refer to in real time through the smart glasses, enabling them to provide consistent, high-quality customer service to customers.

[1703] The hardware used includes smart glasses (e.g., Google Glass), and the software includes SQLite (database management) and OpenAI API (generative AI model), which allows the system to instantly generate and display the optimal conversation flow based on the conversation with the customer.

[1704] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1705] Step 1:

[1706] The server stores past talk flows in a database. Specifically, it collects logs of past sales conversations and analyzes them to extract patterns of successful sales conversations. The input is the log data of past sales conversations, and the output is the patterns of successful talk flows that are saved in the database.

[1707] Step 2:

[1708] The server loads the specifications and features of the new service into the AI ​​as text data. Specifically, the detailed content and characteristics of the new service are input in text format, and the AI ​​learns from them. The input is the text data of the specifications and features of the new service, and the output is the AI ​​in a state where it has learned information about the new service.

[1709] Step 3:

[1710] The server uses artificial intelligence to generate the optimal proposed talk flow. Specifically, it uses natural language processing technology to generate the optimal talk flow based on the accumulated talk flows and the specifications and features of the new service. The inputs are past successful talk flows and the specifications and features of the new service, and the output is the optimal proposed talk flow.

[1711] Step 4:

[1712] The terminal (a visual device worn by the store clerk) displays the generated optimal proposed talk flow in real time. Specifically, it receives the optimal proposed talk flow sent from the server and displays it on the display of the visual device. The input is the optimal proposed talk flow sent from the server, and the output is the talk flow displayed on the visual device.

[1713] Step 5:

[1714] The user (store clerk) converses with the customer while referring to the optimal proposed talk flow displayed through a visual device. Specifically, based on the customer's comments and reactions, the user sends information to the server to generate the next talk flow. The input is the customer's comments and reactions, and the output is the information to generate the next talk flow sent to the server.

[1715] Step 6:

[1716] The server generates the next talk flow based on the conversation with the customer. Specifically, it analyzes the customer's remarks and reactions, and artificial intelligence generates the next talk flow accordingly. The input is data on the customer's remarks and reactions, and the output is the next talk flow generated and sent to the visual device.

[1717] Step 7:

[1718] The terminal (visual device) displays the generated next talk flow in real time. Specifically, it receives the next talk flow sent from the server and displays it on the display of the visual device. The input is the next talk flow sent from the server, and the output is the next talk flow displayed on the visual device.

[1719] Example 2

[1720] Next, a description will be given of Example 2 of Form Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1721] Conventional talk flow generation systems have had difficulty generating talk flows that quickly and accurately reflect the specifications and features of new services. Furthermore, the accumulation of past talk flows and the use of analysis results were not fully utilized, making it difficult to generate talk flows efficiently. This resulted in the issue of creating talk flows when introducing new services taking a great deal of time and effort.

[1722] 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.

[1723] In this invention, the server includes means for storing talk flows that have been developed in the past, means for loading specifications and features of new services into an artificial intelligence, means for receiving specifications and manuals of new services as text data and analyzing them using a text analysis tool, means for inputting the analysis results into a generative AI model and training the AI ​​to understand the content, means for generating talk flows based on the content understood by the AI, and means for providing the generated talk flows. This makes it possible to generate talk flows that quickly and accurately reflect the specifications and features of new services.

[1724] "Talk flows developed in the past" refer to the flow and patterns of dialogue that have been previously executed, and are data that can be accumulated and reused.

[1725] "Specifications and features of new services" refers to information that provides a detailed description and characteristics of the newly provided services, and is the basic data for generating talk flows based on this information.

[1726] "Artificial intelligence" refers to computer systems that mimic human intelligence using techniques such as machine learning and natural language processing.

[1727] "Text data" refers to information expressed as a string of characters, such as specifications, manuals, and other documents expressed in digital format.

[1728] "Text analysis tools" refer to software and algorithms used to analyze text data and extract meaning and structure.

[1729] A "generative AI model" refers to an artificial intelligence model that generates new information or responses based on input data.

[1730] "Talk flow" indicates the flow and pattern of a dialogue, and is a blueprint for smoothly advancing interaction with users.

[1731] "Analysis results" refers to information about the meaning and structure of data obtained by text analysis tools.

[1732] "Learning" refers to the process by which artificial intelligence finds patterns and rules based on input data to improve future predictions and responses.

[1733] "Providing" refers to the act of making the generated talk flow available to users or other systems.

[1734] The present invention relates to a system for generating talk flows that quickly and accurately reflect the specifications and features of new services. A specific embodiment of this system will be described below.

[1735] System configuration

[1736] Subject: Server

[1737] The server receives the specifications and manuals for the new service as text data. Specifically, it receives an HTTP request from the user or device and saves the text data. Next, the server analyzes this text data using a text analysis tool (e.g., NLTK or SpaCy). Based on the analysis results, the server inputs the data into a generative AI model (e.g., GPT-4) and trains the AI ​​to understand the content. Finally, the server generates a talk flow based on the content understood by the AI ​​and provides the generated talk flow to the user or device.

[1738] Subject: Terminal

[1739] The device inputs the specifications and manuals for the new service as text data. The user selects a file using the device's file selection dialog, and the device sends this data to the server. When the server returns the analysis results, the device inputs them into the generative AI model, training the AI ​​to understand the content. Finally, the device displays the talk flow generated by the generative AI model to the user.

[1740] Subject: User

[1741] The user inputs the specifications and manuals for the new service as text data into their device. The user then sends this data to the server via their device, which then analyzes it using a text analysis tool. Once the analysis results are returned from the server, the user inputs the data into the generative AI model via their device, training the AI ​​to understand the content. Finally, the user checks the conversation flow generated by the AI.

[1742] Specific examples

[1743] Example 1: Loading the specifications of a new service into AI

[1744] The user has a PDF specification for the new service. This PDF is converted to text data and entered into the device. The device sends this data to the server, which performs text analysis using NLTK. Based on the analysis results, the server inputs the data into GPT-4, training the AI ​​to understand the content. Finally, the server provides the user with the AI-generated conversation flow.

[1745] Example 2: Loading a new service manual into AI

[1746] The user has a manual for the new service in Word format. The user converts this Word file into text data and inputs it into the device. The device then sends this data to the server, which performs text analysis using SpaCy. Based on the analysis results, the server inputs the data into GPT-4, training the AI ​​to understand the content. Finally, the server provides the user with the AI-generated conversation flow.

[1747] Prompt Sentence Examples

[1748] "Please input the specifications of the new service as text data and analyze it using NLTK. Input the analysis results into GPT-4 and train the AI ​​to understand the content. Finally, please provide the AI-generated conversation flow."

[1749] In this way, the present invention enables the creation of talk flows that quickly and accurately reflect the specifications and features of new services.

[1750] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1751] Step 1:

[1752] The server receives text data such as specifications or manuals for new services from users or terminals. Specifically, the server receives an HTTP request and saves the text data. The input is the text data for specifications or manuals for new services, and the output is the saved text data.

[1753] Step 2:

[1754] The server analyzes the received text data using a text analysis tool (e.g., NLTK or SpaCy). Specifically, the server calls the text analysis tool to perform tokenization, part-of-speech tagging, dependency analysis, etc. on the text data. The input is the saved text data, and the output is the analysis results.

[1755] Step 3:

[1756] The server converts the results of text analysis into a format for input to a generative AI model (e.g., GPT-4). Specifically, the server converts the analysis results into an appropriate data format, such as JSON. The input is the result of text analysis, and the output is the converted data format.

[1757] Step 4:

[1758] The server inputs the converted data into the generative AI model and trains the AI ​​to understand the content. Specifically, the server calls the API of the generative AI model and sends the data. The input is the converted data format, and the output is the content understood by the AI.

[1759] Step 5:

[1760] The server generates a talk flow based on what the generative AI model understands. Specifically, the server receives a response from the AI ​​and executes the logic to construct the talk flow. The input is what the AI ​​understands, and the output is the generated talk flow.

[1761] Step 6:

[1762] The server provides the generated talk flow to the user or terminal. Specifically, the server generates an HTTP response and sends data including the talk flow. The input is the generated talk flow, and the output is the talk flow provided to the user or terminal.

[1763] (Application example 2)

[1764] Next, a description will be given of Application Example 2 of Form Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1765] Whenever a new service is introduced, it is difficult to help users understand its specifications and how to use it. In particular, for complex services and multi-functional platforms, users need a guide to use the service properly. However, with conventional methods, simply referring to manuals or FAQs is insufficient, and there is an issue of making it difficult for users to quickly access the information they are looking for.

[1766] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[1767] In this invention, the server includes means for storing previously deployed talk flows, means for loading specifications and features of new services into an AI, means for generating an optimal proposed talk flow tailored to the service, means for generating answers to user questions, and means for answering user questions based on the generated talk flows, thereby enabling users to smoothly understand and use new services.

[1768] "Talk flow developed in the past" refers to the flow and patterns of conversations that have been carried out in the past, and by storing these, they can be used to help with future conversations.

[1769] "Means of having AI read the specifications and features of new services" refers to a method of inputting detailed descriptions and characteristics of new services into AI as text data, allowing the AI ​​to understand their content.

[1770] "Means for generating optimal proposed talk flow" refers to a method in which AI automatically creates the most suitable dialogue flow for users based on the specifications and features of a new service.

[1771] "Means for generating answers to user questions" refers to the method by which AI receives questions from users and generates appropriate answers to them.

[1772] "Means of answering user questions based on the generated talk flow" refers to a method of providing specific answers to user questions based on the dialogue flow generated by AI.

[1773] A system for implementing this invention is configured as follows: First, the server has a means for storing talk flows that have been developed in the past. This means allows the flow and patterns of previously executed conversations to be saved and used for future conversations.

[1774] The next step involves having the AI ​​learn the specifications and features of the new service. This involves inputting a detailed description and characteristics of the new service into the AI ​​as text data, allowing the AI ​​to understand the content. Specifically, a text analysis tool is used to analyze the manual for the new service, and the AI ​​is then trained on the results.

[1775] Furthermore, the server has a means for generating optimal proposed talk flows, in which AI automatically creates the most suitable dialogue flow for users based on the specifications and features of the new service. It also includes a means for generating answers to users' questions using AI-based natural language processing technology.

[1776] The method for generating answers to user questions involves receiving questions from users and having AI generate appropriate answers to them. It also includes a method for answering user questions based on the generated talk flow, providing specific answers to user questions based on the dialogue flow generated by the AI.

[1777] Hardware and software used

[1778] Hardware: Smartphone

[1779] Software: OpenAI API, Python

[1780] Data processing and calculation

[1781] 1. Reading text data: Read the manual for the new service as a text file.

[1782] 2. Training the AI: Use OpenAI's API to train the AI ​​to learn the contents of the manual.

[1783] 3. Generate talk flow: The AI ​​generates a talk flow based on the manual and saves it in JSON format.

[1784] 4. Generating answers to user questions: Upon receiving a user question, the AI ​​generates an appropriate answer based on the conversation flow.

[1785] Specific examples

[1786] If a user asks, "How do I search for new content?", the AI ​​will answer, based on the contents of the manual, "Enter keywords in the search bar and press the search button."

[1787] Prompt Sentence Examples

[1788] Please answer the user's questions based on the following conversation flow.

[1789] Talk flow:

[1790] 1. How to find new content

[1791] 2. How to view content

[1792] 3. How to set up your account

[1793] User Question: How do I search for new content?

[1794] In this way, users can smoothly understand and use new services.

[1795] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1796] Step 1:

[1797] The server reads the manual for the new service as a text file.

[1798] Input: New service manual (text file)

[1799] Output: Text data

[1800] Specific operation: The server reads the manual for the new service from the specified file path and stores the contents in memory as text data.

[1801] Step 2:

[1802] The server uses a text analysis tool to analyze the read text data.

[1803] Input: Text data

[1804] Output: Analysis results (structured text data)

[1805] Specific operation: The server uses a text analysis tool (for example, a natural language processing library) to analyze the text data and extract sentence structures and keywords.

[1806] Step 3:

[1807] The server trains the AI ​​model based on the analysis results.

[1808] Input: Analysis results (structured text data)

[1809] Output: Trained AI model

[1810] Specific operation: The server uses the OpenAI API to input the analysis results into the AI ​​model, allowing the AI ​​to learn to understand the specifications and features of the new service.

[1811] Step 4:

[1812] The server generates the optimal proposed talk flow using the trained AI model.

[1813] Input: Trained AI model

[1814] Output: Optimal proposal talk flow (JSON format)

[1815] Specific operation: The server inputs the usage scenario of the new service into the AI ​​model, and the AI ​​generates the optimal dialogue flow. The generated dialogue flow is saved in JSON format.

[1816] Step 5:

[1817] The user enters a question using a smartphone.

[1818] Input: User question (text)

[1819] Output: None (passed on to next step)

[1820] Specific operation: The user opens the application on their smartphone and enters a question in text format.

[1821] Step 6:

[1822] The server receives the user's question and generates an answer based on the generated talk flow.

[1823] Input: User question (text), optimal suggestion talk flow (JSON format)

[1824] Output: AI response (text)

[1825] Specific operation: The server inputs the user's question into the AI ​​model and generates an appropriate answer while referring to the generated talk flow.

[1826] Step 7:

[1827] The server sends the generated answer to the user's smartphone.

[1828] Input: AI response (text)

[1829] Output: The answer displayed on the user's smartphone

[1830] Specific operation: The server sends the generated answer to the user's smartphone and displays it in the user's application.

[1831] In this way, users can smoothly understand and use new services.

[1832] Example 3

[1833] Next, a third embodiment of the third embodiment will be described. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1834] Conventional talk flow generation systems have had difficulty quickly and effectively generating talk flows that are optimal for the specifications and features of new services. Furthermore, there were problems with the time it took for users to input information about new services, and with variations in the quality of the generated talk flows. This could lead to ineffective service introductions and proposals, and could result in lower customer satisfaction.

[1835] The specific processing by the specific processing unit 290 of the data processing device 12 in the third embodiment is realized by the following means.

[1836] In this invention, the server includes means for storing previously developed talk flows, means for loading specifications and features of a new service into an AI, means for generating an optimal proposed talk flow tailored to the service, means for a user to input the specifications and features of the new service, means for the server to receive the input information and send a prompt to the generative AI model, means for the server to select a relevant talk flow from a database of stored talk flows, means for the server to generate a new talk flow using the generative AI model, and means for the server to provide the generated talk flow to the user. This makes it possible to quickly and effectively generate a talk flow optimal for the specifications and features of a new service, reducing the user's effort and providing a high-quality talk flow.

[1837] "Past-developed talk flows" refer to previously used dialogue flows and patterns, which can be stored in a database to make them reusable.

[1838] "Specifications and features of new services" refers to detailed descriptions and characteristics of newly provided services, and is information used to generate optimal talk flows.

[1839] "Artificial intelligence" refers to the technology that allows computer systems to mimic human intelligence to learn, reason, and self-correct, and is used in this case to generate talk flow.

[1840] "Optimal proposal talk flow" refers to the dialogue flow that is most suited to the specifications and features of a new service, and is intended to effectively introduce the service to customers.

[1841] "Means for users to input specifications and features of new services" refers to an interface or method by which users can provide information about new services to the system.

[1842] "Means for the server to receive input information and send prompts to the generative AI model" refers to a method for receiving information provided by a user and issuing instructions to the generative AI model based on that information.

[1843] "Means for the server to select relevant talk flows from a database of accumulated talk flows" refers to a method for selecting past talk flows stored in a database that are relevant to the new service.

[1844] "Means for the server to generate a new talk flow using a generative AI model" refers to a method for creating a new talk flow using a generative AI model based on a selected talk flow.

[1845] The term "means by which the server provides the generated talk flow to the user" refers to a method for displaying or transmitting the generated talk flow to the user.

[1846] MODE FOR CARRYING OUT THE INVENTION

[1847] The present invention relates to a system for generating talk flows that are optimal for the specifications and features of new services. Specific embodiments of this system will be described below.

[1848] System configuration

[1849] This system consists of three main components: a server, a terminal, and a user. The server contains a database of previously deployed talk flows, a generative AI model (e.g., OpenAI's GPT-4), and a talk flow generation algorithm. The terminal provides an interface for users to input specifications and features of new services.

[1850] Program processing

[1851] The user uses the terminal to input the specifications and features of the new service. For example, the user may input information such as "This is an eco-friendly product made from renewable materials." The terminal then sends this information to the server.

[1852] The server receives the information sent by the user and creates a prompt to send to the generative AI model based on that information. For example, it generates a prompt such as, "Please generate a talk flow to introduce a new eco-friendly product. This product is made from environmentally friendly materials and is highly energy efficient."

[1853] Next, the server selects a relevant talk flow from a database of stored talk flows based on the prompt sentence. This selection is done using natural language processing technology. For example, a talk flow that emphasizes the benefits of eco-friendly products or a talk flow that suggests an environmentally friendly lifestyle may be selected.

[1854] The server generates a new talk flow based on the selected talk flow using a generative AI model. The generative AI model understands the context and flow of the talk flow and reconstructs it in a way that is optimal for the new service.

[1855] The generated talk flow might look something like this:

[1856] 1. "Hello, today I'd like to introduce you to a new eco-friendly product."

[1857] 2. "This product is made from renewable materials and is manufactured with maximum energy efficiency."

[1858] 3. "Furthermore, using this product will help reduce energy consumption in daily life."

[1859] Finally, the server sends the generated talk flow to the user's device, where the user can view the talk flow on the device screen and make corrections or adjustments as necessary.

[1860] Specific examples

[1861] For example, if the new service is a product that emphasizes "ecology," the user can input the following prompt into the generative AI model:

[1862] Example prompt sentence:

[1863] "Generate a talk flow to introduce a new eco-friendly product. This product is made from environmentally friendly materials and is highly energy efficient."

[1864] Based on this prompt, the server generates an optimal talk flow and provides it to the user. In this way, the user can quickly and effectively generate a talk flow that is optimal for the specifications and features of the new service. The flow of the specification process in the third embodiment will be explained with reference to Figure 15.

[1865] Program processing flow

[1866] Step 1:

[1867] Users input specifications and features of new services

[1868] The user uses the terminal to input the specifications and features of the new service. For example, the user might input "It's an eco-friendly product made from renewable materials." This input data is sent to the server in text format.

[1869] input:

[1870] Specifications and features of the new service (e.g., "It is an ecological product and uses renewable materials")

[1871] output:

[1872] Text data sent to the server

[1873] Step 2:

[1874] The server receives the input information and sends a prompt to the generative AI model.

[1875] The server receives the information sent by the user and creates a prompt to send to the generative AI model based on that information. For example, it generates a prompt such as, "Please generate a talk flow to introduce a new eco-friendly product. This product is made from environmentally friendly materials and is highly energy efficient."

[1876] input:

[1877] Text data sent by the user

[1878] Data processing:

[1879] Generate prompt sentences (e.g., "Generate a talk flow to introduce a new eco-friendly product. This product is made from environmentally friendly materials and is highly energy efficient.")

[1880] output:

[1881] Generated prompt statement

[1882] Step 3:

[1883] The server selects relevant talk flows from its database of accumulated talk flows.

[1884] The server uses natural language processing technology to select a relevant talk flow from a database of stored talk flows based on the generated prompt. For example, a talk flow emphasizing the benefits of eco-friendly products or a talk flow proposing an environmentally friendly lifestyle may be selected.

[1885] input:

[1886] Generated prompt statement

[1887] Data processing:

[1888] Select relevant talk flows using natural language processing technology

[1889] output:

[1890] Selected Talk Flow

[1891] Step 4:

[1892] The server generates a new talk flow using the generative AI model.

[1893] The server generates a new talk flow based on the selected talk flow using a generative AI model. The generative AI model understands the context and flow of the talk flow and reconstructs it in a way that is optimal for the new service.

[1894] input:

[1895] Selected Talk Flow

[1896] Data processing:

[1897] Generate new conversation flows using generative AI models

[1898] output:

[1899] The new talk flow generated

[1900] Step 5:

[1901] The server provides the generated talk flow to the user.

[1902] The server sends the generated talk flow to the user's device, where the user can check the talk flow on the device screen and make corrections or adjustments as necessary.

[1903] input:

[1904] The new talk flow generated

[1905] output:

[1906] Dialogue flow sent to the user's device

[1907] Specific examples of operation

[1908] For example, if the new service is a product that emphasizes "ecology," the user can input the following prompt into the generative AI model:

[1909] Example prompt sentence:

[1910] "Generate a talk flow to introduce a new eco-friendly product. This product is made from environmentally friendly materials and is highly energy efficient."

[1911] Based on this prompt, the server generates an optimal talk flow and provides it to the user. In this way, the user can quickly and effectively generate a talk flow that is optimal for the specifications and features of a new service.

[1912] (Application example 3)

[1913] Next, a description will be given of Application Example 3 of Form Example 3. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1914] Conventional talk flow generation systems have had difficulty quickly and accurately generating optimal talk flows that match the specifications and features of new services. Furthermore, they were unable to provide users with the information they wanted in real time, making improving the user experience a challenge.

[1915] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 3 is realized by the following means. In this invention, the server includes means for storing talk flows developed in the past, means for loading specifications and features of new services into AI, means for generating an optimal proposed talk flow tailored to the service, means for inputting features of products searched by the user, and means for displaying the generated talk flow to the user. This makes it possible to provide the information desired by the user in real time and improve the user experience.

[1916] "Past talk flows" refers to previously used dialogue flows and patterns, and are data that can be accumulated and reused.

[1917] "New service specifications" refers to information that indicates the detailed design, functions, and characteristics of a newly provided service.

[1918] "Features" is information that indicates the characteristics and features of a service or product.

[1919] "AI" is an abbreviation for artificial intelligence, a system that analyzes data and makes decisions using technologies such as machine learning and natural language processing.

[1920] "Optimal proposal talk flow" is a process that generates the most appropriate dialogue flow based on the specifications and features of a specific service or product.

[1921] "Characteristics of the product searched by the user" refers to information that indicates the characteristics or features of the product that the user searches for on the online platform.

[1922] "Generated talk flow" refers to the flow and patterns of dialogue generated by AI based on the specifications and features of the new service.

[1923] The "means for displaying to the user" refers to a method or technology for visually presenting the generated talk flow to the user.

[1924] A system for implementing this invention includes a server, a user terminal, and an AI model. The server stores previously deployed talk flows and loads the specifications and features of new services into the AI. The user terminal inputs the features of the product searched by the user and displays the generated talk flows.

[1925] Hardware and software used

[1926] Hardware: Servers, smartphones

[1927] Software: Python, OpenAI API

[1928] Data processing and calculation

[1929] 1. Data entry: The user uses a smartphone to enter the characteristics of the product they are searching for, for example, "products made with renewable energy, recyclable packaging."

[1930] 2. Prompt generation: Based on the input product characteristics, the server generates a prompt sentence to be input into the generative AI model.

[1931] 3. AI model call: The server uses OpenAI's API to send prompts to a generative AI model (e.g., GPT-3) to generate the optimal conversation flow.

[1932] 4. Displaying the results: The server sends the generated talk flow to the user's terminal and displays it to the user.

[1933] Specific examples

[1934] If a user searches for eco-friendly products that have the characteristics of "products that use renewable energy and recyclable packaging," the following prompt sentence will be generated:

[1935] Prompt Sentence Examples

[1936] Generate a talk flow about a new eco-friendly product. Product features: Renewable energy-based product, recyclable packaging

[1937] The following is an example of a generated talk flow:

[1938] This product is made using renewable energy, making it an eco-friendly choice, and the packaging is recyclable, reducing waste. Perfect for those who want to live an eco-conscious life.

[1939] In this way, the information desired by the user can be provided in real time, improving the user experience.

[1940] The flow of the specific processing in Application Example 3 will be described with reference to FIG.

[1941] Step 1:

[1942] The user uses a smartphone to input the characteristics of the product they are searching for, such as "products that use renewable energy, recyclable packaging." The input data is sent to the server in text format.

[1943] Step 2:

[1944] The server receives the product features sent by the user. Based on the received text data, it generates a prompt sentence to be input into the generative AI model. Specifically, it creates a prompt sentence that includes the product features.

[1945] Step 3:

[1946] The server calls the OpenAI API using the generated prompt sentence. The prompt sentence is sent to the API, and the optimal talk flow is generated by a generative AI model (e.g., GPT-3). The input is the prompt sentence, and the output is the generated talk flow.

[1947] Step 4:

[1948] The server receives the generated talk flow and converts it into an appropriate format for display to the user. Specifically, it converts text data into a format that is easy for the user to understand.

[1949] Step 5:

[1950] The server sends the formatted talk flow to the user's smartphone. The user's device displays the received talk flow on its screen. The user can check the displayed talk flow and obtain detailed product information.

[1951] This series of processes allows users to receive the optimal talk flow in real time based on the characteristics of the product they searched for.

[1952] 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.

[1953] "Example 1"

[1954] As one embodiment of the present invention, a system is provided that includes an emotion engine in which AI recognizes a user's emotions. In this system, the emotion engine recognizes the user's emotions in real time during a dialogue with the user and generates an optimal talk flow according to the emotion. For example, if the user expresses joy, the emotion engine conveys that information to the AI, and the AI ​​generates a talk flow corresponding to joy.

[1955] "Example 2"

[1956] In another embodiment of the present invention, a system is provided in which an emotion engine includes means for recognizing emotions from the tone of a user's voice. This system analyzes audio information such as the tone, volume, and speaking rate of the user's voice, and estimates the user's emotional state from the results. For example, if the user's voice tone is high and the speaking rate is fast, it is determined that the user is likely excited.

[1957] "Example 3"

[1958] In a further embodiment of the present invention, a system is provided that includes a means for adjusting a talk flow according to a recognized user's emotion. In this system, an AI recognizes the user's emotion and generates a talk flow that is optimal for that emotion. For example, if the user expresses anger, the AI ​​uses that information to generate a talk flow that calms the user.

[1959] The processing flow of each embodiment will be described below.

[1960] "Example 1"

[1961] Step 1: A user interaction is initiated.

[1962] Step 2: The emotion engine recognizes the user's emotions in real time.

[1963] Step 3: The recognized emotional information is conveyed to the AI.

[1964] Step 4: The AI ​​generates the optimal conversation flow based on the emotion.

[1965] "Example 2"

[1966] Step 1: The user's voice information (tone of voice, volume, speaking rate, etc.) is analyzed by the emotion engine.

[1967] Step 2: The user's emotional state is inferred from the analysis results.

[1968] Step 3: The estimated emotional information is conveyed to the AI.

[1969] Step 4: The AI ​​generates the optimal conversation flow based on the emotion.

[1970] "Example 3"

[1971] Step 1: A user interaction is initiated.

[1972] Step 2: The emotion engine recognizes the user's emotions in real time.

[1973] Step 3: The recognized emotional information is conveyed to the AI.

[1974] Step 4: The AI ​​generates the optimal conversation flow based on the emotion.

[1975] Step 5: The generated talk flow is presented to the user.

[1976] Example 1

[1977] Next, a description will be given of Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1978] Conventional talk flow generation systems were unable to effectively utilize past conversation data, making it difficult to generate optimal talk flows in real time that matched the user's emotions. Furthermore, there was a lack of a way to quickly generate talk flows that reflected the specifications and features of new services. This meant that improvements in sales efficiency and customer satisfaction were not fully achieved.

[1979] 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.

[1980] In this invention, the server includes means for storing talk flows developed in the past, means for inputting specifications and features of new services into an artificial intelligence, means for generating an optimal proposed talk flow tailored to the service, means for collecting log data of past sales conversations, means for analyzing the collected log data to extract patterns of successful sales conversations, means for recognizing user emotions in real time, and means for generating an optimal talk flow based on the recognized emotion information. This makes it possible to effectively utilize patterns of successful sales conversations in the past and generate an optimal talk flow in real time that corresponds to the user's emotions.

[1981] "Talk flow" refers to a series of statements and responses in a dialogue with a user.

[1982] "Artificial intelligence" refers to the technology that enables computer systems to mimic human intelligence to learn, reason, perceive, and make decisions.

[1983] "Log Data" refers to a series of events and activities that a system or application records during its operation.

[1984] "Natural language processing" refers to the technology that enables computers to understand, gen...

Claims

1. A means of storing talk flows that have been developed in the past; A means to load the specifications and features of the new service into artificial intelligence, A means to generate proposal talk flows for new services, a means for collecting log data of past sales conversations; A means for analyzing the collected log data to extract patterns of successful sales conversations by inputting specific prompt sentences into a generative AI model; A means for recognizing the emotion of the user by analyzing the tone of the user's voice, the volume of the user's voice, and the speaking rate of the user's voice; means for selecting a talk flow based on the emotion; means for displaying the selected talk flow on a user's terminal; A system including:

2. The system of claim 1 , wherein the means for feeding the specifications and the features into an artificial intelligence feeds the specifications and the features as text data.

3. The system of claim 1 , wherein the means for generating the proposed talk flow generates the proposed talk flow using artificial intelligence-based natural language processing. The system of claim 1 comprising:

Images