system

Abstract

Provide a system. 【Solution means】 Means for acquiring an electronic communication history, Means for preprocessing the acquired electronic communication history, Means for learning an individual generation model based on the preprocessed electronic communication history, Means for analyzing a question, Means for selecting an appropriate generation model based on the analysis result, Means for generating an answer using the selected generation model, Means for providing the generated answer, A system including the above.

Description

Technical Field

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

Background Art

[0002] Patent Document 1 discloses a method for controlling a persona chatbot, which is performed by at least one processor, including steps of receiving a user utterance, adding the user utterance to a prompt including an instruction sentence related to an explanation of the chatbot's 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

Summary of the Invention

Problems to be Solved by the Invention

[0004] In enterprises, insufficient communication between departments and inefficient information sharing have become problems. In such a situation, it has become an issue that it is necessary to respond again to questions that have already been answered in the past, or that the knowledge of employees who have left the company is lost. Furthermore, as the personalization of business knowledge progresses, there are many situations where new employees struggle to respond, which may lead to a decrease in productivity. This invention is intended to address these problems and promote the efficiency of business operations and information sharing between departments.

Means for Solving the Problems

[0005] The system acquires electronic communication history, preprocesses it, and trains a generative model that includes individual knowledge and skills. Then, for subsequent questions, it selects the appropriate generative model based on the analysis results and automatically generates answers. This system enables efficient work execution by using digital clones of employees to efficiently utilize past knowledge and prioritize responses to new questions. Furthermore, it allows for the sharing of acquired knowledge across the entire company while protecting personal information. The generated answers also include the decision-making process, ensuring reliable information provision.

[0006] "Electronic communication history" refers to records of past communications conducted via chat or messaging applications within a company.

[0007] "Preprocessing" refers to the process of removing unnecessary information and personally identifiable information from collected electronic communication history and preparing the data into an analyzable format.

[0008] A "generative model" refers to an artificial intelligence model that has been trained to mimic the knowledge and skills of a specific employee based on historical data.

[0009] "Means for analyzing questions" refers to technologies that use natural language processing techniques to understand questions entered by users and provide functions for interpreting their intent.

[0010] "Means of generating answers" refers to the process of automatically constructing answers to input questions using a selected generative model.

[0011] "Means of delivery" refers to the technical means of returning the generated response to the user and communicating it in a usable format. [Brief explanation of the drawing]

[0012] [Figure 1] This is a conceptual diagram showing an example of the configuration of a data processing system according to the first embodiment. [Figure 2] This is a conceptual diagram showing an example of the essential functions of a data processing device and a smart device according to the first embodiment. [Figure 3] This is a conceptual diagram showing an example of the configuration of a data processing system according to the second embodiment. [Figure 4] This is a conceptual diagram showing an example of the main functions of a data processing device and smart glasses according to the second embodiment. [Figure 5] This is a conceptual diagram showing an example of the configuration of a data processing system according to the third embodiment. [Figure 6] This is a conceptual diagram showing an example of the main functions of a data processing device and a headset-type terminal according to the third embodiment. [Figure 7] This is a conceptual diagram showing an example of the configuration of a data processing system according to the fourth embodiment. [Figure 8] This is a conceptual diagram showing an example of the main functions of a data processing device and a robot according to the fourth embodiment. [Figure 9] This shows an emotion map where multiple emotions are mapped. [Figure 10] This shows an emotion map where multiple emotions are mapped. [Figure 11] This is a sequence diagram showing the processing flow of the data processing system in Example 1. [Figure 12] This is a sequence diagram showing the processing flow of the data processing system in Application Example 1. [Figure 13] This is a sequence diagram showing the processing flow of the data processing system in Example 2, which incorporates an emotion engine. [Figure 14] This is a sequence diagram showing the processing flow of the data processing system in Application Example 2, which combines an emotion engine. [Modes for carrying out the invention]

[0013] Hereinafter, an example of an embodiment of the system relating to the technology of this disclosure will be described with reference to the attached drawings.

[0014] First, the terms used in the following description will be explained.

[0015] In the following embodiments, a labeled processor (hereinafter simply referred to as "processor") may be a single arithmetic unit or a combination of multiple arithmetic units. Also, the processor may be a single type of arithmetic unit or a combination of multiple types of arithmetic units. Examples of arithmetic units 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), and the like.

[0016] In the following embodiments, a labeled RAM (Random Access Memory) is a memory in which information is temporarily stored and is used as a work memory by the processor.

[0017] In the following embodiments, a labeled storage is one or more non-volatile storage devices that store various programs and various parameters, etc. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), magnetic disks (e.g., hard disks), or magnetic tapes, etc.

[0018] In the following embodiments, a labeled communication I / F (Interface) is an interface including a communication processor and 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), or Bluetooth (registered trademark), etc.

[0019] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B." That is, "A and / or B" means that it may be A alone, or B alone, or a combination of A and B. Furthermore, in this specification, the same concept as "A and / or B" applies when expressing three or more things linked by "and / or."

[0020] [First Embodiment]

[0021] Figure 1 shows an example of the configuration of the data processing system 10 according to the first embodiment.

[0022] As shown in Figure 1, the 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.

[0023] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0024] The smart device 14 comprises a computer 36, a reception device 38, an output device 40, a camera 42, and a communication interface 44. The computer 36 comprises a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The reception device 38, output device 40, and camera 42 are also connected to the bus 52.

[0025] The reception device 38 is equipped with a touch panel 38A and a microphone 38B, etc., and receives user input. The touch panel 38A receives user input by detecting contact with an object (e.g., a pen or finger). The microphone 38B receives user input by detecting the user's voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and 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.

[0026] 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 perceptible to the user 20 (e.g., audio and / or text). The display 40A displays visible information such as text and images according to instructions from the processor 46. The speaker 40B outputs audio according to instructions from the processor 46. The camera 42 is a small digital camera equipped with an optical system such as a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0027] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various types of information between processor 46 and processor 28 via network 54.

[0028] Figure 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0029] As shown in Figure 2, in the data processing device 12, a specific processing 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" related to the technology of this 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 according to the specific processing program 56 executed on the RAM 30.

[0030] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0031] In the smart device 14, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. The reception output program 60 is used in conjunction with a 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 processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[0032] Next, the specific processing performed by the specific processing unit 290 of the data processing device 12 will be described. In the following description, the data processing device 12 will be referred to as the "server" and the smart device 14 as the "terminal".

[0033] This invention is a system developed to efficiently utilize employees' business knowledge and know-how to improve operational efficiency. The system aims to create a generative model for each employee using the company's past electronic communication history and to automatically answer questions related to their work.

[0034] System Overview

[0035] The server first collects internal electronic communication history and appropriately stores this information in a database. The stored data is then preprocessed and used as training material for generative models. Preprocessing includes normalization of text data, noise reduction, and protection of personal information.

[0036] The server uses pre-processed data to train specialized generative models for each employee. This builds models that reflect each employee's knowledge and skills. Through these models, the system can automatically generate responses by mimicking past experiences and answers provided by the employees.

[0037] When a user inputs a question into the system via a terminal, the server analyzes the question and understands its intent. Based on this analysis, the server selects the optimal generative model. The selected model generates an answer to the analyzed question, referencing similar past cases and related information in the process.

[0038] The generated answers are sent from the server to the user's terminal. Users can then proceed with their work based on the reliable answers provided. This system enables effective knowledge sharing and consistent application of work knowledge, particularly across different departments and with new employees.

[0039] Specific example

[0040] For example, if a user asks for a solution to a technical problem that has arisen in a project, the server searches past electronic communication history for similar problems and generates an answer using an appropriate generative model based on that history. This allows users to solve problems quickly and accurately based on relevant information. Furthermore, since information previously provided by former employees can also be utilized, there is the advantage of ensuring that organizational knowledge is continuously available.

[0041] The following describes the processing flow.

[0042] Step 1:

[0043] The server collects internal electronic communication history. This data is obtained from chat applications, emails, and other sources and stored in a database. During collection, metadata such as the speaker, date and time, and content are also recorded.

[0044] Step 2:

[0045] The server preprocesses the collected communication history. This process includes text normalization, removal of irrelevant information, and filtering of personal information. The preprocessed data is then organized into a format suitable for training generative models.

[0046] Step 3:

[0047] The server uses pre-processed data to train generative models for each employee. During this training process, natural language processing techniques are used to analyze the content and patterns of employees' past statements, thereby generating models that reflect their individual knowledge and skills.

[0048] Step 4:

[0049] The user enters a question into the system from a terminal. The terminal receives the user's input, formats the data, and sends it to the server.

[0050] Step 5:

[0051] The server analyzes the received question. This analysis uses natural language processing to understand the intent and main points of the question, extracting relevant keywords and context.

[0052] Step 6:

[0053] The server selects the optimal generative model based on the analysis results. Here, it identifies and selects the model of the employee with the most relevant knowledge to the question.

[0054] Step 7:

[0055] The server generates answers using the selected generative model. This generation process involves constructing highly accurate answers by referencing data from similar past questions and a knowledge base.

[0056] Step 8:

[0057] The server sends the generated response to the user's device. The user can then receive the response on their device and use it for their work.

[0058] (Example 1)

[0059] Next, we will describe Example 1. In the following description, the data processing device 12 will be referred to as the "server," and the smart device 14 will be referred to as the "terminal."

[0060] In modern companies, there is a challenge in effectively sharing the business knowledge and know-how of each employee and obtaining information necessary to carry out tasks quickly and accurately. Furthermore, there is a need to effectively utilize data while paying attention to the protection of personal information.

[0061] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0062] In this invention, the server includes means for collecting and storing electronic communication history, means for normalizing the collected electronic communication history and removing noise and personal information, and means for learning individual generative models based on the pre-processed history. This enables the automatic generation of answers by utilizing the knowledge accumulated within the company and based on the specialized knowledge of each employee.

[0063] "Electronic communication history" refers to the record of communications such as emails and chat messages conducted within an organization.

[0064] "Normalization" is the process of converting text data into a unified and consistent format, which improves data quality and processing efficiency.

[0065] "Noise reduction" is a technique that removes unnecessary information and errors from text data, and is performed to improve the usefulness of the data.

[0066] A "personalized generative model" is a customized AI model tailored to a specific employee or situation, possessing the ability to generate answers by learning from past knowledge and experience.

[0067] A "user question" is a request from a user to provide information or solve a problem, and is usually written in natural language.

[0068] "Understanding intent" is the process of analyzing a user's question and identifying the underlying requirements and objectives.

[0069] "Referring to past cases" is a method of using data accumulated in the past to derive useful information and answers to current questions.

[0070] "User terminal" refers to a computing device used by a user to access the system, and includes PCs, smartphones, and other devices.

[0071] In this invention, the server first collects internal electronic communication history. This history is collected using the company's email system and chat platform. This collection process utilizes common APIs and data retrieval scripts. The server stores the collected data in a relational database management system (RDBMS), specifically MySQL® or PostgreSQL. The database functions as a foundation for efficiently storing large amounts of historical information and for later analysis and utilization.

[0072] Next, the server preprocesses the stored electronic communication history. This preprocessing utilizes Python NLP libraries such as NLTK and spaCy. Using these libraries, the server normalizes the text data, removing noise and personal information. This process ensures the data is clean and suitable for analysis. For example, it anonymizes names and email addresses while removing special characters and extraneous tags.

[0073] Once the data is ready, the server uses deep learning frameworks such as TENSORFLOW® or PyTorch to train specialized generative models for each employee. This builds generative models that reflect each employee's business knowledge and past experience. These generative models are designed to mimic the specialized knowledge of different employees and generate answers specific to their particular tasks.

[0074] When a user inputs a business-related question into the system via their terminal, the server analyzes the question using natural language processing technology and selects the optimal generative model. Using the selected model, the server generates an answer related to the question, referencing similar past cases. The generated answer is then provided to the user's terminal in real time.

[0075] For example, if a user sends a prompt such as, "Please tell me about past examples of data protection policies for a new project," the server will refer to the communication history of similar past projects, select an appropriate model, and provide a response. In this process, reliable information, including specific methods and points to note regarding the project, can be quickly conveyed to the user.

[0076] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0077] Step 1:

[0078] The server automatically collects electronic communication history from the company's electronic communication platform. Specifically, it retrieves data using APIs from mail servers and chat tools. The input is raw communication data, and the output is raw data stored in an RDBMS. This data is then saved for further processing.

[0079] Step 2:

[0080] The server preprocesses the collected electronic communication history. The input is the raw data saved in step 1. Specifically, it normalizes the data using Python's NLP library, removes noise and HTML tags, and anonymizes personal information. The output is clean, analyzable text data.

[0081] Step 3:

[0082] The server trains a generative AI model for each employee based on pre-processed data. The input is the clean data generated in step 2. In this process, the model is trained using TensorFlow or PyTorch, and a custom model is built that reflects the employee's business knowledge. The output is a generative AI model tailored to each employee.

[0083] Step 4:

[0084] Users input business-related questions into the system using a terminal. Specifically, they send prompt messages using a web form. The input is a question written in natural language, and the output is the query data sent to the server.

[0085] Step 5:

[0086] The server analyzes the question received from the user and understands its intent. The input is the question received in step 4. Using a natural language processing algorithm, it identifies the intent and selects the optimal generative model. The output is the selected generative AI model.

[0087] Step 6:

[0088] The server generates an answer to the question using the selected generative model. The input is the generative AI model obtained in step 5 and the content of the question. The server generates a precise answer while referring to past electronic communication history. The output is the answer data provided to the user.

[0089] Step 7:

[0090] The server-generated response is sent to the user's terminal. The input is the response data from the server, and the output is the answer information displayed on the terminal. This allows the user to quickly use it as a reference for their work.

[0091] (Application Example 1)

[0092] Next, we will explain Application Example 1. In the following explanation, the data processing device 12 will be referred to as the "server," and the smart device 14 will be referred to as the "terminal."

[0093] In modern cities, there is a need for methods to respond quickly and accurately to inquiries and problems from citizens. In particular, the quality of administrative services can be improved by effectively utilizing past cases and examples of responses. However, existing systems have the challenge of not being able to fully utilize past data, resulting in time-consuming individual responses.

[0094] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0095] In this invention, the server includes means for acquiring electronic communication records, means for preprocessing the acquired electronic communication records, and means for learning individual generative models based on the preprocessed electronic communication records. This enables administrative agencies to provide automated responses to inquiries from citizens, referencing past cases.

[0096] "Electronic communication records" refer to logs of emails and messages exchanged within a company or organization, and are a collection of information that includes a history of business-related communication.

[0097] "Preprocessing" refers to the process of removing noise from collected data and ensuring data integrity, and includes operations such as anonymization and normalization of confidential information.

[0098] A "generative model" is an algorithmic model that uses machine learning to learn from a specific dataset and automatically perform tasks such as text generation.

[0099] An "inquiry" is a question or request that a user or citizen enters into a system to obtain information, and it seeks to provide information about a specific event.

[0100] "Analysis results" refer to the information obtained when analyzing the input data, and are used as a reference for selecting a generative model and generating responses.

[0101] "Case data" refers to information that records past cases and how they were handled, and it is data that supports approaches to new problems.

[0102] "Administrative agencies" are organizations of the national or local government that provide public services and support the lives of communities and citizens.

[0103] To implement this invention, a server must first collect electronic communication records. The server obtains data from existing communication logs and preprocesses it. In the preprocessing, natural language processing tools such as NLTK and spaCy are used to remove noise from the collected data and to perform normalization and anonymization to maintain data integrity. Next, a generative model is trained using Python and the Hugging Face Transformers library based on the preprocessed data. The generative model has the ability to generate specific answers based on past data.

[0104] When a user submits a query to the system via a web interface or similar means, the query is first forwarded to the server. The server analyzes the query and selects the optimal generative model based on the analysis results. In this process, past case data is referenced, and the response is supplemented based on similar cases.

[0105] Users can quickly receive automatically generated answers through this process, with the aim of improving public services. For example, if a user asks, "Please tell me about the new recycling program," the server can use past inquiry history to provide details about the program and information about upcoming related events. In this way, the server uses a generation AI model to optimize answers in response to various inquiries.

[0106] Examples of specific prompt messages include the following:

[0107] "Please provide information about the new recycling program."

[0108] "Could you please provide details about similar environmental events that have been held in the past?"

[0109] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0110] Step 1:

[0111] The server retrieves electronic communication records within an organization. Input is data from the organization's mail server or messaging platform. It then stores the communication logs in a database. Output is the set of stored electronic communication records.

[0112] Step 2:

[0113] The server preprocesses the acquired electronic communication records. The input is the electronic communication records saved in step 1. The processing involves denoising, normalization, and anonymization using NLTK or spaCy. This generates analyzable text data while protecting privacy. The output is the preprocessed, clean text data.

[0114] Step 3:

[0115] The server trains a generative model based on preprocessed data. The input is the text data obtained in step 2. A specialized generative model is built using the Hugging Face Transformers library. The output is the trained generative model.

[0116] Step 4:

[0117] The user sends a query to the system from their terminal. The input is the query text entered by the user. The output is the query text received by the server.

[0118] Step 5:

[0119] The server analyzes the query received from the user. The input is the query text received by the server in step 4. The server uses a natural language processing engine to analyze the text and understand its intent. The output is the analysis result.

[0120] Step 6:

[0121] The server selects an appropriate generative model based on the analysis results. The input is the analysis results obtained in step 5. The selection algorithm determines the most relevant generative model. The output is the selected generative model.

[0122] Step 7:

[0123] The server generates answers using the selected generative model. The input consists of the selected generative model and its analysis results. Past case data is also referenced to reinforce the answers. The output is the generated answer text.

[0124] Step 8:

[0125] The user's device receives the response provided by the server. The input is the response text sent from the server. Based on this information, the user can quickly obtain the desired information. The output is the response displayed to the user.

[0126] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[0127] This invention provides an interactive response system that takes user emotions into account by combining an emotion engine. The system uses electronic communication history to construct individual generative models and reflects user emotions in the question analysis and response generation processes.

[0128] System Overview

[0129] The server first collects and stores past electronic communication history within the company into a database. The stored data is preprocessed to remove personally identifiable information and reduce noise. This ensures that a safe and effective dataset is used to train the generative model.

[0130] Based on this pre-processed data, the server trains a generative model tailored to each employee. This results in a highly accurate model that mimics the experience and expertise of each individual employee. The trained model becomes the key to generating appropriate answers in response to user questions.

[0131] When a user enters and submits a question from their device, the server analyzes that question. This analysis utilizes an emotion engine to extract emotions from the user's text. This emotion information is used to understand the intent behind the question and is reflected in the analysis results.

[0132] As a result, the server selects the optimal generative model and generates responses that are adjusted to the user's emotions. The generated responses are crafted in a tone that takes the user's emotions into consideration, thereby improving user satisfaction and convenience. For example, if signs of stress are detected, the response will be provided in a more considerate tone.

[0133] Ultimately, the server sends this generated response to the user's terminal, providing the user with the information they need in an appropriate format. This system not only leverages the expertise of individual employees but also enables responses that take user emotions into consideration, thereby improving operational efficiency and the quality of information exchange.

[0134] Specific example

[0135] For example, when a user asks a question about project delays, the sentiment engine may detect anxiety from the text. In this case, the server uses a generative model of an employee with past project management expertise to generate and provide a response that includes specific measures to alleviate anxiety and encouraging messages. This allows the user to move on to the next step with confidence.

[0136] The following describes the processing flow.

[0137] Step 1:

[0138] The server collects electronic communication history from within the company. It primarily retrieves data from internal chat tools and email logs and stores it in a database. The stored information includes metadata such as the speaker, date and time, and content.

[0139] Step 2:

[0140] The server preprocesses the collected electronic communication history. This processing includes text normalization and removal of unnecessary data, and in particular, filtering to remove personally identifiable information.

[0141] Step 3:

[0142] The server uses pre-processed data to train a generative model for each employee. Using natural language processing algorithms, it builds a model that reflects each employee's job knowledge and skills. This model formation allows the server to mimic each employee's areas of expertise and response patterns.

[0143] Step 4:

[0144] Users enter their questions using their terminals and send them to the server. The question input screen allows for questions in text format, and users can freely submit inquiries.

[0145] Step 5:

[0146] The server receives the user's question and uses an emotion engine to analyze the emotions contained in the question. This extracts the emotional context from the text in the question and helps understand the user's psychological state.

[0147] Step 6:

[0148] The server selects the optimal generative model based on the question content and sentiment analysis results. The selection criteria prioritize models created by employees with relevant knowledge to the question. It also obtains feedback to adjust the tone of the response according to the sentiment.

[0149] Step 7:

[0150] The server generates responses using the selected generative model. At this time, based on the sentiment analysis results, the responses are generated in a tone that matches the user's emotions. Relevant information from past data is referenced, and necessary information is added and the tone adjusted.

[0151] Step 8:

[0152] The server sends the generated response to the terminal, delivering it to the user. The user can then review the response on the terminal and use it to solve their work-related problems. This allows users to solve problems efficiently.

[0153] (Example 2)

[0154] Next, we will describe Example 2. In the following description, the data processing device 12 will be referred to as the "server" and the smart device 14 as the "terminal".

[0155] Traditional interactive systems often provide only uniform answers to user questions, failing to adequately consider user emotions and intentions, resulting in low user satisfaction. Furthermore, obtaining accurate and appropriate answers to questions requiring specialized knowledge was difficult.

[0156] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[0157] In this invention, the server includes means for acquiring electronic communication data, means for preprocessing the acquired data and removing personal information, means for learning individual knowledge models based on the preprocessed data, means for analyzing questions and extracting emotional information, means for selecting an appropriate knowledge model based on the analysis results, means for generating answers in an emotionally appropriate tone using the selected knowledge model, and means for providing the generated answers to the user. This makes it possible to provide high-quality answers that take into account the user's emotions while leveraging expertise.

[0158] "Electronic communication data" refers to information such as emails and messages exchanged via the internet and other communication networks.

[0159] "Preprocessing" refers to the process of removing personally identifiable information from electronic communication data and reducing noise to improve data quality.

[0160] A "knowledge model" refers to a generative AI model trained using a specific dataset, and is an artificial intelligence algorithm that possesses a deep understanding of a particular subject.

[0161] "Emotional information" refers to data that indicates the type and degree of emotion extracted from users' questions and statements.

[0162] "Tone" refers to the wording and facial expressions used in the generated response, and by adjusting them according to the user's emotions, it is an element that influences the nuance and how the response is received.

[0163] "Analysis" refers to the process of using natural language processing techniques to analyze the information contained in a question in detail and understand its intent and content.

[0164] In this embodiment of the invention, the interactive response system employs multiple technical means to provide responses that take into account the user's emotions.

[0165] The server collects electronic communication data via the internet and other communication networks and stores it in a database. This collected data is first pre-processed. This pre-processing uses specialized software, such as Python scripts, to remove personally identifiable information and reduce noise, thereby improving data quality. This process prepares the data for training.

[0166] Next, the server uses the pre-processed data to train a knowledge model specific to each employee, creating an AI model. This training process utilizes machine learning frameworks such as TensorFlow and PyTorch. The server builds a highly accurate model that reflects the employee's work knowledge and experience, which then forms the basis for subsequent question answering.

[0167] When a user enters a question using their device, the text data is sent to the server and analyzed through natural language processing. The server then activates an emotion engine and extracts emotional information from the text using an emotion analysis API. This emotional information is crucial for understanding the intent of the question and selecting the optimal knowledge model.

[0168] The generated responses are adjusted in tone to reflect the user's emotions. For example, if the user is feeling anxious, the response will be adjusted to include elements of warmth and encouragement. This improves both user satisfaction and work efficiency.

[0169] As a concrete example, suppose a user enters the prompt, "I would like specific advice regarding project delays." The server processes this input, selects a knowledge model with data on similar past cases, and generates a reassuring response. This response includes specific measures and words of encouragement, allowing the user to confidently move on to the next step.

[0170] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0171] Step 1:

[0172] The server acquires electronic communication data via the internet or other communication networks. The input includes raw data such as emails and messages. This data is stored in a database. At this stage, the acquired electronic communication data is output as is.

[0173] Step 2:

[0174] The server preprocesses the stored electronic communication data. The input is the raw data obtained in step 1. In this process, specialized software is used to remove personal information and apply algorithms to reduce noise. This generates safe and clean data, which becomes the output for the next step.

[0175] Step 3:

[0176] The server trains a generative AI model tailored to each employee based on pre-processed data. The input is the clean data obtained in step 2. Machine learning frameworks such as TensorFlow and PyTorch are used to train this data and build a knowledge model. The output at this stage is a generative AI model tailored to each employee.

[0177] Step 4:

[0178] The user enters a question via a terminal and sends it to the server. The input is the user's question, which is free-form text. At this point, the entered prompt text is passed directly to the next step.

[0179] Step 5:

[0180] When the server receives a question from a user, it analyzes it using natural language processing tools. The input is the prompt sentence from step 4, and word segmentation and part-of-speech tagging are performed, followed by analysis to understand the intent. The results of this analysis are output, and sentiment extraction is then performed.

[0181] Step 6:

[0182] The server uses an emotion engine to extract emotional information from the question text. The input is the analysis result from step 5. The type and degree of emotion are identified using the emotion analysis API, which is then used in the next step. At this point, this emotional information is output.

[0183] Step 7:

[0184] The server selects the optimal generative AI model based on the analysis results and extracted sentiment information. The inputs are the analysis results from step 5 and the sentiment information from step 6. Considering these, a suitable knowledge model is selected, and the answer is generated by that knowledge model.

[0185] Step 8:

[0186] The server uses the selected generative AI model to generate responses in a tone that matches the user's emotions. The input is the knowledge model selected in step 7, and the output is the response with the adjusted tone. Based on the analyzed emotional information, a response is created that reflects the optimal tone and content for the user.

[0187] Step 9:

[0188] The server sends the final generated response to the user's device. The input is the response generated in step 8. The user receives this information on their device and can obtain the necessary knowledge and advice. The output is the adjusted-tone response displayed on the user's device.

[0189] (Application Example 2)

[0190] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as a "server" and the smart device 14 as a "terminal".

[0191] In caregiving settings, accurate communication tailored to the emotions and circumstances of those receiving care is essential, but achieving this is not easy. In particular, accurately understanding a person's emotions and responding accordingly is a challenging task. Therefore, there is a need to develop systems that enable care staff and family members to respond appropriately based on the emotions of those receiving care.

[0192] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[0193] In this invention, the server includes means for acquiring electronic communication history, means for pre-processing the acquired electronic communication history, means for learning individual generative models based on the pre-processed electronic communication history, means for analyzing questions, means for analyzing the user's emotions, means for selecting an appropriate generative model based on the analysis results and emotions, means for generating an emotionally appropriate response using the selected generative model, and means for providing the generated response. This enables responses based on the emotions of the person receiving care, thereby improving the quality of care.

[0194] "Electronic communication history" refers to a record of messages and communication content transmitted using communication devices.

[0195] "Preprocessing" is the process of removing noise and unnecessary information from acquired data and converting it into a format suitable for analysis and model training.

[0196] A "generative model" is an algorithm that has been trained to produce an appropriate output for a given input.

[0197] "Methods for analyzing questions" refer to technologies for understanding the content of questions entered by users and identifying their intent.

[0198] "Methods for analyzing emotions" refer to technologies that extract emotions from users' statements and texts and understand their emotional state.

[0199] A "generated answer" is a response to a user's question, created by a generative model.

[0200] "Means of provision" refers to the methods and technologies used to present the generated answers to the user.

[0201] The system of this invention primarily consists of a server and a user terminal. The server first acquires electronic communication history and preprocesses it. Preprocessing includes removing personally identifiable information and reducing noise from the collected digital data. This results in a clean dataset suitable for analysis and training generative models. The hardware used is a cloud server, and the software used for data analysis includes Python and R.

[0202] Based on pre-processed data, the server trains individual generative AI models. This involves creating user-specific models and utilizing machine learning algorithms to mimic their expertise and experience. Deep learning frameworks such as TensorFlow and PyTorch are used in this process.

[0203] When a user enters a question from their device, the device sends that data to the server. The server analyzes the received question and activates an emotion analysis engine to extract the user's emotions. This operation can be performed using natural language processing techniques, and libraries such as NLTK and spaCy are used for this purpose.

[0204] Based on the sentiment analysis results, the server selects the most appropriate generative model and generates a response that matches the user's emotions. This response generation process adjusts the tone and content of the response to meet the user's needs. For example, if stress or anxiety is detected, the server provides more reassuring feedback.

[0205] The generated responses are sent from the server to the user's terminal, where the user receives them either visually or audibly. This system enables personalized responses that take the user's emotions into consideration, making it applicable in fields such as elderly care.

[0206] For example, if an elderly person makes a statement to the terminal such as "I'm not feeling very well today," the server can analyze it and respond with encouraging words or suggestions for simple exercises to help them relax. An example of a prompt to input into the generative AI model could be, "Analyze the user's recent emotional state and suggest the most appropriate care method."

[0207] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0208] Step 1:

[0209] The server retrieves electronic communication history data. The input consists of messages and communication data previously sent by the user. This data is retrieved from the database and prepared for the next preprocessing step.

[0210] Step 2:

[0211] The server preprocesses the acquired data. It removes personally identifiable information from the input data and reduces unnecessary noise. This results in a clean dataset suitable for training. Specifically, it performs text normalization and filtering.

[0212] Step 3:

[0213] The server trains a generative AI model based on preprocessed data. The input is a clean dataset, and the output is an individual generative model. The model is trained using a machine learning framework such as TensorFlow.

[0214] Step 4:

[0215] The user enters a question from their terminal and sends it to the server. The input is the user's question text, which the terminal forwards to the server.

[0216] Step 5:

[0217] The server parses the received question. The input is the question text. Natural language processing techniques are used to understand the meaning of the text and identify the necessary context. The output is the parsed intent information.

[0218] Step 6:

[0219] The server analyzes the user's emotions using an emotion analysis engine. The input is the question text, and emotions are extracted using an emotion analysis API, etc. The output is emotion information (joy, anxiety, sadness, etc.).

[0220] Step 7:

[0221] The server selects an appropriate generative model based on the question and sentiment information. It takes the analyzed intent and sentiment information as input and determines the most suitable generative model. The output is the selected generative model.

[0222] Step 8:

[0223] The server generates emotion-sensitive responses using the selected generative model. Here, the tone and content of the responses are adjusted to match the user's emotions based on the input analysis data. The output is the generated response.

[0224] Step 9:

[0225] The server sends the generated response to the user's terminal. The terminal then provides the user with the received response in text or audio format. The output is the response to the user.

[0226] 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 user input for the result of the specific processing. The control unit 46A transmits the audio data indicating 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.

[0227] Data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of data generation model 58 is ChatGPT (registered trademark) (Internet search).<URL: https: / / openai.com / blog / chatgpt> ), Gemini (registered trademark) (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0228] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and the specific processing may also be performed by the smart device 14.

[0229] [Second Embodiment]

[0230] Figure 3 shows an example of the configuration of the data processing system 210 according to the second embodiment.

[0231] As shown in Figure 3, the data processing system 210 includes a data processing device 12 and smart glasses 214. An example of the data processing device 12 is a server.

[0232] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0233] The smart glasses 214 include a computer 36, a microphone 238, a speaker 240, a camera 42, and a communication interface 44. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, and camera 42 are also connected to the bus 52.

[0234] The microphone 238 receives voice signals from the user 20 and receives instructions from the user 20. The microphone 238 captures the voice signals from the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[0235] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the area around the user 20 (for example, an imaging range defined by a field of view equivalent to the width of a typical healthy person's field of vision).

[0236] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[0237] Figure 4 shows an example of the main functions of the data processing device 12 and the smart glasses 214. As shown in Figure 4, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[0238] The specific processing program 56 is an example of a "program" relating to the technology of this 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.

[0239] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0240] In the smart glasses 214, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. 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 processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[0241] Next, the identification processing performed by the identification processing unit 290 of the data processing device 12 will be described. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal".

[0242] This invention is a system developed to efficiently utilize employees' business knowledge and know-how to improve operational efficiency. The system aims to create a generative model for each employee using the company's past electronic communication history and to automatically answer questions related to their work.

[0243] System Overview

[0244] The server first collects internal electronic communication history and appropriately stores this information in a database. The stored data is then preprocessed and used as training material for generative models. Preprocessing includes normalization of text data, noise reduction, and protection of personal information.

[0245] The server uses pre-processed data to train specialized generative models for each employee. This builds models that reflect each employee's knowledge and skills. Through these models, the system can automatically generate responses by mimicking past experiences and answers provided by the employees.

[0246] When a user inputs a question into the system via a terminal, the server analyzes the question and understands its intent. Based on this analysis, the server selects the optimal generative model. The selected model generates an answer to the analyzed question, referencing similar past cases and related information in the process.

[0247] The generated answers are sent from the server to the user's terminal. Users can then proceed with their work based on the reliable answers provided. This system enables effective knowledge sharing and consistent application of work knowledge, particularly across different departments and with new employees.

[0248] Specific example

[0249] For example, if a user asks for a solution to a technical problem that has arisen in a project, the server searches past electronic communication history for similar problems and generates an answer using an appropriate generative model based on that history. This allows users to solve problems quickly and accurately based on relevant information. Furthermore, since information previously provided by former employees can also be utilized, there is the advantage of ensuring that organizational knowledge is continuously available.

[0250] The following describes the processing flow.

[0251] Step 1:

[0252] The server collects internal electronic communication history. This data is obtained from chat applications, emails, and other sources and stored in a database. During collection, metadata such as the speaker, date and time, and content are also recorded.

[0253] Step 2:

[0254] The server preprocesses the collected communication history. This process includes text normalization, removal of irrelevant information, and filtering of personal information. The preprocessed data is then organized into a format suitable for training generative models.

[0255] Step 3:

[0256] The server uses pre-processed data to train generative models for each employee. During this training process, natural language processing techniques are used to analyze the content and patterns of employees' past statements, thereby generating models that reflect their individual knowledge and skills.

[0257] Step 4:

[0258] The user enters a question into the system from a terminal. The terminal receives the user's input, formats the data, and sends it to the server.

[0259] Step 5:

[0260] The server analyzes the received question. This analysis uses natural language processing to understand the intent and main points of the question, extracting relevant keywords and context.

[0261] Step 6:

[0262] The server selects the optimal generative model based on the analysis results. Here, it identifies and selects the model of the employee with the most relevant knowledge to the question.

[0263] Step 7:

[0264] The server generates answers using the selected generative model. This generation process involves constructing highly accurate answers by referencing data from similar past questions and a knowledge base.

[0265] Step 8:

[0266] The server sends the generated response to the user's device. The user can then receive the response on their device and use it for their work.

[0267] (Example 1)

[0268] Next, we will describe Example 1. In the following description, the data processing device 12 will be referred to as the "server," and the smart glasses 214 will be referred to as the "terminal."

[0269] In modern companies, there is a challenge in effectively sharing the business knowledge and know-how of each employee and obtaining information necessary to carry out tasks quickly and accurately. Furthermore, there is a need to effectively utilize data while paying attention to the protection of personal information.

[0270] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0271] In this invention, the server includes means for collecting and storing electronic communication history, means for normalizing the collected electronic communication history and removing noise and personal information, and means for learning individual generative models based on the pre-processed history. This enables the automatic generation of answers by utilizing the knowledge accumulated within the company and based on the specialized knowledge of each employee.

[0272] "Electronic communication history" refers to the record of communications such as emails and chat messages conducted within an organization.

[0273] "Normalization" is the process of converting text data into a unified and consistent format, which improves data quality and processing efficiency.

[0274] "Noise reduction" is a technique that removes unnecessary information and errors from text data, and is performed to improve the usefulness of the data.

[0275] A "personalized generative model" is a customized AI model tailored to a specific employee or situation, possessing the ability to generate answers by learning from past knowledge and experience.

[0276] A "user question" is a request from a user to provide information or solve a problem, and is usually written in natural language.

[0277] "Understanding intent" is the process of analyzing a user's question and identifying the underlying requirements and objectives.

[0278] "Referring to past cases" is a method of using data accumulated in the past to derive useful information and answers to current questions.

[0279] "User terminal" refers to a computing device used by a user to access the system, and includes PCs, smartphones, and other devices.

[0280] In this invention, the server first collects internal electronic communication history. This history is collected using the company's email system and chat platform. This collection process utilizes common APIs and data retrieval scripts. The server stores the collected data in a relational database management system (RDBMS), specifically MySQL or PostgreSQL. The database functions as a foundation for efficiently storing large amounts of historical information and for later analysis and utilization.

[0281] Next, the server preprocesses the saved electronic communication history. For preprocessing, NLTK and spaCy, which are NLP libraries in the Python language, are utilized. Using these libraries, the server normalizes the text data and removes noise and personal information. Through this process, the data becomes clean and in a form suitable for analysis. For example, while removing special characters and extra tags, personal names and email addresses are anonymized.

[0282] After the data preparation is complete, the server uses TensorFlow or PyTorch as deep learning frameworks to train specialized generation models for each employee. Thereby, generation models reflecting the business knowledge and past response experiences of each employee are constructed. These generation models are for imitating the specialized knowledge possessed by different employees and generating answers specialized for specific businesses.

[0283] When a user inputs a business-related question into the system via a terminal, the server analyzes the question using natural language processing technology and selects the optimal generation model. The server uses the selected model to generate an answer related to the question while referring to past similar cases. Subsequently, the generated answer is provided to the user's terminal in real time.

[0284] As a specific example, when a user sends a prompt sentence such as "Please tell me about past cases regarding data protection policies in new projects", the server refers to the communication history related to past similar projects, selects an appropriate model, and provides an answer. In this process, reliable information including specific methods and precautions regarding the project can be quickly conveyed to the user.

[0285] The flow of the specific process in Example 1 will be described using FIG. 11.

[0286] Step 1:

[0287] The server automatically collects electronic communication history from the company's electronic communication platform. Specifically, it retrieves data using APIs from mail servers and chat tools. The input is raw communication data, and the output is raw data stored in an RDBMS. This data is then saved for further processing.

[0288] Step 2:

[0289] The server preprocesses the collected electronic communication history. The input is the raw data saved in step 1. Specifically, it normalizes the data using Python's NLP library, removes noise and HTML tags, and anonymizes personal information. The output is clean, analyzable text data.

[0290] Step 3:

[0291] The server trains a generative AI model for each employee based on pre-processed data. The input is the clean data generated in step 2. In this process, the model is trained using TensorFlow or PyTorch, and a custom model is built that reflects the employee's business knowledge. The output is a generative AI model tailored to each employee.

[0292] Step 4:

[0293] Users input business-related questions into the system using a terminal. Specifically, they send prompt messages using a web form. The input is a question written in natural language, and the output is the query data sent to the server.

[0294] Step 5:

[0295] The server analyzes the question received from the user and understands its intent. The input is the question received in step 4. Using a natural language processing algorithm, it identifies the intent and selects the optimal generative model. The output is the selected generative AI model.

[0296] Step 6:

[0297] The server generates an answer to the question using the selected generative model. The input is the generative AI model obtained in step 5 and the content of the question. The server generates a precise answer while referring to past electronic communication history. The output is the answer data provided to the user.

[0298] Step 7:

[0299] The server-generated response is sent to the user's terminal. The input is the response data from the server, and the output is the answer information displayed on the terminal. This allows the user to quickly use it as a reference for their work.

[0300] (Application Example 1)

[0301] Next, we will explain Application Example 1. In the following explanation, the data processing device 12 will be referred to as the "server," and the smart glasses 214 will be referred to as the "terminal."

[0302] In modern cities, there is a need for methods to respond quickly and accurately to inquiries and problems from citizens. In particular, the quality of administrative services can be improved by effectively utilizing past cases and examples of responses. However, existing systems have the challenge of not being able to fully utilize past data, resulting in time-consuming individual responses.

[0303] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0304] In this invention, the server includes means for acquiring electronic communication records, means for preprocessing the acquired electronic communication records, and means for learning individual generative models based on the preprocessed electronic communication records. This enables administrative agencies to provide automated responses to inquiries from citizens, referencing past cases.

[0305] "Electronic communication records" refer to the logs of emails and messages conducted within a company or organization, and are a collection of information including the communication history in business operations.

[0306] "Preprocessing" refers to the process of removing noise from the collected data and ensuring data consistency, and includes operations such as anonymization and normalization of confidential information.

[0307] "Generative model" refers to an algorithm model that uses machine learning to learn from a specific dataset and automatically executes tasks such as text generation.

[0308] "Query" refers to the questions or requests that users or citizens input into the system to obtain information, and is something that requests the provision of information regarding a specific event.

[0309] "Analysis result" refers to the information obtained when analyzing the input data, and is used as a reference for the selection of the generative model and the generation of answers.

[0310] "Case data" refers to the information recording cases that occurred in the past and the corresponding methods, and is data that supports approaches to new problems.

[0311] "Administrative agency" refers to the organizations of the country or local government that provide public services and support the life of the region and citizens.

[0312] To implement this invention, first, the server needs to collect electronic communication records. The server obtains data from existing communication logs and preprocesses it. In preprocessing, natural language processing tools such as NLTK and spaCy are used to remove noise from the collected data and perform normalization and anonymization processes to maintain data consistency. Next, based on the preprocessed data, a generative model is learned using libraries such as Python and Hugging Face Transformers. The generative model has the ability to generate specific answers based on past data.

[0313] When a user submits a query to the system via a web interface or similar means, the query is first forwarded to the server. The server analyzes the query and selects the optimal generative model based on the analysis results. In this process, past case data is referenced, and the response is supplemented based on similar cases.

[0314] Users can quickly receive automatically generated answers through this process, with the aim of improving public services. For example, if a user asks, "Please tell me about the new recycling program," the server can use past inquiry history to provide details about the program and information about upcoming related events. In this way, the server uses a generation AI model to optimize answers in response to various inquiries.

[0315] Examples of specific prompt messages include the following:

[0316] "Please provide information about the new recycling program."

[0317] "Could you please provide details about similar environmental events that have been held in the past?"

[0318] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0319] Step 1:

[0320] The server retrieves electronic communication records within an organization. Input is data from the organization's mail server or messaging platform. It then stores the communication logs in a database. Output is the set of stored electronic communication records.

[0321] Step 2:

[0322] The server preprocesses the acquired electronic communication records. The input is the electronic communication records saved in step 1. The processing involves denoising, normalization, and anonymization using NLTK or spaCy. This generates analyzable text data while protecting privacy. The output is the preprocessed, clean text data.

[0323] Step 3:

[0324] The server trains a generative model based on preprocessed data. The input is the text data obtained in step 2. A specialized generative model is built using the Hugging Face Transformers library. The output is the trained generative model.

[0325] Step 4:

[0326] The user sends a query to the system from their terminal. The input is the query text entered by the user. The output is the query text received by the server.

[0327] Step 5:

[0328] The server analyzes the query received from the user. The input is the query text received by the server in step 4. The server uses a natural language processing engine to analyze the text and understand its intent. The output is the analysis result.

[0329] Step 6:

[0330] The server selects an appropriate generative model based on the analysis results. The input is the analysis results obtained in step 5. The selection algorithm determines the most relevant generative model. The output is the selected generative model.

[0331] Step 7:

[0332] The server generates answers using the selected generative model. The input consists of the selected generative model and its analysis results. Past case data is also referenced to reinforce the answers. The output is the generated answer text.

[0333] Step 8:

[0334] The user's device receives the response provided by the server. The input is the response text sent from the server. Based on this information, the user can quickly obtain the desired information. The output is the response displayed to the user.

[0335] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[0336] This invention provides an interactive response system that takes user emotions into account by combining an emotion engine. The system uses electronic communication history to construct individual generative models and reflects user emotions in the question analysis and response generation processes.

[0337] System Overview

[0338] The server first collects and stores past electronic communication history within the company into a database. The stored data is preprocessed to remove personally identifiable information and reduce noise. This ensures that a safe and effective dataset is used to train the generative model.

[0339] Based on this pre-processed data, the server trains a generative model tailored to each employee. This results in a highly accurate model that mimics the experience and expertise of each individual employee. The trained model becomes the key to generating appropriate answers in response to user questions.

[0340] When a user enters and submits a question from their device, the server analyzes that question. This analysis utilizes an emotion engine to extract emotions from the user's text. This emotion information is used to understand the intent behind the question and is reflected in the analysis results.

[0341] As a result, the server selects the optimal generative model and generates responses that are adjusted to the user's emotions. The generated responses are crafted in a tone that takes the user's emotions into consideration, thereby improving user satisfaction and convenience. For example, if signs of stress are detected, the response will be provided in a more considerate tone.

[0342] Ultimately, the server sends this generated response to the user's terminal, providing the user with the information they need in an appropriate format. This system not only leverages the expertise of individual employees but also enables responses that take user emotions into consideration, thereby improving operational efficiency and the quality of information exchange.

[0343] Specific example

[0344] For example, when a user asks a question about project delays, the sentiment engine may detect anxiety from the text. In this case, the server uses a generative model of an employee with past project management expertise to generate and provide a response that includes specific measures to alleviate anxiety and encouraging messages. This allows the user to move on to the next step with confidence.

[0345] The following describes the processing flow.

[0346] Step 1:

[0347] The server collects electronic communication history from within the company. It primarily retrieves data from internal chat tools and email logs and stores it in a database. The stored information includes metadata such as the speaker, date and time, and content.

[0348] Step 2:

[0349] The server preprocesses the collected electronic communication history. This processing includes text normalization and removal of unnecessary data, and in particular, filtering to remove personally identifiable information.

[0350] Step 3:

[0351] The server uses pre-processed data to train a generative model for each employee. Using natural language processing algorithms, it builds a model that reflects each employee's job knowledge and skills. This model formation allows the server to mimic each employee's areas of expertise and response patterns.

[0352] Step 4:

[0353] Users enter their questions using their terminals and send them to the server. The question input screen allows for questions in text format, and users can freely submit inquiries.

[0354] Step 5:

[0355] The server receives the user's question and uses an emotion engine to analyze the emotions contained in the question. This extracts the emotional context from the text in the question and helps understand the user's psychological state.

[0356] Step 6:

[0357] The server selects the optimal generative model based on the question content and sentiment analysis results. The selection criteria prioritize models created by employees with relevant knowledge to the question. It also obtains feedback to adjust the tone of the response according to the sentiment.

[0358] Step 7:

[0359] The server generates responses using the selected generative model. At this time, based on the sentiment analysis results, the responses are generated in a tone that matches the user's emotions. Relevant information from past data is referenced, and necessary information is added and the tone adjusted.

[0360] Step 8:

[0361] The server sends the generated response to the terminal, delivering it to the user. The user can then review the response on the terminal and use it to solve their work-related problems. This allows users to solve problems efficiently.

[0362] (Example 2)

[0363] Next, we will describe Example 2. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal".

[0364] Traditional interactive systems often provide only uniform answers to user questions, failing to adequately consider user emotions and intentions, resulting in low user satisfaction. Furthermore, obtaining accurate and appropriate answers to questions requiring specialized knowledge was difficult.

[0365] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[0366] In this invention, the server includes means for acquiring electronic communication data, means for preprocessing the acquired data and removing personal information, means for learning individual knowledge models based on the preprocessed data, means for analyzing questions and extracting emotional information, means for selecting an appropriate knowledge model based on the analysis results, means for generating answers in an emotionally appropriate tone using the selected knowledge model, and means for providing the generated answers to the user. This makes it possible to provide high-quality answers that take into account the user's emotions while leveraging expertise.

[0367] "Electronic communication data" refers to information such as emails and messages exchanged via the internet and other communication networks.

[0368] "Preprocessing" refers to the process of removing personally identifiable information from electronic communication data and reducing noise to improve data quality.

[0369] A "knowledge model" refers to a generative AI model trained using a specific dataset, and is an artificial intelligence algorithm that possesses a deep understanding of a particular subject.

[0370] "Emotional information" refers to data that indicates the type and degree of emotion extracted from users' questions and statements.

[0371] "Tone" refers to the wording and facial expressions used in the generated response, and by adjusting them according to the user's emotions, it is an element that influences the nuance and how the response is received.

[0372] "Analysis" refers to the process of using natural language processing techniques to analyze the information contained in a question in detail and understand its intent and content.

[0373] In this embodiment of the invention, the interactive response system employs multiple technical means to provide responses that take into account the user's emotions.

[0374] The server collects electronic communication data via the internet and other communication networks and stores it in a database. This collected data is first pre-processed. This pre-processing uses specialized software, such as Python scripts, to remove personally identifiable information and reduce noise, thereby improving data quality. This process prepares the data for training.

[0375] Next, the server uses the pre-processed data to train a knowledge model specific to each employee, creating an AI model. This training process utilizes machine learning frameworks such as TensorFlow and PyTorch. The server builds a highly accurate model that reflects the employee's work knowledge and experience, which then forms the basis for subsequent question answering.

[0376] When a user enters a question using their device, the text data is sent to the server and analyzed through natural language processing. The server then activates an emotion engine and extracts emotional information from the text using an emotion analysis API. This emotional information is crucial for understanding the intent of the question and selecting the optimal knowledge model.

[0377] The generated responses are adjusted in tone to reflect the user's emotions. For example, if the user is feeling anxious, the response will be adjusted to include elements of warmth and encouragement. This improves both user satisfaction and work efficiency.

[0378] As a concrete example, suppose a user enters the prompt, "I would like specific advice regarding project delays." The server processes this input, selects a knowledge model with data on similar past cases, and generates a reassuring response. This response includes specific measures and words of encouragement, allowing the user to confidently move on to the next step.

[0379] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0380] Step 1:

[0381] The server acquires electronic communication data via the internet or other communication networks. The input includes raw data such as emails and messages. This data is stored in a database. At this stage, the acquired electronic communication data is output as is.

[0382] Step 2:

[0383] The server preprocesses the stored electronic communication data. The input is the raw data obtained in step 1. In this process, specialized software is used to remove personal information and apply algorithms to reduce noise. This generates safe and clean data, which becomes the output for the next step.

[0384] Step 3:

[0385] The server trains a generative AI model tailored to each employee based on pre-processed data. The input is the clean data obtained in step 2. Machine learning frameworks such as TensorFlow and PyTorch are used to train this data and build a knowledge model. The output at this stage is a generative AI model tailored to each employee.

[0386] Step 4:

[0387] The user enters a question via a terminal and sends it to the server. The input is the user's question, which is free-form text. At this point, the entered prompt text is passed directly to the next step.

[0388] Step 5:

[0389] When the server receives a question from a user, it analyzes it using natural language processing tools. The input is the prompt sentence from step 4, and word segmentation and part-of-speech tagging are performed, followed by analysis to understand the intent. The results of this analysis are output, and sentiment extraction is then performed.

[0390] Step 6:

[0391] The server uses an emotion engine to extract emotional information from the question text. The input is the analysis result from step 5. The type and degree of emotion are identified using the emotion analysis API, which is then used in the next step. At this point, this emotional information is output.

[0392] Step 7:

[0393] The server selects the optimal generative AI model based on the analysis results and extracted sentiment information. The inputs are the analysis results from step 5 and the sentiment information from step 6. Considering these, a suitable knowledge model is selected, and the answer is generated by that knowledge model.

[0394] Step 8:

[0395] The server uses the selected generative AI model to generate responses in a tone that matches the user's emotions. The input is the knowledge model selected in step 7, and the output is the response with the adjusted tone. Based on the analyzed emotional information, a response is created that reflects the optimal tone and content for the user.

[0396] Step 9:

[0397] The server sends the final generated response to the user's device. The input is the response generated in step 8. The user receives this information on their device and can obtain the necessary knowledge and advice. The output is the adjusted-tone response displayed on the user's device.

[0398] (Application Example 2)

[0399] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as the "server," and the smart glasses 214 will be referred to as the "terminal."

[0400] In caregiving settings, accurate communication tailored to the emotions and circumstances of those receiving care is essential, but achieving this is not easy. In particular, accurately understanding a person's emotions and responding accordingly is a challenging task. Therefore, there is a need to develop systems that enable care staff and family members to respond appropriately based on the emotions of those receiving care.

[0401] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[0402] In this invention, the server includes means for acquiring electronic communication history, means for pre-processing the acquired electronic communication history, means for learning individual generative models based on the pre-processed electronic communication history, means for analyzing questions, means for analyzing the user's emotions, means for selecting an appropriate generative model based on the analysis results and emotions, means for generating an emotionally appropriate response using the selected generative model, and means for providing the generated response. This enables responses based on the emotions of the person receiving care, thereby improving the quality of care.

[0403] "Electronic communication history" refers to a record of messages and communication content transmitted using communication devices.

[0404] "Preprocessing" is the process of removing noise and unnecessary information from acquired data and converting it into a format suitable for analysis and model training.

[0405] A "generative model" is an algorithm that has been trained to produce an appropriate output for a given input.

[0406] "Methods for analyzing questions" refer to technologies for understanding the content of questions entered by users and identifying their intent.

[0407] "Methods for analyzing emotions" refer to technologies that extract emotions from users' statements and texts and understand their emotional state.

[0408] A "generated answer" is a response to a user's question, created by a generative model.

[0409] "Means of provision" refers to the methods and technologies used to present the generated answers to the user.

[0410] The system of this invention primarily consists of a server and a user terminal. The server first acquires electronic communication history and preprocesses it. Preprocessing includes removing personally identifiable information and reducing noise from the collected digital data. This results in a clean dataset suitable for analysis and training generative models. The hardware used is a cloud server, and the software used for data analysis includes Python and R.

[0411] Based on pre-processed data, the server trains individual generative AI models. This involves creating user-specific models and utilizing machine learning algorithms to mimic their expertise and experience. Deep learning frameworks such as TensorFlow and PyTorch are used in this process.

[0412] When a user enters a question from their device, the device sends that data to the server. The server analyzes the received question and activates an emotion analysis engine to extract the user's emotions. This operation can be performed using natural language processing techniques, and libraries such as NLTK and spaCy are used for this purpose.

[0413] Based on the sentiment analysis results, the server selects the most appropriate generative model and generates a response that matches the user's emotions. This response generation process adjusts the tone and content of the response to meet the user's needs. For example, if stress or anxiety is detected, the server provides more reassuring feedback.

[0414] The generated responses are sent from the server to the user's terminal, where the user receives them either visually or audibly. This system enables personalized responses that take the user's emotions into consideration, making it applicable in fields such as elderly care.

[0415] For example, if an elderly person makes a statement to the terminal such as "I'm not feeling very well today," the server can analyze it and respond with encouraging words or suggestions for simple exercises to help them relax. An example of a prompt to input into the generative AI model could be, "Analyze the user's recent emotional state and suggest the most appropriate care method."

[0416] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0417] Step 1:

[0418] The server retrieves electronic communication history data. The input consists of messages and communication data previously sent by the user. This data is retrieved from the database and prepared for the next preprocessing step.

[0419] Step 2:

[0420] The server preprocesses the acquired data. It removes personally identifiable information from the input data and reduces unnecessary noise. This results in a clean dataset suitable for training. Specifically, it performs text normalization and filtering.

[0421] Step 3:

[0422] The server trains a generative AI model based on preprocessed data. The input is a clean dataset, and the output is an individual generative model. The model is trained using a machine learning framework such as TensorFlow.

[0423] Step 4:

[0424] The user enters a question from their terminal and sends it to the server. The input is the user's question text, which the terminal forwards to the server.

[0425] Step 5:

[0426] The server parses the received question. The input is the question text. Natural language processing techniques are used to understand the meaning of the text and identify the necessary context. The output is the parsed intent information.

[0427] Step 6:

[0428] The server analyzes the user's emotions using an emotion analysis engine. The input is the question text, and emotions are extracted using an emotion analysis API, etc. The output is emotion information (joy, anxiety, sadness, etc.).

[0429] Step 7:

[0430] The server selects an appropriate generative model based on the question and sentiment information. It takes the analyzed intent and sentiment information as input and determines the most suitable generative model. The output is the selected generative model.

[0431] Step 8:

[0432] The server generates emotion-sensitive responses using the selected generative model. Here, the tone and content of the responses are adjusted to match the user's emotions based on the input analysis data. The output is the generated response.

[0433] Step 9:

[0434] The server sends the generated response to the user's terminal. The terminal then provides the user with the received response in text or audio format. The output is the response to the user.

[0435] 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 user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[0436] Data generation model 58 is a type of so-called generative AI (Artificial Intelligence). One example of data generation model 58 is ChatGPT (Internet search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0437] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and the specific processing may also be performed by the smart glasses 214.

[0438] [Third Embodiment]

[0439] Figure 5 shows an example of the configuration of the data processing system 310 according to the third embodiment.

[0440] As shown in Figure 5, the data processing system 310 includes a data processing device 12 and a headset terminal 314. An example of the data processing device 12 is a server.

[0441] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0442] The headset terminal 314 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication interface 44, and a display 343. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, camera 42, and display 343 are also connected to the bus 52.

[0443] The microphone 238 receives voice signals from the user 20 and receives instructions from the user 20. The microphone 238 captures the voice signals from the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[0444] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the area around the user 20 (for example, an imaging range defined by a field of view equivalent to the width of a typical healthy person's field of vision).

[0445] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[0446] Figure 6 shows an example of the main functions of the data processing device 12 and the headset terminal 314. As shown in Figure 6, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[0447] The specific processing program 56 is an example of a "program" relating to the technology of this 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.

[0448] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0449] In the headset terminal 314, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. 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 processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[0450] Next, the specific processing performed by the specific processing unit 290 of the data processing device 12 will be described. In the following description, the data processing device 12 will be referred to as the "server" and the headset terminal 314 will be referred to as the "terminal".

[0451] This invention is a system developed to efficiently utilize employees' business knowledge and know-how to improve operational efficiency. The system aims to create a generative model for each employee using the company's past electronic communication history and to automatically answer questions related to their work.

[0452] System Overview

[0453] The server first collects internal electronic communication history and appropriately stores this information in a database. The stored data is then preprocessed and used as training material for generative models. Preprocessing includes normalization of text data, noise reduction, and protection of personal information.

[0454] The server uses pre-processed data to train specialized generative models for each employee. This builds models that reflect each employee's knowledge and skills. Through these models, the system can automatically generate responses by mimicking past experiences and answers provided by the employees.

[0455] When a user inputs a question into the system via a terminal, the server analyzes the question and understands its intent. Based on this analysis, the server selects the optimal generative model. The selected model generates an answer to the analyzed question, referencing similar past cases and related information in the process.

[0456] The generated answers are sent from the server to the user's terminal. Users can then proceed with their work based on the reliable answers provided. This system enables effective knowledge sharing and consistent application of work knowledge, particularly across different departments and with new employees.

[0457] Specific example

[0458] For example, if a user asks for a solution to a technical problem that has arisen in a project, the server searches past electronic communication history for similar problems and generates an answer using an appropriate generative model based on that history. This allows users to solve problems quickly and accurately based on relevant information. Furthermore, since information previously provided by former employees can also be utilized, there is the advantage of ensuring that organizational knowledge is continuously available.

[0459] The following describes the processing flow.

[0460] Step 1:

[0461] The server collects internal electronic communication history. This data is obtained from chat applications, emails, and other sources and stored in a database. During collection, metadata such as the speaker, date and time, and content are also recorded.

[0462] Step 2:

[0463] The server preprocesses the collected communication history. This process includes text normalization, removal of irrelevant information, and filtering of personal information. The preprocessed data is then organized into a format suitable for training generative models.

[0464] Step 3:

[0465] The server uses pre-processed data to train generative models for each employee. During this training process, natural language processing techniques are used to analyze the content and patterns of employees' past statements, thereby generating models that reflect their individual knowledge and skills.

[0466] Step 4:

[0467] The user enters a question into the system from a terminal. The terminal receives the user's input, formats the data, and sends it to the server.

[0468] Step 5:

[0469] The server analyzes the received question. This analysis uses natural language processing to understand the intent and main points of the question, extracting relevant keywords and context.

[0470] Step 6:

[0471] The server selects the optimal generative model based on the analysis results. Here, it identifies and selects the model of the employee with the most relevant knowledge to the question.

[0472] Step 7:

[0473] The server generates answers using the selected generative model. This generation process involves constructing highly accurate answers by referencing data from similar past questions and a knowledge base.

[0474] Step 8:

[0475] The server sends the generated response to the user's device. The user can then receive the response on their device and use it for their work.

[0476] (Example 1)

[0477] Next, we will describe Example 1. In the following description, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[0478] In modern companies, there is a challenge in effectively sharing the business knowledge and know-how of each employee and obtaining information necessary to carry out tasks quickly and accurately. Furthermore, there is a need to effectively utilize data while paying attention to the protection of personal information.

[0479] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0480] In this invention, the server includes means for collecting and storing electronic communication history, means for normalizing the collected electronic communication history and removing noise and personal information, and means for learning individual generative models based on the pre-processed history. This enables the automatic generation of answers by utilizing the knowledge accumulated within the company and based on the specialized knowledge of each employee.

[0481] "Electronic communication history" refers to the record of communications such as emails and chat messages conducted within an organization.

[0482] "Normalization" is the process of converting text data into a unified and consistent format, which improves data quality and processing efficiency.

[0483] "Noise reduction" is a technique that removes unnecessary information and errors from text data, and is performed to improve the usefulness of the data.

[0484] A "personalized generative model" is a customized AI model tailored to a specific employee or situation, possessing the ability to generate answers by learning from past knowledge and experience.

[0485] A "user question" is a request from a user to provide information or solve a problem, and is usually written in natural language.

[0486] "Understanding intent" is the process of analyzing a user's question and identifying the underlying requirements and objectives.

[0487] "Referring to past cases" is a method of using data accumulated in the past to derive useful information and answers to current questions.

[0488] "User terminal" refers to a computing device used by a user to access the system, and includes PCs, smartphones, and other devices.

[0489] In this invention, the server first collects internal electronic communication history. This history is collected using the company's email system and chat platform. This collection process utilizes common APIs and data retrieval scripts. The server stores the collected data in a relational database management system (RDBMS), specifically MySQL or PostgreSQL. The database functions as a foundation for efficiently storing large amounts of historical information and for later analysis and utilization.

[0490] Next, the server preprocesses the stored electronic communication history. This preprocessing utilizes Python NLP libraries such as NLTK and spaCy. Using these libraries, the server normalizes the text data, removing noise and personal information. This process ensures the data is clean and suitable for analysis. For example, it anonymizes names and email addresses while removing special characters and extraneous tags.

[0491] Once the data is ready, the server uses TensorFlow or PyTorch as a deep learning framework to train specialized generative models for each employee. This builds generative models that reflect each employee's job knowledge and past experience. These generative models are designed to mimic the specialized knowledge of different employees and generate answers specific to their particular tasks.

[0492] When a user inputs a business-related question into the system via their terminal, the server analyzes the question using natural language processing technology and selects the optimal generative model. Using the selected model, the server generates an answer related to the question, referencing similar past cases. The generated answer is then provided to the user's terminal in real time.

[0493] For example, if a user sends a prompt such as, "Please tell me about past examples of data protection policies for a new project," the server will refer to the communication history of similar past projects, select an appropriate model, and provide a response. In this process, reliable information, including specific methods and points to note regarding the project, can be quickly conveyed to the user.

[0494] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0495] Step 1:

[0496] The server automatically collects electronic communication history from the company's electronic communication platform. Specifically, it retrieves data using APIs from mail servers and chat tools. The input is raw communication data, and the output is raw data stored in an RDBMS. This data is then saved for further processing.

[0497] Step 2:

[0498] The server preprocesses the collected electronic communication history. The input is the raw data saved in step 1. Specifically, it normalizes the data using Python's NLP library, removes noise and HTML tags, and anonymizes personal information. The output is clean, analyzable text data.

[0499] Step 3:

[0500] The server trains a generative AI model for each employee based on pre-processed data. The input is the clean data generated in step 2. In this process, the model is trained using TensorFlow or PyTorch, and a custom model is built that reflects the employee's business knowledge. The output is a generative AI model tailored to each employee.

[0501] Step 4:

[0502] Users input business-related questions into the system using a terminal. Specifically, they send prompt messages using a web form. The input is a question written in natural language, and the output is the query data sent to the server.

[0503] Step 5:

[0504] The server analyzes the question received from the user and understands its intent. The input is the question received in step 4. Using a natural language processing algorithm, it identifies the intent and selects the optimal generative model. The output is the selected generative AI model.

[0505] Step 6:

[0506] The server generates an answer to the question using the selected generative model. The input is the generative AI model obtained in step 5 and the content of the question. The server generates a precise answer while referring to past electronic communication history. The output is the answer data provided to the user.

[0507] Step 7:

[0508] The server-generated response is sent to the user's terminal. The input is the response data from the server, and the output is the answer information displayed on the terminal. This allows the user to quickly use it as a reference for their work.

[0509] (Application Example 1)

[0510] Next, we will explain Application Example 1. In the following explanation, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[0511] In modern cities, there is a need for methods to respond quickly and accurately to inquiries and problems from citizens. In particular, the quality of administrative services can be improved by effectively utilizing past cases and examples of responses. However, existing systems have the challenge of not being able to fully utilize past data, resulting in time-consuming individual responses.

[0512] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0513] In this invention, the server includes means for acquiring electronic communication records, means for preprocessing the acquired electronic communication records, and means for learning individual generative models based on the preprocessed electronic communication records. This enables administrative agencies to provide automated responses to inquiries from citizens, referencing past cases.

[0514] "Electronic communication records" refer to logs of emails and messages exchanged within a company or organization, and are a collection of information that includes a history of business-related communication.

[0515] "Preprocessing" refers to the process of removing noise from collected data and ensuring data integrity, and includes operations such as anonymization and normalization of confidential information.

[0516] A "generative model" is an algorithmic model that uses machine learning to learn from a specific dataset and automatically perform tasks such as text generation.

[0517] An "inquiry" is a question or request that a user or citizen enters into a system to obtain information, and it seeks to provide information about a specific event.

[0518] "Analysis results" refer to the information obtained when analyzing the input data, and are used as a reference for selecting a generative model and generating responses.

[0519] "Case data" refers to information that records past cases and how they were handled, and it is data that supports approaches to new problems.

[0520] "Administrative agencies" are organizations of the national or local government that provide public services and support the lives of communities and citizens.

[0521] To implement this invention, a server must first collect electronic communication records. The server obtains data from existing communication logs and preprocesses it. In the preprocessing, natural language processing tools such as NLTK and spaCy are used to remove noise from the collected data and to perform normalization and anonymization to maintain data integrity. Next, a generative model is trained using Python and the Hugging Face Transformers library based on the preprocessed data. The generative model has the ability to generate specific answers based on past data.

[0522] When a user submits a query to the system via a web interface or similar means, the query is first forwarded to the server. The server analyzes the query and selects the optimal generative model based on the analysis results. In this process, past case data is referenced, and the response is supplemented based on similar cases.

[0523] Users can quickly receive automatically generated answers through this process, with the aim of improving public services. For example, if a user asks, "Please tell me about the new recycling program," the server can use past inquiry history to provide details about the program and information about upcoming related events. In this way, the server uses a generation AI model to optimize answers in response to various inquiries.

[0524] Examples of specific prompt messages include the following:

[0525] "Please provide information about the new recycling program."

[0526] "Could you please provide details about similar environmental events that have been held in the past?"

[0527] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0528] Step 1:

[0529] The server retrieves electronic communication records within an organization. Input is data from the organization's mail server or messaging platform. It then stores the communication logs in a database. Output is the set of stored electronic communication records.

[0530] Step 2:

[0531] The server preprocesses the acquired electronic communication records. The input is the electronic communication records saved in step 1. The processing involves denoising, normalization, and anonymization using NLTK or spaCy. This generates analyzable text data while protecting privacy. The output is the preprocessed, clean text data.

[0532] Step 3:

[0533] The server trains a generative model based on preprocessed data. The input is the text data obtained in step 2. A specialized generative model is built using the Hugging Face Transformers library. The output is the trained generative model.

[0534] Step 4:

[0535] The user sends a query to the system from their terminal. The input is the query text entered by the user. The output is the query text received by the server.

[0536] Step 5:

[0537] The server analyzes the query received from the user. The input is the query text received by the server in step 4. The server uses a natural language processing engine to analyze the text and understand its intent. The output is the analysis result.

[0538] Step 6:

[0539] The server selects an appropriate generative model based on the analysis results. The input is the analysis results obtained in step 5. The selection algorithm determines the most relevant generative model. The output is the selected generative model.

[0540] Step 7:

[0541] The server generates answers using the selected generative model. The input consists of the selected generative model and its analysis results. Past case data is also referenced to reinforce the answers. The output is the generated answer text.

[0542] Step 8:

[0543] The user's device receives the response provided by the server. The input is the response text sent from the server. Based on this information, the user can quickly obtain the desired information. The output is the response displayed to the user.

[0544] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[0545] This invention provides an interactive response system that takes user emotions into account by combining an emotion engine. The system uses electronic communication history to construct individual generative models and reflects user emotions in the question analysis and response generation processes.

[0546] System Overview

[0547] The server first collects and stores past electronic communication history within the company into a database. The stored data is preprocessed to remove personally identifiable information and reduce noise. This ensures that a safe and effective dataset is used to train the generative model.

[0548] Based on this pre-processed data, the server trains a generative model tailored to each employee. This results in a highly accurate model that mimics the experience and expertise of each individual employee. The trained model becomes the key to generating appropriate answers in response to user questions.

[0549] When a user enters and submits a question from their device, the server analyzes that question. This analysis utilizes an emotion engine to extract emotions from the user's text. This emotion information is used to understand the intent behind the question and is reflected in the analysis results.

[0550] As a result, the server selects the optimal generative model and generates responses that are adjusted to the user's emotions. The generated responses are crafted in a tone that takes the user's emotions into consideration, thereby improving user satisfaction and convenience. For example, if signs of stress are detected, the response will be provided in a more considerate tone.

[0551] Ultimately, the server sends this generated response to the user's terminal, providing the user with the information they need in an appropriate format. This system not only leverages the expertise of individual employees but also enables responses that take user emotions into consideration, thereby improving operational efficiency and the quality of information exchange.

[0552] Specific example

[0553] For example, when a user asks a question about project delays, the sentiment engine may detect anxiety from the text. In this case, the server uses a generative model of an employee with past project management expertise to generate and provide a response that includes specific measures to alleviate anxiety and encouraging messages. This allows the user to move on to the next step with confidence.

[0554] The following describes the processing flow.

[0555] Step 1:

[0556] The server collects electronic communication history from within the company. It primarily retrieves data from internal chat tools and email logs and stores it in a database. The stored information includes metadata such as the speaker, date and time, and content.

[0557] Step 2:

[0558] The server preprocesses the collected electronic communication history. This processing includes text normalization and removal of unnecessary data, and in particular, filtering to remove personally identifiable information.

[0559] Step 3:

[0560] The server uses pre-processed data to train a generative model for each employee. Using natural language processing algorithms, it builds a model that reflects each employee's job knowledge and skills. This model formation allows the server to mimic each employee's areas of expertise and response patterns.

[0561] Step 4:

[0562] Users enter their questions using their terminals and send them to the server. The question input screen allows for questions in text format, and users can freely submit inquiries.

[0563] Step 5:

[0564] The server receives the user's question and uses an emotion engine to analyze the emotions contained in the question. This extracts the emotional context from the text in the question and helps understand the user's psychological state.

[0565] Step 6:

[0566] The server selects the optimal generative model based on the question content and sentiment analysis results. The selection criteria prioritize models created by employees with relevant knowledge to the question. It also obtains feedback to adjust the tone of the response according to the sentiment.

[0567] Step 7:

[0568] The server generates responses using the selected generative model. At this time, based on the sentiment analysis results, the responses are generated in a tone that matches the user's emotions. Relevant information from past data is referenced, and necessary information is added and the tone adjusted.

[0569] Step 8:

[0570] The server sends the generated response to the terminal, delivering it to the user. The user can then review the response on the terminal and use it to solve their work-related problems. This allows users to solve problems efficiently.

[0571] (Example 2)

[0572] Next, we will describe Example 2. In the following description, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[0573] Traditional interactive systems often provide only uniform answers to user questions, failing to adequately consider user emotions and intentions, resulting in low user satisfaction. Furthermore, obtaining accurate and appropriate answers to questions requiring specialized knowledge was difficult.

[0574] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[0575] In this invention, the server includes means for acquiring electronic communication data, means for preprocessing the acquired data and removing personal information, means for learning individual knowledge models based on the preprocessed data, means for analyzing questions and extracting emotional information, means for selecting an appropriate knowledge model based on the analysis results, means for generating answers in an emotionally appropriate tone using the selected knowledge model, and means for providing the generated answers to the user. This makes it possible to provide high-quality answers that take into account the user's emotions while leveraging expertise.

[0576] "Electronic communication data" refers to information such as emails and messages exchanged via the internet and other communication networks.

[0577] "Preprocessing" refers to the process of removing personally identifiable information from electronic communication data and reducing noise to improve data quality.

[0578] A "knowledge model" refers to a generative AI model trained using a specific dataset, and is an artificial intelligence algorithm that possesses a deep understanding of a particular subject.

[0579] "Emotional information" refers to data that indicates the type and degree of emotion extracted from users' questions and statements.

[0580] "Tone" refers to the wording and facial expressions used in the generated response, and by adjusting them according to the user's emotions, it is an element that influences the nuance and how the response is received.

[0581] "Analysis" refers to the process of using natural language processing techniques to analyze the information contained in a question in detail and understand its intent and content.

[0582] In this embodiment of the invention, the interactive response system employs multiple technical means to provide responses that take into account the user's emotions.

[0583] The server collects electronic communication data via the internet and other communication networks and stores it in a database. This collected data is first pre-processed. This pre-processing uses specialized software, such as Python scripts, to remove personally identifiable information and reduce noise, thereby improving data quality. This process prepares the data for training.

[0584] Next, the server uses the pre-processed data to train a knowledge model specific to each employee, creating an AI model. This training process utilizes machine learning frameworks such as TensorFlow and PyTorch. The server builds a highly accurate model that reflects the employee's work knowledge and experience, which then forms the basis for subsequent question answering.

[0585] When a user enters a question using their device, the text data is sent to the server and analyzed through natural language processing. The server then activates an emotion engine and extracts emotional information from the text using an emotion analysis API. This emotional information is crucial for understanding the intent of the question and selecting the optimal knowledge model.

[0586] The generated responses are adjusted in tone to reflect the user's emotions. For example, if the user is feeling anxious, the response will be adjusted to include elements of warmth and encouragement. This improves both user satisfaction and work efficiency.

[0587] As a concrete example, suppose a user enters the prompt, "I would like specific advice regarding project delays." The server processes this input, selects a knowledge model with data on similar past cases, and generates a reassuring response. This response includes specific measures and words of encouragement, allowing the user to confidently move on to the next step.

[0588] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0589] Step 1:

[0590] The server acquires electronic communication data via the internet or other communication networks. The input includes raw data such as emails and messages. This data is stored in a database. At this stage, the acquired electronic communication data is output as is.

[0591] Step 2:

[0592] The server preprocesses the stored electronic communication data. The input is the raw data obtained in step 1. In this process, specialized software is used to remove personal information and apply algorithms to reduce noise. This generates safe and clean data, which becomes the output for the next step.

[0593] Step 3:

[0594] The server trains a generative AI model tailored to each employee based on pre-processed data. The input is the clean data obtained in step 2. Machine learning frameworks such as TensorFlow and PyTorch are used to train this data and build a knowledge model. The output at this stage is a generative AI model tailored to each employee.

[0595] Step 4:

[0596] The user enters a question via a terminal and sends it to the server. The input is the user's question, which is free-form text. At this point, the entered prompt text is passed directly to the next step.

[0597] Step 5:

[0598] When the server receives a question from a user, it analyzes it using natural language processing tools. The input is the prompt sentence from step 4, and word segmentation and part-of-speech tagging are performed, followed by analysis to understand the intent. The results of this analysis are output, and sentiment extraction is then performed.

[0599] Step 6:

[0600] The server uses an emotion engine to extract emotional information from the question text. The input is the analysis result from step 5. The type and degree of emotion are identified using the emotion analysis API, which is then used in the next step. At this point, this emotional information is output.

[0601] Step 7:

[0602] The server selects the optimal generative AI model based on the analysis results and extracted sentiment information. The inputs are the analysis results from step 5 and the sentiment information from step 6. Considering these, a suitable knowledge model is selected, and the answer is generated by that knowledge model.

[0603] Step 8:

[0604] The server uses the selected generative AI model to generate responses in a tone that matches the user's emotions. The input is the knowledge model selected in step 7, and the output is the response with the adjusted tone. Based on the analyzed emotional information, a response is created that reflects the optimal tone and content for the user.

[0605] Step 9:

[0606] The server sends the final generated response to the user's device. The input is the response generated in step 8. The user receives this information on their device and can obtain the necessary knowledge and advice. The output is the adjusted-tone response displayed on the user's device.

[0607] (Application Example 2)

[0608] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[0609] In caregiving settings, accurate communication tailored to the emotions and circumstances of those receiving care is essential, but achieving this is not easy. In particular, accurately understanding a person's emotions and responding accordingly is a challenging task. Therefore, there is a need to develop systems that enable care staff and family members to respond appropriately based on the emotions of those receiving care.

[0610] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[0611] In this invention, the server includes means for acquiring electronic communication history, means for pre-processing the acquired electronic communication history, means for learning individual generative models based on the pre-processed electronic communication history, means for analyzing questions, means for analyzing the user's emotions, means for selecting an appropriate generative model based on the analysis results and emotions, means for generating an emotionally appropriate response using the selected generative model, and means for providing the generated response. This enables responses based on the emotions of the person receiving care, thereby improving the quality of care.

[0612] "Electronic communication history" refers to a record of messages and communication content transmitted using communication devices.

[0613] "Preprocessing" is the process of removing noise and unnecessary information from acquired data and converting it into a format suitable for analysis and model training.

[0614] A "generative model" is an algorithm that has been trained to produce an appropriate output for a given input.

[0615] "Methods for analyzing questions" refer to technologies for understanding the content of questions entered by users and identifying their intent.

[0616] "Methods for analyzing emotions" refer to technologies that extract emotions from users' statements and texts and understand their emotional state.

[0617] A "generated answer" is a response to a user's question, created by a generative model.

[0618] "Means of provision" refers to the methods and technologies used to present the generated answers to the user.

[0619] The system of this invention primarily consists of a server and a user terminal. The server first acquires electronic communication history and preprocesses it. Preprocessing includes removing personally identifiable information and reducing noise from the collected digital data. This results in a clean dataset suitable for analysis and training generative models. The hardware used is a cloud server, and the software used for data analysis includes Python and R.

[0620] Based on pre-processed data, the server trains individual generative AI models. This involves creating user-specific models and utilizing machine learning algorithms to mimic their expertise and experience. Deep learning frameworks such as TensorFlow and PyTorch are used in this process.

[0621] When a user enters a question from their device, the device sends that data to the server. The server analyzes the received question and activates an emotion analysis engine to extract the user's emotions. This operation can be performed using natural language processing techniques, and libraries such as NLTK and spaCy are used for this purpose.

[0622] Based on the sentiment analysis results, the server selects the most appropriate generative model and generates a response that matches the user's emotions. This response generation process adjusts the tone and content of the response to meet the user's needs. For example, if stress or anxiety is detected, the server provides more reassuring feedback.

[0623] The generated responses are sent from the server to the user's terminal, where the user receives them either visually or audibly. This system enables personalized responses that take the user's emotions into consideration, making it applicable in fields such as elderly care.

[0624] For example, if an elderly person makes a statement to the terminal such as "I'm not feeling very well today," the server can analyze it and respond with encouraging words or suggestions for simple exercises to help them relax. An example of a prompt to input into the generative AI model could be, "Analyze the user's recent emotional state and suggest the most appropriate care method."

[0625] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0626] Step 1:

[0627] The server retrieves electronic communication history data. The input consists of messages and communication data previously sent by the user. This data is retrieved from the database and prepared for the next preprocessing step.

[0628] Step 2:

[0629] The server preprocesses the acquired data. It removes personally identifiable information from the input data and reduces unnecessary noise. This results in a clean dataset suitable for training. Specifically, it performs text normalization and filtering.

[0630] Step 3:

[0631] The server trains a generative AI model based on preprocessed data. The input is a clean dataset, and the output is an individual generative model. The model is trained using a machine learning framework such as TensorFlow.

[0632] Step 4:

[0633] The user enters a question from their terminal and sends it to the server. The input is the user's question text, which the terminal forwards to the server.

[0634] Step 5:

[0635] The server parses the received question. The input is the question text. Natural language processing techniques are used to understand the meaning of the text and identify the necessary context. The output is the parsed intent information.

[0636] Step 6:

[0637] The server analyzes the user's emotions using an emotion analysis engine. The input is the question text, and emotions are extracted using an emotion analysis API, etc. The output is emotion information (joy, anxiety, sadness, etc.).

[0638] Step 7:

[0639] The server selects an appropriate generative model based on the question and sentiment information. It takes the analyzed intent and sentiment information as input and determines the most suitable generative model. The output is the selected generative model.

[0640] Step 8:

[0641] The server generates emotion-sensitive responses using the selected generative model. Here, the tone and content of the responses are adjusted to match the user's emotions based on the input analysis data. The output is the generated response.

[0642] Step 9:

[0643] The server sends the generated response to the user's terminal. The terminal then provides the user with the received response in text or audio format. The output is the response to the user.

[0644] The specific processing unit 290 transmits the result of the specific processing to the headset terminal 314. In the headset terminal 314, the control unit 46A causes the speaker 240 and display 343 to output the result of the specific processing. The microphone 238 acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[0645] Data generation model 58 is a type of so-called generative AI (Artificial Intelligence). One example of data generation model 58 is ChatGPT (Internet search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0646] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and specific processing may also be performed by the headset terminal 314.

[0647] [Fourth Embodiment]

[0648] Figure 7 shows an example of the configuration of the data processing system 410 according to the fourth embodiment.

[0649] As shown in Figure 7, the data processing system 410 includes a data processing device 12 and a robot 414. An example of the data processing device 12 is a server.

[0650] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0651] The robot 414 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication interface 44, and a controlled object 443. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, camera 42, and controlled object 443 are also connected to the bus 52.

[0652] The microphone 238 receives voice signals from the user 20 and receives instructions from the user 20. The microphone 238 captures the voice signals from the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[0653] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the area around the user 20 (for example, an imaging range defined by a field of view equivalent to the width of a typical healthy person's field of vision).

[0654] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[0655] The controlled object 443 includes a display device, LEDs in the eyes, and motors that drive the arms, hands, and feet. The posture and gestures of the robot 414 are controlled by controlling the motors of the arms, hands, and feet. Some of the robot 414's emotions can be expressed by controlling these motors. Furthermore, the robot 414's facial expressions can also be expressed by controlling the illumination state of the LEDs in its eyes.

[0656] Figure 8 shows an example of the main functions of the data processing device 12 and the robot 414. As shown in Figure 8, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[0657] The specific processing program 56 is an example of a "program" relating to the technology of this 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.

[0658] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0659] In robot 414, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. 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 processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[0660] Next, the specific processing performed by the specific processing unit 290 of the data processing device 12 will be described. In the following description, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[0661] This invention is a system developed to efficiently utilize employees' business knowledge and know-how to improve operational efficiency. The system aims to create a generative model for each employee using the company's past electronic communication history and to automatically answer questions related to their work.

[0662] System Overview

[0663] The server first collects internal electronic communication history and appropriately stores this information in a database. The stored data is then preprocessed and used as training material for generative models. Preprocessing includes normalization of text data, noise reduction, and protection of personal information.

[0664] The server uses pre-processed data to train specialized generative models for each employee. This builds models that reflect each employee's knowledge and skills. Through these models, the system can automatically generate responses by mimicking past experiences and answers provided by the employees.

[0665] When a user inputs a question into the system via a terminal, the server analyzes the question and understands its intent. Based on this analysis, the server selects the optimal generative model. The selected model generates an answer to the analyzed question, referencing similar past cases and related information in the process.

[0666] The generated answers are sent from the server to the user's terminal. Users can then proceed with their work based on the reliable answers provided. This system enables effective knowledge sharing and consistent application of work knowledge, particularly across different departments and with new employees.

[0667] Specific example

[0668] For example, if a user asks for a solution to a technical problem that has arisen in a project, the server searches past electronic communication history for similar problems and generates an answer using an appropriate generative model based on that history. This allows users to solve problems quickly and accurately based on relevant information. Furthermore, since information previously provided by former employees can also be utilized, there is the advantage of ensuring that organizational knowledge is continuously available.

[0669] The following describes the processing flow.

[0670] Step 1:

[0671] The server collects internal electronic communication history. This data is obtained from chat applications, emails, and other sources and stored in a database. During collection, metadata such as the speaker, date and time, and content are also recorded.

[0672] Step 2:

[0673] The server preprocesses the collected communication history. This process includes text normalization, removal of irrelevant information, and filtering of personal information. The preprocessed data is then organized into a format suitable for training generative models.

[0674] Step 3:

[0675] The server uses pre-processed data to train generative models for each employee. During this training process, natural language processing techniques are used to analyze the content and patterns of employees' past statements, thereby generating models that reflect their individual knowledge and skills.

[0676] Step 4:

[0677] The user enters a question into the system from a terminal. The terminal receives the user's input, formats the data, and sends it to the server.

[0678] Step 5:

[0679] The server analyzes the received question. This analysis uses natural language processing to understand the intent and main points of the question, extracting relevant keywords and context.

[0680] Step 6:

[0681] The server selects the optimal generative model based on the analysis results. Here, it identifies and selects the model of the employee with the most relevant knowledge to the question.

[0682] Step 7:

[0683] The server generates answers using the selected generative model. This generation process involves constructing highly accurate answers by referencing data from similar past questions and a knowledge base.

[0684] Step 8:

[0685] The server sends the generated response to the user's device. The user can then receive the response on their device and use it for their work.

[0686] (Example 1)

[0687] Next, we will describe Example 1. In the following description, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[0688] In modern companies, there is a challenge in effectively sharing the business knowledge and know-how of each employee and obtaining information necessary to carry out tasks quickly and accurately. Furthermore, there is a need to effectively utilize data while paying attention to the protection of personal information.

[0689] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0690] In this invention, the server includes means for collecting and storing electronic communication history, means for normalizing the collected electronic communication history and removing noise and personal information, and means for learning individual generative models based on the pre-processed history. This enables the automatic generation of answers by utilizing the knowledge accumulated within the company and based on the specialized knowledge of each employee.

[0691] "Electronic communication history" refers to the record of communications such as emails and chat messages conducted within an organization.

[0692] "Normalization" is the process of converting text data into a unified and consistent format, which improves data quality and processing efficiency.

[0693] "Noise reduction" is a technique that removes unnecessary information and errors from text data, and is performed to improve the usefulness of the data.

[0694] A "personalized generative model" is a customized AI model tailored to a specific employee or situation, possessing the ability to generate answers by learning from past knowledge and experience.

[0695] A "user question" is a request from a user to provide information or solve a problem, and is usually written in natural language.

[0696] "Understanding intent" is the process of analyzing a user's question and identifying the underlying requirements and objectives.

[0697] "Referring to past cases" is a method of using data accumulated in the past to derive useful information and answers to current questions.

[0698] "User terminal" refers to a computing device used by a user to access the system, and includes PCs, smartphones, and other devices.

[0699] In this invention, the server first collects internal electronic communication history. This history is collected using the company's email system and chat platform. This collection process utilizes common APIs and data retrieval scripts. The server stores the collected data in a relational database management system (RDBMS), specifically MySQL or PostgreSQL. The database functions as a foundation for efficiently storing large amounts of historical information and for later analysis and utilization.

[0700] Next, the server preprocesses the stored electronic communication history. This preprocessing utilizes Python NLP libraries such as NLTK and spaCy. Using these libraries, the server normalizes the text data, removing noise and personal information. This process ensures the data is clean and suitable for analysis. For example, it anonymizes names and email addresses while removing special characters and extraneous tags.

[0701] Once the data is ready, the server uses TensorFlow or PyTorch as a deep learning framework to train specialized generative models for each employee. This builds generative models that reflect each employee's job knowledge and past experience. These generative models are designed to mimic the specialized knowledge of different employees and generate answers specific to their particular tasks.

[0702] When a user inputs a business-related question into the system via their terminal, the server analyzes the question using natural language processing technology and selects the optimal generative model. Using the selected model, the server generates an answer related to the question, referencing similar past cases. The generated answer is then provided to the user's terminal in real time.

[0703] For example, if a user sends a prompt such as, "Please tell me about past examples of data protection policies for a new project," the server will refer to the communication history of similar past projects, select an appropriate model, and provide a response. In this process, reliable information, including specific methods and points to note regarding the project, can be quickly conveyed to the user.

[0704] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0705] Step 1:

[0706] The server automatically collects electronic communication history from the company's electronic communication platform. Specifically, it retrieves data using APIs from mail servers and chat tools. The input is raw communication data, and the output is raw data stored in an RDBMS. This data is then saved for further processing.

[0707] Step 2:

[0708] The server preprocesses the collected electronic communication history. The input is the raw data saved in step 1. Specifically, it normalizes the data using Python's NLP library, removes noise and HTML tags, and anonymizes personal information. The output is clean, analyzable text data.

[0709] Step 3:

[0710] The server trains a generative AI model for each employee based on pre-processed data. The input is the clean data generated in step 2. In this process, the model is trained using TensorFlow or PyTorch, and a custom model is built that reflects the employee's business knowledge. The output is a generative AI model tailored to each employee.

[0711] Step 4:

[0712] Users input business-related questions into the system using a terminal. Specifically, they send prompt messages using a web form. The input is a question written in natural language, and the output is the query data sent to the server.

[0713] Step 5:

[0714] The server analyzes the question received from the user and understands its intent. The input is the question received in step 4. Using a natural language processing algorithm, it identifies the intent and selects the optimal generative model. The output is the selected generative AI model.

[0715] Step 6:

[0716] The server generates an answer to the question using the selected generative model. The input is the generative AI model obtained in step 5 and the content of the question. The server generates a precise answer while referring to past electronic communication history. The output is the answer data provided to the user.

[0717] Step 7:

[0718] The server-generated response is sent to the user's terminal. The input is the response data from the server, and the output is the answer information displayed on the terminal. This allows the user to quickly use it as a reference for their work.

[0719] (Application Example 1)

[0720] Next, we will explain Application Example 1. In the following explanation, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[0721] In modern cities, there is a need for methods to respond quickly and accurately to inquiries and problems from citizens. In particular, the quality of administrative services can be improved by effectively utilizing past cases and examples of responses. However, existing systems have the challenge of not being able to fully utilize past data, resulting in time-consuming individual responses.

[0722] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0723] In this invention, the server includes means for acquiring electronic communication records, means for preprocessing the acquired electronic communication records, and means for learning individual generative models based on the preprocessed electronic communication records. This enables administrative agencies to provide automated responses to inquiries from citizens, referencing past cases.

[0724] "Electronic communication records" refer to logs of emails and messages exchanged within a company or organization, and are a collection of information that includes a history of business-related communication.

[0725] "Preprocessing" refers to the process of removing noise from collected data and ensuring data integrity, and includes operations such as anonymization and normalization of confidential information.

[0726] A "generative model" is an algorithmic model that uses machine learning to learn from a specific dataset and automatically perform tasks such as text generation.

[0727] An "inquiry" is a question or request that a user or citizen enters into a system to obtain information, and it seeks to provide information about a specific event.

[0728] "Analysis results" refer to the information obtained when analyzing the input data, and are used as a reference for selecting a generative model and generating responses.

[0729] "Case data" refers to information that records past cases and how they were handled, and it is data that supports approaches to new problems.

[0730] "Administrative agencies" are organizations of the national or local government that provide public services and support the lives of communities and citizens.

[0731] To implement this invention, a server must first collect electronic communication records. The server obtains data from existing communication logs and preprocesses it. In the preprocessing, natural language processing tools such as NLTK and spaCy are used to remove noise from the collected data and to perform normalization and anonymization to maintain data integrity. Next, a generative model is trained using Python and the Hugging Face Transformers library based on the preprocessed data. The generative model has the ability to generate specific answers based on past data.

[0732] When a user submits a query to the system via a web interface or similar means, the query is first forwarded to the server. The server analyzes the query and selects the optimal generative model based on the analysis results. In this process, past case data is referenced, and the response is supplemented based on similar cases.

[0733] Users can quickly receive automatically generated answers through this process, with the aim of improving public services. For example, if a user asks, "Please tell me about the new recycling program," the server can use past inquiry history to provide details about the program and information about upcoming related events. In this way, the server uses a generation AI model to optimize answers in response to various inquiries.

[0734] Examples of specific prompt messages include the following:

[0735] "Please provide information about the new recycling program."

[0736] "Could you please provide details about similar environmental events that have been held in the past?"

[0737] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0738] Step 1:

[0739] The server retrieves electronic communication records within an organization. Input is data from the organization's mail server or messaging platform. It then stores the communication logs in a database. Output is the set of stored electronic communication records.

[0740] Step 2:

[0741] The server preprocesses the acquired electronic communication records. The input is the electronic communication records saved in step 1. The processing involves denoising, normalization, and anonymization using NLTK or spaCy. This generates analyzable text data while protecting privacy. The output is the preprocessed, clean text data.

[0742] Step 3:

[0743] The server trains a generative model based on preprocessed data. The input is the text data obtained in step 2. A specialized generative model is built using the Hugging Face Transformers library. The output is the trained generative model.

[0744] Step 4:

[0745] The user sends a query to the system from their terminal. The input is the query text entered by the user. The output is the query text received by the server.

[0746] Step 5:

[0747] The server analyzes the query received from the user. The input is the query text received by the server in step 4. The server uses a natural language processing engine to analyze the text and understand its intent. The output is the analysis result.

[0748] Step 6:

[0749] The server selects an appropriate generative model based on the analysis results. The input is the analysis results obtained in step 5. The selection algorithm determines the most relevant generative model. The output is the selected generative model.

[0750] Step 7:

[0751] The server generates answers using the selected generative model. The input consists of the selected generative model and its analysis results. Past case data is also referenced to reinforce the answers. The output is the generated answer text.

[0752] Step 8:

[0753] The user's device receives the response provided by the server. The input is the response text sent from the server. Based on this information, the user can quickly obtain the desired information. The output is the response displayed to the user.

[0754] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[0755] This invention provides an interactive response system that takes user emotions into account by combining an emotion engine. The system uses electronic communication history to construct individual generative models and reflects user emotions in the question analysis and response generation processes.

[0756] System Overview

[0757] The server first collects and stores past electronic communication history within the company into a database. The stored data is preprocessed to remove personally identifiable information and reduce noise. This ensures that a safe and effective dataset is used to train the generative model.

[0758] Based on this pre-processed data, the server trains a generative model tailored to each employee. This results in a highly accurate model that mimics the experience and expertise of each individual employee. The trained model becomes the key to generating appropriate answers in response to user questions.

[0759] When a user enters and submits a question from their device, the server analyzes that question. This analysis utilizes an emotion engine to extract emotions from the user's text. This emotion information is used to understand the intent behind the question and is reflected in the analysis results.

[0760] As a result, the server selects the optimal generative model and generates responses that are adjusted to the user's emotions. The generated responses are crafted in a tone that takes the user's emotions into consideration, thereby improving user satisfaction and convenience. For example, if signs of stress are detected, the response will be provided in a more considerate tone.

[0761] Ultimately, the server sends this generated response to the user's terminal, providing the user with the information they need in an appropriate format. This system not only leverages the expertise of individual employees but also enables responses that take user emotions into consideration, thereby improving operational efficiency and the quality of information exchange.

[0762] Specific example

[0763] For example, when a user asks a question about project delays, the sentiment engine may detect anxiety from the text. In this case, the server uses a generative model of an employee with past project management expertise to generate and provide a response that includes specific measures to alleviate anxiety and encouraging messages. This allows the user to move on to the next step with confidence.

[0764] The following describes the processing flow.

[0765] Step 1:

[0766] The server collects electronic communication history from within the company. It primarily retrieves data from internal chat tools and email logs and stores it in a database. The stored information includes metadata such as the speaker, date and time, and content.

[0767] Step 2:

[0768] The server preprocesses the collected electronic communication history. This processing includes text normalization and removal of unnecessary data, and in particular, filtering to remove personally identifiable information.

[0769] Step 3:

[0770] The server uses pre-processed data to train a generative model for each employee. Using natural language processing algorithms, it builds a model that reflects each employee's job knowledge and skills. This model formation allows the server to mimic each employee's areas of expertise and response patterns.

[0771] Step 4:

[0772] Users enter their questions using their terminals and send them to the server. The question input screen allows for questions in text format, and users can freely submit inquiries.

[0773] Step 5:

[0774] The server receives the user's question and uses an emotion engine to analyze the emotions contained in the question. This extracts the emotional context from the text in the question and helps understand the user's psychological state.

[0775] Step 6:

[0776] The server selects the optimal generative model based on the question content and sentiment analysis results. The selection criteria prioritize models created by employees with relevant knowledge to the question. It also obtains feedback to adjust the tone of the response according to the sentiment.

[0777] Step 7:

[0778] The server generates responses using the selected generative model. At this time, based on the sentiment analysis results, the responses are generated in a tone that matches the user's emotions. Relevant information from past data is referenced, and necessary information is added and the tone adjusted.

[0779] Step 8:

[0780] The server sends the generated response to the terminal, delivering it to the user. The user can then review the response on the terminal and use it to solve their work-related problems. This allows users to solve problems efficiently.

[0781] (Example 2)

[0782] Next, we will describe Example 2. In the following description, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[0783] Traditional interactive systems often provide only uniform answers to user questions, failing to adequately consider user emotions and intentions, resulting in low user satisfaction. Furthermore, obtaining accurate and appropriate answers to questions requiring specialized knowledge was difficult.

[0784] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[0785] In this invention, the server includes means for acquiring electronic communication data, means for preprocessing the acquired data and removing personal information, means for learning individual knowledge models based on the preprocessed data, means for analyzing questions and extracting emotional information, means for selecting an appropriate knowledge model based on the analysis results, means for generating answers in an emotionally appropriate tone using the selected knowledge model, and means for providing the generated answers to the user. This makes it possible to provide high-quality answers that take into account the user's emotions while leveraging expertise.

[0786] "Electronic communication data" refers to information such as emails and messages exchanged via the internet and other communication networks.

[0787] "Preprocessing" refers to the process of removing personally identifiable information from electronic communication data and reducing noise to improve data quality.

[0788] A "knowledge model" refers to a generative AI model trained using a specific dataset, and is an artificial intelligence algorithm that possesses a deep understanding of a particular subject.

[0789] "Emotional information" refers to data that indicates the type and degree of emotion extracted from users' questions and statements.

[0790] "Tone" refers to the wording and facial expressions used in the generated response, and by adjusting them according to the user's emotions, it is an element that influences the nuance and how the response is received.

[0791] "Analysis" refers to the process of using natural language processing techniques to analyze the information contained in a question in detail and understand its intent and content.

[0792] In this embodiment of the invention, the interactive response system employs multiple technical means to provide responses that take into account the user's emotions.

[0793] The server collects electronic communication data via the internet and other communication networks and stores it in a database. This collected data is first pre-processed. This pre-processing uses specialized software, such as Python scripts, to remove personally identifiable information and reduce noise, thereby improving data quality. This process prepares the data for training.

[0794] Next, the server uses the pre-processed data to train a knowledge model specific to each employee, creating an AI model. This training process utilizes machine learning frameworks such as TensorFlow and PyTorch. The server builds a highly accurate model that reflects the employee's work knowledge and experience, which then forms the basis for subsequent question answering.

[0795] When a user enters a question using their device, the text data is sent to the server and analyzed through natural language processing. The server then activates an emotion engine and extracts emotional information from the text using an emotion analysis API. This emotional information is crucial for understanding the intent of the question and selecting the optimal knowledge model.

[0796] The generated responses are adjusted in tone to reflect the user's emotions. For example, if the user is feeling anxious, the response will be adjusted to include elements of warmth and encouragement. This improves both user satisfaction and work efficiency.

[0797] As a concrete example, suppose a user enters the prompt, "I would like specific advice regarding project delays." The server processes this input, selects a knowledge model with data on similar past cases, and generates a reassuring response. This response includes specific measures and words of encouragement, allowing the user to confidently move on to the next step.

[0798] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0799] Step 1:

[0800] The server acquires electronic communication data via the internet or other communication networks. The input includes raw data such as emails and messages. This data is stored in a database. At this stage, the acquired electronic communication data is output as is.

[0801] Step 2:

[0802] The server preprocesses the stored electronic communication data. The input is the raw data obtained in step 1. In this process, specialized software is used to remove personal information and apply algorithms to reduce noise. This generates safe and clean data, which becomes the output for the next step.

[0803] Step 3:

[0804] The server trains a generative AI model tailored to each employee based on pre-processed data. The input is the clean data obtained in step 2. Machine learning frameworks such as TensorFlow and PyTorch are used to train this data and build a knowledge model. The output at this stage is a generative AI model tailored to each employee.

[0805] Step 4:

[0806] The user enters a question via a terminal and sends it to the server. The input is the user's question, which is free-form text. At this point, the entered prompt text is passed directly to the next step.

[0807] Step 5:

[0808] When the server receives a question from a user, it analyzes it using natural language processing tools. The input is the prompt sentence from step 4, and word segmentation and part-of-speech tagging are performed, followed by analysis to understand the intent. The results of this analysis are output, and sentiment extraction is then performed.

[0809] Step 6:

[0810] The server uses an emotion engine to extract emotional information from the question text. The input is the analysis result from step 5. The type and degree of emotion are identified using the emotion analysis API, which is then used in the next step. At this point, this emotional information is output.

[0811] Step 7:

[0812] The server selects the optimal generative AI model based on the analysis results and extracted sentiment information. The inputs are the analysis results from step 5 and the sentiment information from step 6. Considering these, a suitable knowledge model is selected, and the answer is generated by that knowledge model.

[0813] Step 8:

[0814] The server uses the selected generative AI model to generate responses in a tone that matches the user's emotions. The input is the knowledge model selected in step 7, and the output is the response with the adjusted tone. Based on the analyzed emotional information, a response is created that reflects the optimal tone and content for the user.

[0815] Step 9:

[0816] The server sends the final generated response to the user's device. The input is the response generated in step 8. The user receives this information on their device and can obtain the necessary knowledge and advice. The output is the adjusted-tone response displayed on the user's device.

[0817] (Application Example 2)

[0818] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[0819] In caregiving settings, accurate communication tailored to the emotions and circumstances of those receiving care is essential, but achieving this is not easy. In particular, accurately understanding a person's emotions and responding accordingly is a challenging task. Therefore, there is a need to develop systems that enable care staff and family members to respond appropriately based on the emotions of those receiving care.

[0820] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[0821] In this invention, the server includes means for acquiring electronic communication history, means for pre-processing the acquired electronic communication history, means for learning individual generative models based on the pre-processed electronic communication history, means for analyzing questions, means for analyzing the user's emotions, means for selecting an appropriate generative model based on the analysis results and emotions, means for generating an emotionally appropriate response using the selected generative model, and means for providing the generated response. This enables responses based on the emotions of the person receiving care, thereby improving the quality of care.

[0822] "Electronic communication history" refers to a record of messages and communication content transmitted using communication devices.

[0823] "Preprocessing" is the process of removing noise and unnecessary information from acquired data and converting it into a format suitable for analysis and model training.

[0824] A "generative model" is an algorithm that has been trained to produce an appropriate output for a given input.

[0825] "Methods for analyzing questions" refer to technologies for understanding the content of questions entered by users and identifying their intent.

[0826] "Methods for analyzing emotions" refer to technologies that extract emotions from users' statements and texts and understand their emotional state.

[0827] A "generated answer" is a response to a user's question, created by a generative model.

[0828] "Means of provision" refers to the methods and technologies used to present the generated answers to the user.

[0829] The system of this invention primarily consists of a server and a user terminal. The server first acquires electronic communication history and preprocesses it. Preprocessing includes removing personally identifiable information and reducing noise from the collected digital data. This results in a clean dataset suitable for analysis and training generative models. The hardware used is a cloud server, and the software used for data analysis includes Python and R.

[0830] Based on pre-processed data, the server trains individual generative AI models. This involves creating user-specific models and utilizing machine learning algorithms to mimic their expertise and experience. Deep learning frameworks such as TensorFlow and PyTorch are used in this process.

[0831] When a user enters a question from their device, the device sends that data to the server. The server analyzes the received question and activates an emotion analysis engine to extract the user's emotions. This operation can be performed using natural language processing techniques, and libraries such as NLTK and spaCy are used for this purpose.

[0832] Based on the sentiment analysis results, the server selects the most appropriate generative model and generates a response that matches the user's emotions. This response generation process adjusts the tone and content of the response to meet the user's needs. For example, if stress or anxiety is detected, the server provides more reassuring feedback.

[0833] The generated responses are sent from the server to the user's terminal, where the user receives them either visually or audibly. This system enables personalized responses that take the user's emotions into consideration, making it applicable in fields such as elderly care.

[0834] For example, if an elderly person makes a statement to the terminal such as "I'm not feeling very well today," the server can analyze it and respond with encouraging words or suggestions for simple exercises to help them relax. An example of a prompt to input into the generative AI model could be, "Analyze the user's recent emotional state and suggest the most appropriate care method."

[0835] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0836] Step 1:

[0837] The server retrieves electronic communication history data. The input consists of messages and communication data previously sent by the user. This data is retrieved from the database and prepared for the next preprocessing step.

[0838] Step 2:

[0839] The server preprocesses the acquired data. It removes personally identifiable information from the input data and reduces unnecessary noise. This results in a clean dataset suitable for training. Specifically, it performs text normalization and filtering.

[0840] Step 3:

[0841] The server trains a generative AI model based on preprocessed data. The input is a clean dataset, and the output is an individual generative model. The model is trained using a machine learning framework such as TensorFlow.

[0842] Step 4:

[0843] The user enters a question from their terminal and sends it to the server. The input is the user's question text, which the terminal forwards to the server.

[0844] Step 5:

[0845] The server parses the received question. The input is the question text. Natural language processing techniques are used to understand the meaning of the text and identify the necessary context. The output is the parsed intent information.

[0846] Step 6:

[0847] The server analyzes the user's emotions using an emotion analysis engine. The input is the question text, and emotions are extracted using an emotion analysis API, etc. The output is emotion information (joy, anxiety, sadness, etc.).

[0848] Step 7:

[0849] The server selects an appropriate generative model based on the question and sentiment information. It takes the analyzed intent and sentiment information as input and determines the most suitable generative model. The output is the selected generative model.

[0850] Step 8:

[0851] The server generates emotion-sensitive responses using the selected generative model. Here, the tone and content of the responses are adjusted to match the user's emotions based on the input analysis data. The output is the generated response.

[0852] Step 9:

[0853] The server sends the generated response to the user's terminal. The terminal then provides the user with the received response in text or audio format. The output is the response to the user.

[0854] The specific processing unit 290 transmits the result of the specific processing to the robot 414. In the robot 414, the control unit 46A causes the speaker 240 and the controlled object 443 to output the result of the specific processing. The microphone 238 acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[0855] Data generation model 58 is a type of so-called generative AI (Artificial Intelligence). One example of data generation model 58 is ChatGPT (Internet search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0856] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and the specific processing may also be performed by the robot 414.

[0857] Furthermore, the emotion identification model 59, acting as an emotion engine, may determine the user's emotion according to a specific mapping. Specifically, the emotion identification model 59 may determine the user's emotion according to a specific mapping, which is an emotion map (see Figure 9). Similarly, the emotion identification model 59 may also determine the robot's emotion, and the identification processing unit 290 may perform identification processing using the robot's emotion.

[0858] Figure 9 shows an emotion map 400 in which multiple emotions are mapped. In the emotion map 400, emotions are arranged in concentric circles radiating from the center. The closer to the center of the concentric circles, the more primitive the emotions are located. Further out of the concentric circles, emotions representing states and actions arising from mental states are located. Emotion is a concept that includes feelings and mental states. On the left side of the concentric circles, emotions that are generally generated from reactions occurring in the brain are located. On the right side of the concentric circles, emotions that are generally induced by situational judgment are located. Above and below the concentric circles, emotions that are generally generated from reactions occurring in the brain and induced by situational judgment are located. In addition, the emotion of "pleasure" is located on the upper side of the concentric circles, and the emotion of "displeasure" is located on the lower side. Thus, in the emotion map 400, multiple emotions are mapped based on the structure in which emotions arise, and emotions that are likely to occur simultaneously are mapped close together.

[0859] These emotions are distributed at the 3 o'clock position on the Emotion Map 400, and usually fluctuate between feelings of security and anxiety. In the right half of the Emotion Map 400, situational awareness takes precedence over internal feelings, resulting in a calm impression.

[0860] The inside of the Emotion Map 400 represents inner thoughts, while the outside represents actions. Therefore, the further you go from the outside of the Emotion Map 400, the more visible (expressed in actions) your emotions become.

[0861] Here, human emotions are based on various balances, such as posture and blood sugar levels. When these balances deviate from the ideal, it results in discomfort, and when they approach the ideal, it results in pleasure. Similarly, in robots, cars, motorcycles, etc., emotions can be created based on various balances, such as posture and battery level. When these balances deviate from the ideal, it results in discomfort, and when they approach the ideal, it results in pleasure. The emotion map can be generated, for example, based on Dr. Mitsuyoshi's emotion map (Research on a system for analyzing brain physiological signals of speech emotion recognition and emotion, Tokushima University, doctoral dissertation: https: / / ci.nii.ac.jp / naid / 500000375379). The left half of the emotion map contains emotions belonging to a region called "response," where sensation is dominant. The right half of the emotion map contains emotions belonging to a region called "situation," where situational awareness is dominant.

[0862] The emotion map defines two emotions that promote learning. One is the emotion around the middle of the negative "repentance" and "reflection" on the situation side. In other words, it is when the robot experiences negative emotions such as "I never want to feel this way again" or "I don't want to be scolded again." The other is the emotion around the positive "desire" on the reaction side. In other words, it is when the robot has positive feelings such as "I want more" or "I want to know more."

[0863] The emotion identification model 59 inputs user input into a pre-trained neural network, obtains emotion values ​​representing each emotion shown in the emotion map 400, and determines the user's emotion. This neural network is pre-trained based on multiple training data sets, which are combinations of user input and emotion values ​​representing each emotion shown in the emotion map 400. Furthermore, this neural network is trained so that emotions located close together have similar values, as shown in the emotion map 900 in Figure 10. Figure 10 shows an example where multiple emotions such as "reassured," "calm," and "confident" have similar emotion values.

[0864] The above description primarily focuses on the functions of the data processing device 12 in relation to this disclosure. However, the system related to this disclosure is not necessarily implemented on a server. The system related to this disclosure may be implemented as a general information processing system. This disclosure may be implemented, for example, as a software program that runs on a personal computer or as an application that runs on a smartphone. The method related to this disclosure may be provided to users in SaaS (Software as a Service) format.

[0865] In the above embodiment, an example was given in which a specific process is performed by a single computer 22. However, the technology of this disclosure is not limited thereto, and a distributed processing of the specific process may be performed by multiple computers, including computer 22. For example, a data generation model 58 may be provided in an external device of the data processing device 12, and the external device may generate data according to the input data.

[0866] In the above embodiment, an example was given in which the specific processing program 56 is stored in the storage 32, but the technology of this disclosure is not limited thereto. For example, the specific processing program 56 may be stored in a portable, computer-readable, non-temporary storage medium such as a USB (Universal Serial Bus) memory. The specific processing program 56 stored in the non-temporary storage medium is installed in the computer 22 of the data processing device 12. The processor 28 executes specific processing according to the specific processing program 56.

[0867] Alternatively, the specific processing program 56 may be stored in a storage device such as a server connected to the data processing device 12 via the network 54, and the specific processing program 56 may be downloaded and installed on the computer 22 in response to a request from the data processing device 12.

[0868] Furthermore, it is not necessary to store the entirety of the specific processing program 56 in a storage device such as a server connected to the data processing device 12 via the network 54, or to store the entirety of the specific processing program 56 in the storage 32; it is acceptable to store only a portion of the specific processing program 56.

[0869] The following types of processors can be used as hardware resources to perform specific processing. Examples of processors include a CPU, a general-purpose processor that functions as a hardware resource to perform specific processing by executing software, i.e., a program. Other examples of processors include dedicated electrical circuits, such as FPGAs (Field-Programmable Gate Arrays), PLDs (Programmable Logic Devices), or ASICs (Application Specific Integrated Circuits), which have circuit configurations specifically designed to perform specific processing. All of these processors have built-in or connected memory, and all of them perform specific processing by using memory.

[0870] The hardware resource that performs a specific process may consist of one of these various processors, or it may consist of a combination of two or more processors of the same or different types (for example, a combination of multiple FPGAs, or a combination of a CPU and an FPGA). Alternatively, the hardware resource that performs a specific process may consist of a single processor.

[0871] Examples of configurations using a single processor include, firstly, a configuration in which one or more CPUs and software are combined to form a single processor, and this processor functions as a hardware resource that performs a specific process. Secondly, there is a configuration using a processor that realizes the functions of the entire system, including multiple hardware resources that perform a specific process, on a single IC chip, as exemplified by SoCs (System-on-a-chip). In this way, a specific process is realized using one or more of the above types of processors as hardware resources.

[0872] Furthermore, the hardware structure of these various processors can more specifically utilize electrical circuits that combine circuit elements such as semiconductor devices. Also, the specific processing described above is merely an example. Therefore, it goes without saying that unnecessary steps can be deleted, new steps added, or the processing order rearranged, as long as it does not deviate from the main purpose.

[0873] The descriptions and illustrations presented above are detailed explanations of the technical aspects of this disclosure and are merely examples of the technical aspects. For example, the above descriptions of the structure, function, operation, and effect are examples of the structure, function, operation, and effect of the technical aspects of this disclosure. Therefore, it goes without saying that you may delete unnecessary parts, add new elements, or replace elements in the descriptions and illustrations presented above, as long as you do not deviate from the essence of the technical aspects of this disclosure. Furthermore, in order to avoid confusion and facilitate understanding of the technical aspects of this disclosure, explanations of common technical knowledge and the like that do not require special explanation to enable the implementation of the technical aspects of this disclosure have been omitted from the descriptions and illustrations presented above.

[0874] All documents, patent applications, and technical standards described herein are incorporated by reference to the same extent as if each individual document, patent application, and technical standard were specifically and individually noted to be incorporated by reference.

[0875] The following is further disclosed regarding the embodiments described above.

[0876] (Claim 1)

[0877] Means for obtaining electronic communication history,

[0878] A means for pre-processing the acquired electronic communication history,

[0879] A means for learning individual generative models based on pre-processed electronic communication history,

[0880] Means for analyzing questions,

[0881] A means for selecting an appropriate generative model based on the analysis results,

[0882] A means of generating an answer using a selected generative model,

[0883] Means for providing the generated answer,

[0884] A system that includes this.

[0885] (Claim 2)

[0886] The system according to claim 1, which removes personally identifiable information when collecting data.

[0887] (Claim 3)

[0888] The system according to claim 1, wherein the generated response includes a decision process.

[0889] "Example 1"

[0890] (Claim 1)

[0891] Means for collecting and storing electronic communication history,

[0892] A means of normalizing collected electronic communication history and removing noise and personal information,

[0893] A means for training individual generative models based on preprocessed history,

[0894] A means of analyzing user questions to understand their intent,

[0895] A means for selecting an appropriate generative model based on the analyzed results,

[0896] A means of generating answers by referencing past cases using a selected generative model,

[0897] A means of providing the generated response to the user's device,

[0898] A system that includes this.

[0899] (Claim 2)

[0900] The system according to claim 1, which includes a process for anonymizing personally identifiable information when collecting data.

[0901] (Claim 3)

[0902] The system according to claim 1, which includes a decision process that references past similar cases when generating answers to user questions.

[0903] "Application Example 1"

[0904] (Claim 1)

[0905] Means for obtaining electronic communication records,

[0906] A means for pre-processing acquired electronic communication records,

[0907] A means for learning individual generative models based on pre-processed electronic communication records,

[0908] Means for analyzing inquiries,

[0909] A means for selecting an appropriate generative model based on the analysis results,

[0910] A means of generating an answer using a selected generative model,

[0911] Means for providing the generated answer,

[0912] A means to reinforce the answers generated by referring to past case data,

[0913] A means of automatically generating information based on inquiries from citizens,

[0914] A system that includes this.

[0915] (Claim 2)

[0916] The system according to claim 1, which removes personally identifiable information when collecting data.

[0917] (Claim 3)

[0918] The system according to claim 1, wherein the generated response includes a decision-making process.

[0919] "Example 2 of combining an emotion engine"

[0920] (Claim 1)

[0921] Means for acquiring electronic communication data,

[0922] A means of preprocessing acquired electronic communication data and removing personal information,

[0923] A means for training individual knowledge models based on preprocessed data,

[0924] A means of analyzing questions and extracting emotional information,

[0925] A means of selecting an appropriate knowledge model based on the results of analysis and extraction,

[0926] A means of generating responses in an emotionally appropriate tone using a selected knowledge model,

[0927] A means of providing the generated answers to the user,

[0928] A system that includes this.

[0929] (Claim 2)

[0930] The system according to claim 1, which takes emotion into consideration when parsing prompt text.

[0931] (Claim 3)

[0932] The system according to claim 1, wherein the generated response includes adjustments based on the user's emotions.

[0933] "Application example 2 when combining with an emotional engine"

[0934] (Claim 1)

[0935] Means for obtaining electronic communication history,

[0936] A means for pre-processing the acquired electronic communication history,

[0937] A means for learning individual generative models based on pre-processed electronic communication history,

[0938] Means for analyzing questions,

[0939] A means of analyzing user emotions,

[0940] A means of selecting an appropriate generative model based on the analysis results and emotions,

[0941] A means of generating emotionally appropriate responses using a selected generative model,

[0942] Means for providing the generated answer,

[0943] A system that includes this.

[0944] (Claim 2)

[0945] The system according to claim 1, which removes personally identifiable information when collecting data.

[0946] (Claim 3)

[0947] The system according to claim 1, wherein the generated response includes an emotional response process. [Explanation of Symbols]

[0948] 10, 210, 310, 410 Data Processing Systems 12 Data Processing Devices 14 Smart Devices 214 Smart Glasses 314 Headset-type terminal 414 Robots< / url:> < / url:> < / url:> < / url:>

Claims

1. Means for obtaining electronic communication history, A means for pre-processing the acquired electronic communication history, A means for learning individual generative models based on pre-processed electronic communication history, Means for analyzing questions, A means for selecting an appropriate generative model based on the analysis results, A means of generating an answer using a selected generative model, Means for providing the generated answer, A system that includes this.

2. The system according to claim 1, which removes personally identifiable information when collecting data.

3. The system according to claim 1, wherein the generated response includes a decision process.

Images