System

Abstract

A system is provided.SOLUTION: A system comprising: means for collecting data about past and present experts; means for processing and structuring the data; means for storing the structured data in a database; means for scoring experts from the database that are suitable for a particular discussion purpose; and means for generating models of the scored experts and conducting a discussion.SELECTED DRAWING: Figure 1

Description

[Technical Field]

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

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

[0003] [Patent Document 1] Japanese Patent Publication No. 2022-180282 Summary of the Invention [Problem to be solved by the invention]

[0004] Modern companies face complex problems and challenges that require the utilization of diverse perspectives and knowledge. However, it is difficult to directly utilize the knowledge and experience of specific experts or great figures, which limits high-quality brainstorming and the generation of new business ideas. Furthermore, there is a problem in that it is not possible to combine the knowledge of great figures of the past and current experts across time and space. This invention aims to solve these problems and provide a system that enables high-quality discussions and the generation of new business ideas. [Means for solving the problem]

[0005] The present invention provides a system including means for collecting data on past and present experts, means for processing and structuring the data, means for storing the structured data in a database, means for scoring experts from the database that are suitable for a specific discussion purpose, and means for generating models of the scored experts and conducting discussions.

[0006] Specifically, it uses natural language processing technology to extract keywords related to specific discussion objectives and then scores relevant experts based on this. The generated expert model then uses a generative AI model to mimic the expert's speech, enabling high-quality discussions. This allows users to combine the knowledge of past and present experts to effectively generate new business ideas.

[0007] "Data collection means" refers to means for obtaining data about a particular person from various sources.

[0008] "Means for processing and structuring data" refers to means for analyzing, organizing, and converting acquired data into a format that can be stored in a database.

[0009] A "means for storing in a database" is a means for storing structured data in a digital storage device and making it accessible at a later time.

[0010] A "scoring means" is a means for assessing and ranking the suitability of experts in relation to a particular discussion purpose.

[0011] The "means for holding a discussion" is a means for using a selected expert model to respond to questions from users and advance a discussion.

[0012] A "generative AI model" is a model that uses artificial intelligence technology to imitate the speech and thought patterns of a specific person.

[0013] "Natural language processing technology" is a set of technologies for analyzing, understanding, and generating human language.

[0014] "Keyword extraction" refers to the process of identifying and extracting important words and phrases from input text data. [Brief explanation of the drawings]

[0015] [Figure 1] 1 is a conceptual diagram showing an example of the configuration of a data processing system according to a first embodiment. [Figure 2] 1 is a conceptual diagram showing an example of main functions of a data processing device and a smart device according to a first embodiment. [Figure 3] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a second embodiment. [Figure 4] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and smart glasses according to a second embodiment. [Figure 5] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a third embodiment. [Figure 6] FIG. 11 is a conceptual diagram showing an example of main functions of a data processing device and a headset-type terminal according to a third embodiment. [Figure 7] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a fourth embodiment. [Figure 8] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and a robot according to a fourth embodiment. [Figure 9] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 10] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 11] FIG. 3 is a sequence diagram showing a processing flow of the data processing system according to the first embodiment. [Figure 12] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 1. [Figure 13]FIG. 10 is a sequence diagram showing the flow of processing in the data processing system according to the second embodiment when an emotion engine is combined. [Figure 14] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 2 when an emotion engine is combined. DETAILED DESCRIPTION OF THE INVENTION

[0016] An example of an embodiment of a system according to the technology of the present disclosure will be described below with reference to the accompanying drawings.

[0017] First, the terms used in the following description will be explained.

[0018] In the following embodiments, a coded processor (hereinafter simply referred to as a "processor") may be a single arithmetic device or a combination of multiple arithmetic devices. Furthermore, a processor may be a single type of arithmetic device or a combination of multiple types of arithmetic devices. Examples of arithmetic devices include a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), a GPGPU (General-Purpose computing on Graphics Processing Units), and an APU (Accelerated Processing Unit).

[0019] In the following embodiments, a coded RAM (Random Access Memory) is a memory in which information is temporarily stored and is used as a working memory by a processor.

[0020] In the following embodiments, the coded storage is one or more non-volatile storage devices that store various programs, various parameters, etc. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), magnetic disks (e.g., hard disks), and magnetic tapes.

[0021] In the following embodiments, a communication I / F (Interface) with a symbol is an interface including a communication processor, an antenna, etc. The communication I / F controls communication between multiple computers. Examples of communication standards applied to the communication I / F include wireless communication standards including 5G (5th Generation Mobile Communication System), Wi-Fi (registered trademark), Bluetooth (registered trademark), etc.

[0022] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B." In other words, "A and / or B" means that it may be only A, only B, or a combination of A and B. Furthermore, in this specification, the same concept as "A and / or B" is also applied when three or more things are expressed connected by "and / or."

[0023] [First embodiment]

[0024] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0025] 1, a data processing system 10 includes a data processing device 12 and a smart device 14. An example of the data processing device 12 is a server.

[0026] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0027] The smart device 14 includes a computer 36, a reception device 38, an output device 40, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The reception device 38, the output device 40, and the camera 42 are also connected to the bus 52.

[0028] The reception device 38 includes a touch panel 38A, a microphone 38B, and the like, and receives user input. The touch panel 38A detects contact with an indicator (for example, a pen or a finger) to receive user input by the touch of the indicator. The microphone 38B detects the user's voice to receive user input by voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the data indicating the user input.

[0029] The output device 40 includes a display 40A and a speaker 40B, and presents data to the user 20 by outputting the data in a form of expression that the user 20 can perceive (for example, audio and / or text). The display 40A displays visible information such as text and images in accordance with instructions from the processor 46. The speaker 40B outputs audio in accordance with instructions from the processor 46. The camera 42 is a compact digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0030] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 control the exchange of various information between the processor 46 and the processor 28 via the network 54.

[0031] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0032] 2, in the data processing device 12, a specific process is performed by the processor 28. A specific processing program 56 is stored in the storage 32. The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific process is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0033] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0034] In the smart device 14, the processor 46 performs the reception output process. The storage 50 stores a reception output program 60. The reception output program 60 is used in conjunction with the specific processing program 56 by the data processing system 10. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0035] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0036] The present invention is a system that systematically collects and analyzes information about past and present experts, and utilizes the results to realize high-quality brainstorming. Specific embodiments of the present invention are described below.

[0037] 1. Data collection and database creation

[0038] The first process performed by the server is data collection. Specifically, the server collects data on famous people and experts from sources such as online articles, academic papers, and book databases through scraping or APIs. The collected data is stored in a temporary storage area.

[0039] The server then cleanses the collected data, removes noise, and standardizes the data format. It then uses natural language processing (NLP) technology to tokenize the data and extract entities (people, dates, places, etc.). Based on the extracted entities, it generates profile data of famous people and experts, which are then stored in a database.

[0040] 2. Language Analysis and Model Fine-tuning

[0041] The server analyzes the profile data stored in the database. Specifically, it uses a natural language processing engine to analyze the text data, and performs grammatical analysis and sentiment analysis. Based on the results of this analysis, the speech patterns and styles of each great person or expert are added to the data model.

[0042] The server then inputs the analyzed data into a generative AI model (e.g., GPT-4) and performs transfer learning, which gives the model the ability to imitate a specific person. The fine-tuned model is then stored in storage.

[0043] 3. Scoring system and expert selection

[0044] The user inputs the purpose of the discussion through the device interface. For example, they can set a purpose such as "brainstorm ideas for a new product." The input purpose is then sent from the device to the server.

[0045] The server analyzes the purpose and extracts relevant keywords. It then searches the database for relevant figures and experts based on the keywords and ranks them using a scoring algorithm. The scoring results are sent to the user's device, where the user can review them and select the experts they want to include.

[0046] 4. Brainstorming

[0047] The server launches the generative AI models of the selected experts. The user inputs questions or opinions through the device. For example, if the user asks, "Tell me your ideas for new environmental technologies," the server interprets the question and sends it to the appropriate expert model. The generative AI models generate responses in the style of each expert, synthesize them, and send them to the device.

[0048] Specific examples

[0049] Consider a case where a user sets the theme as "Business ideas related to new environmental technologies." The server scores experts knowledgeable in environmental technologies from the database, and for example, Mr. A, Mr. B, and Mr. C receive high scores. The user checks the scoring results and selects Mr. A, Mr. B, and Mr. C. The server then launches the models of the selected experts, and when the user types "I would like to discuss new product strategies," the server responds in a synthesized form with each expert. In this way, the user can obtain new business ideas from multiple perspectives.

[0050] As described above, this system combines the knowledge of great people from the past and current experts to provide a concrete means for achieving high-quality brainstorming.

[0051] The processing flow will be explained below.

[0052] Step 1:

[0053] The server collects data about famous people and experts from various sources (book databases, web articles, academic papers, etc.) using scraping technology and APIs. The collected data is stored in a temporary storage area.

[0054] Step 2:

[0055] The server cleanses the data in the temporary storage area, removing noise and standardizing the data format, then uses natural language processing (NLP) techniques to tokenize the data and extract entities (people, dates, places, etc.).

[0056] Step 3:

[0057] The server generates profile data of famous people and experts based on the extracted entities, including their names, fields of expertise, and representative achievements. The generated profile data is stored in a database.

[0058] Step 4:

[0059] The server analyzes the profile data stored in the database, performs grammatical analysis and sentiment analysis, and adds the speech patterns and styles of each great person or expert to the data model based on the analysis results.

[0060] Step 5:

[0061] The server inputs the analyzed data into a generative AI model (e.g., GPT-4) for transfer learning. This process allows the generative AI model to mimic the speech and thought patterns of a specific person. Once the fine-tuned model is complete, it is stored in storage.

[0062] Step 6:

[0063] The user uses the device interface to input the purpose of the discussion, such as "brainstorm ideas for a new product." The input purpose is then sent from the device to the server.

[0064] Step 7:

[0065] The server analyzes the purpose of the discussion and extracts relevant keywords, then searches a database for relevant figures and experts based on the keywords.

[0066] Step 8:

[0067] The server uses a scoring algorithm to evaluate the search results and rank them in order of relevance, and the results are sent to the user's device, where the user can review and select the experts they want to include.

[0068] Step 9:

[0069] The server launches a generative AI model of a selected expert, and the user inputs questions or opinions from their device and sends them to the server.

[0070] Step 10:

[0071] The server interprets the user's input and sends questions and opinions to the appropriate expert model. The generative AI model generates responses in each expert's style, synthesizes them, and sends them to the device.

[0072] Step 11:

[0073] The terminal displays the response from the server on the user interface, and the user can check the displayed results and continue brainstorming by entering further questions or opinions as needed.

[0074] The above processing flow allows users to generate new business ideas through discussions with great people and experts from across time and space.

[0075] Example 1

[0076] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0077] Conventional brainstorming systems have difficulty effectively utilizing the knowledge of past and present experts. Furthermore, they lack the means to efficiently select experts suited to specific discussion purposes and achieve high-quality brainstorming. As a result, the quality of discussions declines, and it is difficult to obtain new ideas and insights.

[0078] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0079] In this invention, the server includes means for collecting data from information sources such as online articles, papers, and books, means for cleansing the data and extracting entities using natural language processing technology, means for storing the structured data in a database, means for inputting the purpose of the discussion and extracting related keywords, means for scoring experts suitable for the specific purpose of the discussion from the database, means for generating and launching a generative AI model of the selected expert based on the scoring results, and means for receiving questions from users via the terminal, sending the questions to the generative AI model, generating responses, and sending them to the terminal. This makes it possible to combine the knowledge of past and present experts and efficiently realize multifaceted and advanced brainstorming.

[0080] "Articles on the Internet" refers to text data and news articles posted on websites.

[0081] A "paper" refers to a document that describes the results of academic research or a document that reports the results of research.

[0082] "Book" means a printed document that has been published as a publication.

[0083] "Means of collecting data" refers to the methods and tools used to obtain the required information from the source.

[0084] "Data cleansing methods" refer to methods and tools used to remove noise and inaccurate information from collected data and format it.

[0085] "Natural language processing technology" refers to technology that enables computers to understand and process human language.

[0086] "Entity extraction" refers to methods and tools that extract important information, such as specific people, dates, or places, from text data.

[0087] "Structured data" refers to data that has been organized and arranged in a format suitable for a database.

[0088] "Means of storing data in a database" refers to the methods and tools used to store data in a database.

[0089] The "purpose of the discussion" refers to the goal or theme you want to achieve when having the discussion.

[0090] "Keyword extraction methods" refers to methods and tools for finding related words and phrases from text data.

[0091] A "scoring method" refers to a method or tool that evaluates the importance or relevance of data based on specific criteria.

[0092] A "generative AI model" refers to an artificial intelligence model that has the ability to perform specific tasks based on training data.

[0093] "Means for receiving questions from users" refers to the methods and tools by which the system obtains questions and opinions entered by users.

[0094] "Means for sending questions to a generative AI model" refers to methods or tools for communicating a user's questions to a generative AI model.

[0095] "Means for generating a response" refers to the methods and tools that a generative AI model uses to create an answer to a user's question.

[0096] "Means for sending to terminal" refers to a method or tool for transmitting the generated response to the user's terminal.

[0097] The present invention is a system that systematically collects and analyzes information about past and present experts, and utilizes the results to realize high-quality brainstorming. Specific embodiments of the present invention are described below.

[0098] Data collection and database creation

[0099] The server first collects information from online articles, academic papers, book databases, etc. Specifically, the server uses web scraping tools such as Python's BeautifulSoup and Scrapy to obtain information about famous people and experts. For academic papers, the server also uses the Google Scholar API to collect data. This collected data is then stored in a temporary storage area (for example, a database such as MySQL).

[0100] Next, the server cleanses the data using the Pandas library. Specifically, it removes unnecessary data and noise, converts date and time formats, and normalizes text. The cleansed data is then tokenized using natural language processing (NLP) techniques to extract entities (people, dates, places, etc.). This analysis is performed using libraries such as NLTK and spaCy. Finally, the structured data is stored in a database such as PostgreSQL.

[0101] Language analysis and model fine-tuning

[0102] The server analyzes the profile data stored in the database. For the analysis, it uses a natural language processing engine such as BERT to perform grammatical analysis and sentiment analysis. Based on the results of this analysis, speech patterns and styles are added to the data model.

[0103] The server then inputs the analyzed data into a generative AI model such as GPT-4 to perform transfer learning. This transfer learning is performed using PyTorch or TensorFlow. The generative AI model, which has acquired the ability to imitate a specific person through transfer learning, is stored in storage such as AWS S3.

[0104] Scoring system and expert selection

[0105] The user inputs the purpose of the discussion through the terminal interface, for example, "Brainstorm ideas for a new product." The input purpose is sent from the terminal to the server.

[0106] The server analyzes the purpose and extracts relevant keywords. It then uses SQL queries to search for relevant experts from the database and scores them using TF-IDF and other machine learning algorithms. The scoring results are sent to the device, where the user can review them and select the desired expert.

[0107] Brainstorming

[0108] The server launches a generative AI model of the selected expert. This is done using container orchestration tools such as Docker or Kubernetes. The user inputs a question or opinion through the terminal. For example, they might input, "Tell me your ideas for new environmental technologies." The server interprets the question and sends it to the appropriate expert model. The generative AI model generates a response in the expert's style, synthesizes it, and sends it to the terminal.

[0109] Specific examples

[0110] If a user selects the theme "Business ideas related to new environmental technologies," the server scores experts knowledgeable in environmental technologies from the database, and for example, Mr. A, Mr. B, and Mr. C receive high scores. The user checks the scoring results and selects Mr. A, Mr. B, and Mr. C. The server then activates the model of the selected experts, and when the user enters "I would like to discuss new product strategies," the server generates a response in a synthesized form from each expert and sends it to the terminal.

[0111] In this way, the system of the present invention combines the knowledge of past and present experts and provides a concrete means for realizing high-quality brainstorming.

[0112] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0113] Step 1: Data collection

[0114] The server collects data from online articles, academic papers, book databases, and other sources. Specifically, it uses web scraping tools such as Python's BeautifulSoup and Scrapy to obtain information about famous people and experts. For academic papers, it also uses the Google Scholar API to obtain paper data. Inputs include URLs and API endpoints, and the output is the obtained raw data stored in a temporary storage area. This process collects a large amount of text data.

[0115] Step 2: Data cleansing

[0116] The server cleanses the collected data. Specifically, it uses the Pandas library to manipulate data frames and remove unnecessary data and noise. This step also performs date and time format conversions and text normalization. The input is raw data stored in a temporary storage area, and the output is clean, cleaned data. This process improves the quality of the data and makes subsequent analysis easier.

[0117] Step 3: Data analysis using natural language processing

[0118] The server analyzes the cleansed data using natural language processing techniques. Specifically, it uses libraries such as NLTK and spaCy to tokenize the data and extract entities (people, dates, places, etc.). The input is the cleansed data, and the output is extracted entity data. This processing structures the text data and generates expert profile data.

[0119] Step 4: Store in the database

[0120] The server stores the extracted entity data in a database such as PostgreSQL. The input is the entity data, and the output is a structured database entry. This process allows for efficient management of expert information and facilitates subsequent search and analysis.

[0121] Step 5: Input and Analysis of Desired Objectives

[0122] The user inputs the purpose of the discussion through the terminal interface. For example, they might input "brainstorm ideas for a new product." The input is the discussion purpose entered by the user, and the output is text data sent to the server. This step transmits the user's discussion purpose to the server.

[0123] Step 6: Keyword extraction and search

[0124] The server analyzes the received discussion purpose and extracts relevant keywords. It then searches for relevant experts from the database using an SQL query. The inputs are the analyzed text data and the database, and the output is a list of relevant experts. This process identifies experts suitable for the discussion purpose.

[0125] Step 7: Scoring

[0126] The server scores experts using a scoring algorithm (e.g., TF-IDF or other machine learning algorithm) against the relevant data. The input is a list of experts and relevant data, and the output is a scored list of experts. This process prioritizes the appropriate experts.

[0127] Step 8: Selecting experts

[0128] The user checks the scoring results on their device and selects the experts they want to participate in. The input is a list of scored experts, and the output is information about the selected experts sent to the server. This process allows the user to select the experts they want.

[0129] Step 9: Launching the expert model

[0130] The server launches the generative AI model of the selected expert. Specifically, the model is launched using a container orchestration tool such as Docker or Kubernetes. The input is information about the selected expert, and the output is the launched generative AI model. This process puts the expert model into operation.

[0131] Step 10: Brainstorming and response generation

[0132] A user inputs a question or opinion through a device. For example, they might input, "Tell me your ideas for new environmental technologies." The input is the user's question text, and the output is sent to the generative AI model. The generative AI model then generates a response based on the question and sends it in a synthesized form to the user's device. This process allows the user to obtain high-quality opinions and suggestions.

[0133] (Application example 1)

[0134] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0135] The development of autonomous vehicles requires advanced algorithms and safety measures, but there is a lack of effective means to integrate expert knowledge. There is also a need for efficient methods to perform simulations based on expert knowledge and to verify new ideas and technical solutions.

[0136] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0137] In this invention, the server includes means for collecting data on past and present experts, means for processing and structuring the data, means for storing the structured data in a database, means for scoring experts suitable for a specific discussion purpose from the database, means for generating a model of the scored experts and holding discussions, and means for running a simulation based on the expert model. This makes it possible to effectively integrate the latest expert knowledge on autonomous vehicles and verify optimal operation algorithms and safety measures through simulation.

[0138] "Past and present expert data" is a collection of information about people who have past and present expertise and experience in a particular field.

[0139] "Means of collection" refers to the technical means used to obtain data from online papers, articles, book databases, etc.

[0140] "Means of processing and structuring" refers to technical means of systematically organizing collected data by processing it through data cleansing, tokenization, entity extraction, etc.

[0141] "Means of storing data in a database" means a technical means of efficiently storing structured data and keeping it in a state that can be quickly retrieved and used as needed.

[0142] An "expert appropriate for a particular purpose" is someone with particular knowledge and experience in a field relevant to the topic or goal of the discussion.

[0143] A "scoring means" is a technological means that uses specific criteria or algorithms to assess and rank the suitability or relevance of Experts.

[0144] "Means for generating a model of a scored expert and engaging in discussions" refers to the technical means for creating a generative AI model that mimics the style and speech patterns of a scored expert and using it to engage in discussions and responses.

[0145] A "generative AI model" is an artificial intelligence model that uses machine learning technology to mimic the knowledge and speaking style of experts.

[0146] "Means for performing simulations" refers to the technical means for verifying the operation algorithms and safety measures of autonomous vehicles in a virtual environment using the knowledge of the generated expert model.

[0147] The present invention provides a system for leveraging expert knowledge in the development and validation of autonomous vehicles. The system uses the following major hardware and software components:

[0148] 1. Data Collection

[0149] The server collects expert data on autonomous driving technology from academic papers, technical articles, and various book databases on the Internet. Specifically, it retrieves data using the URL Fetch API and performs web scraping using Beautiful Soup (Python).

[0150] 2. Data Analysis and Natural Language Processing

[0151] To process and structure the collected data, the server uses natural language processing techniques, specifically NLP libraries such as spaCy and NLTK for data cleansing, tokenization, and entity extraction, and then stores the structured data in a database.

[0152] 3. Expert Selection and Fine-Tuning

[0153] To select the most suitable experts for a specific discussion purpose, the server searches the data in the database and evaluates their relevance using scoring algorithms such as TF-IDF and BERT. Using the associated expert data, the GPT-4 model is fine-tuned using Hugging Face Transformers, resulting in a generative AI model that mimics each expert's speaking patterns and style.

[0154] 4. Simulation and Verification

[0155] Users can input questions about autonomous driving or the purpose of their discussion through their devices. For example, "What are the safety measures for autonomous vehicles in urban areas?" The server generates a response based on the relevant expert model. This response is used to run simulations in a virtual environment to verify optimal driving algorithms and safety measures.

[0156] Specific examples

[0157] For example, if a user types in "What are the safety measures for self-driving vehicles in urban areas?", the system will act as follows:

[0158] 1. Prompt statement generation (server):

[0159] User Query: "What are the safety measures for autonomous vehicles in urban areas?"

[0160] Model Prompt: "The system will discuss safety measures for autonomous vehicles in urban areas. Please suggest the best solution based on past and current expert knowledge."

[0161] 2. Generative AI model response (server):

[0162] Selected expert models respond to questions and their answers are used in the simulation.

[0163] 3. Simulation and result verification (server):

[0164] Response-based simulations are run in a virtual environment to verify optimal operating algorithms and safety measures.

[0165] This allows users to gain a multifaceted perspective on the development and operation management of autonomous vehicles while integrating the latest expert knowledge. By effectively utilizing the knowledge of past and present experts, this system can significantly improve the safety and efficiency of autonomous driving technology.

[0166] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0167] Step 1:

[0168] The server collects data related to autonomous driving technology. It uses the URL Fetch API to retrieve academic papers and technical articles from the Internet and performs web scraping using Beautiful Soup (Python). The input is a list of web page URLs, and the output is text data related to the scraped experts.

[0169] Step 2:

[0170] The server analyzes the collected text data using natural language processing technology. Specifically, the server uses spaCy or NLTK to cleanse, tokenize, and extract entities from the data. The input is scraped text data, and the output is structured data (such as entity information).

[0171] Step 3:

[0172] The server stores the structured data in a database, specifically by inserting well-organized data into a SQL or NoSQL database. The input is structured data, and the output is an updated database.

[0173] Step 4:

[0174] The server scores experts suitable for a specific discussion purpose. It uses natural language processing technology to extract keywords related to the discussion purpose from the user (e.g., "Safety measures for autonomous vehicles in urban areas") and evaluates their relevance using TF-IDF and BERT. The input is the discussion purpose text from the user and expert data in the database, and the output is a list of scored experts.

[0175] Step 5:

[0176] The server generates a scored expert model and conducts the discussion. Using Hugging Face Transformers, it performs transfer learning on the GPT-4 model to create a generative AI model that mimics the expert's speech patterns and style. The input is the scored expert data, and the output is the generated expert model.

[0177] Step 6:

[0178] Users input questions about autonomous driving or the purpose of the discussion through their terminal. For example, they might input, "What are the safety measures for autonomous vehicles in urban areas?" The input is the question text, and the output is the request data sent to the server.

[0179] Step 7:

[0180] The server analyzes the user's question, generates a prompt sentence based on the generated expert model, and inputs the prompt sentence to the expert model to obtain a response. The input is the user's question text and the generated expert model, and the output is the response of the expert model.

[0181] Step 8:

[0182] The server runs a simulation based on the expert model's responses and verifies the results. Simulation software is used to experiment with operation algorithms and safety measures in a virtual environment. The input is the expert model's response text, and the output is the simulation results.

[0183] Step 9:

[0184] The server provides the simulation results to the user. The results are displayed on the terminal and presented in a format that is easy for the user to understand. The input is the simulation result data, and the output is the result provided to the user.

[0185] Furthermore, an emotion engine that estimates the user's emotion may be combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59 and perform identification processing using the user's emotion.

[0186] This invention combines a system that collects and analyzes information on past and present experts and performs high-quality brainstorming based on the results with an emotion engine that recognizes and analyzes the user's emotions. This system enables responses and discussions to proceed according to the user's emotional state, resulting in more effective discussions.

[0187] 1. Data collection and database creation

[0188] The server first collects data about famous people and experts from sources such as online book databases, web articles, and academic papers. To achieve this, it uses scraping technology and APIs. The collected data is then stored in a temporary storage area.

[0189] The server then cleanses the collected data, removing noise and standardizing the data format. It then uses natural language processing (NLP) technology to tokenize the data and extract entities (people, dates, places, etc.). Based on the extracted entities, it generates profile data of famous people and experts, which are then stored in a database.

[0190] 2. Language Analysis and Model Fine-tuning

[0191] The server analyzes the profile data stored in the database. Specifically, it uses a natural language processing engine to analyze the text data, and performs grammatical analysis and sentiment analysis. Based on the results of this analysis, the speech patterns and styles of each great person or expert are added to the data model.

[0192] The server then inputs the analyzed data into a generative AI model (e.g., GPT-4) for transfer learning. Through this process, the generative AI model acquires the ability to mimic a specific person. Once the fine-tuned model is complete, it is stored in storage.

[0193] 3. Scoring system and expert selection

[0194] Users input the purpose of the discussion through the device interface. For example, they set a specific purpose such as "brainstorm ideas for a new product." The input purpose is sent from the device to the server.

[0195] The server analyzes the purpose and extracts relevant keywords.The server then searches a database for relevant famous people and experts based on the keywords, and uses a scoring algorithm to evaluate and rank them.The scoring results are sent to the terminal, where the user can review them and select the experts they want to participate.

[0196] 4. Utilizing the Emotion Engine

[0197] The server uses an emotion engine in discussions between the user and the expert model to analyze voice and text inputs from the user and recognize the user's emotional state (e.g., joy, anger, sadness, surprise).

[0198] 5. Brainstorming

[0199] The server launches the generative AI models of the selected experts. The user inputs questions or opinions through the device. For example, if the user asks, "Tell me your ideas for new environmental technologies," the server interprets the question and sends it to the appropriate expert model. The generative AI models generate responses in the style of each expert, synthesize them, and send them to the device.

[0200] Furthermore, the emotion engine recognizes the user's emotions, allowing the server to adjust the content and tone of the expert's response. For example, if the user is feeling impatient, the server can provide a calm response to enhance the effectiveness of the discussion.

[0201] Specific examples

[0202] If a user sets the theme as "Business ideas related to new environmental technologies," the server scores experts knowledgeable in environmental technologies from the database, and for example, experts A, B, and C receive high scores. The user checks the scoring results and selects experts A, B, or C. The server launches the models of the selected experts, and when the user asks, "I would like to discuss new product strategies," each expert responds in a synthesized form.

[0203] The server's built-in emotion engine detects the user's emotional state and adjusts the expert model's response as needed. For example, if the user is emotional, the expert model's response can be adjusted to a calm and peaceful tone to facilitate a smooth discussion.

[0204] As described above, this system combines the knowledge of great people from the past and current experts with the user's emotional state, providing a concrete means for achieving high-quality brainstorming and the creation of new business ideas.

[0205] The processing flow will be explained below.

[0206] Step 1:

[0207] The server collects data about famous people and experts from various sources (book databases, web articles, academic papers, etc.) using scraping technology and APIs. The collected data is stored in a temporary storage area.

[0208] Step 2:

[0209] The server cleanses the data in the temporary storage area, removing noise and standardizing the data format, then uses natural language processing (NLP) techniques to tokenize the data and extract entities (people, dates, places, etc.).

[0210] Step 3:

[0211] The server generates profile data of famous people and experts based on the extracted entities, including their names, fields of expertise, and representative achievements. The generated profile data is stored in a database.

[0212] Step 4:

[0213] The server analyzes the profile data stored in the database. Specifically, it uses a natural language processing engine to perform grammatical analysis and sentiment analysis of the text data. Based on the results of this analysis, the speech patterns and styles of each great person or expert are added to the data model.

[0214] Step 5:

[0215] The server inputs the analyzed data into a generative AI model (e.g., GPT-4) and performs transfer learning. This allows the generative AI model to acquire the ability to imitate a specific person. Once fine-tuned, the model is stored in storage.

[0216] Step 6:

[0217] The user inputs the purpose of the discussion using the device interface. For example, they set a purpose such as "brainstorm ideas for a new product." The input purpose is sent from the device to the server.

[0218] Step 7:

[0219] The server analyzes the purpose of the input discussion and extracts relevant keywords, then searches the database for relevant figures and experts based on the keywords.

[0220] Step 8:

[0221] The server uses a scoring algorithm to evaluate the search results and rank the most relevant experts, which are then sent to the device where the user can review and select the experts they want to participate.

[0222] Step 9:

[0223] The server runs a generative AI model of selected experts, which runs on a dedicated cloud infrastructure.

[0224] Step 10:

[0225] Users can input questions or opinions into their devices and send them to the server. For example, they can ask, "Tell me your ideas for new environmental technologies."

[0226] Step 11:

[0227] The server interprets the user's input and sends the question to the appropriate expert model, which then generates a response in the speaking style of each expert.

[0228] Step 12:

[0229] An emotion engine built into the server analyzes the user's voice and text input to assess their emotional state, for example detecting anger or impatience in the user's voice.

[0230] Step 13:

[0231] The server adjusts the content and tone of the generative AI model's response based on the user's emotional state as recognized by the emotion engine. For example, if the user is angry, the expert model's response will be changed to a calmer tone.

[0232] Step 14:

[0233] The server composes the final response and sends it to the terminal.

[0234] Step 15:

[0235] The terminal displays the response from the server on the user interface, and the user can check the displayed results and continue brainstorming by entering further questions or opinions as needed.

[0236] Through these processes, users can engage in discussions with great figures and experts from across time and space, and realize high-quality discussions that are adjusted according to their emotional state.

[0237] Example 2

[0238] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0239] In today's knowledge-based society, it is extremely important to conduct high-quality discussions by utilizing the knowledge of past and present experts. However, the process of extracting relevant information from vast amounts of data and selecting appropriate experts is difficult, especially when it comes to taking user emotions into account. This has led to a growing need for a system that recognizes the user's emotional state and generates responses based on the expert's style.

[0240] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0241] In this invention, the server includes means for collecting data on past and present experts, means for cleansing the data and analyzing it using natural language processing technology, means for storing structured data generated from the data in a database, means for scoring experts suitable for a specific discussion purpose from the database, means for generating a model of the scored experts and holding a discussion, means for recognizing the emotional state of a user during the discussion and adjusting the progress of the discussion, and means for generating expert responses using the generative AI model during the discussion. This makes it possible to conduct high-quality discussions by making the most of expert knowledge and effectively progressing the discussion in accordance with the emotional state of the user.

[0242] "Data Collection Tools" are tools used to gather information about past and present professionals from the Internet and other databases.

[0243] "Data cleansing means" refers to a means for removing noise from collected data and standardizing the data format.

[0244] "Natural language processing technology" is a technology that analyzes text data and performs tokenization, entity extraction, sentiment analysis, etc.

[0245] A "database" is a system that stores the generated structured data and retrieves and uses it when needed.

[0246] A "scoring tool" is a tool for searching a database for experts suitable for a particular discussion purpose, assessing their relevance, and ranking them.

[0247] An "expert model generation means" is a means for creating a generative AI model that mimics a scored expert.

[0248] The "discussion tool" is a tool that uses a generative AI model to generate responses based on the expert's style and engages in discussions between the user and the expert.

[0249] The "emotion recognition means" is a means for recognizing the emotional state of users during a discussion and adjusting the content and tone of the discussion.

[0250] A "generative AI model" is an artificial intelligence technology that can learn from large amounts of text data and generate text for specific tasks.

[0251] This invention is a system that collects and analyzes information about past and present experts, and uses the results to conduct high-quality discussions. By combining it with an emotion engine that recognizes and analyzes the user's emotions, it is possible to respond and progress the discussion in accordance with the user's emotional state. Specific embodiments for implementing this system are described below.

[0252] Data collection

[0253] The server first collects data about experts from sources such as online book databases, web articles, and academic papers. This is done using scraping tools such as BeautifulSoup and Scrapy, and APIs such as the Google Books API. The collected data is then stored in a temporary storage area. Specifically, the server collects titles, author names, citations, etc. from the book database and stores them in the temporary storage area.

[0254] Data Cleansing and Analysis

[0255] The collected data is cleansed by the server. Specifically, the Pandas library is used to fill in missing data and remove inaccurate data. Natural language processing (NLP) techniques are then used to tokenize the data, and entities (people, dates, places, etc.) are extracted using tools such as SpaCy. Expert profile data is generated based on the extracted entities, and this data is stored in a MySQL database.

[0256] Fine-tuning the model

[0257] The server analyzes the profile data stored in the database. It uses natural language processing engines such as BERT and GPT-3 to perform grammatical and sentiment analysis of the text data, and adds each expert's speech patterns and style to the data model. The analyzed data is then input into a generative AI model (e.g., GPT-4) for transfer learning. Once the fine-tuned model is complete, it is stored in Amazon S3 storage.

[0258] Scoring and Expert Selection

[0259] The user inputs the purpose of the discussion through the device interface. For example, they can set a specific purpose such as "brainstorm ideas for a new product." The input purpose is sent from the device to the server. The server analyzes the purpose and extracts related keywords. It then searches for appropriate experts in a database and uses a scoring algorithm to evaluate their relevance and rank them. The scoring results are sent to the device, where the user can review them and select the experts they want to participate in.

[0260] Emotion recognition and discussion moderation

[0261] The server uses an emotion engine in the discussion between the user and the expert model. The emotion engine uses the Google Cloud Natural Language API and Microsoft Azure Text Analytics API to analyze the user's voice and text input and recognize their emotional state (joy, anger, sadness, surprise, etc.). Based on the recognized emotion, the content and tone of the expert model's response are adjusted.

[0262] Brainstorming

[0263] The server launches a generative AI model of a selected expert. The user inputs a question or opinion through their device. For example, they might ask, "Tell me your ideas for new environmental technologies." The server interprets the question and sends it to the appropriate expert model. The generative AI model generates a response in the style of each expert, synthesizes it, and sends it to the device. An emotion engine recognizes the user's emotions and adjusts the expert model's response as needed.

[0264] An example of a specific prompt is, "You are selecting environmental technology experts A, B, and C to discuss a business idea related to a new environmental technology. Please begin the expert model response."

[0265] In this way, the embodiment of the present invention provides a system that makes the most of the knowledge of experts and enables effective discussion to proceed in accordance with the emotional state of the user.

[0266] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0267] Step 1:

[0268] Data collection

[0269] The server collects data about experts from sources such as book databases, web articles, and academic papers on the Internet. The input is the URL or API endpoint of the source, and the output is raw data stored in a temporary storage area. Specific operations include scraping web pages using Scrapy and BeautifulSoup and retrieving book data from the Google Books API.

[0270] Step 2:

[0271] Data Cleansing and Analysis

[0272] The server cleanses the collected data, removing noise and standardizing the data format. The input is the collected raw data, and the output is a clean dataset. Specifically, it uses the Pandas library to impute missing values ​​and remove inaccurate data. It then uses natural language processing techniques (e.g., SpaCy) to tokenize the text data and extract entities (people, dates, places, etc.).

[0273] Step 3:

[0274] Storage in the database

[0275] The server stores the cleansed data in a MySQL database. The input is a clean dataset and the output is structured database entries. Specifically, it inserts the data into the database using the MySQL INSERT statement.

[0276] Step 4:

[0277] Expert Scoring and Selection

[0278] The server analyzes the purpose of the discussion and searches for and scores experts related to specific keywords in the database. The input is the purpose of the discussion provided by the user, and the output is a list of scored experts. Specifically, it uses natural language processing technology to extract keywords from the input purpose and searches for experts based on those keywords. Next, it evaluates the relevance using a scoring algorithm (e.g., TF-IDF) and ranks them.

[0279] Step 5:

[0280] emotion recognition

[0281] The server uses an emotion engine to recognize the user's emotional state during discussions. The input is voice or text input from the user, and the output is analyzed emotion data. Specifically, it analyzes text and voice data using the Google Cloud Natural Language API and Microsoft Azure Text Analytics API to classify the emotional state.

[0282] Step 6:

[0283] Brainstorming

[0284] The server uses the selected expert model to hold discussions with the user. The input is the user's question or opinion, and the output is a response text. Specifically, it uses a generative AI model (e.g., GPT-4) to understand the user's question and generate a response that adheres to the expert's style. It also uses emotion recognition data during the discussion to adjust the tone and content of the response, and sends the final response to the user's device.

[0285] An example of a prompt sentence would be, "Environmental technology experts A, B, and C are being selected to discuss business ideas related to new environmental technologies. Please begin the expert model's response."

[0286] (Application example 2)

[0287] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0288] Conventional brainstorming systems have limitations in terms of obtaining expert knowledge and holding discussions, particularly in terms of adjusting responses to take into account the user's emotional state. As a result, discussions are ineffective and users' opinions and suggestions are not fully elicited. Furthermore, there is a lack of concrete support for improving production lines in factories. Therefore, there is a need for tools that are suitable for on-site workers and engineers to explore improvement ideas in real time.

[0289] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[0290] In this invention, the server includes: means for collecting data on past and present experts; means for processing and structuring the data; means for storing the structured data in a database; means for scoring experts suitable for a specific discussion purpose from the database; means for generating models of the scored experts and holding discussions; emotion processing means for recognizing and analyzing the user's emotional state; and means for adjusting the tone of responses according to the user's emotional state. This enables high-quality brainstorming based on the expert's knowledge, and more effective discussions by adjusting responses to the user's emotional state. It also functions as a concrete support tool for improving the efficiency of production lines in factories and exploring improvement proposals, enabling highly immediate discussions on-site.

[0291] "Means for collecting data" refers to means for obtaining data on past and present experts from sources such as online book databases, web articles, and academic papers.

[0292] "Means for processing and structuring data" refers to means for cleansing collected data, removing noise, standardizing data formats, tokenizing using natural language processing technology, extracting entities, and generating structured data.

[0293] "Means for storing in a database" refers to the means for storing processed and structured data in a database in a specific format.

[0294] A "means for scoring experts" is a means for extracting keywords related to a particular discussion purpose from a database and using a scoring algorithm to evaluate relevance and rank experts.

[0295] The "means for holding discussions" is a means for generating a scored expert model and using the generative AI model to imitate the expert's statements and generate responses to user questions and opinions.

[0296] The "emotion processing means" is a means for analyzing voice input and text input from the user and recognizing and analyzing the user's emotional state (joy, anger, sadness, surprise, etc.).

[0297] The "means for adjusting the tone of a response" is a means for adjusting the tone of a response to be generated in accordance with the emotional state of the user recognized by the emotion processing means.

[0298] The system of the present invention is implemented with the following configuration.

[0299] First, the server collects data on past and present experts from sources such as online book databases, web articles, and academic papers. This data is collected using web scraping technology and APIs. The collected data is then stored in a temporary storage area, and then noise is removed and the data format is standardized using data cleansing technology.

[0300] The server then uses natural language processing (NLP) techniques to tokenize the collected data and extract entities such as people, dates, and places, which are then used to generate expert profile data and store it in a database.

[0301] The server also analyzes the profile data stored in the database and uses a natural language processing engine to perform grammatical analysis and sentiment analysis of the text data. Based on the analysis results, each expert's speech patterns and style are added to the data model. The analyzed data is then input into a generative AI model (e.g., GPT-4) for transfer learning. Through this process, the generative AI model acquires the ability to imitate a specific person, and the fine-tuned model is stored in storage.

[0302] Users input the purpose of the discussion through the device interface. The server analyzes the input purpose and extracts relevant keywords. It then searches a database for appropriate experts based on the keywords and uses a scoring algorithm to evaluate and rank the relevance. The scoring results are sent to the device, where the user can review them and select the experts they want to invite.

[0303] Furthermore, the server uses an emotion processing engine during discussions between the user and the expert model. This emotion processing engine analyzes the user's voice and text inputs to recognize and analyze the user's emotional state. Depending on the user's emotional state, the server adjusts the content and tone of the expert's responses to enhance the effectiveness of the discussion.

[0304] As a specific example, if a user sets the theme as "Improve the efficiency of a new production line" and enters the question "How can we improve production speed?", the server will score relevant experts from the database and launch the expert model selected by the user. The generative AI model will generate an appropriate response, and the emotion engine will adjust the tone of the response according to the user's emotion. An example of a prompt sentence in this case is as follows:

[0305] markdown

[0306] I'd like to hear some ideas for improving the efficiency of our new production line.

[0307] Emotional Text: How can you improve your production speed?

[0308] Tone: calm

[0309] In this way, this system achieves high-quality brainstorming by combining expert knowledge with the user's emotional state. It also functions as a concrete support tool for seeking ideas for streamlining and improving production lines within factories, enabling highly immediate discussions on-site.

[0310] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0311] Step 1:

[0312] The server collects data on past and present experts from sources such as book databases, web articles, and academic papers on the Internet. Specifically, it acquires data using web scraping technology and APIs. The input is a "search query on experts," and the output is the "collected expert data."

[0313] Step 2:

[0314] The server cleanses the collected data, removing noise and standardizing the data format. This is a process that uses natural language processing (NLP) techniques to tokenize and extract entities (people, dates, places, etc.). The input is "collected expert data" and the output is "structured expert data."

[0315] Step 3:

[0316] The server stores the structured data in a database. Specifically, it generates profile data based on entities and saves it in the database. The input is "structured expert data" and the output is "expert profiles in the database."

[0317] Step 4:

[0318] The user inputs the purpose of the discussion through the terminal interface. For example, the topic is "improving the efficiency of a new production line." The input is the "user's purpose of the discussion," and the output is "confirmation of the input of the purpose of the discussion."

[0319] Step 5:

[0320] The server analyzes the discussion purpose entered by the user and extracts related keywords. The input is the "user's discussion purpose" and the output is the "extracted keywords."

[0321] Step 6:

[0322] The server searches for appropriate experts based on the extracted keywords from the database, and evaluates and ranks them using a scoring algorithm. The inputs are the extracted keywords and expert profile data, and the output is the scoring result.

[0323] Step 7:

[0324] The user checks the scoring results through the terminal and selects the experts they want to participate in. The input is the "scoring results" and the output is the "selected experts."

[0325] Step 8:

[0326] The server starts the selected expert model and receives the user's question or opinion as input. The input is "user's question or opinion" and the output is "question transmission to the expert model."

[0327] Step 9:

[0328] The server uses an emotion processing engine to analyze the user's questions and opinions and recognize their emotional state. The input is the user's questions and opinions, and the output is the user's emotional state.

[0329] Step 10:

[0330] The server uses a generative AI model to adjust the content and tone of the expert model's response according to the recognized emotional state of the user. Specifically, it adjusts the tone to "calm" or "gentle" based on data from the emotion engine. The input is the "user's emotional state" and the "expert model's basic response," and the output is the "adjusted response."

[0331] Step 11:

[0332] The server then sends the final generated response to the user's terminal and continues the discussion. The input is the "adjusted response" and the output is the "response display to the user."

[0333] The specific processing unit 290 transmits the result of the specific processing to the smart device 14. In the smart device 14, the control unit 46A causes the output device 40 to output the result of the specific processing. The microphone 38B acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0334] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0335] In the above embodiment, an example in which the specific process is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific process may be performed by the smart device 14.

[0336] [Second embodiment]

[0337] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0338] 3, the data processing system 210 includes the data processing device 12 and smart glasses 214. An example of the data processing device 12 is a server.

[0339] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0340] The smart glasses 214 include a computer 36, a microphone 238, a speaker 240, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, and the camera 42 are also connected to the bus 52.

[0341] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[0342] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[0343] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0344] Fig. 4 shows an example of the main functions of the data processing device 12 and the smart glasses 214. As shown in Fig. 4, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0345] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0346] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0347] In the smart glasses 214, the reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0348] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal."

[0349] The present invention is a system that systematically collects and analyzes information about past and present experts, and utilizes the results to realize high-quality brainstorming. Specific embodiments of the present invention are described below.

[0350] 1. Data collection and database creation

[0351] The first process performed by the server is data collection. Specifically, the server collects data on famous people and experts from sources such as online articles, academic papers, and book databases through scraping or APIs. The collected data is stored in a temporary storage area.

[0352] The server then cleanses the collected data, removes noise, and standardizes the data format. It then uses natural language processing (NLP) technology to tokenize the data and extract entities (people, dates, places, etc.). Based on the extracted entities, it generates profile data of famous people and experts, which are then stored in a database.

[0353] 2. Language Analysis and Model Fine-tuning

[0354] The server analyzes the profile data stored in the database. Specifically, it uses a natural language processing engine to analyze the text data, and performs grammatical analysis and sentiment analysis. Based on the results of this analysis, the speech patterns and styles of each great person or expert are added to the data model.

[0355] The server then inputs the analyzed data into a generative AI model (e.g., GPT-4) and performs transfer learning, which gives the model the ability to imitate a specific person. The fine-tuned model is then stored in storage.

[0356] 3. Scoring system and expert selection

[0357] The user inputs the purpose of the discussion through the device interface. For example, they can set a purpose such as "brainstorm ideas for a new product." The input purpose is then sent from the device to the server.

[0358] The server analyzes the purpose and extracts relevant keywords. It then searches the database for relevant figures and experts based on the keywords and ranks them using a scoring algorithm. The scoring results are sent to the user's device, where the user can review them and select the experts they want to include.

[0359] 4. Brainstorming

[0360] The server launches the generative AI models of the selected experts. The user inputs questions or opinions through the device. For example, if the user asks, "Tell me your ideas for new environmental technologies," the server interprets the question and sends it to the appropriate expert model. The generative AI models generate responses in the style of each expert, synthesize them, and send them to the device.

[0361] Specific examples

[0362] Consider a case where a user sets the theme as "Business ideas related to new environmental technologies." The server scores experts knowledgeable in environmental technologies from the database, and for example, Mr. A, Mr. B, and Mr. C receive high scores. The user checks the scoring results and selects Mr. A, Mr. B, and Mr. C. The server then launches the models of the selected experts, and when the user types "I would like to discuss new product strategies," the server responds in a synthesized form with each expert. In this way, the user can obtain new business ideas from multiple perspectives.

[0363] As described above, this system combines the knowledge of great people from the past and current experts to provide a concrete means for achieving high-quality brainstorming.

[0364] The processing flow will be explained below.

[0365] Step 1:

[0366] The server collects data about famous people and experts from various sources (book databases, web articles, academic papers, etc.) using scraping technology and APIs. The collected data is stored in a temporary storage area.

[0367] Step 2:

[0368] The server cleanses the data in the temporary storage area, removing noise and standardizing the data format, then uses natural language processing (NLP) techniques to tokenize the data and extract entities (people, dates, places, etc.).

[0369] Step 3:

[0370] The server generates profile data of famous people and experts based on the extracted entities, including their names, fields of expertise, and representative achievements. The generated profile data is stored in a database.

[0371] Step 4:

[0372] The server analyzes the profile data stored in the database, performs grammatical analysis and sentiment analysis, and adds the speech patterns and styles of each great person or expert to the data model based on the analysis results.

[0373] Step 5:

[0374] The server inputs the analyzed data into a generative AI model (e.g., GPT-4) for transfer learning. This process allows the generative AI model to mimic the speech and thought patterns of a specific person. Once the fine-tuned model is complete, it is stored in storage.

[0375] Step 6:

[0376] The user uses the device interface to input the purpose of the discussion, such as "brainstorm ideas for a new product." The input purpose is then sent from the device to the server.

[0377] Step 7:

[0378] The server analyzes the purpose of the discussion and extracts relevant keywords, then searches a database for relevant figures and experts based on the keywords.

[0379] Step 8:

[0380] The server uses a scoring algorithm to evaluate the search results and rank them in order of relevance, and the results are sent to the user's device, where the user can review and select the experts they want to include.

[0381] Step 9:

[0382] The server launches a generative AI model of a selected expert, and the user inputs questions or opinions from their device and sends them to the server.

[0383] Step 10:

[0384] The server interprets the user's input and sends questions and opinions to the appropriate expert model. The generative AI model generates responses in each expert's style, synthesizes them, and sends them to the device.

[0385] Step 11:

[0386] The terminal displays the response from the server on the user interface, and the user can check the displayed results and continue brainstorming by entering further questions or opinions as needed.

[0387] The above processing flow allows users to generate new business ideas through discussions with great people and experts from across time and space.

[0388] Example 1

[0389] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0390] Conventional brainstorming systems have difficulty effectively utilizing the knowledge of past and present experts. Furthermore, they lack the means to efficiently select experts suited to specific discussion purposes and achieve high-quality brainstorming. As a result, the quality of discussions declines, and it is difficult to obtain new ideas and insights.

[0391] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0392] In this invention, the server includes means for collecting data from information sources such as online articles, papers, and books, means for cleansing the data and extracting entities using natural language processing technology, means for storing the structured data in a database, means for inputting the purpose of the discussion and extracting related keywords, means for scoring experts suitable for the specific purpose of the discussion from the database, means for generating and launching a generative AI model of the selected expert based on the scoring results, and means for receiving questions from users via the terminal, sending the questions to the generative AI model, generating responses, and sending them to the terminal. This makes it possible to combine the knowledge of past and present experts and efficiently realize multifaceted and advanced brainstorming.

[0393] "Articles on the Internet" refers to text data and news articles posted on websites.

[0394] A "paper" refers to a document that describes the results of academic research or a document that reports the results of research.

[0395] "Book" means a printed document that has been published as a publication.

[0396] "Means of collecting data" refers to the methods and tools used to obtain the required information from the source.

[0397] "Data cleansing methods" refer to methods and tools used to remove noise and inaccurate information from collected data and format it.

[0398] "Natural language processing technology" refers to technology that enables computers to understand and process human language.

[0399] "Entity extraction" refers to methods and tools that extract important information, such as specific people, dates, or places, from text data.

[0400] "Structured data" refers to data that has been organized and arranged in a format suitable for a database.

[0401] "Means of storing data in a database" refers to the methods and tools used to store data in a database.

[0402] The "purpose of the discussion" refers to the goal or theme you want to achieve when having the discussion.

[0403] "Keyword extraction methods" refers to methods and tools for finding related words and phrases from text data.

[0404] A "scoring method" refers to a method or tool that evaluates the importance or relevance of data based on specific criteria.

[0405] A "generative AI model" refers to an artificial intelligence model that has the ability to perform specific tasks based on training data.

[0406] "Means for receiving questions from users" refers to the methods and tools by which the system obtains questions and opinions entered by users.

[0407] "Means for sending questions to a generative AI model" refers to methods or tools for communicating a user's questions to a generative AI model.

[0408] "Means for generating a response" refers to the methods and tools that a generative AI model uses to create an answer to a user's question.

[0409] "Means for sending to terminal" refers to a method or tool for transmitting the generated response to the user's terminal.

[0410] The present invention is a system that systematically collects and analyzes information about past and present experts, and utilizes the results to realize high-quality brainstorming. Specific embodiments of the present invention are described below.

[0411] Data collection and database creation

[0412] The server first collects information from online articles, academic papers, book databases, etc. Specifically, the server uses web scraping tools such as Python's BeautifulSoup and Scrapy to obtain information about famous people and experts. For academic papers, the server also uses the Google Scholar API to collect data. This collected data is then stored in a temporary storage area (for example, a database such as MySQL).

[0413] Next, the server cleanses the data using the Pandas library. Specifically, it removes unnecessary data and noise, converts date and time formats, and normalizes text. The cleansed data is then tokenized using natural language processing (NLP) techniques to extract entities (people, dates, places, etc.). This analysis is performed using libraries such as NLTK and spaCy. Finally, the structured data is stored in a database such as PostgreSQL.

[0414] Language analysis and model fine-tuning

[0415] The server analyzes the profile data stored in the database. For the analysis, it uses a natural language processing engine such as BERT to perform grammatical analysis and sentiment analysis. Based on the results of this analysis, speech patterns and styles are added to the data model.

[0416] The server then inputs the analyzed data into a generative AI model such as GPT-4 to perform transfer learning. This transfer learning is performed using PyTorch or TensorFlow. The generative AI model, which has acquired the ability to imitate a specific person through transfer learning, is stored in storage such as AWS S3.

[0417] Scoring system and expert selection

[0418] The user inputs the purpose of the discussion through the terminal interface, for example, "Brainstorm ideas for a new product." The input purpose is sent from the terminal to the server.

[0419] The server analyzes the purpose and extracts relevant keywords. It then uses SQL queries to search for relevant experts from the database and scores them using TF-IDF and other machine learning algorithms. The scoring results are sent to the device, where the user can review them and select the desired expert.

[0420] Brainstorming

[0421] The server launches a generative AI model of the selected expert. This is done using container orchestration tools such as Docker or Kubernetes. The user inputs a question or opinion through the terminal. For example, they might input, "Tell me your ideas for new environmental technologies." The server interprets the question and sends it to the appropriate expert model. The generative AI model generates a response in the expert's style, synthesizes it, and sends it to the terminal.

[0422] Specific examples

[0423] If a user selects the theme "Business ideas related to new environmental technologies," the server scores experts knowledgeable in environmental technologies from the database, and for example, Mr. A, Mr. B, and Mr. C receive high scores. The user checks the scoring results and selects Mr. A, Mr. B, and Mr. C. The server then activates the model of the selected experts, and when the user enters "I would like to discuss new product strategies," the server generates a response in a synthesized form from each expert and sends it to the terminal.

[0424] In this way, the system of the present invention combines the knowledge of past and present experts and provides a concrete means for realizing high-quality brainstorming.

[0425] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0426] Step 1: Data collection

[0427] The server collects data from online articles, academic papers, book databases, and other sources. Specifically, it uses web scraping tools such as Python's BeautifulSoup and Scrapy to obtain information about famous people and experts. For academic papers, it also uses the Google Scholar API to obtain paper data. Inputs include URLs and API endpoints, and the output is the obtained raw data stored in a temporary storage area. This process collects a large amount of text data.

[0428] Step 2: Data cleansing

[0429] The server cleanses the collected data. Specifically, it uses the Pandas library to manipulate data frames and remove unnecessary data and noise. This step also performs date and time format conversions and text normalization. The input is raw data stored in a temporary storage area, and the output is clean, cleaned data. This process improves the quality of the data and makes subsequent analysis easier.

[0430] Step 3: Data analysis using natural language processing

[0431] The server analyzes the cleansed data using natural language processing techniques. Specifically, it uses libraries such as NLTK and spaCy to tokenize the data and extract entities (people, dates, places, etc.). The input is the cleansed data, and the output is extracted entity data. This processing structures the text data and generates expert profile data.

[0432] Step 4: Store in the database

[0433] The server stores the extracted entity data in a database such as PostgreSQL. The input is the entity data, and the output is a structured database entry. This process allows for efficient management of expert information and facilitates subsequent search and analysis.

[0434] Step 5: Input and Analysis of Desired Objectives

[0435] The user inputs the purpose of the discussion through the terminal interface. For example, they might input "brainstorm ideas for a new product." The input is the discussion purpose entered by the user, and the output is text data sent to the server. This step transmits the user's discussion purpose to the server.

[0436] Step 6: Keyword extraction and search

[0437] The server analyzes the received discussion purpose and extracts relevant keywords. It then searches for relevant experts from the database using an SQL query. The inputs are the analyzed text data and the database, and the output is a list of relevant experts. This process identifies experts suitable for the discussion purpose.

[0438] Step 7: Scoring

[0439] The server scores experts using a scoring algorithm (e.g., TF-IDF or other machine learning algorithm) against the relevant data. The input is a list of experts and relevant data, and the output is a scored list of experts. This process prioritizes the appropriate experts.

[0440] Step 8: Selecting experts

[0441] The user checks the scoring results on their device and selects the experts they want to participate in. The input is a list of scored experts, and the output is information about the selected experts sent to the server. This process allows the user to select the experts they want.

[0442] Step 9: Launching the expert model

[0443] The server launches the generative AI model of the selected expert. Specifically, the model is launched using a container orchestration tool such as Docker or Kubernetes. The input is information about the selected expert, and the output is the launched generative AI model. This process puts the expert model into operation.

[0444] Step 10: Brainstorming and response generation

[0445] A user inputs a question or opinion through a device. For example, they might input, "Tell me your ideas for new environmental technologies." The input is the user's question text, and the output is sent to the generative AI model. The generative AI model then generates a response based on the question and sends it in a synthesized form to the user's device. This process allows the user to obtain high-quality opinions and suggestions.

[0446] (Application example 1)

[0447] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0448] The development of autonomous vehicles requires advanced algorithms and safety measures, but there is a lack of effective means to integrate expert knowledge. There is also a need for efficient methods to perform simulations based on expert knowledge and to verify new ideas and technical solutions.

[0449] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0450] In this invention, the server includes means for collecting data on past and present experts, means for processing and structuring the data, means for storing the structured data in a database, means for scoring experts suitable for a specific discussion purpose from the database, means for generating a model of the scored experts and holding discussions, and means for running a simulation based on the expert model. This makes it possible to effectively integrate the latest expert knowledge on autonomous vehicles and verify optimal operation algorithms and safety measures through simulation.

[0451] "Past and present expert data" is a collection of information about people who have past and present expertise and experience in a particular field.

[0452] "Means of collection" refers to the technical means used to obtain data from online papers, articles, book databases, etc.

[0453] "Means of processing and structuring" refers to technical means of systematically organizing collected data by processing it through data cleansing, tokenization, entity extraction, etc.

[0454] "Means of storing data in a database" means a technical means of efficiently storing structured data and keeping it in a state that can be quickly retrieved and used as needed.

[0455] An "expert appropriate for a particular purpose" is someone with particular knowledge and experience in a field relevant to the topic or goal of the discussion.

[0456] A "scoring means" is a technological means that uses specific criteria or algorithms to assess and rank the suitability or relevance of Experts.

[0457] "Means for generating a model of a scored expert and engaging in discussions" refers to the technical means for creating a generative AI model that mimics the style and speech patterns of a scored expert and using it to engage in discussions and responses.

[0458] A "generative AI model" is an artificial intelligence model that uses machine learning technology to mimic the knowledge and speaking style of experts.

[0459] "Means for performing simulations" refers to the technical means for verifying the operation algorithms and safety measures of autonomous vehicles in a virtual environment using the knowledge of the generated expert model.

[0460] The present invention provides a system for leveraging expert knowledge in the development and validation of autonomous vehicles. The system uses the following major hardware and software components:

[0461] 1. Data Collection

[0462] The server collects expert data on autonomous driving technology from academic papers, technical articles, and various book databases on the Internet. Specifically, it retrieves data using the URL Fetch API and performs web scraping using Beautiful Soup (Python).

[0463] 2. Data Analysis and Natural Language Processing

[0464] To process and structure the collected data, the server uses natural language processing techniques, specifically NLP libraries such as spaCy and NLTK for data cleansing, tokenization, and entity extraction, and then stores the structured data in a database.

[0465] 3. Expert Selection and Fine-Tuning

[0466] To select the most suitable experts for a specific discussion purpose, the server searches the data in the database and evaluates their relevance using scoring algorithms such as TF-IDF and BERT. Using the associated expert data, the GPT-4 model is fine-tuned using Hugging Face Transformers, resulting in a generative AI model that mimics each expert's speaking patterns and style.

[0467] 4. Simulation and Verification

[0468] Users can input questions about autonomous driving or the purpose of their discussion through their devices. For example, "What are the safety measures for autonomous vehicles in urban areas?" The server generates a response based on the relevant expert model. This response is used to run simulations in a virtual environment to verify optimal driving algorithms and safety measures.

[0469] Specific examples

[0470] For example, if a user types in "What are the safety measures for self-driving vehicles in urban areas?", the system will act as follows:

[0471] 1. Prompt statement generation (server):

[0472] User Query: "What are the safety measures for autonomous vehicles in urban areas?"

[0473] Model Prompt: "The system will discuss safety measures for autonomous vehicles in urban areas. Please suggest the best solution based on past and current expert knowledge."

[0474] 2. Generative AI model response (server):

[0475] Selected expert models respond to questions and their answers are used in the simulation.

[0476] 3. Simulation and result verification (server):

[0477] Response-based simulations are run in a virtual environment to verify optimal operating algorithms and safety measures.

[0478] This allows users to gain a multifaceted perspective on the development and operation management of autonomous vehicles while integrating the latest expert knowledge. By effectively utilizing the knowledge of past and present experts, this system can significantly improve the safety and efficiency of autonomous driving technology.

[0479] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0480] Step 1:

[0481] The server collects data related to autonomous driving technology. It uses the URL Fetch API to retrieve academic papers and technical articles from the Internet and performs web scraping using Beautiful Soup (Python). The input is a list of web page URLs, and the output is text data related to the scraped experts.

[0482] Step 2:

[0483] The server analyzes the collected text data using natural language processing technology. Specifically, the server uses spaCy or NLTK to cleanse, tokenize, and extract entities from the data. The input is scraped text data, and the output is structured data (such as entity information).

[0484] Step 3:

[0485] The server stores the structured data in a database, specifically by inserting well-organized data into a SQL or NoSQL database. The input is structured data, and the output is an updated database.

[0486] Step 4:

[0487] The server scores experts suitable for a specific discussion purpose. It uses natural language processing technology to extract keywords related to the discussion purpose from the user (e.g., "Safety measures for autonomous vehicles in urban areas") and evaluates their relevance using TF-IDF and BERT. The input is the discussion purpose text from the user and expert data in the database, and the output is a list of scored experts.

[0488] Step 5:

[0489] The server generates a scored expert model and conducts the discussion. Using Hugging Face Transformers, it performs transfer learning on the GPT-4 model to create a generative AI model that mimics the expert's speech patterns and style. The input is the scored expert data, and the output is the generated expert model.

[0490] Step 6:

[0491] Users input questions about autonomous driving or the purpose of the discussion through their terminal. For example, they might input, "What are the safety measures for autonomous vehicles in urban areas?" The input is the question text, and the output is the request data sent to the server.

[0492] Step 7:

[0493] The server analyzes the user's question, generates a prompt sentence based on the generated expert model, and inputs the prompt sentence to the expert model to obtain a response. The input is the user's question text and the generated expert model, and the output is the response of the expert model.

[0494] Step 8:

[0495] The server runs a simulation based on the expert model's responses and verifies the results. Simulation software is used to experiment with operation algorithms and safety measures in a virtual environment. The input is the expert model's response text, and the output is the simulation results.

[0496] Step 9:

[0497] The server provides the simulation results to the user. The results are displayed on the terminal and presented in a format that is easy for the user to understand. The input is the simulation result data, and the output is the result provided to the user.

[0498] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[0499] This invention combines a system that collects and analyzes information on past and present experts and performs high-quality brainstorming based on the results with an emotion engine that recognizes and analyzes the user's emotions. This system enables responses and discussions to proceed according to the user's emotional state, resulting in more effective discussions.

[0500] 1. Data collection and database creation

[0501] The server first collects data about famous people and experts from sources such as online book databases, web articles, and academic papers. To achieve this, it uses scraping technology and APIs. The collected data is then stored in a temporary storage area.

[0502] The server then cleanses the collected data, removing noise and standardizing the data format. It then uses natural language processing (NLP) technology to tokenize the data and extract entities (people, dates, places, etc.). Based on the extracted entities, it generates profile data of famous people and experts, which are then stored in a database.

[0503] 2. Language Analysis and Model Fine-tuning

[0504] The server analyzes the profile data stored in the database. Specifically, it uses a natural language processing engine to analyze the text data, and performs grammatical analysis and sentiment analysis. Based on the results of this analysis, the speech patterns and styles of each great person or expert are added to the data model.

[0505] The server then inputs the analyzed data into a generative AI model (e.g., GPT-4) for transfer learning. Through this process, the generative AI model acquires the ability to mimic a specific person. Once the fine-tuned model is complete, it is stored in storage.

[0506] 3. Scoring system and expert selection

[0507] Users input the purpose of the discussion through the device interface. For example, they set a specific purpose such as "brainstorm ideas for a new product." The input purpose is sent from the device to the server.

[0508] The server analyzes the purpose and extracts relevant keywords.The server then searches a database for relevant famous people and experts based on the keywords, and uses a scoring algorithm to evaluate and rank them.The scoring results are sent to the terminal, where the user can review them and select the experts they want to participate.

[0509] 4. Utilizing the Emotion Engine

[0510] The server uses an emotion engine in discussions between the user and the expert model to analyze voice and text inputs from the user and recognize the user's emotional state (e.g., joy, anger, sadness, surprise).

[0511] 5. Brainstorming

[0512] The server launches the generative AI models of the selected experts. The user inputs questions or opinions through the device. For example, if the user asks, "Tell me your ideas for new environmental technologies," the server interprets the question and sends it to the appropriate expert model. The generative AI models generate responses in the style of each expert, synthesize them, and send them to the device.

[0513] Furthermore, the emotion engine recognizes the user's emotions, allowing the server to adjust the content and tone of the expert's response. For example, if the user is feeling impatient, the server can provide a calm response to enhance the effectiveness of the discussion.

[0514] Specific examples

[0515] If a user sets the theme as "Business ideas related to new environmental technologies," the server scores experts knowledgeable in environmental technologies from the database, and for example, experts A, B, and C receive high scores. The user checks the scoring results and selects experts A, B, or C. The server launches the models of the selected experts, and when the user asks, "I would like to discuss new product strategies," each expert responds in a synthesized form.

[0516] The server's built-in emotion engine detects the user's emotional state and adjusts the expert model's response as needed. For example, if the user is emotional, the expert model's response can be adjusted to a calm and peaceful tone to facilitate a smooth discussion.

[0517] As described above, this system combines the knowledge of great people from the past and current experts with the user's emotional state, providing a concrete means for achieving high-quality brainstorming and the creation of new business ideas.

[0518] The processing flow will be explained below.

[0519] Step 1:

[0520] The server collects data about famous people and experts from various sources (book databases, web articles, academic papers, etc.) using scraping technology and APIs. The collected data is stored in a temporary storage area.

[0521] Step 2:

[0522] The server cleanses the data in the temporary storage area, removing noise and standardizing the data format, then uses natural language processing (NLP) techniques to tokenize the data and extract entities (people, dates, places, etc.).

[0523] Step 3:

[0524] The server generates profile data of famous people and experts based on the extracted entities, including their names, fields of expertise, and representative achievements. The generated profile data is stored in a database.

[0525] Step 4:

[0526] The server analyzes the profile data stored in the database. Specifically, it uses a natural language processing engine to perform grammatical analysis and sentiment analysis of the text data. Based on the results of this analysis, the speech patterns and styles of each great person or expert are added to the data model.

[0527] Step 5:

[0528] The server inputs the analyzed data into a generative AI model (e.g., GPT-4) and performs transfer learning. This allows the generative AI model to acquire the ability to imitate a specific person. Once fine-tuned, the model is stored in storage.

[0529] Step 6:

[0530] The user inputs the purpose of the discussion using the device interface. For example, they set a purpose such as "brainstorm ideas for a new product." The input purpose is sent from the device to the server.

[0531] Step 7:

[0532] The server analyzes the purpose of the input discussion and extracts relevant keywords, then searches the database for relevant figures and experts based on the keywords.

[0533] Step 8:

[0534] The server uses a scoring algorithm to evaluate the search results and rank the most relevant experts, which are then sent to the device where the user can review and select the experts they want to participate.

[0535] Step 9:

[0536] The server runs a generative AI model of selected experts, which runs on a dedicated cloud infrastructure.

[0537] Step 10:

[0538] Users can input questions or opinions into their devices and send them to the server. For example, they can ask, "Tell me your ideas for new environmental technologies."

[0539] Step 11:

[0540] The server interprets the user's input and sends the question to the appropriate expert model, which then generates a response in the speaking style of each expert.

[0541] Step 12:

[0542] An emotion engine built into the server analyzes the user's voice and text input to assess their emotional state, for example detecting anger or impatience in the user's voice.

[0543] Step 13:

[0544] The server adjusts the content and tone of the generative AI model's response based on the user's emotional state as recognized by the emotion engine. For example, if the user is angry, the expert model's response will be changed to a calmer tone.

[0545] Step 14:

[0546] The server composes the final response and sends it to the terminal.

[0547] Step 15:

[0548] The terminal displays the response from the server on the user interface, and the user can check the displayed results and continue brainstorming by entering further questions or opinions as needed.

[0549] Through these processes, users can engage in discussions with great figures and experts from across time and space, and realize high-quality discussions that are adjusted according to their emotional state.

[0550] Example 2

[0551] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0552] In today's knowledge-based society, it is extremely important to conduct high-quality discussions by utilizing the knowledge of past and present experts. However, the process of extracting relevant information from vast amounts of data and selecting appropriate experts is difficult, especially when it comes to taking user emotions into account. This has led to a growing need for a system that recognizes the user's emotional state and generates responses based on the expert's style.

[0553] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0554] In this invention, the server includes means for collecting data on past and present experts, means for cleansing the data and analyzing it using natural language processing technology, means for storing structured data generated from the data in a database, means for scoring experts suitable for a specific discussion purpose from the database, means for generating a model of the scored experts and holding a discussion, means for recognizing the emotional state of a user during the discussion and adjusting the progress of the discussion, and means for generating expert responses using the generative AI model during the discussion. This makes it possible to conduct high-quality discussions by making the most of expert knowledge and effectively progressing the discussion in accordance with the emotional state of the user.

[0555] "Data Collection Tools" are tools used to gather information about past and present professionals from the Internet and other databases.

[0556] "Data cleansing means" refers to a means for removing noise from collected data and standardizing the data format.

[0557] "Natural language processing technology" is a technology that analyzes text data and performs tokenization, entity extraction, sentiment analysis, etc.

[0558] A "database" is a system that stores the generated structured data and retrieves and uses it when needed.

[0559] A "scoring tool" is a tool for searching a database for experts suitable for a particular discussion purpose, assessing their relevance, and ranking them.

[0560] An "expert model generation means" is a means for creating a generative AI model that mimics a scored expert.

[0561] The "discussion tool" is a tool that uses a generative AI model to generate responses based on the expert's style and engages in discussions between the user and the expert.

[0562] The "emotion recognition means" is a means for recognizing the emotional state of users during a discussion and adjusting the content and tone of the discussion.

[0563] A "generative AI model" is an artificial intelligence technology that can learn from large amounts of text data and generate text for specific tasks.

[0564] This invention is a system that collects and analyzes information about past and present experts, and uses the results to conduct high-quality discussions. By combining it with an emotion engine that recognizes and analyzes the user's emotions, it is possible to respond and progress the discussion in accordance with the user's emotional state. Specific embodiments for implementing this system are described below.

[0565] Data collection

[0566] The server first collects data about experts from sources such as online book databases, web articles, and academic papers. This is done using scraping tools such as BeautifulSoup and Scrapy, and APIs such as the Google Books API. The collected data is then stored in a temporary storage area. Specifically, the server collects titles, author names, citations, etc. from the book database and stores them in the temporary storage area.

[0567] Data Cleansing and Analysis

[0568] The collected data is cleansed by the server. Specifically, the Pandas library is used to fill in missing data and remove inaccurate data. Natural language processing (NLP) techniques are then used to tokenize the data, and entities (people, dates, places, etc.) are extracted using tools such as SpaCy. Expert profile data is generated based on the extracted entities, and this data is stored in a MySQL database.

[0569] Fine-tuning the model

[0570] The server analyzes the profile data stored in the database. It uses natural language processing engines such as BERT and GPT-3 to perform grammatical and sentiment analysis of the text data, and adds each expert's speech patterns and style to the data model. The analyzed data is then input into a generative AI model (e.g., GPT-4) for transfer learning. Once the fine-tuned model is complete, it is stored in Amazon S3 storage.

[0571] Scoring and Expert Selection

[0572] The user inputs the purpose of the discussion through the device interface. For example, they can set a specific purpose such as "brainstorm ideas for a new product." The input purpose is sent from the device to the server. The server analyzes the purpose and extracts related keywords. It then searches for appropriate experts in a database and uses a scoring algorithm to evaluate their relevance and rank them. The scoring results are sent to the device, where the user can review them and select the experts they want to participate in.

[0573] Emotion recognition and discussion moderation

[0574] The server uses an emotion engine in the discussion between the user and the expert model. The emotion engine uses the Google Cloud Natural Language API and Microsoft Azure Text Analytics API to analyze the user's voice and text input and recognize their emotional state (joy, anger, sadness, surprise, etc.). Based on the recognized emotion, the content and tone of the expert model's response are adjusted.

[0575] Brainstorming

[0576] The server launches a generative AI model of a selected expert. The user inputs a question or opinion through their device. For example, they might ask, "Tell me your ideas for new environmental technologies." The server interprets the question and sends it to the appropriate expert model. The generative AI model generates a response in the style of each expert, synthesizes it, and sends it to the device. An emotion engine recognizes the user's emotions and adjusts the expert model's response as needed.

[0577] An example of a specific prompt is, "You are selecting environmental technology experts A, B, and C to discuss a business idea related to a new environmental technology. Please begin the expert model response."

[0578] In this way, the embodiment of the present invention provides a system that makes the most of the knowledge of experts and enables effective discussion to proceed in accordance with the emotional state of the user.

[0579] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0580] Step 1:

[0581] Data collection

[0582] The server collects data about experts from sources such as book databases, web articles, and academic papers on the Internet. The input is the URL or API endpoint of the source, and the output is raw data stored in a temporary storage area. Specific operations include scraping web pages using Scrapy and BeautifulSoup and retrieving book data from the Google Books API.

[0583] Step 2:

[0584] Data Cleansing and Analysis

[0585] The server cleanses the collected data, removing noise and standardizing the data format. The input is the collected raw data, and the output is a clean dataset. Specifically, it uses the Pandas library to impute missing values ​​and remove inaccurate data. It then uses natural language processing techniques (e.g., SpaCy) to tokenize the text data and extract entities (people, dates, places, etc.).

[0586] Step 3:

[0587] Storage in the database

[0588] The server stores the cleansed data in a MySQL database. The input is a clean dataset and the output is structured database entries. Specifically, it inserts the data into the database using the MySQL INSERT statement.

[0589] Step 4:

[0590] Expert Scoring and Selection

[0591] The server analyzes the purpose of the discussion and searches for and scores experts related to specific keywords in the database. The input is the purpose of the discussion provided by the user, and the output is a list of scored experts. Specifically, it uses natural language processing technology to extract keywords from the input purpose and searches for experts based on those keywords. Next, it evaluates the relevance using a scoring algorithm (e.g., TF-IDF) and ranks them.

[0592] Step 5:

[0593] emotion recognition

[0594] The server uses an emotion engine to recognize the user's emotional state during discussions. The input is voice or text input from the user, and the output is analyzed emotion data. Specifically, it analyzes text and voice data using the Google Cloud Natural Language API and Microsoft Azure Text Analytics API to classify the emotional state.

[0595] Step 6:

[0596] Brainstorming

[0597] The server uses the selected expert model to hold discussions with the user. The input is the user's question or opinion, and the output is a response text. Specifically, it uses a generative AI model (e.g., GPT-4) to understand the user's question and generate a response that adheres to the expert's style. It also uses emotion recognition data during the discussion to adjust the tone and content of the response, and sends the final response to the user's device.

[0598] An example of a prompt sentence would be, "Environmental technology experts A, B, and C are being selected to discuss business ideas related to new environmental technologies. Please begin the expert model's response."

[0599] (Application example 2)

[0600] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0601] Conventional brainstorming systems have limitations in terms of obtaining expert knowledge and holding discussions, particularly in terms of adjusting responses to take into account the user's emotional state. As a result, discussions are ineffective and users' opinions and suggestions are not fully elicited. Furthermore, there is a lack of concrete support for improving production lines in factories. Therefore, there is a need for tools that are suitable for on-site workers and engineers to explore improvement ideas in real time.

[0602] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[0603] In this invention, the server includes: means for collecting data on past and present experts; means for processing and structuring the data; means for storing the structured data in a database; means for scoring experts suitable for a specific discussion purpose from the database; means for generating models of the scored experts and holding discussions; emotion processing means for recognizing and analyzing the user's emotional state; and means for adjusting the tone of responses according to the user's emotional state. This enables high-quality brainstorming based on the expert's knowledge, and more effective discussions by adjusting responses to the user's emotional state. It also functions as a concrete support tool for improving the efficiency of production lines in factories and exploring improvement proposals, enabling highly immediate discussions on-site.

[0604] "Means for collecting data" refers to means for obtaining data on past and present experts from sources such as online book databases, web articles, and academic papers.

[0605] "Means for processing and structuring data" refers to means for cleansing collected data, removing noise, standardizing data formats, tokenizing using natural language processing technology, extracting entities, and generating structured data.

[0606] "Means for storing in a database" refers to the means for storing processed and structured data in a database in a specific format.

[0607] A "means for scoring experts" is a means for extracting keywords related to a particular discussion purpose from a database and using a scoring algorithm to evaluate relevance and rank experts.

[0608] The "means for holding discussions" is a means for generating a scored expert model and using the generative AI model to imitate the expert's statements and generate responses to user questions and opinions.

[0609] The "emotion processing means" is a means for analyzing voice input and text input from the user and recognizing and analyzing the user's emotional state (joy, anger, sadness, surprise, etc.).

[0610] The "means for adjusting the tone of a response" is a means for adjusting the tone of a response to be generated in accordance with the emotional state of the user recognized by the emotion processing means.

[0611] The system of the present invention is implemented with the following configuration.

[0612] First, the server collects data on past and present experts from sources such as online book databases, web articles, and academic papers. This data is collected using web scraping technology and APIs. The collected data is then stored in a temporary storage area, and then noise is removed and the data format is standardized using data cleansing technology.

[0613] The server then uses natural language processing (NLP) techniques to tokenize the collected data and extract entities such as people, dates, and places, which are then used to generate expert profile data and store it in a database.

[0614] The server also analyzes the profile data stored in the database and uses a natural language processing engine to perform grammatical analysis and sentiment analysis of the text data. Based on the analysis results, each expert's speech patterns and style are added to the data model. The analyzed data is then input into a generative AI model (e.g., GPT-4) for transfer learning. Through this process, the generative AI model acquires the ability to imitate a specific person, and the fine-tuned model is stored in storage.

[0615] Users input the purpose of the discussion through the device interface. The server analyzes the input purpose and extracts relevant keywords. It then searches a database for appropriate experts based on the keywords and uses a scoring algorithm to evaluate and rank the relevance. The scoring results are sent to the device, where the user can review them and select the experts they want to invite.

[0616] Furthermore, the server uses an emotion processing engine during discussions between the user and the expert model. This emotion processing engine analyzes the user's voice and text inputs to recognize and analyze the user's emotional state. Depending on the user's emotional state, the server adjusts the content and tone of the expert's responses to enhance the effectiveness of the discussion.

[0617] As a specific example, if a user sets the theme as "Improve the efficiency of a new production line" and enters the question "How can we improve production speed?", the server will score relevant experts from the database and launch the expert model selected by the user. The generative AI model will generate an appropriate response, and the emotion engine will adjust the tone of the response according to the user's emotion. An example of a prompt sentence in this case is as follows:

[0618] markdown

[0619] I'd like to hear some ideas for improving the efficiency of our new production line.

[0620] Emotional Text: How can you improve your production speed?

[0621] Tone: calm

[0622] In this way, this system achieves high-quality brainstorming by combining expert knowledge with the user's emotional state. It also functions as a concrete support tool for seeking ideas for streamlining and improving production lines within factories, enabling highly immediate discussions on-site.

[0623] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0624] Step 1:

[0625] The server collects data on past and present experts from sources such as book databases, web articles, and academic papers on the Internet. Specifically, it acquires data using web scraping technology and APIs. The input is a "search query on experts," and the output is the "collected expert data."

[0626] Step 2:

[0627] The server cleanses the collected data, removing noise and standardizing the data format. This is a process that uses natural language processing (NLP) techniques to tokenize and extract entities (people, dates, places, etc.). The input is "collected expert data" and the output is "structured expert data."

[0628] Step 3:

[0629] The server stores the structured data in a database. Specifically, it generates profile data based on entities and saves it in the database. The input is "structured expert data" and the output is "expert profiles in the database."

[0630] Step 4:

[0631] The user inputs the purpose of the discussion through the terminal interface. For example, the topic is "improving the efficiency of a new production line." The input is the "user's purpose of the discussion," and the output is "confirmation of the input of the purpose of the discussion."

[0632] Step 5:

[0633] The server analyzes the discussion purpose entered by the user and extracts related keywords. The input is the "user's discussion purpose" and the output is the "extracted keywords."

[0634] Step 6:

[0635] The server searches for appropriate experts based on the extracted keywords from the database, and evaluates and ranks them using a scoring algorithm. The inputs are the extracted keywords and expert profile data, and the output is the scoring result.

[0636] Step 7:

[0637] The user checks the scoring results through the terminal and selects the experts they want to participate in. The input is the "scoring results" and the output is the "selected experts."

[0638] Step 8:

[0639] The server starts the selected expert model and receives the user's question or opinion as input. The input is "user's question or opinion" and the output is "question transmission to the expert model."

[0640] Step 9:

[0641] The server uses an emotion processing engine to analyze the user's questions and opinions and recognize their emotional state. The input is the user's questions and opinions, and the output is the user's emotional state.

[0642] Step 10:

[0643] The server uses a generative AI model to adjust the content and tone of the expert model's response according to the recognized emotional state of the user. Specifically, it adjusts the tone to "calm" or "gentle" based on data from the emotion engine. The input is the "user's emotional state" and the "expert model's basic response," and the output is the "adjusted response."

[0644] Step 11:

[0645] The server then sends the final generated response to the user's terminal and continues the discussion. The input is the "adjusted response" and the output is the "response display to the user."

[0646] The specific processing unit 290 transmits the result of the specific processing to the smart glasses 214. In the smart glasses 214, the control unit 46A causes the speaker 240 to output the result of the specific processing. The microphone 238 acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0647] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0648] In the above embodiment, an example in which the specific processing is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the smart glasses 214.

[0649] [Third embodiment]

[0650] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[0651] 5, the data processing system 310 includes the data processing device 12 and a headset terminal 314. An example of the data processing device 12 is a server.

[0652] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0653] The headset type terminal 314 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a display 343. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the display 343 are also connected to the bus 52.

[0654] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[0655] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[0656] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0657] Fig. 6 shows an example of the main functions of the data processing device 12 and the headset type terminal 314. As shown in Fig. 6, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0658] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0659] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0660] In the headset type terminal 314, a reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0661] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as the "server" and the headset type terminal 314 will be referred to as the "terminal."

[0662] The present invention is a system that systematically collects and analyzes information about past and present experts, and utilizes the results to realize high-quality brainstorming. Specific embodiments of the present invention are described below.

[0663] 1. Data collection and database creation

[0664] The first process performed by the server is data collection. Specifically, the server collects data on famous people and experts from sources such as online articles, academic papers, and book databases through scraping or APIs. The collected data is stored in a temporary storage area.

[0665] The server then cleanses the collected data, removes noise, and standardizes the data format. It then uses natural language processing (NLP) technology to tokenize the data and extract entities (people, dates, places, etc.). Based on the extracted entities, it generates profile data of famous people and experts, which are then stored in a database.

[0666] 2. Language Analysis and Model Fine-tuning

[0667] The server analyzes the profile data stored in the database. Specifically, it uses a natural language processing engine to analyze the text data, and performs grammatical analysis and sentiment analysis. Based on the results of this analysis, the speech patterns and styles of each great person or expert are added to the data model.

[0668] The server then inputs the analyzed data into a generative AI model (e.g., GPT-4) and performs transfer learning, which gives the model the ability to imitate a specific person. The fine-tuned model is then stored in storage.

[0669] 3. Scoring system and expert selection

[0670] The user inputs the purpose of the discussion through the device interface. For example, they can set a purpose such as "brainstorm ideas for a new product." The input purpose is then sent from the device to the server.

[0671] The server analyzes the purpose and extracts relevant keywords. It then searches the database for relevant figures and experts based on the keywords and ranks them using a scoring algorithm. The scoring results are sent to the user's device, where the user can review them and select the experts they want to include.

[0672] 4. Brainstorming

[0673] The server launches the generative AI models of the selected experts. The user inputs questions or opinions through the device. For example, if the user asks, "Tell me your ideas for new environmental technologies," the server interprets the question and sends it to the appropriate expert model. The generative AI models generate responses in the style of each expert, synthesize them, and send them to the device.

[0674] Specific examples

[0675] Consider a case where a user sets the theme as "Business ideas related to new environmental technologies." The server scores experts knowledgeable in environmental technologies from the database, and for example, Mr. A, Mr. B, and Mr. C receive high scores. The user checks the scoring results and selects Mr. A, Mr. B, and Mr. C. The server then launches the models of the selected experts, and when the user types "I would like to discuss new product strategies," the server responds in a synthesized form with each expert. In this way, the user can obtain new business ideas from multiple perspectives.

[0676] As described above, this system combines the knowledge of great people from the past and current experts to provide a concrete means for achieving high-quality brainstorming.

[0677] The processing flow will be explained below.

[0678] Step 1:

[0679] The server collects data about famous people and experts from various sources (book databases, web articles, academic papers, etc.) using scraping technology and APIs. The collected data is stored in a temporary storage area.

[0680] Step 2:

[0681] The server cleanses the data in the temporary storage area, removing noise and standardizing the data format, then uses natural language processing (NLP) techniques to tokenize the data and extract entities (people, dates, places, etc.).

[0682] Step 3:

[0683] The server generates profile data of famous people and experts based on the extracted entities, including their names, fields of expertise, and representative achievements. The generated profile data is stored in a database.

[0684] Step 4:

[0685] The server analyzes the profile data stored in the database, performs grammatical analysis and sentiment analysis, and adds the speech patterns and styles of each great person or expert to the data model based on the analysis results.

[0686] Step 5:

[0687] The server inputs the analyzed data into a generative AI model (e.g., GPT-4) for transfer learning. This process allows the generative AI model to mimic the speech and thought patterns of a specific person. Once the fine-tuned model is complete, it is stored in storage.

[0688] Step 6:

[0689] The user uses the device interface to input the purpose of the discussion, such as "brainstorm ideas for a new product." The input purpose is then sent from the device to the server.

[0690] Step 7:

[0691] The server analyzes the purpose of the discussion and extracts relevant keywords, then searches a database for relevant figures and experts based on the keywords.

[0692] Step 8:

[0693] The server uses a scoring algorithm to evaluate the search results and rank them in order of relevance, and the results are sent to the user's device, where the user can review and select the experts they want to include.

[0694] Step 9:

[0695] The server launches a generative AI model of a selected expert, and the user inputs questions or opinions from their device and sends them to the server.

[0696] Step 10:

[0697] The server interprets the user's input and sends questions and opinions to the appropriate expert model. The generative AI model generates responses in each expert's style, synthesizes them, and sends them to the device.

[0698] Step 11:

[0699] The terminal displays the response from the server on the user interface, and the user can check the displayed results and continue brainstorming by entering further questions or opinions as needed.

[0700] The above processing flow allows users to generate new business ideas through discussions with great people and experts from across time and space.

[0701] Example 1

[0702] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[0703] Conventional brainstorming systems have difficulty effectively utilizing the knowledge of past and present experts. Furthermore, they lack the means to efficiently select experts suited to specific discussion purposes and achieve high-quality brainstorming. As a result, the quality of discussions declines, and it is difficult to obtain new ideas and insights.

[0704] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0705] In this invention, the server includes means for collecting data from information sources such as online articles, papers, and books, means for cleansing the data and extracting entities using natural language processing technology, means for storing the structured data in a database, means for inputting the purpose of the discussion and extracting related keywords, means for scoring experts suitable for the specific purpose of the discussion from the database, means for generating and launching a generative AI model of the selected expert based on the scoring results, and means for receiving questions from users via the terminal, sending the questions to the generative AI model, generating responses, and sending them to the terminal. This makes it possible to combine the knowledge of past and present experts and efficiently realize multifaceted and advanced brainstorming.

[0706] "Articles on the Internet" refers to text data and news articles posted on websites.

[0707] A "paper" refers to a document that describes the results of academic research or a document that reports the results of research.

[0708] "Book" means a printed document that has been published as a publication.

[0709] "Means of collecting data" refers to the methods and tools used to obtain the required information from the source.

[0710] "Data cleansing methods" refer to methods and tools used to remove noise and inaccurate information from collected data and format it.

[0711] "Natural language processing technology" refers to technology that enables computers to understand and process human language.

[0712] "Entity extraction" refers to methods and tools that extract important information, such as specific people, dates, or places, from text data.

[0713] "Structured data" refers to data that has been organized and arranged in a format suitable for a database.

[0714] "Means of storing data in a database" refers to the methods and tools used to store data in a database.

[0715] The "purpose of the discussion" refers to the goal or theme you want to achieve when having the discussion.

[0716] "Keyword extraction methods" refers to methods and tools for finding related words and phrases from text data.

[0717] A "scoring method" refers to a method or tool that evaluates the importance or relevance of data based on specific criteria.

[0718] A "generative AI model" refers to an artificial intelligence model that has the ability to perform specific tasks based on training data.

[0719] "Means for receiving questions from users" refers to the methods and tools by which the system obtains questions and opinions entered by users.

[0720] "Means for sending questions to a generative AI model" refers to methods or tools for communicating a user's questions to a generative AI model.

[0721] "Means for generating a response" refers to the methods and tools that a generative AI model uses to create an answer to a user's question.

[0722] "Means for sending to terminal" refers to a method or tool for transmitting the generated response to the user's terminal.

[0723] The present invention is a system that systematically collects and analyzes information about past and present experts, and utilizes the results to realize high-quality brainstorming. Specific embodiments of the present invention are described below.

[0724] Data collection and database creation

[0725] The server first collects information from online articles, academic papers, book databases, etc. Specifically, the server uses web scraping tools such as Python's BeautifulSoup and Scrapy to obtain information about famous people and experts. For academic papers, the server also uses the Google Scholar API to collect data. This collected data is then stored in a temporary storage area (for example, a database such as MySQL).

[0726] Next, the server cleanses the data using the Pandas library. Specifically, it removes unnecessary data and noise, converts date and time formats, and normalizes text. The cleansed data is then tokenized using natural language processing (NLP) techniques to extract entities (people, dates, places, etc.). This analysis is performed using libraries such as NLTK and spaCy. Finally, the structured data is stored in a database such as PostgreSQL.

[0727] Language analysis and model fine-tuning

[0728] The server analyzes the profile data stored in the database. For the analysis, it uses a natural language processing engine such as BERT to perform grammatical analysis and sentiment analysis. Based on the results of this analysis, speech patterns and styles are added to the data model.

[0729] The server then inputs the analyzed data into a generative AI model such as GPT-4 to perform transfer learning. This transfer learning is performed using PyTorch or TensorFlow. The generative AI model, which has acquired the ability to imitate a specific person through transfer learning, is stored in storage such as AWS S3.

[0730] Scoring system and expert selection

[0731] The user inputs the purpose of the discussion through the terminal interface, for example, "Brainstorm ideas for a new product." The input purpose is sent from the terminal to the server.

[0732] The server analyzes the purpose and extracts relevant keywords. It then uses SQL queries to search for relevant experts from the database and scores them using TF-IDF and other machine learning algorithms. The scoring results are sent to the device, where the user can review them and select the desired expert.

[0733] Brainstorming

[0734] The server launches a generative AI model of the selected expert. This is done using container orchestration tools such as Docker or Kubernetes. The user inputs a question or opinion through the terminal. For example, they might input, "Tell me your ideas for new environmental technologies." The server interprets the question and sends it to the appropriate expert model. The generative AI model generates a response in the expert's style, synthesizes it, and sends it to the terminal.

[0735] Specific examples

[0736] If a user selects the theme "Business ideas related to new environmental technologies," the server scores experts knowledgeable in environmental technologies from the database, and for example, Mr. A, Mr. B, and Mr. C receive high scores. The user checks the scoring results and selects Mr. A, Mr. B, and Mr. C. The server then activates the model of the selected experts, and when the user enters "I would like to discuss new product strategies," the server generates a response in a synthesized form from each expert and sends it to the terminal.

[0737] In this way, the system of the present invention combines the knowledge of past and present experts and provides a concrete means for realizing high-quality brainstorming.

[0738] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0739] Step 1: Data collection

[0740] The server collects data from online articles, academic papers, book databases, and other sources. Specifically, it uses web scraping tools such as Python's BeautifulSoup and Scrapy to obtain information about famous people and experts. For academic papers, it also uses the Google Scholar API to obtain paper data. Inputs include URLs and API endpoints, and the output is the obtained raw data stored in a temporary storage area. This process collects a large amount of text data.

[0741] Step 2: Data cleansing

[0742] The server cleanses the collected data. Specifically, it uses the Pandas library to manipulate data frames and remove unnecessary data and noise. This step also performs date and time format conversions and text normalization. The input is raw data stored in a temporary storage area, and the output is clean, cleaned data. This process improves the quality of the data and makes subsequent analysis easier.

[0743] Step 3: Data analysis using natural language processing

[0744] The server analyzes the cleansed data using natural language processing techniques. Specifically, it uses libraries such as NLTK and spaCy to tokenize the data and extract entities (people, dates, places, etc.). The input is the cleansed data, and the output is extracted entity data. This processing structures the text data and generates expert profile data.

[0745] Step 4: Store in the database

[0746] The server stores the extracted entity data in a database such as PostgreSQL. The input is the entity data, and the output is a structured database entry. This process allows for efficient management of expert information and facilitates subsequent search and analysis.

[0747] Step 5: Input and Analysis of Desired Objectives

[0748] The user inputs the purpose of the discussion through the terminal interface. For example, they might input "brainstorm ideas for a new product." The input is the discussion purpose entered by the user, and the output is text data sent to the server. This step transmits the user's discussion purpose to the server.

[0749] Step 6: Keyword extraction and search

[0750] The server analyzes the received discussion purpose and extracts relevant keywords. It then searches for relevant experts from the database using an SQL query. The inputs are the analyzed text data and the database, and the output is a list of relevant experts. This process identifies experts suitable for the discussion purpose.

[0751] Step 7: Scoring

[0752] The server scores experts using a scoring algorithm (e.g., TF-IDF or other machine learning algorithm) against the relevant data. The input is a list of experts and relevant data, and the output is a scored list of experts. This process prioritizes the appropriate experts.

[0753] Step 8: Selecting experts

[0754] The user checks the scoring results on their device and selects the experts they want to participate in. The input is a list of scored experts, and the output is information about the selected experts sent to the server. This process allows the user to select the experts they want.

[0755] Step 9: Launching the expert model

[0756] The server launches the generative AI model of the selected expert. Specifically, the model is launched using a container orchestration tool such as Docker or Kubernetes. The input is information about the selected expert, and the output is the launched generative AI model. This process puts the expert model into operation.

[0757] Step 10: Brainstorming and response generation

[0758] A user inputs a question or opinion through a device. For example, they might input, "Tell me your ideas for new environmental technologies." The input is the user's question text, and the output is sent to the generative AI model. The generative AI model then generates a response based on the question and sends it in a synthesized form to the user's device. This process allows the user to obtain high-quality opinions and suggestions.

[0759] (Application example 1)

[0760] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[0761] The development of autonomous vehicles requires advanced algorithms and safety measures, but there is a lack of effective means to integrate expert knowledge. There is also a need for efficient methods to perform simulations based on expert knowledge and to verify new ideas and technical solutions.

[0762] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0763] In this invention, the server includes means for collecting data on past and present experts, means for processing and structuring the data, means for storing the structured data in a database, means for scoring experts suitable for a specific discussion purpose from the database, means for generating a model of the scored experts and holding discussions, and means for running a simulation based on the expert model. This makes it possible to effectively integrate the latest expert knowledge on autonomous vehicles and verify optimal operation algorithms and safety measures through simulation.

[0764] "Past and present expert data" is a collection of information about people who have past and present expertise and experience in a particular field.

[0765] "Means of collection" refers to the technical means used to obtain data from online papers, articles, book databases, etc.

[0766] "Means of processing and structuring" refers to technical means of systematically organizing collected data by processing it through data cleansing, tokenization, entity extraction, etc.

[0767] "Means of storing data in a database" means a technical means of efficiently storing structured data and keeping it in a state that can be quickly retrieved and used as needed.

[0768] An "expert appropriate for a particular purpose" is someone with particular knowledge and experience in a field relevant to the topic or goal of the discussion.

[0769] A "scoring means" is a technological means that uses specific criteria or algorithms to assess and rank the suitability or relevance of Experts.

[0770] "Means for generating a model of a scored expert and engaging in discussions" refers to the technical means for creating a generative AI model that mimics the style and speech patterns of a scored expert and using it to engage in discussions and responses.

[0771] A "generative AI model" is an artificial intelligence model that uses machine learning technology to mimic the knowledge and speaking style of experts.

[0772] "Means for performing simulations" refers to the technical means for verifying the operation algorithms and safety measures of autonomous vehicles in a virtual environment using the knowledge of the generated expert model.

[0773] The present invention provides a system for leveraging expert knowledge in the development and validation of autonomous vehicles. The system uses the following major hardware and software components:

[0774] 1. Data Collection

[0775] The server collects expert data on autonomous driving technology from academic papers, technical articles, and various book databases on the Internet. Specifically, it retrieves data using the URL Fetch API and performs web scraping using Beautiful Soup (Python).

[0776] 2. Data Analysis and Natural Language Processing

[0777] To process and structure the collected data, the server uses natural language processing techniques, specifically NLP libraries such as spaCy and NLTK for data cleansing, tokenization, and entity extraction, and then stores the structured data in a database.

[0778] 3. Expert Selection and Fine-Tuning

[0779] To select the most suitable experts for a specific discussion purpose, the server searches the data in the database and evaluates their relevance using scoring algorithms such as TF-IDF and BERT. Using the associated expert data, the GPT-4 model is fine-tuned using Hugging Face Transformers, resulting in a generative AI model that mimics each expert's speaking patterns and style.

[0780] 4. Simulation and Verification

[0781] Users can input questions about autonomous driving or the purpose of their discussion through their devices. For example, "What are the safety measures for autonomous vehicles in urban areas?" The server generates a response based on the relevant expert model. This response is used to run simulations in a virtual environment to verify optimal driving algorithms and safety measures.

[0782] Specific examples

[0783] For example, if a user types in "What are the safety measures for self-driving vehicles in urban areas?", the system will act as follows:

[0784] 1. Prompt statement generation (server):

[0785] User Query: "What are the safety measures for autonomous vehicles in urban areas?"

[0786] Model Prompt: "The system will discuss safety measures for autonomous vehicles in urban areas. Please suggest the best solution based on past and current expert knowledge."

[0787] 2. Generative AI model response (server):

[0788] Selected expert models respond to questions and their answers are used in the simulation.

[0789] 3. Simulation and result verification (server):

[0790] Response-based simulations are run in a virtual environment to verify optimal operating algorithms and safety measures.

[0791] This allows users to gain a multifaceted perspective on the development and operation management of autonomous vehicles while integrating the latest expert knowledge. By effectively utilizing the knowledge of past and present experts, this system can significantly improve the safety and efficiency of autonomous driving technology.

[0792] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0793] Step 1:

[0794] The server collects data related to autonomous driving technology. It uses the URL Fetch API to retrieve academic papers and technical articles from the Internet and performs web scraping using Beautiful Soup (Python). The input is a list of web page URLs, and the output is text data related to the scraped experts.

[0795] Step 2:

[0796] The server analyzes the collected text data using natural language processing technology. Specifically, the server uses spaCy or NLTK to cleanse, tokenize, and extract entities from the data. The input is scraped text data, and the output is structured data (such as entity information).

[0797] Step 3:

[0798] The server stores the structured data in a database, specifically by inserting well-organized data into a SQL or NoSQL database. The input is structured data, and the output is an updated database.

[0799] Step 4:

[0800] The server scores experts suitable for a specific discussion purpose. It uses natural language processing technology to extract keywords related to the discussion purpose from the user (e.g., "Safety measures for autonomous vehicles in urban areas") and evaluates their relevance using TF-IDF and BERT. The input is the discussion purpose text from the user and expert data in the database, and the output is a list of scored experts.

[0801] Step 5:

[0802] The server generates a scored expert model and conducts the discussion. Using Hugging Face Transformers, it performs transfer learning on the GPT-4 model to create a generative AI model that mimics the expert's speech patterns and style. The input is the scored expert data, and the output is the generated expert model.

[0803] Step 6:

[0804] Users input questions about autonomous driving or the purpose of the discussion through their terminal. For example, they might input, "What are the safety measures for autonomous vehicles in urban areas?" The input is the question text, and the output is the request data sent to the server.

[0805] Step 7:

[0806] The server analyzes the user's question, generates a prompt sentence based on the generated expert model, and inputs the prompt sentence to the expert model to obtain a response. The input is the user's question text and the generated expert model, and the output is the response of the expert model.

[0807] Step 8:

[0808] The server runs a simulation based on the expert model's responses and verifies the results. Simulation software is used to experiment with operation algorithms and safety measures in a virtual environment. The input is the expert model's response text, and the output is the simulation results.

[0809] Step 9:

[0810] The server provides the simulation results to the user. The results are displayed on the terminal and presented in a format that is easy for the user to understand. The input is the simulation result data, and the output is the result provided to the user.

[0811] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[0812] This invention combines a system that collects and analyzes information on past and present experts and performs high-quality brainstorming based on the results with an emotion engine that recognizes and analyzes the user's emotions. This system enables responses and discussions to proceed according to the user's emotional state, resulting in more effective discussions.

[0813] 1. Data collection and database creation

[0814] The server first collects data about famous people and experts from sources such as online book databases, web articles, and academic papers. To achieve this, it uses scraping technology and APIs. The collected data is then stored in a temporary storage area.

[0815] The server then cleanses the collected data, removing noise and standardizing the data format. It then uses natural language processing (NLP) technology to tokenize the data and extract entities (people, dates, places, etc.). Based on the extracted entities, it generates profile data of famous people and experts, which are then stored in a database.

[0816] 2. Language Analysis and Model Fine-tuning

[0817] The server analyzes the profile data stored in the database. Specifically, it uses a natural language processing engine to analyze the text data, and performs grammatical analysis and sentiment analysis. Based on the results of this analysis, the speech patterns and styles of each great person or expert are added to the data model.

[0818] The server then inputs the analyzed data into a generative AI model (e.g., GPT-4) for transfer learning. Through this process, the generative AI model acquires the ability to mimic a specific person. Once the fine-tuned model is complete, it is stored in storage.

[0819] 3. Scoring system and expert selection

[0820] Users input the purpose of the discussion through the device interface. For example, they set a specific purpose such as "brainstorm ideas for a new product." The input purpose is sent from the device to the server.

[0821] The server analyzes the purpose and extracts relevant keywords.The server then searches a database for relevant famous people and experts based on the keywords, and uses a scoring algorithm to evaluate and rank them.The scoring results are sent to the terminal, where the user can review them and select the experts they want to participate.

[0822] 4. Utilizing the Emotion Engine

[0823] The server uses an emotion engine in discussions between the user and the expert model to analyze voice and text inputs from the user and recognize the user's emotional state (e.g., joy, anger, sadness, surprise).

[0824] 5. Brainstorming

[0825] The server launches the generative AI models of the selected experts. The user inputs questions or opinions through the device. For example, if the user asks, "Tell me your ideas for new environmental technologies," the server interprets the question and sends it to the appropriate expert model. The generative AI models generate responses in the style of each expert, synthesize them, and send them to the device.

[0826] Furthermore, the emotion engine recognizes the user's emotions, allowing the server to adjust the content and tone of the expert's response. For example, if the user is feeling impatient, the server can provide a calm response to enhance the effectiveness of the discussion.

[0827] Specific examples

[0828] If a user sets the theme as "Business ideas related to new environmental technologies," the server scores experts knowledgeable in environmental technologies from the database, and for example, experts A, B, and C receive high scores. The user checks the scoring results and selects experts A, B, or C. The server launches the models of the selected experts, and when the user asks, "I would like to discuss new product strategies," each expert responds in a synthesized form.

[0829] The server's built-in emotion engine detects the user's emotional state and adjusts the expert model's response as needed. For example, if the user is emotional, the expert model's response can be adjusted to a calm and peaceful tone to facilitate a smooth discussion.

[0830] As described above, this system combines the knowledge of great people from the past and current experts with the user's emotional state, providing a concrete means for achieving high-quality brainstorming and the creation of new business ideas.

[0831] The processing flow will be explained below.

[0832] Step 1:

[0833] The server collects data about famous people and experts from various sources (book databases, web articles, academic papers, etc.) using scraping technology and APIs. The collected data is stored in a temporary storage area.

[0834] Step 2:

[0835] The server cleanses the data in the temporary storage area, removing noise and standardizing the data format, then uses natural language processing (NLP) techniques to tokenize the data and extract entities (people, dates, places, etc.).

[0836] Step 3:

[0837] The server generates profile data of famous people and experts based on the extracted entities, including their names, fields of expertise, and representative achievements. The generated profile data is stored in a database.

[0838] Step 4:

[0839] The server analyzes the profile data stored in the database. Specifically, it uses a natural language processing engine to perform grammatical analysis and sentiment analysis of the text data. Based on the results of this analysis, the speech patterns and styles of each great person or expert are added to the data model.

[0840] Step 5:

[0841] The server inputs the analyzed data into a generative AI model (e.g., GPT-4) and performs transfer learning. This allows the generative AI model to acquire the ability to imitate a specific person. Once fine-tuned, the model is stored in storage.

[0842] Step 6:

[0843] The user inputs the purpose of the discussion using the device interface. For example, they set a purpose such as "brainstorm ideas for a new product." The input purpose is sent from the device to the server.

[0844] Step 7:

[0845] The server analyzes the purpose of the input discussion and extracts relevant keywords, then searches the database for relevant figures and experts based on the keywords.

[0846] Step 8:

[0847] The server uses a scoring algorithm to evaluate the search results and rank the most relevant experts, which are then sent to the device where the user can review and select the experts they want to participate.

[0848] Step 9:

[0849] The server runs a generative AI model of selected experts, which runs on a dedicated cloud infrastructure.

[0850] Step 10:

[0851] Users can input questions or opinions into their devices and send them to the server. For example, they can ask, "Tell me your ideas for new environmental technologies."

[0852] Step 11:

[0853] The server interprets the user's input and sends the question to the appropriate expert model, which then generates a response in the speaking style of each expert.

[0854] Step 12:

[0855] An emotion engine built into the server analyzes the user's voice and text input to assess their emotional state, for example detecting anger or impatience in the user's voice.

[0856] Step 13:

[0857] The server adjusts the content and tone of the generative AI model's response based on the user's emotional state as recognized by the emotion engine. For example, if the user is angry, the expert model's response will be changed to a calmer tone.

[0858] Step 14:

[0859] The server composes the final response and sends it to the terminal.

[0860] Step 15:

[0861] The terminal displays the response from the server on the user interface, and the user can check the displayed results and continue brainstorming by entering further questions or opinions as needed.

[0862] Through these processes, users can engage in discussions with great figures and experts from across time and space, and realize high-quality discussions that are adjusted according to their emotional state.

[0863] Example 2

[0864] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[0865] In today's knowledge-based society, it is extremely important to conduct high-quality discussions by utilizing the knowledge of past and present experts. However, the process of extracting relevant information from vast amounts of data and selecting appropriate experts is difficult, especially when it comes to taking user emotions into account. This has led to a growing need for a system that recognizes the user's emotional state and generates responses based on the expert's style.

[0866] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0867] In this invention, the server includes means for collecting data on past and present experts, means for cleansing the data and analyzing it using natural language processing technology, means for storing structured data generated from the data in a database, means for scoring experts suitable for a specific discussion purpose from the database, means for generating a model of the scored experts and holding a discussion, means for recognizing the emotional state of a user during the discussion and adjusting the progress of the discussion, and means for generating expert responses using the generative AI model during the discussion. This makes it possible to conduct high-quality discussions by making the most of expert knowledge and effectively progressing the discussion in accordance with the emotional state of the user.

[0868] "Data Collection Tools" are tools used to gather information about past and present professionals from the Internet and other databases.

[0869] "Data cleansing means" refers to a means for removing noise from collected data and standardizing the data format.

[0870] "Natural language processing technology" is a technology that analyzes text data and performs tokenization, entity extraction, sentiment analysis, etc.

[0871] A "database" is a system that stores the generated structured data and retrieves and uses it when needed.

[0872] A "scoring tool" is a tool for searching a database for experts suitable for a particular discussion purpose, assessing their relevance, and ranking them.

[0873] An "expert model generation means" is a means for creating a generative AI model that mimics a scored expert.

[0874] The "discussion tool" is a tool that uses a generative AI model to generate responses based on the expert's style and engages in discussions between the user and the expert.

[0875] The "emotion recognition means" is a means for recognizing the emotional state of users during a discussion and adjusting the content and tone of the discussion.

[0876] A "generative AI model" is an artificial intelligence technology that can learn from large amounts of text data and generate text for specific tasks.

[0877] This invention is a system that collects and analyzes information about past and present experts, and uses the results to conduct high-quality discussions. By combining it with an emotion engine that recognizes and analyzes the user's emotions, it is possible to respond and progress the discussion in accordance with the user's emotional state. Specific embodiments for implementing this system are described below.

[0878] Data collection

[0879] The server first collects data about experts from sources such as online book databases, web articles, and academic papers. This is done using scraping tools such as BeautifulSoup and Scrapy, and APIs such as the Google Books API. The collected data is then stored in a temporary storage area. Specifically, the server collects titles, author names, citations, etc. from the book database and stores them in the temporary storage area.

[0880] Data Cleansing and Analysis

[0881] The collected data is cleansed by the server. Specifically, the Pandas library is used to fill in missing data and remove inaccurate data. Natural language processing (NLP) techniques are then used to tokenize the data, and entities (people, dates, places, etc.) are extracted using tools such as SpaCy. Expert profile data is generated based on the extracted entities, and this data is stored in a MySQL database.

[0882] Fine-tuning the model

[0883] The server analyzes the profile data stored in the database. It uses natural language processing engines such as BERT and GPT-3 to perform grammatical and sentiment analysis of the text data, and adds each expert's speech patterns and style to the data model. The analyzed data is then input into a generative AI model (e.g., GPT-4) for transfer learning. Once the fine-tuned model is complete, it is stored in Amazon S3 storage.

[0884] Scoring and Expert Selection

[0885] The user inputs the purpose of the discussion through the device interface. For example, they can set a specific purpose such as "brainstorm ideas for a new product." The input purpose is sent from the device to the server. The server analyzes the purpose and extracts related keywords. It then searches for appropriate experts in a database and uses a scoring algorithm to evaluate their relevance and rank them. The scoring results are sent to the device, where the user can review them and select the experts they want to participate in.

[0886] Emotion recognition and discussion moderation

[0887] The server uses an emotion engine in the discussion between the user and the expert model. The emotion engine uses the Google Cloud Natural Language API and Microsoft Azure Text Analytics API to analyze the user's voice and text input and recognize their emotional state (joy, anger, sadness, surprise, etc.). Based on the recognized emotion, the content and tone of the expert model's response are adjusted.

[0888] Brainstorming

[0889] The server launches a generative AI model of a selected expert. The user inputs a question or opinion through their device. For example, they might ask, "Tell me your ideas for new environmental technologies." The server interprets the question and sends it to the appropriate expert model. The generative AI model generates a response in the style of each expert, synthesizes it, and sends it to the device. An emotion engine recognizes the user's emotions and adjusts the expert model's response as needed.

[0890] An example of a specific prompt is, "You are selecting environmental technology experts A, B, and C to discuss a business idea related to a new environmental technology. Please begin the expert model response."

[0891] In this way, the embodiment of the present invention provides a system that makes the most of the knowledge of experts and enables effective discussion to proceed in accordance with the emotional state of the user.

[0892] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0893] Step 1:

[0894] Data collection

[0895] The server collects data about experts from sources such as book databases, web articles, and academic papers on the Internet. The input is the URL or API endpoint of the source, and the output is raw data stored in a temporary storage area. Specific operations include scraping web pages using Scrapy and BeautifulSoup and retrieving book data from the Google Books API.

[0896] Step 2:

[0897] Data Cleansing and Analysis

[0898] The server cleanses the collected data, removing noise and standardizing the data format. The input is the collected raw data, and the output is a clean dataset. Specifically, it uses the Pandas library to impute missing values ​​and remove inaccurate data. It then uses natural language processing techniques (e.g., SpaCy) to tokenize the text data and extract entities (people, dates, places, etc.).

[0899] Step 3:

[0900] Storage in the database

[0901] The server stores the cleansed data in a MySQL database. The input is a clean dataset and the output is structured database entries. Specifically, it inserts the data into the database using the MySQL INSERT statement.

[0902] Step 4:

[0903] Expert Scoring and Selection

[0904] The server analyzes the purpose of the discussion and searches for and scores experts related to specific keywords in the database. The input is the purpose of the discussion provided by the user, and the output is a list of scored experts. Specifically, it uses natural language processing technology to extract keywords from the input purpose and searches for experts based on those keywords. Next, it evaluates the relevance using a scoring algorithm (e.g., TF-IDF) and ranks them.

[0905] Step 5:

[0906] emotion recognition

[0907] The server uses an emotion engine to recognize the user's emotional state during discussions. The input is voice or text input from the user, and the output is analyzed emotion data. Specifically, it analyzes text and voice data using the Google Cloud Natural Language API and Microsoft Azure Text Analytics API to classify the emotional state.

[0908] Step 6:

[0909] Brainstorming

[0910] The server uses the selected expert model to hold discussions with the user. The input is the user's question or opinion, and the output is a response text. Specifically, it uses a generative AI model (e.g., GPT-4) to understand the user's question and generate a response that adheres to the expert's style. It also uses emotion recognition data during the discussion to adjust the tone and content of the response, and sends the final response to the user's device.

[0911] An example of a prompt sentence would be, "Environmental technology experts A, B, and C are being selected to discuss business ideas related to new environmental technologies. Please begin the expert model's response."

[0912] (Application example 2)

[0913] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[0914] Conventional brainstorming systems have limitations in terms of obtaining expert knowledge and holding discussions, particularly in terms of adjusting responses to take into account the user's emotional state. As a result, discussions are ineffective and users' opinions and suggestions are not fully elicited. Furthermore, there is a lack of concrete support for improving production lines in factories. Therefore, there is a need for tools that are suitable for on-site workers and engineers to explore improvement ideas in real time.

[0915] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[0916] In this invention, the server includes: means for collecting data on past and present experts; means for processing and structuring the data; means for storing the structured data in a database; means for scoring experts suitable for a specific discussion purpose from the database; means for generating models of the scored experts and holding discussions; emotion processing means for recognizing and analyzing the user's emotional state; and means for adjusting the tone of responses according to the user's emotional state. This enables high-quality brainstorming based on the expert's knowledge, and more effective discussions by adjusting responses to the user's emotional state. It also functions as a concrete support tool for improving the efficiency of production lines in factories and exploring improvement proposals, enabling highly immediate discussions on-site.

[0917] "Means for collecting data" refers to means for obtaining data on past and present experts from sources such as online book databases, web articles, and academic papers.

[0918] "Means for processing and structuring data" refers to means for cleansing collected data, removing noise, standardizing data formats, tokenizing using natural language processing technology, extracting entities, and generating structured data.

[0919] "Means for storing in a database" refers to the means for storing processed and structured data in a database in a specific format.

[0920] A "means for scoring experts" is a means for extracting keywords related to a particular discussion purpose from a database and using a scoring algorithm to evaluate relevance and rank experts.

[0921] The "means for holding discussions" is a means for generating a scored expert model and using the generative AI model to imitate the expert's statements and generate responses to user questions and opinions.

[0922] The "emotion processing means" is a means for analyzing voice input and text input from the user and recognizing and analyzing the user's emotional state (joy, anger, sadness, surprise, etc.).

[0923] The "means for adjusting the tone of a response" is a means for adjusting the tone of a response to be generated in accordance with the emotional state of the user recognized by the emotion processing means.

[0924] The system of the present invention is implemented with the following configuration.

[0925] First, the server collects data on past and present experts from sources such as online book databases, web articles, and academic papers. This data is collected using web scraping technology and APIs. The collected data is then stored in a temporary storage area, and then noise is removed and the data format is standardized using data cleansing technology.

[0926] The server then uses natural language processing (NLP) techniques to tokenize the collected data and extract entities such as people, dates, and places, which are then used to generate expert profile data and store it in a database.

[0927] The server also analyzes the profile data stored in the database and uses a natural language processing engine to perform grammatical analysis and sentiment analysis of the text data. Based on the analysis results, each expert's speech patterns and style are added to the data model. The analyzed data is then input into a generative AI model (e.g., GPT-4) for transfer learning. Through this process, the generative AI model acquires the ability to imitate a specific person, and the fine-tuned model is stored in storage.

[0928] Users input the purpose of the discussion through the device interface. The server analyzes the input purpose and extracts relevant keywords. It then searches a database for appropriate experts based on the keywords and uses a scoring algorithm to evaluate and rank the relevance. The scoring results are sent to the device, where the user can review them and select the experts they want to invite.

[0929] Furthermore, the server uses an emotion processing engine during discussions between the user and the expert model. This emotion processing engine analyzes the user's voice and text inputs to recognize and analyze the user's emotional state. Depending on the user's emotional state, the server adjusts the content and tone of the expert's responses to enhance the effectiveness of the discussion.

[0930] As a specific example, if a user sets the theme as "Improve the efficiency of a new production line" and enters the question "How can we improve production speed?", the server will score relevant experts from the database and launch the expert model selected by the user. The generative AI model will generate an appropriate response, and the emotion engine will adjust the tone of the response according to the user's emotion. An example of a prompt sentence in this case is as follows:

[0931] markdown

[0932] I'd like to hear some ideas for improving the efficiency of our new production line.

[0933] Emotional Text: How can you improve your production speed?

[0934] Tone: calm

[0935] In this way, this system achieves high-quality brainstorming by combining expert knowledge with the user's emotional state. It also functions as a concrete support tool for seeking ideas for streamlining and improving production lines within factories, enabling highly immediate discussions on-site.

[0936] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0937] Step 1:

[0938] The server collects data on past and present experts from sources such as book databases, web articles, and academic papers on the Internet. Specifically, it acquires data using web scraping technology and APIs. The input is a "search query on experts," and the output is the "collected expert data."

[0939] Step 2:

[0940] The server cleanses the collected data, removing noise and standardizing the data format. This is a process that uses natural language processing (NLP) techniques to tokenize and extract entities (people, dates, places, etc.). The input is "collected expert data" and the output is "structured expert data."

[0941] Step 3:

[0942] The server stores the structured data in a database. Specifically, it generates profile data based on entities and saves it in the database. The input is "structured expert data" and the output is "expert profiles in the database."

[0943] Step 4:

[0944] The user inputs the purpose of the discussion through the terminal interface. For example, the topic is "improving the efficiency of a new production line." The input is the "user's purpose of the discussion," and the output is "confirmation of the input of the purpose of the discussion."

[0945] Step 5:

[0946] The server analyzes the discussion purpose entered by the user and extracts related keywords. The input is the "user's discussion purpose" and the output is the "extracted keywords."

[0947] Step 6:

[0948] The server searches for appropriate experts based on the extracted keywords from the database, and evaluates and ranks them using a scoring algorithm. The inputs are the extracted keywords and expert profile data, and the output is the scoring result.

[0949] Step 7:

[0950] The user checks the scoring results through the terminal and selects the experts they want to participate in. The input is the "scoring results" and the output is the "selected experts."

[0951] Step 8:

[0952] The server starts the selected expert model and receives the user's question or opinion as input. The input is "user's question or opinion" and the output is "question transmission to the expert model."

[0953] Step 9:

[0954] The server uses an emotion processing engine to analyze the user's questions and opinions and recognize their emotional state. The input is the user's questions and opinions, and the output is the user's emotional state.

[0955] Step 10:

[0956] The server uses a generative AI model to adjust the content and tone of the expert model's response according to the recognized emotional state of the user. Specifically, it adjusts the tone to "calm" or "gentle" based on data from the emotion engine. The input is the "user's emotional state" and the "expert model's basic response," and the output is the "adjusted response."

[0957] Step 11:

[0958] The server then sends the final generated response to the user's terminal and continues the discussion. The input is the "adjusted response" and the output is the "response display to the user."

[0959] The specific processing unit 290 transmits the result of the specific processing to the headset type terminal 314. In the headset type terminal 314, the control unit 46A causes the speaker 240 and the display 343 to output the result of the specific processing. The microphone 238 acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0960] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0961] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the headset type terminal 314.

[0962] [Fourth embodiment]

[0963] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[0964] 7, a data processing system 410 includes a data processing device 12 and a robot 414. An example of the data processing device 12 is a server.

[0965] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0966] The robot 414 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a control target 443. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the control target 443 are also connected to the bus 52.

[0967] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[0968] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[0969] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0970] The control object 443 includes a display device, LEDs in the eyes, and motors for driving the arms, hands, and feet. The posture and gestures of the robot 414 are controlled by controlling the motors of the arms, hands, and feet. Some of the emotions of the robot 414 can be expressed by controlling these motors. In addition, the facial expressions of the robot 414 can also be expressed by controlling the light emission state of the LEDs in the eyes of the robot 414.

[0971] Fig. 8 shows an example of the main functions of the data processing device 12 and the robot 414. As shown in Fig. 8, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0972] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0973] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0974] In the robot 414, the processor 46 performs the reception output process. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0975] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[0976] The present invention is a system that systematically collects and analyzes information about past and present experts, and utilizes the results to realize high-quality brainstorming. Specific embodiments of the present invention are described below.

[0977] 1. Data collection and database creation

[0978] The first process performed by the server is data collection. Specifically, the server collects data on famous people and experts from sources such as online articles, academic papers, and book databases through scraping or APIs. The collected data is stored in a temporary storage area.

[0979] The server then cleanses the collected data, removes noise, and standardizes the data format. It then uses natural language processing (NLP) technology to tokenize the data and extract entities (people, dates, places, etc.). Based on the extracted entities, it generates profile data of famous people and experts, which are then stored in a database.

[0980] 2. Language Analysis and Model Fine-tuning

[0981] The server analyzes the profile data stored in the database. Specifically, it uses a natural language processing engine to analyze the text data, and performs grammatical analysis and sentiment analysis. Based on the results of this analysis, the speech patterns and styles of each great person or expert are added to the data model.

[0982] The server then inputs the analyzed data into a generative AI model (e.g., GPT-4) and performs transfer learning, which gives the model the ability to imitate a specific person. The fine-tuned model is then stored in storage.

[0983] 3. Scoring system and expert selection

[0984] The user inputs the purpose of the discussion through the device interface. For example, they can set a purpose such as "brainstorm ideas for a new product." The input purpose is then sent from the device to the server.

[0985] The server analyzes the purpose and extracts relevant keywords. It then searches the database for relevant figures and experts based on the keywords and ranks them using a scoring algorithm. The scoring results are sent to the user's device, where the user can review them and select the experts they want to include.

[0986] 4. Brainstorming

[0987] The server launches the generative AI models of the selected experts. The user inputs questions or opinions through the device. For example, if the user asks, "Tell me your ideas for new environmental technologies," the server interprets the question and sends it to the appropriate expert model. The generative AI models generate responses in the style of each expert, synthesize them, and send them to the device.

[0988] Specific examples

[0989] Consider a case where a user sets the theme as "Business ideas related to new environmental technologies." The server scores experts knowledgeable in environmental technologies from the database, and for example, Mr. A, Mr. B, and Mr. C receive high scores. The user checks the scoring results and selects Mr. A, Mr. B, and Mr. C. The server then launches the models of the selected experts, and when the user types "I would like to discuss new product strategies," the server responds in a synthesized form with each expert. In this way, the user can obtain new business ideas from multiple perspectives.

[0990] As described above, this system combines the knowledge of great people from the past and current experts to provide a concrete means for achieving high-quality brainstorming.

[0991] The processing flow will be explained below.

[0992] Step 1:

[0993] The server collects data about famous people and experts from various sources (book databases, web articles, academic papers, etc.) using scraping technology and APIs. The collected data is stored in a temporary storage area.

[0994] Step 2:

[0995] The server cleanses the data in the temporary storage area, removing noise and standardizing the data format, then uses natural language processing (NLP) techniques to tokenize the data and extract entities (people, dates, places, etc.).

[0996] Step 3:

[0997] The server generates profile data of famous people and experts based on the extracted entities, including their names, fields of expertise, and representative achievements. The generated profile data is stored in a database.

[0998] Step 4:

[0999] The server analyzes the profile data stored in the database, performs grammatical analysis and sentiment analysis, and adds the speech patterns and styles of each great person or expert to the data model based on the analysis results.

[1000] Step 5:

[1001] The server inputs the analyzed data into a generative AI model (e.g., GPT-4) for transfer learning. This process allows the generative AI model to mimic the speech and thought patterns of a specific person. Once the fine-tuned model is complete, it is stored in storage.

[1002] Step 6:

[1003] The user uses the device interface to input the purpose of the discussion, such as "brainstorm ideas for a new product." The input purpose is then sent from the device to the server.

[1004] Step 7:

[1005] The server analyzes the purpose of the discussion and extracts relevant keywords, then searches a database for relevant figures and experts based on the keywords.

[1006] Step 8:

[1007] The server uses a scoring algorithm to evaluate the search results and rank them in order of relevance, and the results are sent to the user's device, where the user can review and select the experts they want to include.

[1008] Step 9:

[1009] The server launches a generative AI model of a selected expert, and the user inputs questions or opinions from their device and sends them to the server.

[1010] Step 10:

[1011] The server interprets the user's input and sends questions and opinions to the appropriate expert model. The generative AI model generates responses in each expert's style, synthesizes them, and sends them to the device.

[1012] Step 11:

[1013] The terminal displays the response from the server on the user interface, and the user can check the displayed results and continue brainstorming by entering further questions or opinions as needed.

[1014] The above processing flow allows users to generate new business ideas through discussions with great people and experts from across time and space.

[1015] Example 1

[1016] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1017] Conventional brainstorming systems have difficulty effectively utilizing the knowledge of past and present experts. Furthermore, they lack the means to efficiently select experts suited to specific discussion purposes and achieve high-quality brainstorming. As a result, the quality of discussions declines, and it is difficult to obtain new ideas and insights.

[1018] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[1019] In this invention, the server includes means for collecting data from information sources such as online articles, papers, and books, means for cleansing the data and extracting entities using natural language processing technology, means for storing the structured data in a database, means for inputting the purpose of the discussion and extracting related keywords, means for scoring experts suitable for the specific purpose of the discussion from the database, means for generating and launching a generative AI model of the selected expert based on the scoring results, and means for receiving questions from users via the terminal, sending the questions to the generative AI model, generating responses, and sending them to the terminal. This makes it possible to combine the knowledge of past and present experts and efficiently realize multifaceted and advanced brainstorming.

[1020] "Articles on the Internet" refers to text data and news articles posted on websites.

[1021] A "paper" refers to a document that describes the results of academic research or a document that reports the results of research.

[1022] "Book" means a printed document that has been published as a publication.

[1023] "Means of collecting data" refers to the methods and tools used to obtain the required information from the source.

[1024] "Data cleansing methods" refer to methods and tools used to remove noise and inaccurate information from collected data and format it.

[1025] "Natural language processing technology" refers to technology that enables computers to understand and process human language.

[1026] "Entity extraction" refers to methods and tools that extract important information, such as specific people, dates, or places, from text data.

[1027] "Structured data" refers to data that has been organized and arranged in a format suitable for a database.

[1028] "Means of storing data in a database" refers to the methods and tools used to store data in a database.

[1029] The "purpose of the discussion" refers to the goal or theme you want to achieve when having the discussion.

[1030] "Keyword extraction methods" refers to methods and tools for finding related words and phrases from text data.

[1031] A "scoring method" refers to a method or tool that evaluates the importance or relevance of data based on specific criteria.

[1032] A "generative AI model" refers to an artificial intelligence model that has the ability to perform specific tasks based on training data.

[1033] "Means for receiving questions from users" refers to the methods and tools by which the system obtains questions and opinions entered by users.

[1034] "Means for sending questions to a generative AI model" refers to methods or tools for communicating a user's questions to a generative AI model.

[1035] "Means for generating a response" refers to the methods and tools that a generative AI model uses to create an answer to a user's question.

[1036] "Means for sending to terminal" refers to a method or tool for transmitting the generated response to the user's terminal.

[1037] The present invention is a system that systematically collects and analyzes information about past and present experts, and utilizes the results to realize high-quality brainstorming. Specific embodiments of the present invention are described below.

[1038] Data collection and database creation

[1039] The server first collects information from online articles, academic papers, book databases, etc. Specifically, the server uses web scraping tools such as Python's BeautifulSoup and Scrapy to obtain information about famous people and experts. For academic papers, the server also uses the Google Scholar API to collect data. This collected data is then stored in a temporary storage area (for example, a database such as MySQL).

[1040] Next, the server cleanses the data using the Pandas library. Specifically, it removes unnecessary data and noise, converts date and time formats, and normalizes text. The cleansed data is then tokenized using natural language processing (NLP) techniques to extract entities (people, dates, places, etc.). This analysis is performed using libraries such as NLTK and spaCy. Finally, the structured data is stored in a database such as PostgreSQL.

[1041] Language analysis and model fine-tuning

[1042] The server analyzes the profile data stored in the database. For the analysis, it uses a natural language processing engine such as BERT to perform grammatical analysis and sentiment analysis. Based on the results of this analysis, speech patterns and styles are added to the data model.

[1043] The server then inputs the analyzed data into a generative AI model such as GPT-4 to perform transfer learning. This transfer learning is performed using PyTorch or TensorFlow. The generative AI model, which has acquired the ability to imitate a specific person through transfer learning, is stored in storage such as AWS S3.

[1044] Scoring system and expert selection

[1045] The user inputs the purpose of the discussion through the terminal interface, for example, "Brainstorm ideas for a new product." The input purpose is sent from the terminal to the server.

[1046] The server analyzes the purpose and extracts relevant keywords. It then uses SQL queries to search for relevant experts from the database and scores them using TF-IDF and other machine learning algorithms. The scoring results are sent to the device, where the user can review them and select the desired expert.

[1047] Brainstorming

[1048] The server launches a generative AI model of the selected expert. This is done using container orchestration tools such as Docker or Kubernetes. The user inputs a question or opinion through the terminal. For example, they might input, "Tell me your ideas for new environmental technologies." The server interprets the question and sends it to the appropriate expert model. The generative AI model generates a response in the expert's style, synthesizes it, and sends it to the terminal.

[1049] Specific examples

[1050] If a user selects the theme "Business ideas related to new environmental technologies," the server scores experts knowledgeable in environmental technologies from the database, and for example, Mr. A, Mr. B, and Mr. C receive high scores. The user checks the scoring results and selects Mr. A, Mr. B, and Mr. C. The server then activates the model of the selected experts, and when the user enters "I would like to discuss new product strategies," the server generates a response in a synthesized form from each expert and sends it to the terminal.

[1051] In this way, the system of the present invention combines the knowledge of past and present experts and provides a concrete means for realizing high-quality brainstorming.

[1052] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1053] Step 1: Data collection

[1054] The server collects data from online articles, academic papers, book databases, and other sources. Specifically, it uses web scraping tools such as Python's BeautifulSoup and Scrapy to obtain information about famous people and experts. For academic papers, it also uses the Google Scholar API to obtain paper data. Inputs include URLs and API endpoints, and the output is the obtained raw data stored in a temporary storage area. This process collects a large amount of text data.

[1055] Step 2: Data cleansing

[1056] The server cleanses the collected data. Specifically, it uses the Pandas library to manipulate data frames and remove unnecessary data and noise. This step also performs date and time format conversions and text normalization. The input is raw data stored in a temporary storage area, and the output is clean, cleaned data. This process improves the quality of the data and makes subsequent analysis easier.

[1057] Step 3: Data analysis using natural language processing

[1058] The server analyzes the cleansed data using natural language processing techniques. Specifically, it uses libraries such as NLTK and spaCy to tokenize the data and extract entities (people, dates, places, etc.). The input is the cleansed data, and the output is extracted entity data. This processing structures the text data and generates expert profile data.

[1059] Step 4: Store in the database

[1060] The server stores the extracted entity data in a database such as PostgreSQL. The input is the entity data, and the output is a structured database entry. This process allows for efficient management of expert information and facilitates subsequent search and analysis.

[1061] Step 5: Input and Analysis of Desired Objectives

[1062] The user inputs the purpose of the discussion through the terminal interface. For example, they might input "brainstorm ideas for a new product." The input is the discussion purpose entered by the user, and the output is text data sent to the server. This step transmits the user's discussion purpose to the server.

[1063] Step 6: Keyword extraction and search

[1064] The server analyzes the received discussion purpose and extracts relevant keywords. It then searches for relevant experts from the database using an SQL query. The inputs are the analyzed text data and the database, and the output is a list of relevant experts. This process identifies experts suitable for the discussion purpose.

[1065] Step 7: Scoring

[1066] The server scores experts using a scoring algorithm (e.g., TF-IDF or other machine learning algorithm) against the relevant data. The input is a list of experts and relevant data, and the output is a scored list of experts. This process prioritizes the appropriate experts.

[1067] Step 8: Selecting experts

[1068] The user checks the scoring results on their device and selects the experts they want to participate in. The input is a list of scored experts, and the output is information about the selected experts sent to the server. This process allows the user to select the experts they want.

[1069] Step 9: Launching the expert model

[1070] The server launches the generative AI model of the selected expert. Specifically, the model is launched using a container orchestration tool such as Docker or Kubernetes. The input is information about the selected expert, and the output is the launched generative AI model. This process puts the expert model into operation.

[1071] Step 10: Brainstorming and response generation

[1072] A user inputs a question or opinion through a device. For example, they might input, "Tell me your ideas for new environmental technologies." The input is the user's question text, and the output is sent to the generative AI model. The generative AI model then generates a response based on the question and sends it in a synthesized form to the user's device. This process allows the user to obtain high-quality opinions and suggestions.

[1073] (Application example 1)

[1074] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1075] The development of autonomous vehicles requires advanced algorithms and safety measures, but there is a lack of effective means to integrate expert knowledge. There is also a need for efficient methods to perform simulations based on expert knowledge and to verify new ideas and technical solutions.

[1076] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[1077] In this invention, the server includes means for collecting data on past and present experts, means for processing and structuring the data, means for storing the structured data in a database, means for scoring experts suitable for a specific discussion purpose from the database, means for generating a model of the scored experts and holding discussions, and means for running a simulation based on the expert model. This makes it possible to effectively integrate the latest expert knowledge on autonomous vehicles and verify optimal operation algorithms and safety measures through simulation.

[1078] "Past and present expert data" is a collection of information about people who have past and present expertise and experience in a particular field.

[1079] "Means of collection" refers to the technical means used to obtain data from online papers, articles, book databases, etc.

[1080] "Means of processing and structuring" refers to technical means of systematically organizing collected data by processing it through data cleansing, tokenization, entity extraction, etc.

[1081] "Means of storing data in a database" means a technical means of efficiently storing structured data and keeping it in a state that can be quickly retrieved and used as needed.

[1082] An "expert appropriate for a particular purpose" is someone with particular knowledge and experience in a field relevant to the topic or goal of the discussion.

[1083] A "scoring means" is a technological means that uses specific criteria or algorithms to assess and rank the suitability or relevance of Experts.

[1084] "Means for generating a model of a scored expert and engaging in discussions" refers to the technical means for creating a generative AI model that mimics the style and speech patterns of a scored expert and using it to engage in discussions and responses.

[1085] A "generative AI model" is an artificial intelligence model that uses machine learning technology to mimic the knowledge and speaking style of experts.

[1086] "Means for performing simulations" refers to the technical means for verifying the operation algorithms and safety measures of autonomous vehicles in a virtual environment using the knowledge of the generated expert model.

[1087] The present invention provides a system for leveraging expert knowledge in the development and validation of autonomous vehicles. The system uses the following major hardware and software components:

[1088] 1. Data Collection

[1089] The server collects expert data on autonomous driving technology from academic papers, technical articles, and various book databases on the Internet. Specifically, it retrieves data using the URL Fetch API and performs web scraping using Beautiful Soup (Python).

[1090] 2. Data Analysis and Natural Language Processing

[1091] To process and structure the collected data, the server uses natural language processing techniques, specifically NLP libraries such as spaCy and NLTK for data cleansing, tokenization, and entity extraction, and then stores the structured data in a database.

[1092] 3. Expert Selection and Fine-Tuning

[1093] To select the most suitable experts for a specific discussion purpose, the server searches the data in the database and evaluates their relevance using scoring algorithms such as TF-IDF and BERT. Using the associated expert data, the GPT-4 model is fine-tuned using Hugging Face Transformers, resulting in a generative AI model that mimics each expert's speaking patterns and style.

[1094] 4. Simulation and Verification

[1095] Users can input questions about autonomous driving or the purpose of their discussion through their devices. For example, "What are the safety measures for autonomous vehicles in urban areas?" The server generates a response based on the relevant expert model. This response is used to run simulations in a virtual environment to verify optimal driving algorithms and safety measures.

[1096] Specific examples

[1097] For example, if a user types in "What are the safety measures for self-driving vehicles in urban areas?", the system will act as follows:

[1098] 1. Prompt statement generation (server):

[1099] User Query: "What are the safety measures for autonomous vehicles in urban areas?"

[1100] Model Prompt: "The system will discuss safety measures for autonomous vehicles in urban areas. Please suggest the best solution based on past and current expert knowledge."

[1101] 2. Generative AI model response (server):

[1102] Selected expert models respond to questions and their answers are used in the simulation.

[1103] 3. Simulation and result verification (server):

[1104] Response-based simulations are run in a virtual environment to verify optimal operating algorithms and safety measures.

[1105] This allows users to gain a multifaceted perspective on the development and operation management of autonomous vehicles while integrating the latest expert knowledge. By effectively utilizing the knowledge of past and present experts, this system can significantly improve the safety and efficiency of autonomous driving technology.

[1106] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1107] Step 1:

[1108] The server collects data related to autonomous driving technology. It uses the URL Fetch API to retrieve academic papers and technical articles from the Internet and performs web scraping using Beautiful Soup (Python). The input is a list of web page URLs, and the output is text data related to the scraped experts.

[1109] Step 2:

[1110] The server analyzes the collected text data using natural language processing technology. Specifically, the server uses spaCy or NLTK to cleanse, tokenize, and extract entities from the data. The input is scraped text data, and the output is structured data (such as entity information).

[1111] Step 3:

[1112] The server stores the structured data in a database, specifically by inserting well-organized data into a SQL or NoSQL database. The input is structured data, and the output is an updated database.

[1113] Step 4:

[1114] The server scores experts suitable for a specific discussion purpose. It uses natural language processing technology to extract keywords related to the discussion purpose from the user (e.g., "Safety measures for autonomous vehicles in urban areas") and evaluates their relevance using TF-IDF and BERT. The input is the discussion purpose text from the user and expert data in the database, and the output is a list of scored experts.

[1115] Step 5:

[1116] The server generates a scored expert model and conducts the discussion. Using Hugging Face Transformers, it performs transfer learning on the GPT-4 model to create a generative AI model that mimics the expert's speech patterns and style. The input is the scored expert data, and the output is the generated expert model.

[1117] Step 6:

[1118] Users input questions about autonomous driving or the purpose of the discussion through their terminal. For example, they might input, "What are the safety measures for autonomous vehicles in urban areas?" The input is the question text, and the output is the request data sent to the server.

[1119] Step 7:

[1120] The server analyzes the user's question, generates a prompt sentence based on the generated expert model, and inputs the prompt sentence to the expert model to obtain a response. The input is the user's question text and the generated expert model, and the output is the response of the expert model.

[1121] Step 8:

[1122] The server runs a simulation based on the expert model's responses and verifies the results. Simulation software is used to experiment with operation algorithms and safety measures in a virtual environment. The input is the expert model's response text, and the output is the simulation results.

[1123] Step 9:

[1124] The server provides the simulation results to the user. The results are displayed on the terminal and presented in a format that is easy for the user to understand. The input is the simulation result data, and the output is the result provided to the user.

[1125] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[1126] This invention combines a system that collects and analyzes information on past and present experts and performs high-quality brainstorming based on the results with an emotion engine that recognizes and analyzes the user's emotions. This system enables responses and discussions to proceed according to the user's emotional state, resulting in more effective discussions.

[1127] 1. Data collection and database creation

[1128] The server first collects data about famous people and experts from sources such as online book databases, web articles, and academic papers. To achieve this, it uses scraping technology and APIs. The collected data is then stored in a temporary storage area.

[1129] The server then cleanses the collected data, removing noise and standardizing the data format. It then uses natural language processing (NLP) technology to tokenize the data and extract entities (people, dates, places, etc.). Based on the extracted entities, it generates profile data of famous people and experts, which are then stored in a database.

[1130] 2. Language Analysis and Model Fine-tuning

[1131] The server analyzes the profile data stored in the database. Specifically, it uses a natural language processing engine to analyze the text data, and performs grammatical analysis and sentiment analysis. Based on the results of this analysis, the speech patterns and styles of each great person or expert are added to the data model.

[1132] The server then inputs the analyzed data into a generative AI model (e.g., GPT-4) for transfer learning. Through this process, the generative AI model acquires the ability to mimic a specific person. Once the fine-tuned model is complete, it is stored in storage.

[1133] 3. Scoring system and expert selection

[1134] Users input the purpose of the discussion through the device interface. For example, they set a specific purpose such as "brainstorm ideas for a new product." The input purpose is sent from the device to the server.

[1135] The server analyzes the purpose and extracts relevant keywords.The server then searches a database for relevant famous people and experts based on the keywords, and uses a scoring algorithm to evaluate and rank them.The scoring results are sent to the terminal, where the user can review them and select the experts they want to participate.

[1136] 4. Utilizing the Emotion Engine

[1137] The server uses an emotion engine in discussions between the user and the expert model to analyze voice and text inputs from the user and recognize the user's emotional state (e.g., joy, anger, sadness, surprise).

[1138] 5. Brainstorming

[1139] The server launches the generative AI models of the selected experts. The user inputs questions or opinions through the device. For example, if the user asks, "Tell me your ideas for new environmental technologies," the server interprets the question and sends it to the appropriate expert model. The generative AI models generate responses in the style of each expert, synthesize them, and send them to the device.

[1140] Furthermore, the emotion engine recognizes the user's emotions, allowing the server to adjust the content and tone of the expert's response. For example, if the user is feeling impatient, the server can provide a calm response to enhance the effectiveness of the discussion.

[1141] Specific examples

[1142] If a user sets the theme as "Business ideas related to new environmental technologies," the server scores experts knowledgeable in environmental technologies from the database, and for example, experts A, B, and C receive high scores. The user checks the scoring results and selects experts A, B, or C. The server launches the models of the selected experts, and when the user asks, "I would like to discuss new product strategies," each expert responds in a synthesized form.

[1143] The server's built-in emotion engine detects the user's emotional state and adjusts the expert model's response as needed. For example, if the user is emotional, the expert model's response can be adjusted to a calm and peaceful tone to facilitate a smooth discussion.

[1144] As described above, this system combines the knowledge of great people from the past and current experts with the user's emotional state, providing a concrete means for achieving high-quality brainstorming and the creation of new business ideas.

[1145] The processing flow will be explained below.

[1146] Step 1:

[1147] The server collects data about famous people and experts from various sources (book databases, web articles, academic papers, etc.) using scraping technology and APIs. The collected data is stored in a temporary storage area.

[1148] Step 2:

[1149] The server cleanses the data in the temporary storage area, removing noise and standardizing the data format, then uses natural language processing (NLP) techniques to tokenize the data and extract entities (people, dates, places, etc.).

[1150] Step 3:

[1151] The server generates profile data of famous people and experts based on the extracted entities, including their names, fields of expertise, and representative achievements. The generated profile data is stored in a database.

[1152] Step 4:

[1153] The server analyzes the profile data stored in the database. Specifically, it uses a natural language processing engine to perform grammatical analysis and sentiment analysis of the text data. Based on the results of this analysis, the speech patterns and styles of each great person or expert are added to the data model.

[1154] Step 5:

[1155] The server inputs the analyzed data into a generative AI model (e.g., GPT-4) and performs transfer learning. This allows the generative AI model to acquire the ability to imitate a specific person. Once fine-tuned, the model is stored in storage.

[1156] Step 6:

[1157] The user inputs the purpose of the discussion using the device interface. For example, they set a purpose such as "brainstorm ideas for a new product." The input purpose is sent from the device to the server.

[1158] Step 7:

[1159] The server analyzes the purpose of the input discussion and extracts relevant keywords, then searches the database for relevant figures and experts based on the keywords.

[1160] Step 8:

[1161] The server uses a scoring algorithm to evaluate the search results and rank the most relevant experts, which are then sent to the device where the user can review and select the experts they want to participate.

[1162] Step 9:

[1163] The server runs a generative AI model of selected experts, which runs on a dedicated cloud infrastructure.

[1164] Step 10:

[1165] Users can input questions or opinions into their devices and send them to the server. For example, they can ask, "Tell me your ideas for new environmental technologies."

[1166] Step 11:

[1167] The server interprets the user's input and sends the question to the appropriate expert model, which then generates a response in the speaking style of each expert.

[1168] Step 12:

[1169] An emotion engine built into the server analyzes the user's voice and text input to assess their emotional state, for example detecting anger or impatience in the user's voice.

[1170] Step 13:

[1171] The server adjusts the content and tone of the generative AI model's response based on the user's emotional state as recognized by the emotion engine. For example, if the user is angry, the expert model's response will be changed to a calmer tone.

[1172] Step 14:

[1173] The server composes the final response and sends it to the terminal.

[1174] Step 15:

[1175] The terminal displays the response from the server on the user interface, and the user can check the displayed results and continue brainstorming by entering further questions or opinions as needed.

[1176] Through these processes, users can engage in discussions with great figures and experts from across time and space, and realize high-quality discussions that are adjusted according to their emotional state.

[1177] Example 2

[1178] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1179] In today's knowledge-based society, it is extremely important to conduct high-quality discussions by utilizing the knowledge of past and present experts. However, the process of extracting relevant information from vast amounts of data and selecting appropriate experts is difficult, especially when it comes to taking user emotions into account. This has led to a growing need for a system that recognizes the user's emotional state and generates responses based on the expert's style.

[1180] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[1181] In this invention, the server includes means for collecting data on past and present experts, means for cleansing the data and analyzing it using natural language processing technology, means for storing structured data generated from the data in a database, means for scoring experts suitable for a specific discussion purpose from the database, means for generating a model of the scored experts and holding a discussion, means for recognizing the emotional state of a user during the discussion and adjusting the progress of the discussion, and means for generating expert responses using the generative AI model during the discussion. This makes it possible to conduct high-quality discussions by making the most of expert knowledge and effectively progressing the discussion in accordance with the emotional state of the user.

[1182] "Data Collection Tools" are tools used to gather information about past and present professionals from the Internet and other databases.

[1183] "Data cleansing means" refers to a means for removing noise from collected data and standardizing the data format.

[1184] "Natural language processing technology" is a technology that analyzes text data and performs tokenization, entity extraction, sentiment analysis, etc.

[1185] A "database" is a system that stores the generated structured data and retrieves and uses it when needed.

[1186] A "scoring tool" is a tool for searching a database for experts suitable for a particular discussion purpose, assessing their relevance, and ranking them.

[1187] An "expert model generation means" is a means for creating a generative AI model that mimics a scored expert.

[1188] The "discussion tool" is a tool that uses a generative AI model to generate responses based on the expert's style and engages in discussions between the user and the expert.

[1189] The "emotion recognition means" is a means for recognizing the emotional state of users during a discussion and adjusting the content and tone of the discussion.

[1190] A "generative AI model" is an artificial intelligence technology that can learn from large amounts of text data and generate text for specific tasks.

[1191] This invention is a system that collects and analyzes information about past and present experts, and uses the results to conduct high-quality discussions. By combining it with an emotion engine that recognizes and analyzes the user's emotions, it is possible to respond and progress the discussion in accordance with the user's emotional state. Specific embodiments for implementing this system are described below.

[1192] Data collection

[1193] The server first collects data about experts from sources such as online book databases, web articles, and academic papers. This is done using scraping tools such as BeautifulSoup and Scrapy, and APIs such as the Google Books API. The collected data is then stored in a temporary storage area. Specifically, the server collects titles, author names, citations, etc. from the book database and stores them in the temporary storage area.

[1194] Data Cleansing and Analysis

[1195] The collected data is cleansed by the server. Specifically, the Pandas library is used to fill in missing data and remove inaccurate data. Natural language processing (NLP) techniques are then used to tokenize the data, and entities (people, dates, places, etc.) are extracted using tools such as SpaCy. Expert profile data is generated based on the extracted entities, and this data is stored in a MySQL database.

[1196] Fine-tuning the model

[1197] The server analyzes the profile data stored in the database. It uses natural language processing engines such as BERT and GPT-3 to perform grammatical and sentiment analysis of the text data, and adds each expert's speech patterns and style to the data model. The analyzed data is then input into a generative AI model (e.g., GPT-4) for transfer learning. Once the fine-tuned model is complete, it is stored in Amazon S3 storage.

[1198] Scoring and Expert Selection

[1199] The user inputs the purpose of the discussion through the device interface. For example, they can set a specific purpose such as "brainstorm ideas for a new product." The input purpose is sent from the device to the server. The server analyzes the purpose and extracts related keywords. It then searches for appropriate experts in a database and uses a scoring algorithm to evaluate their relevance and rank them. The scoring results are sent to the device, where the user can review them and select the experts they want to participate in.

[1200] Emotion recognition and discussion moderation

[1201] The server uses an emotion engine in the discussion between the user and the expert model. The emotion engine uses the Google Cloud Natural Language API and Microsoft Azure Text Analytics API to analyze the user's voice and text input and recognize their emotional state (joy, anger, sadness, surprise, etc.). Based on the recognized emotion, the content and tone of the expert model's response are adjusted.

[1202] Brainstorming

[1203] The server launches a generative AI model of a selected expert. The user inputs a question or opinion through their device. For example, they might ask, "Tell me your ideas for new environmental technologies." The server interprets the question and sends it to the appropriate expert model. The generative AI model generates a response in the style of each expert, synthesizes it, and sends it to the device. An emotion engine recognizes the user's emotions and adjusts the expert model's response as needed.

[1204] An example of a specific prompt is, "You are selecting environmental technology experts A, B, and C to discuss a business idea related to a new environmental technology. Please begin the expert model response."

[1205] In this way, the embodiment of the present invention provides a system that makes the most of the knowledge of experts and enables effective discussion to proceed in accordance with the emotional state of the user.

[1206] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1207] Step 1:

[1208] Data collection

[1209] The server collects data about experts from sources such as book databases, web articles, and academic papers on the Internet. The input is the URL or API endpoint of the source, and the output is raw data stored in a temporary storage area. Specific operations include scraping web pages using Scrapy and BeautifulSoup and retrieving book data from the Google Books API.

[1210] Step 2:

[1211] Data Cleansing and Analysis

[1212] The server cleanses the collected data, removing noise and standardizing the data format. The input is the collected raw data, and the output is a clean dataset. Specifically, it uses the Pandas library to impute missing values ​​and remove inaccurate data. It then uses natural language processing techniques (e.g., SpaCy) to tokenize the text data and extract entities (people, dates, places, etc.).

[1213] Step 3:

[1214] Storage in the database

[1215] The server stores the cleansed data in a MySQL database. The input is a clean dataset and the output is structured database entries. Specifically, it inserts the data into the database using the MySQL INSERT statement.

[1216] Step 4:

[1217] Expert Scoring and Selection

[1218] The server analyzes the purpose of the discussion and searches for and scores experts related to specific keywords in the database. The input is the purpose of the discussion provided by the user, and the output is a list of scored experts. Specifically, it uses natural language processing technology to extract keywords from the input purpose and searches for experts based on those keywords. Next, it evaluates the relevance using a scoring algorithm (e.g., TF-IDF) and ranks them.

[1219] Step 5:

[1220] emotion recognition

[1221] The server uses an emotion engine to recognize the user's emotional state during discussions. The input is voice or text input from the user, and the output is analyzed emotion data. Specifically, it analyzes text and voice data using the Google Cloud Natural Language API and Microsoft Azure Text Analytics API to classify the emotional state.

[1222] Step 6:

[1223] Brainstorming

[1224] The server uses the selected expert model to hold discussions with the user. The input is the user's question or opinion, and the output is a response text. Specifically, it uses a generative AI model (e.g., GPT-4) to understand the user's question and generate a response that adheres to the expert's style. It also uses emotion recognition data during the discussion to adjust the tone and content of the response, and sends the final response to the user's device.

[1225] An example of a prompt sentence would be, "Environmental technology experts A, B, and C are being selected to discuss business ideas related to new environmental technologies. Please begin the expert model's response."

[1226] (Application example 2)

[1227] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1228] Conventional brainstorming systems have limitations in terms of obtaining expert knowledge and holding discussions, particularly in terms of adjusting responses to take into account the user's emotional state. As a result, discussions are ineffective and users' opinions and suggestions are not fully elicited. Furthermore, there is a lack of concrete support for improving production lines in factories. Therefore, there is a need for tools that are suitable for on-site workers and engineers to explore improvement ideas in real time.

[1229] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[1230] In this invention, the server includes: means for collecting data on past and present experts; means for processing and structuring the data; means for storing the structured data in a database; means for scoring experts suitable for a specific discussion purpose from the database; means for generating models of the scored experts and holding discussions; emotion processing means for recognizing and analyzing the user's emotional state; and means for adjusting the tone of responses according to the user's emotional state. This enables high-quality brainstorming based on the expert's knowledge, and more effective discussions by adjusting responses to the user's emotional state. It also functions as a concrete support tool for improving the efficiency of production lines in factories and exploring improvement proposals, enabling highly immediate discussions on-site.

[1231] "Means for collecting data" refers to means for obtaining data on past and present experts from sources such as online book databases, web articles, and academic papers.

[1232] "Means for processing and structuring data" refers to means for cleansing collected data, removing noise, standardizing data formats, tokenizing using natural language processing technology, extracting entities, and generating structured data.

[1233] "Means for storing in a database" refers to the means for storing processed and structured data in a database in a specific format.

[1234] A "means for scoring experts" is a means for extracting keywords related to a particular discussion purpose from a database and using a scoring algorithm to evaluate relevance and rank experts.

[1235] The "means for holding discussions" is a means for generating a scored expert model and using the generative AI model to imitate the expert's statements and generate responses to user questions and opinions.

[1236] The "emotion processing means" is a means for analyzing voice input and text input from the user and recognizing and analyzing the user's emotional state (joy, anger, sadness, surprise, etc.).

[1237] The "means for adjusting the tone of a response" is a means for adjusting the tone of a response to be generated in accordance with the emotional state of the user recognized by the emotion processing means.

[1238] The system of the present invention is implemented with the following configuration.

[1239] First, the server collects data on past and present experts from sources such as online book databases, web articles, and academic papers. This data is collected using web scraping technology and APIs. The collected data is then stored in a temporary storage area, and then noise is removed and the data format is standardized using data cleansing technology.

[1240] The server then uses natural language processing (NLP) techniques to tokenize the collected data and extract entities such as people, dates, and places, which are then used to generate expert profile data and store it in a database.

[1241] The server also analyzes the profile data stored in the database and uses a natural language processing engine to perform grammatical analysis and sentiment analysis of the text data. Based on the analysis results, each expert's speech patterns and style are added to the data model. The analyzed data is then input into a generative AI model (e.g., GPT-4) for transfer learning. Through this process, the generative AI model acquires the ability to imitate a specific person, and the fine-tuned model is stored in storage.

[1242] Users input the purpose of the discussion through the device interface. The server analyzes the input purpose and extracts relevant keywords. It then searches a database for appropriate experts based on the keywords and uses a scoring algorithm to evaluate and rank the relevance. The scoring results are sent to the device, where the user can review them and select the experts they want to invite.

[1243] Furthermore, the server uses an emotion processing engine during discussions between the user and the expert model. This emotion processing engine analyzes the user's voice and text inputs to recognize and analyze the user's emotional state. Depending on the user's emotional state, the server adjusts the content and tone of the expert's responses to enhance the effectiveness of the discussion.

[1244] As a specific example, if a user sets the theme as "Improve the efficiency of a new production line" and enters the question "How can we improve production speed?", the server will score relevant experts from the database and launch the expert model selected by the user. The generative AI model will generate an appropriate response, and the emotion engine will adjust the tone of the response according to the user's emotion. An example of a prompt sentence in this case is as follows:

[1245] markdown

[1246] I'd like to hear some ideas for improving the efficiency of our new production line.

[1247] Emotional Text: How can you improve your production speed?

[1248] Tone: calm

[1249] In this way, this system achieves high-quality brainstorming by combining expert knowledge with the user's emotional state. It also functions as a concrete support tool for seeking ideas for streamlining and improving production lines within factories, enabling highly immediate discussions on-site.

[1250] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1251] Step 1:

[1252] The server collects data on past and present experts from sources such as book databases, web articles, and academic papers on the Internet. Specifically, it acquires data using web scraping technology and APIs. The input is a "search query on experts," and the output is the "collected expert data."

[1253] Step 2:

[1254] The server cleanses the collected data, removing noise and standardizing the data format. This is a process that uses natural language processing (NLP) techniques to tokenize and extract entities (people, dates, places, etc.). The input is "collected expert data" and the output is "structured expert data."

[1255] Step 3:

[1256] The server stores the structured data in a database. Specifically, it generates profile data based on entities and saves it in the database. The input is "structured expert data" and the output is "expert profiles in the database."

[1257] Step 4:

[1258] The user inputs the purpose of the discussion through the terminal interface. For example, the topic is "improving the efficiency of a new production line." The input is the "user's purpose of the discussion," and the output is "confirmation of the input of the purpose of the discussion."

[1259] Step 5:

[1260] The server analyzes the discussion purpose entered by the user and extracts related keywords. The input is the "user's discussion purpose" and the output is the "extracted keywords."

[1261] Step 6:

[1262] The server searches for appropriate experts based on the extracted keywords from the database, and evaluates and ranks them using a scoring algorithm. The inputs are the extracted keywords and expert profile data, and the output is the scoring result.

[1263] Step 7:

[1264] The user checks the scoring results through the terminal and selects the experts they want to participate in. The input is the "scoring results" and the output is the "selected experts."

[1265] Step 8:

[1266] The server starts the selected expert model and receives the user's question or opinion as input. The input is "user's question or opinion" and the output is "question transmission to the expert model."

[1267] Step 9:

[1268] The server uses an emotion processing engine to analyze the user's questions and opinions and recognize their emotional state. The input is the user's questions and opinions, and the output is the user's emotional state.

[1269] Step 10:

[1270] The server uses a generative AI model to adjust the content and tone of the expert model's response according to the recognized emotional state of the user. Specifically, it adjusts the tone to "calm" or "gentle" based on data from the emotion engine. The input is the "user's emotional state" and the "expert model's basic response," and the output is the "adjusted response."

[1271] Step 11:

[1272] The server then sends the final generated response to the user's terminal and continues the discussion. The input is the "adjusted response" and the output is the "response display to the user."

[1273] The specific processing unit 290 transmits the result of the specific processing to the robot 414. In the robot 414, the control unit 46A causes the speaker 240 and the control target 443 to output the result of the specific processing. The microphone 238 acquires voice indicating a user input regarding the result of the specific processing. The control unit 46A transmits voice data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the voice data.

[1274] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1275] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the robot 414.

[1276] The emotion identification model 59 as an emotion engine may determine the user's emotion according to a specific mapping. Specifically, the emotion identification model 59 may determine the user's emotion according to an emotion map (see FIG. 9), which is a specific mapping. Similarly, the emotion identification model 59 may determine the robot's emotion, and the identification processing unit 290 may perform identification processing using the robot's emotion.

[1277] FIG. 9 is a diagram illustrating an emotion map 400 on which multiple emotions are mapped. In the emotion map 400, emotions are arranged in concentric circles radiating from the center. Emotions closer to the center of the concentric circles are more primitive. Emotions representing states and actions arising from a state of mind are arranged on the outer edges of the concentric circles. The concept of emotion includes both affect and mental states. Emotions generally generated from reactions occurring in the brain are arranged on the left side of the concentric circles. Emotions generally induced by situational judgment are arranged on the right side of the concentric circles. Emotions generally generated from reactions occurring in the brain and induced by situational judgment are arranged on the upper and lower sides of the concentric circles. Furthermore, the emotion of "pleasure" is arranged on the upper side of the concentric circles, and the emotion of "discomfort" is arranged on the lower side. In this way, in the emotion map 400, multiple emotions are mapped based on the structure by which emotions are generated, and emotions that tend to occur simultaneously are mapped close to each other.

[1278] These emotions are distributed in the 3 o'clock direction on emotion map 400, and typically fluctuate between relief and anxiety. In the right half of emotion map 400, situational awareness dominates over internal sensations, resulting in a sense of calm.

[1279] The inside of emotion map 400 represents what is going on in the mind, and the outside of emotion map 400 represents behavior, so the further you go outside emotion map 400, the more visible the emotions become (the more they are expressed in behavior).

[1280] Human emotions are based on various balances, such as posture and blood sugar levels. When these balances deviate from the ideal, a state of discomfort is indicated, and when they approach the ideal, a state of pleasure is indicated. Emotions can also be created for robots, automobiles, and motorcycles, based on various balances, such as posture and remaining battery life. When these balances deviate from the ideal, a state of discomfort is indicated, and when they approach the ideal, a state of pleasure is indicated. An emotion map can be generated, for example, based on Dr. Mitsuyoshi's emotion map (Research on Voice Emotion Recognition and Emotional Brain Physiological Signal Analysis Systems, Tokushima University, Doctoral Dissertation: https: / / ci.nii.ac.jp / naid / 500000375379). The left half of the emotion map lists emotions belonging to the "reaction" domain, where sensation is dominant. The right half of the emotion map lists emotions belonging to the "situation" domain, where situational awareness is dominant.

[1281] The emotion map defines two emotions that promote learning. One is a negative emotion on the situation side, around the middle of "repentance" or "reflection." In other words, this occurs when the robot experiences negative emotions such as "I never want to feel this way again" or "I don't want to be scolded again." The other is a positive emotion on the response side, around "desire." In other words, this occurs when the robot experiences positive feelings such as "I want more" or "I want to know more."

[1282] The emotion identification model 59 inputs user input into a pre-trained neural network, obtains emotion values ​​indicating each emotion shown in the emotion map 400, and determines the user's emotion. This neural network is pre-trained based on multiple pieces of training data that are combinations of user input and emotion values ​​indicating each emotion shown in the emotion map 400. Furthermore, this neural network is trained so that emotions that are located close to each other have similar values, as in the emotion map 900 shown in FIG. 10. FIG. 10 shows an example in which multiple emotions, "relieved," "calm," and "reassuring," have similar emotion values.

[1283] The system according to the present disclosure has been described above mainly with respect to the functions of the data processing device 12, but the system according to the present disclosure is not necessarily implemented on a server. The system according to the present disclosure may be implemented as a general information processing system. The present disclosure may be implemented, for example, as a software program running on a personal computer or an application running on a smartphone, etc. The method according to the present disclosure may be provided to users in the form of SaaS (Software as a Service).

[1284] In the above embodiment, an example was given in which the specific processing is performed by one computer 22, but the technology of the present disclosure is not limited to this, and the specific processing may be distributed and performed by a plurality of computers including the computer 22. For example, the data generation model 58 may be provided in an external device of the data processing device 12, and data may be generated in the external device in accordance with input data.

[1285] In the above embodiment, an example in which the specific processing program 56 is stored in the storage 32 has been described, but the technology of the present disclosure is not limited to this. For example, the specific processing program 56 may be stored in a portable, computer-readable, non-transitory storage medium such as a USB (Universal Serial Bus) memory. The specific processing program 56 stored in the non-transitory storage medium is installed in the computer 22 of the data processing device 12. The processor 28 executes the specific processing in accordance with the specific processing program 56.

[1286] 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.

[1287] It is not necessary to store all of the specific processing program 56 in a storage device such as a server connected to the data processing device 12 via the network 54, or to store all of the specific processing program 56 in the storage 32; only a portion of the specific processing program 56 may be stored.

[1288] The hardware resource for executing a specific process can be any of the following processors: An example of a processor is a CPU, which is a general-purpose processor that functions as a hardware resource for executing a specific process by executing software, i.e., a program. Another example of a processor is a dedicated electrical circuit, such as an FPGA (Field-Programmable Gate Array), a PLD (Programmable Logic Device), or an ASIC (Application Specific Integrated Circuit), which is a processor with a circuit configuration designed specifically for executing a specific process. Each processor has built-in or connected memory, and each processor uses the memory to execute the specific process.

[1289] The hardware resource that executes the specific processing may be configured with one of these various processors, or may be configured with a combination of two or more processors of the same or different types (for example, a combination of multiple FPGAs, or a combination of a CPU and an FPGA). Also, the hardware resource that executes the specific processing may be a single processor.

[1290] As an example of a system configured with a single processor, first, one processor is configured by combining one or more CPUs and software, and this processor functions as a hardware resource that executes a specific process. Second, there is a system that uses a processor that realizes the functions of an entire system including multiple hardware resources that execute a specific process on a single IC chip, as typified by SoC (System-on-a-chip). In this way, a specific process is realized using one or more of the above-mentioned various processors as hardware resources.

[1291] Furthermore, the hardware structure of these various processors can be, more specifically, an electric circuit that combines circuit elements such as semiconductor devices. The specific processing described above is merely an example. Therefore, it goes without saying that unnecessary steps may be deleted, new steps may be added, or the processing order may be rearranged, without departing from the spirit of the invention.

[1292] The above-described description and illustrations are a detailed explanation of the parts related to the technology of the present disclosure and are merely an example of the technology of the present disclosure. For example, the above description of the configuration, functions, actions, and effects is an explanation of an example of the configuration, functions, actions, and effects of the parts related to the technology of the present disclosure. Therefore, it goes without saying that unnecessary parts may be deleted, new elements may be added, or replacements may be made to the above-described description and illustrations within the scope of the gist of the technology of the present disclosure. Furthermore, to avoid confusion and facilitate understanding of the parts related to the technology of the present disclosure, the above-described description and illustrations omit explanations of common technical knowledge that do not require particular explanation to enable the implementation of the technology of the present disclosure.

[1293] All publications, patent applications, and technical standards mentioned in this specification are herein incorporated by reference to the same extent as if each individual publication, patent application, or technical standard was specifically and individually indicated to be incorporated by reference.

[1294] The following is further disclosed regarding the above embodiment.

[1295] (Claim 1)

[1296] a means of collecting data on past and present professionals;

[1297] means for processing and structuring said data;

[1298] means for storing the structured data in a database;

[1299] A means for scoring experts from said database who are suitable for a particular discussion purpose;

[1300] A system including means for generating and discussing models of the scored experts.

[1301] (Claim 2)

[1302] 2. The system of claim 1, wherein the means for conducting the discussion utilizes a generative AI model to mimic the speech of an expert.

[1303] (Claim 3)

[1304] The system of claim 1, wherein the means for scoring experts suitable for the specific discussion purpose uses natural language processing technology to extract keywords related to the purpose and evaluate their relevance.

[1305] "Example 1"

[1306] (Claim 1)

[1307] A means of collecting data from sources such as internet articles, papers, books, etc.

[1308] means for cleansing the data and extracting entities using natural language processing techniques;

[1309] means for storing the structured data in a database;

[1310] A means for inputting the purpose of the discussion and extracting related keywords;

[1311] A means for scoring experts from said database who are suitable for a particular discussion purpose;

[1312] A means for generating and launching a generative AI model of an expert selected based on the scoring result;

[1313] The system includes means for receiving a question from a user through the terminal, sending the question to a generative AI model, generating a response, and sending the response to the terminal.

[1314] (Claim 2)

[1315] The system of claim 1, characterized in that it uses a generative AI model to mimic the speech of experts and respond to user questions.

[1316] (Claim 3)

[1317] The system according to claim 1, characterized in that it uses natural language processing technology to extract keywords related to the purpose of the discussion and evaluate the relevance.

[1318] "Application Example 1"

[1319] (Claim 1)

[1320] a means of collecting data on past and present professionals;

[1321] means for processing and structuring said data;

[1322] means for storing the structured data in a database;

[1323] A means for scoring experts from said database who are suitable for a particular discussion purpose;

[1324] means for generating and discussing the scored expert models;

[1325] A system including means for performing a simulation based on said expert model.

[1326] (Claim 2)

[1327] 2. The system of claim 1, wherein the system utilizes a generative AI model to mimic expert speech.

[1328] (Claim 3)

[1329] The system of claim 1, wherein the means for scoring experts suitable for the specific discussion purpose uses natural language processing technology to extract keywords related to the purpose and evaluate their relevance.

[1330] "Example 2: Combining Emotion Engines"

[1331] (Claim 1)

[1332] a means of collecting data on past and present professionals;

[1333] means for cleansing the data and analyzing it using natural language processing techniques;

[1334] means for storing structured data generated from the data in a database;

[1335] A means for scoring experts from said database who are suitable for a particular discussion purpose;

[1336] means for generating and discussing the scored expert models;

[1337] means for recognizing the emotional state of the user during the discussion and adjusting the progress of the discussion;

[1338] and generating expert responses using a generative AI model during said discussion.

[1339] A system including:

[1340] (Claim 2)

[1341] 2. The system of claim 1, wherein the system utilizes a generative AI model to mimic expert speech.

[1342] (Claim 3)

[1343] The system of claim 1, wherein the means for scoring experts suitable for the specific discussion purpose uses natural language processing technology to extract terms related to the purpose and evaluate their relevance.

[1344] "Application example 2 when combining emotion engines"

[1345] (Claim 1)

[1346] a means of collecting data on past and present professionals;

[1347] means for processing and structuring said data;

[1348] means for storing the structured data in a database;

[1349] A means for scoring experts from said database who are suitable for a particular discussion purpose;

[1350] means for generating and discussing the scored expert models;

[1351] an emotion processing means for recognizing and analyzing the user's emotional state;

[1352] means for adjusting the tone of the response depending on the emotional state of the user;

[1353] A system including:

[1354] (Claim 2)

[1355] 2. The system of claim 1, wherein the means for conducting the discussion utilizes a generative AI model to mimic the speech of an expert.

[1356] (Claim 3)

[1357] The system of claim 1, wherein the means for scoring experts suitable for the specific discussion purpose uses natural language processing technology to extract keywords related to the purpose and evaluate their relevance. [Explanation of symbols]

[1358] 10, 210, 310, 410 Data Processing Systems 12 Data Processing Device 14 Smart Devices 214 Smart Glasses 314 Headset-type terminal 414 Robot< / url:> < / url:> < / url:> < / url:>

Claims

1. a means of collecting data on past and present professionals; means for processing and structuring said data; means for storing the structured data in a database; A means for scoring experts from said database who are suitable for a particular discussion purpose; A system including means for generating and discussing models of the scored experts.

2. 2. The system of claim 1, wherein the means for conducting the discussion utilizes a generative AI model to mimic the speech of an expert.

3. The system of claim 1, wherein the means for scoring experts suitable for the specific discussion purpose uses natural language processing technology to extract keywords related to the purpose and evaluate relevance.

Images