system

Abstract

We provide the system. [Solution] A method for extracting and pre-processing company data, Means for acquiring and integrating reliable external knowledge data, Methods for training generative models, A means of generating a response to a query from a user, A system that includes means for continuously improving the model.

Description

Technical Field

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

Background Art

[0002] Patent Document 1 discloses a method for controlling a persona chatbot, which is performed by at least one processor and includes steps of receiving a user utterance, adding the user utterance to a prompt including an instruction sentence related to an explanation of a character of the chatbot, encoding the prompt, and inputting the encoded prompt into a language model to generate a chatbot utterance in response to the user utterance.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] Conventional generative AI models have a problem that it is difficult to provide information suitable for specific business operations and situations of companies and organizations because they give responses based on general knowledge. In addition, there are problems of providing misinformation and security risks, and thus accurate and reliable responses are required. Furthermore, there is also a problem that efficiency has not advanced in customer service and customer support, and costs cannot be reduced. To solve these problems, it is necessary to develop a generative AI model specialized for business operations.

Means for Solving the Problems

[0005] This invention provides a system comprising means for extracting and preprocessing proprietary data, means for acquiring and integrating reliable external knowledge data, means for training a generative model, means for generating responses to user queries, and means for continuously improving the model. This enables companies to generate accurate responses based on information specific to their operations, thereby improving the efficiency and reducing costs of customer support. In particular, by adjusting the generative model using user feedback information and further training it using industry-specific data, a higher level of adaptability and accuracy can be achieved.

[0006] "Company data" refers to unique information and datasets related to business operations that are owned by a company or organization.

[0007] "External knowledge data" refers to datasets containing general or specialized knowledge provided by reliable third parties.

[0008] A "generative model" refers to the structure or system of artificial intelligence that uses machine learning algorithms to generate responses and interactions based on input information.

[0009] A "user" refers to an individual or representative of an organization who uses the system to enter queries and receive responses.

[0010] A "query" refers to a question or command that a user enters to request information from a system.

[0011] "Feedback information" refers to information that users use to provide evaluations and opinions about the system's performance and responsiveness.

[0012] "Specialized data" refers to specialized datasets related to specific industries or business operations. [Brief explanation of the drawing]

[0013] [Figure 1] This is a conceptual diagram showing an example of the configuration of a data processing system according to the first embodiment. [Figure 2] This is a conceptual diagram showing an example of the essential functions of a data processing device and a smart device according to the first embodiment. [Figure 3] This is a conceptual diagram showing an example of the configuration of a data processing system according to the second embodiment. [Figure 4] This is a conceptual diagram showing an example of the main functions of a data processing device and smart glasses according to the second embodiment. [Figure 5] This is a conceptual diagram showing an example of the configuration of a data processing system according to the third embodiment. [Figure 6] This is a conceptual diagram showing an example of the main functions of a data processing device and a headset-type terminal according to the third embodiment. [Figure 7] This is a conceptual diagram showing an example of the configuration of a data processing system according to the fourth embodiment. [Figure 8] This is a conceptual diagram showing an example of the main functions of a data processing device and a robot according to the fourth embodiment. [Figure 9] This shows an emotion map where multiple emotions are mapped. [Figure 10] This shows an emotion map where multiple emotions are mapped. [Figure 11] This is a sequence diagram showing the processing flow of the data processing system in Example 1. [Figure 12] This is a sequence diagram showing the processing flow of the data processing system in Application Example 1. [Figure 13] This is a sequence diagram showing the processing flow of the data processing system in Example 2, which incorporates an emotion engine. [Figure 14] This is a sequence diagram showing the processing flow of the data processing system in Application Example 2, which combines an emotion engine. [Modes for carrying out the invention]

[0014] An example of an embodiment of the system according to the technology of the present disclosure will be described below with reference to the accompanying drawings.

[0015] First, the terms used in the following description will be explained.

[0016] In the following embodiments, a tagged processor (hereinafter simply referred to as "processor") may be a single arithmetic unit or a combination of multiple arithmetic units. Also, the processor may be a single type of arithmetic unit or a combination of multiple types of arithmetic units. Examples of arithmetic units include a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), a GPGPU (General-Purpose computing on Graphics Processing Units), an APU (Accelerated Processing Unit), and the like.

[0017] In the following embodiments, a tagged RAM (Random Access Memory) is a memory in which information is temporarily stored and is used as a work memory by the processor.

[0018] In the following embodiments, a tagged storage is one or more non-volatile storage devices that store various programs and various parameters, etc. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), magnetic disks (e.g., hard disks), or magnetic tapes, etc.

[0019] In the following embodiments, the signed communication interface (I / F) is an interface that includes a communication processor and an antenna, etc. The communication interface manages communication between multiple computers. Examples of communication standards applicable to the communication interface include wireless communication standards such as 5G (5th Generation Mobile Communication System), Wi-Fi (registered trademark), or Bluetooth (registered trademark).

[0020] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B." That is, "A and / or B" means that it may be A alone, or B alone, or a combination of A and B. Furthermore, in this specification, the same concept as "A and / or B" applies when expressing three or more things linked by "and / or."

[0021] [First Embodiment]

[0022] Figure 1 shows an example of the configuration of the data processing system 10 according to the first embodiment.

[0023] As shown in Figure 1, the data processing system 10 includes a data processing device 12 and a smart device 14. An example of the data processing device 12 is a server.

[0024] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0025] The smart device 14 comprises a computer 36, a reception device 38, an output device 40, a camera 42, and a communication interface 44. The computer 36 comprises a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The reception device 38, output device 40, and camera 42 are also connected to the bus 52.

[0026] The reception device 38 is equipped with a touch panel 38A and a microphone 38B, etc., and receives user input. The touch panel 38A receives user input by detecting contact with an object (e.g., a pen or finger). The microphone 38B receives user input by detecting the user's voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the data indicating the user input.

[0027] The output device 40 includes a display 40A and a speaker 40B, and presents data to the user 20 by outputting the data in a form perceptible to the user 20 (e.g., audio and / or text). The display 40A displays visible information such as text and images according to instructions from the processor 46. The speaker 40B outputs audio according to instructions from the processor 46. The camera 42 is a small digital camera equipped with an optical system such as a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0028] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various types of information between processor 46 and processor 28 via network 54.

[0029] Figure 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0030] As shown in Figure 2, in the data processing device 12, a specific processing is performed by the processor 28. A specific processing program 56 is stored in the storage 32. The specific processing program 56 is an example of a "program" related to the technology of this disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 according to the specific processing program 56 executed on the RAM 30.

[0031] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0032] In the smart device 14, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. The reception output program 60 is used in conjunction with a specific processing program 56 by the data processing system 10. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[0033] Next, the specific processing performed by the specific processing unit 290 of the data processing device 12 will be described. In the following description, the data processing device 12 will be referred to as the "server" and the smart device 14 as the "terminal".

[0034] One embodiment of this invention is a system that enables companies and organizations to use customized generative AI models specifically for their own use. The basic components of this system and their operation are described below.

[0035] The central element of this system is the process of integrating internal data with external knowledge data to effectively train generative models. The server's first task is to collect data related to the company's operations and customer interactions and preprocess it appropriately. Preprocessing includes data cleaning and formatting standardization. This makes it easier for the model to learn efficiently.

[0036] Next, the server acquires knowledge data from trusted external data sources and integrates it with the company's own data. This includes legal information, industry standards, and general knowledge. Using this constructed dataset, the server trains a generative model. During training, the model is optimized by referring to feedback, particularly to reduce misinformation and improve the accuracy of responses.

[0037] The terminal accepts queries from users. However, since each user has different needs, the server selects information relevant to the user's specific background and edits the response accordingly during the response generation stage. Through this process, the system enables fast and accurate support.

[0038] For example, if a company receives an inquiry about a new product, the user enters this information into the system. The query received by the terminal is sent to a server, where a trained model analyzes the information by combining the company's data with external knowledge data. As a result, an accurate answer regarding the product's characteristics and warranty is generated and sent back to the user via the terminal.

[0039] This process allows companies to reduce the risk of misinformation, increase customer satisfaction, and efficiently manage costs.

[0040] The following describes the processing flow.

[0041] Step 1:

[0042] The server extracts company data from the database. This process includes past customer inquiry data and product-related documents.

[0043] Step 2:

[0044] The server preprocesses the extracted data. Specifically, this involves cleaning the data, imputing missing values, and standardizing the data format. This ensures data consistency.

[0045] Step 3:

[0046] The server connects to reliable external data sources and retrieves the necessary knowledge data. This data includes laws, industry standards, and general knowledge, which are then integrated as pre-cleaned information.

[0047] Step 4:

[0048] The server integrates processed internal data with external data and begins training the generative AI model. At this stage, the model is optimized using supervised learning and reinforcement learning techniques.

[0049] Step 5:

[0050] The server evaluates the training results based on validation data and checks the model's accuracy and performance. If insufficient, it adjusts the training parameters and runs the training again.

[0051] Step 6:

[0052] Users enter their questions through a contact form or chat interface.

[0053] Step 7:

[0054] The terminal receives the user's query and sends it to the server.

[0055] Step 8:

[0056] The server parses the received queries and prepares the data to be input into the model in the most optimal format. At this time, relevant company data and knowledge data are selected and used.

[0057] Step 9:

[0058] The server uses a generated AI model to produce a response from the prepared data. This response is structured in a format that matches the information requested by the user.

[0059] Step 10:

[0060] The server sends the generated response back to the terminal.

[0061] Step 11:

[0062] The terminal displays a response to the user. The user is presented with information that is easy to understand, accurate, and timely.

[0063] Step 12:

[0064] The user provides feedback on the response they received via the terminal.

[0065] Step 13:

[0066] The server receives and analyzes user feedback, accumulating it as data that helps improve the model's accuracy. Once sufficient feedback has been collected, the model is retrained to further improve the system's accuracy.

[0067] (Example 1)

[0068] Next, we will describe Example 1. In the following description, the data processing device 12 will be referred to as the "server," and the smart device 14 will be referred to as the "terminal."

[0069] The goal of using generative AI models is to effectively integrate company-specific data with reliable external data to provide fast, accurate, and optimized responses for each user. A challenge with conventional technologies is that such data integration and individualized responses are not performed efficiently.

[0070] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0071] In this invention, the server includes means for collecting and preprocessing data, means for acquiring and integrating external information, and means for training a generative model. This makes it possible to provide fast and accurate responses to user inquiries while maintaining data consistency.

[0072] "Means of collecting and pre-processing data" refers to methods used by a server to gather necessary information and prepare it in a format that can be analyzed.

[0073] "Methods for acquiring and integrating external information" refers to methods of extracting information from reliable external sources and transforming it into a format that can be used together with one's own company's information.

[0074] "Means for training generative models" are methods for training a model using collected and pre-processed data to improve the accuracy of its responses.

[0075] "Data consistency" means that information obtained from different sources is not contradictory and is in a coherent state.

[0076] "Providing a fast and accurate response" refers to the ability to generate and deliver reliable answers to user inquiries in the shortest possible time.

[0077] To implement this invention, a server plays a central role. The server first collects internal company data using a data management system. Database management software such as PostgreSQL or MySQL® is used in this process. The collected data is preprocessed using a Python script to impute missing data and remove outliers.

[0078] Next, the server obtains external information via the internet, extracting necessary information from reliable public APIs and databases. This data is integrated with the internal data using ETL tools. Using the integrated dataset, the server trains a generative AI model using deep learning libraries such as TENSORFLOW® and PyTorch. During this training process, the model is optimized based on past feedback.

[0079] The terminal receives prompt messages from the user and sends them to the server. For example, if the prompt message is "Please tell me more about the environmental impact of the new product," the server uses a trained model to analyze internal data and external information to generate the optimal response. As a result, detailed and accurate information is quickly provided to the user via the terminal.

[0080] The information generated in this way allows companies to provide highly accurate support to users while improving the overall efficiency of their business processes.

[0081] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0082] Step 1:

[0083] The server collects internal data. The server uses a database management system (PostgreSQL or MySQL) to retrieve business data from the company. The input consists of business records and customer interaction data, and the output is a database containing this data. This information is preprocessed using a Python script to handle missing or outlier values.

[0084] Step 2:

[0085] The server retrieves and integrates external information. The server obtains legal information and general knowledge from trusted public APIs and external databases. The input is the endpoints of the external APIs and databases, and the output is the retrieved external knowledge data. This data is integrated with internal data using ETL tools to create a consistent dataset.

[0086] Step 3:

[0087] The server trains the generative AI model. The server uses a unified dataset to train the model using deep learning libraries such as TensorFlow and PyTorch. The input is the unified dataset, and the output is the trained generative AI model. During model training, parameters are optimized by referencing past feedback to improve the model's accuracy.

[0088] Step 4:

[0089] The terminal accepts prompt messages from the user. The user enters queries through the terminal's interface and sends them to the server. The input is the prompt message entered by the user, and the output is the query sent to the server. For example, a specific question such as "Please tell me about the product's warranty period" is possible.

[0090] Step 5:

[0091] The server generates responses using a trained model. Based on the user's prompt and integrated data, the server uses its trained model to generate accurate answers to questions. The input is the prompt and integrated dataset, and the output is the generated response. The server edits the response to provide information optimized for the user's background.

[0092] Step 6:

[0093] The terminal provides the user with the generated response. The response from the server is transmitted to the user through the terminal. The input is the response data from the server, and the output is the information provided to the end user. This process allows the user to receive quick and accurate answers.

[0094] (Application Example 1)

[0095] Next, we will explain Application Example 1. In the following explanation, the data processing device 12 will be referred to as the "server," and the smart device 14 will be referred to as the "terminal."

[0096] In physical stores, there is a challenge in providing accurate information to customers in real time in response to their inquiries. This challenge can lead to decreased customer satisfaction and, consequently, lost sales opportunities. Furthermore, general inquiry response systems cannot provide accurate responses that integrate with external information. The objective of this invention is to provide an effective method for solving this problem.

[0097] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0098] In this invention, the server includes means for extracting and pre-processing company information, means for acquiring and integrating reliable external knowledge information, means for training a generative model, and means for performing speech recognition in real time and presenting the results on a visual display device. This makes it possible to provide accurate information immediately in response to customer inquiries in physical stores.

[0099] "Company information" refers to data that is independently collected and managed by a specific organization, such as business-related data and customer information.

[0100] "External knowledge information" refers to information from external sources, such as industry standards and legal information, obtained from reliable third parties or publicly available databases.

[0101] A "generative model" is an algorithmic system that uses machine learning techniques to generate information based on data such as text and audio.

[0102] "Training" is the process by which a model uses data to optimize its performance in order to improve its performance towards a defined objective.

[0103] "Real-time speech recognition" is a technology that converts input audio signals into text in real time.

[0104] A "visual display device" refers to hardware equipment, including displays and screens, used to visually present information to users.

[0105] The system implementing this invention aims to provide rapid and accurate information in response to customer inquiries. The system mainly consists of a server, a terminal including a visual display device worn by the user, and a generation model.

[0106] The server first extracts the company's own information and performs preprocessing such as data cleaning and formatting standardization. Next, it obtains external knowledge information from reliable data sources and integrates it with the company's information. This integrated data is used to train generative models, which undergo specialized training for specific customer responses.

[0107] The user wears a visual display device such as smart glasses. The terminal enables voice input and uses speech recognition technology such as Google® Speech-to-Text API to transcribe customer inquiries into text in real time. The recognized text is sent to a server, where a trained generative model analyzes the data and generates the most appropriate response.

[0108] The generated response is displayed on the visual display device, and the user uses it to communicate the necessary information to the customer. For example, if a customer asks, "Is this product heat resistant?", the system quickly displays information about the product's heat resistance and communicates it visually to the store clerk.

[0109] An example of a prompt is one that specifically states the user's question, such as "Product name: Product name, Attribute: Please tell me about the heat resistance performance." Using this prompt, the generative AI model can provide an accurate response.

[0110] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0111] Step 1:

[0112] The server retrieves company information from a database, cleans that information, and standardizes its format. The input is raw company information, and the output is data in a standardized format. Specifically, it removes duplicate data and converts data in different formats to a unified format.

[0113] Step 2:

[0114] The server collects external knowledge information from reliable data sources and integrates it with pre-processed internal information. The input consists of external knowledge information and pre-processed internal information, and the output is an integrated dataset. At this stage, the relevance of the information is verified, and the necessary data is selected and integrated.

[0115] Step 3:

[0116] The server trains a generative AI model using an integrated dataset. The input is the integrated dataset, and the output is the trained generative model. The generative model uses machine learning algorithms to learn patterns from the dataset.

[0117] Step 4:

[0118] The user receives voice input from customers through the device's visual display. The input is voice data, and the output is a text-based query. Specifically, the device uses the Google Speech-to-Text API to convert the voice to text in real time.

[0119] Step 5:

[0120] The server receives a text-based query and generates the optimal response using a trained generative AI model. The input is the query text, and the output is the generated response. The generative model leverages advanced natural language processing to generate appropriate information for the query.

[0121] Step 6:

[0122] The terminal displays the generated response on a visual display device, and the user refers to it to convey information to the customer. The input is the generated response, and the output is the visually presented information. Specifically, text is displayed on the screen, and the user confirms it.

[0123] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[0124] As an embodiment of the present invention, we propose the construction of a system that enables companies to recognize user emotions and respond accordingly in customer service. This system incorporates an emotion engine in addition to conventional generative AI models to simultaneously analyze user queries and the emotions behind them, thereby achieving more appropriate and satisfying responses.

[0125] First, the server trains a generative model using internal data and reliable external data. Furthermore, an emotion engine is developed to analyze emotions from user text input and voice data. This emotion engine uses natural language processing technology to determine the user's emotions (e.g., joy, anger, sadness) in real time.

[0126] The terminal receives queries from the user and extracts sentiment data as part of the response. This sentiment data is sent to the server and analyzed by a pre-trained model. This allows the server to generate the optimal response while taking the user's emotions into consideration.

[0127] As a concrete example, consider a customer support scenario. When a user contacts support while feeling dissatisfied with a product defect, the emotion engine identifies emotions such as "dissatisfaction" or "anger." This allows the server to generate a more careful and sympathetic response. For example, the response might be, "We apologize for the inconvenience. Here are some possible solutions..."

[0128] In this way, this system can improve customer satisfaction by providing customized responses tailored to each user's emotions, going beyond mere information provision. This approach allows companies to offer more effective customer support, potentially contributing to increased brand loyalty.

[0129] The following describes the processing flow.

[0130] Step 1:

[0131] Users enter questions and opinions through contact forms or voice devices. Both text and voice data are accepted as input.

[0132] Step 2:

[0133] The terminal analyzes the user input it receives and extracts both text and audio data. In the case of audio input, speech recognition is performed and the data is converted into text format.

[0134] Step 3:

[0135] The device sends text data to the emotion engine. The emotion engine uses natural language processing techniques to identify emotions from the user's text. In this process, emotion labels such as "joy," "anger," and "sadness" are assigned.

[0136] Step 4:

[0137] The terminal sends a user query containing sentiment data, which is the result of the analysis, to the server. The server integrates the data based on the received sentiment labels and query content.

[0138] Step 5:

[0139] The server uses integrated data to run a generative AI model and generate the optimal response. During this process, it adjusts the wording and tone as needed based on emotion labels.

[0140] Step 6:

[0141] The server sends the generated response back to the terminal.

[0142] Step 7:

[0143] The device displays a response to the user. This response takes the user's emotions into consideration and includes appropriate tone and information.

[0144] Step 8:

[0145] Users provide feedback on the response. This feedback is based on factors such as the speed and accuracy of the response and overall satisfaction.

[0146] Step 9:

[0147] The server collects and analyzes user feedback. This data is used to train future models, contributing to continuous system improvement.

[0148] (Example 2)

[0149] Next, we will describe Example 2. In the following description, the data processing device 12 will be referred to as the "server" and the smart device 14 as the "terminal".

[0150] In modern business customer service, effectively understanding user emotions and providing appropriate responses is crucial. However, traditional systems have limitations in their ability to analyze user emotions accurately in real time and generate optimal responses. This results in decreased customer satisfaction and hinders the improvement of brand loyalty.

[0151] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[0152] In this invention, the server includes means for extracting and preprocessing proprietary data, means for acquiring and integrating reliable external knowledge data, means for training a generative model, means for analyzing sentiment in response to user queries, means for generating responses based on the analyzed sentiment information, and means for continuously improving the model. This enables companies to understand user sentiment and generate optimal responses based on it, thereby improving customer satisfaction and strengthening brand loyalty.

[0153] "Company data" refers to all information owned and collected by a company in the course of its operations, including past interactions with customers and documents generated within the company.

[0154] "External knowledge data" refers to reliable information obtained from outside the company and used in its activities, including publicly available databases and information provided by third-party organizations.

[0155] A "generative model" is a form of artificial intelligence technology that is trained on collected data and has the ability to generate appropriate responses to user input.

[0156] "Sentiment analysis" refers to the process of identifying and analyzing emotions from user input, and is a technology that is performed in real time using natural language processing techniques.

[0157] "Response generation" is the process of constructing and providing an appropriate response based on the user's query and the emotions behind it.

[0158] "Model improvement" refers to the ongoing process of tuning and optimizing generative models and sentiment analysis engines to improve their performance and accuracy.

[0159] "Feedback information" refers to information based on reactions and evaluations obtained from users and the environment, and is data used to improve and optimize the system.

[0160] "Specialized data" refers to information carefully selected according to specific industries or applications, and is data used to maximize the use of its characteristics in training generative models.

[0161] This invention provides a system for companies to recognize user emotions in customer service and provide appropriate responses. This system integrates a generative AI model and an emotion analysis engine to analyze user queries and provide the optimal response.

[0162] First, the server collects internal data as well as reliable external knowledge data. This data is used to train generative AI models and sentiment analysis engines. The generative AI models can generate optimal responses while leveraging natural language processing techniques to discern user emotions and intentions.

[0163] Specific software includes natural language processing libraries and machine learning frameworks, and these tools are used to build and continuously improve generative AI models and sentiment analysis engines.

[0164] The terminal is a device that constantly receives queries from users, analyzing user text and voice data in real time to detect emotional data. By introducing an emotional analysis engine, it accurately determines the user's emotions such as "joy," "anger," and "sadness" and sends this information to the server.

[0165] As a concrete example, consider a case where a user inquires about a product defect. In this case, the device identifies the user's feelings of dissatisfaction and transmits that data to the server. Based on this emotional data, the server uses a generative AI model to generate a response such as, "We apologize for the inconvenience. To help resolve your issue, please try the following steps."

[0166] An example of a prompt to input into a generative AI model might be: "If the user's emotion is 'anger,' generate a careful and sympathetic response."

[0167] This system allows companies to comprehensively understand user emotions and improve the quality of their responses, ultimately leading to a significant increase in customer satisfaction.

[0168] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0169] Step 1:

[0170] The server collects and preprocesses internal data and reliable external knowledge data. Input data includes raw data, logs, and information extracted from external databases. This data is cleansed, normalized, and processed into a format suitable for training generative AI models. The output is data in a format usable as a training dataset.

[0171] Step 2:

[0172] The server trains a generative AI model and an emotion analysis engine. The input is the training dataset formatted in step 1. The model is trained using machine learning algorithms to learn responses to user input. The output is the trained generative AI model and emotion analysis engine.

[0173] Step 3:

[0174] The terminal receives queries from the user. The input is text or audio data provided by the user. The terminal extracts the necessary information from this query and converts it into a standardized input data format through text analysis or speech recognition. The output is the parsed text data.

[0175] Step 4:

[0176] The device analyzes emotions using the parsed text data. The input is the text data obtained in step 3. It runs the emotion analysis engine and identifies the user's emotional state (e.g., joy, anger, sadness). The output is a dataset containing the detected emotions.

[0177] Step 5:

[0178] The server receives sentiment analysis results and generates a response using a generative AI model. The input consists of user query data and sentiment data. The generative AI model determines the optimal response to the input and constructs appropriate phrasing and information based on the prompt. The output is the response sentence to be provided to the user.

[0179] Step 6:

[0180] The terminal provides the user with the response from the server. The input is the response statement generated in step 5. The terminal presents this response to the user visually or aurally. The output is the response information received by the user.

[0181] (Application Example 2)

[0182] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as a "server" and the smart device 14 as a "terminal".

[0183] Traditional content distribution services have not adequately captured users' emotions through responses and content recommendations, making it a challenge to improve user satisfaction. In particular, there is a need for a means to extract and provide appropriate content based on the emotions users experience while viewing.

[0184] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[0185] In this invention, the server includes means for recognizing and analyzing the user's emotions, means for providing appropriate responses and recommendations based on the analyzed user's emotions, and means for constructing a set of learning data using industry-specific specialized data. This makes it possible to optimize the user's viewing experience based on emotions and improve satisfaction.

[0186] "Company information" refers to internal data and knowledge owned by a company, and is generally obtained from databases and documents that exist within the organization.

[0187] "External knowledge" refers to reliable information and data obtained from outside the company, and is based on third-party organizations, publicly available statistics, research data, etc.

[0188] A "generative model" refers to an artificial intelligence algorithm trained to generate appropriate responses or outputs in response to user inquiries or other inputs.

[0189] "User" refers to an individual or group that operates or uses the system and is interested in its output.

[0190] "Means for recognizing and analyzing emotions" refers to technologies that analyze a user's emotions from facial expressions, voice, or text data and identify their emotional state.

[0191] "Content recommendation" refers to a method of presenting optimal information and entertainment based on the recognized user's state and preferences.

[0192] A "server" refers to a central processing unit that handles data processing, algorithm execution, and communication with users over a network.

[0193] "Optimizing the viewing experience" refers to making adjustments and providing content tailored to the user's situation in order to increase their satisfaction and interest when consuming media content.

[0194] The system for carrying out the present invention includes three main elements: a server, a terminal, and a user.

[0195] The server first acquires its own information and reliable knowledge data obtained from external sources, and then integrates them. At this time, it uses the integrated dataset to train a generative AI model. The generative AI model is an algorithm for generating appropriate responses to user queries. The server also maintains an emotion engine that recognizes and analyzes user emotions in real time. This engine utilizes machine learning libraries such as TensorFlow and works in conjunction with natural language processing techniques to analyze emotional information.

[0196] The terminal functions as an interface with the user and sends user input to the server. The terminal is integrated into a smartphone, smart glasses, or similar device and utilizes a camera and microphone to collect the user's facial expressions and voice in real time for emotion analysis.

[0197] Users can feel that the content and information they receive through this system are most appropriate to their current emotions. Recommendations generated by the server based on emotion analysis present the most suitable content according to the user's state at that time. For example, when a user is smiling, humorous movies or TV shows can be recommended.

[0198] For example, if the emotion engine detects a user smiling while watching enjoyable content, a prompt message such as "Recommend a movie to watch next that is optimized for the current emotional tone" is input to the AI ​​model, which then provides a list of specific movies that are appropriate for that situation.

[0199] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0200] Step 1:

[0201] The device collects the user's facial expressions and voice. Using its built-in camera and microphone, the device records the user's facial expressions and voice tone in real time. This video and audio data is input to the device and sent to a server for emotion analysis.

[0202] Step 2:

[0203] The server analyzes the received video and audio data using an emotion engine. The emotion engine uses natural language processing techniques and machine learning algorithms to identify the user's emotional state (e.g., joy, sadness, surprise). Specifically, it quantifies the emotional tone based on the analysis results. This analysis result becomes the output, and the process proceeds to the next processing step.

[0204] Step 3:

[0205] The server inputs prompt sentences into a generating AI model based on the analyzed sentiment data, and then generates responses and content recommendations. For example, a prompt sentence such as "The user is happy, so suggest some interesting and fun movies" might be generated. The generating AI model processes this prompt and outputs the most suitable list of movies.

[0206] Step 4:

[0207] The server sends the generated response and content recommendations to the terminal. The terminal presents the received information to the user. As output from the terminal, the user can view, select, and watch the movie list and recommended content displayed on the screen.

[0208] Step 5:

[0209] The system collects user feedback and sends it to the server. The device records user feedback such as viewing time for selected content and post-viewing survey results, and sends this data to the server. This feedback is used to adjust the recommendation algorithm for future content.

[0210] The specific processing unit 290 transmits the result of the specific processing to the smart device 14. In the smart device 14, the control unit 46A causes the output device 40 to output the result of the specific processing. The microphone 38B acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0211] Data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of data generation model 58 is ChatGPT (registered trademark) (Internet search).<URL: https: / / openai.com / blog / chatgpt> ), Gemini (registered trademark) (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0212] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and the specific processing may also be performed by the smart device 14.

[0213] [Second Embodiment]

[0214] Figure 3 shows an example of the configuration of the data processing system 210 according to the second embodiment.

[0215] As shown in Figure 3, the data processing system 210 includes a data processing device 12 and smart glasses 214. An example of the data processing device 12 is a server.

[0216] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0217] The smart glasses 214 include a computer 36, a microphone 238, a speaker 240, a camera 42, and a communication interface 44. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, and camera 42 are also connected to the bus 52.

[0218] The microphone 238 receives voice signals from the user 20 and receives instructions from the user 20. The microphone 238 captures the voice signals from the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[0219] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the area around the user 20 (for example, an imaging range defined by a field of view equivalent to the width of a typical healthy person's field of vision).

[0220] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[0221] Figure 4 shows an example of the main functions of the data processing device 12 and the smart glasses 214. As shown in Figure 4, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[0222] The specific processing program 56 is an example of a "program" relating to the technology of this disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0223] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0224] In the smart glasses 214, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[0225] Next, the identification processing performed by the identification processing unit 290 of the data processing device 12 will be described. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal".

[0226] One embodiment of this invention is a system that enables companies and organizations to use customized generative AI models specifically for their own use. The basic components of this system and their operation are described below.

[0227] The central element of this system is the process of integrating internal data with external knowledge data to effectively train generative models. The server's first task is to collect data related to the company's operations and customer interactions and preprocess it appropriately. Preprocessing includes data cleaning and formatting standardization. This makes it easier for the model to learn efficiently.

[0228] Next, the server acquires knowledge data from trusted external data sources and integrates it with the company's own data. This includes legal information, industry standards, and general knowledge. Using this constructed dataset, the server trains a generative model. During training, the model is optimized by referring to feedback, particularly to reduce misinformation and improve the accuracy of responses.

[0229] The terminal accepts queries from users. However, since each user has different needs, the server selects information relevant to the user's specific background and edits the response accordingly during the response generation stage. Through this process, the system enables fast and accurate support.

[0230] For example, if a company receives an inquiry about a new product, the user enters this information into the system. The query received by the terminal is sent to a server, where a trained model analyzes the information by combining the company's data with external knowledge data. As a result, an accurate answer regarding the product's characteristics and warranty is generated and sent back to the user via the terminal.

[0231] This process allows companies to reduce the risk of misinformation, increase customer satisfaction, and efficiently manage costs.

[0232] The following describes the processing flow.

[0233] Step 1:

[0234] The server extracts company data from the database. This process includes past customer inquiry data and product-related documents.

[0235] Step 2:

[0236] The server preprocesses the extracted data. Specifically, this involves cleaning the data, imputing missing values, and standardizing the data format. This ensures data consistency.

[0237] Step 3:

[0238] The server connects to reliable external data sources and retrieves the necessary knowledge data. This data includes laws, industry standards, and general knowledge, which are then integrated as pre-cleaned information.

[0239] Step 4:

[0240] The server integrates processed internal data with external data and begins training the generative AI model. At this stage, the model is optimized using supervised learning and reinforcement learning techniques.

[0241] Step 5:

[0242] The server evaluates the training results based on validation data and checks the model's accuracy and performance. If insufficient, it adjusts the training parameters and runs the training again.

[0243] Step 6:

[0244] Users enter their questions through a contact form or chat interface.

[0245] Step 7:

[0246] The terminal receives the user's query and sends it to the server.

[0247] Step 8:

[0248] The server parses the received queries and prepares the data to be input into the model in the most optimal format. At this time, relevant company data and knowledge data are selected and used.

[0249] Step 9:

[0250] The server uses a generated AI model to produce a response from the prepared data. This response is structured in a format that matches the information requested by the user.

[0251] Step 10:

[0252] The server sends the generated response back to the terminal.

[0253] Step 11:

[0254] The terminal displays a response to the user. The user is presented with information that is easy to understand, accurate, and timely.

[0255] Step 12:

[0256] The user provides feedback on the response they received via the terminal.

[0257] Step 13:

[0258] The server receives and analyzes user feedback, accumulating it as data that helps improve the model's accuracy. Once sufficient feedback has been collected, the model is retrained to further improve the system's accuracy.

[0259] (Example 1)

[0260] Next, we will describe Example 1. In the following description, the data processing device 12 will be referred to as the "server," and the smart glasses 214 will be referred to as the "terminal."

[0261] The goal of using generative AI models is to effectively integrate company-specific data with reliable external data to provide fast, accurate, and optimized responses for each user. A challenge with conventional technologies is that such data integration and individualized responses are not performed efficiently.

[0262] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0263] In this invention, the server includes means for collecting and preprocessing data, means for acquiring and integrating external information, and means for training a generative model. This makes it possible to provide fast and accurate responses to user inquiries while maintaining data consistency.

[0264] "Means of collecting and pre-processing data" refers to methods used by a server to gather necessary information and prepare it in a format that can be analyzed.

[0265] "Methods for acquiring and integrating external information" refers to methods of extracting information from reliable external sources and transforming it into a format that can be used together with one's own company's information.

[0266] "Means for training generative models" are methods for training a model using collected and pre-processed data to improve the accuracy of its responses.

[0267] "Data consistency" means that information obtained from different sources is not contradictory and is in a coherent state.

[0268] "Providing a fast and accurate response" refers to the ability to generate and deliver reliable answers to user inquiries in the shortest possible time.

[0269] To implement this invention, a server plays a central role. The server first collects internal company data using a data management system. Database management software such as PostgreSQL or MySQL is used in this process. The collected data is preprocessed using a Python script to impute missing data and remove outliers.

[0270] Next, the server obtains external information via the internet, extracting necessary data from reliable public APIs and databases. This data is then integrated with the internal data using ETL tools. Using the integrated dataset, the server trains a generative AI model using deep learning libraries such as TensorFlow and PyTorch. During this training process, the model is optimized based on past feedback.

[0271] The terminal receives prompt messages from the user and sends them to the server. For example, if the prompt message is "Please tell me more about the environmental impact of the new product," the server uses a trained model to analyze internal data and external information to generate the optimal response. As a result, detailed and accurate information is quickly provided to the user via the terminal.

[0272] The information generated in this way allows companies to provide highly accurate support to users while improving the overall efficiency of their business processes.

[0273] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0274] Step 1:

[0275] The server collects internal data. The server uses a database management system (PostgreSQL or MySQL) to retrieve business data from the company. The input consists of business records and customer interaction data, and the output is a database containing this data. This information is preprocessed using a Python script to handle missing or outlier values.

[0276] Step 2:

[0277] The server retrieves and integrates external information. The server obtains legal information and general knowledge from reliable public APIs and external databases. The input is the endpoints of external APIs or databases, and the output is the acquired external knowledge data. This data is integrated with internal data using an ETL tool to create a consistent dataset.

[0278] Step 3:

[0279] The server trains a generative AI model. The server uses the integrated dataset to train the model using deep learning libraries such as TensorFlow or PyTorch. The input is the integrated dataset, and the output is the trained generative AI model. During model training, past feedback is referred to optimize parameters and improve model accuracy.

[0280] Step 4:

[0281] The terminal receives a prompt sentence from the user. The user enters a query through the terminal interface and sends it to the server. The input is the prompt sentence entered by the user, and the output is the query sent to the server. For example, specific questions such as "Please tell me about the product's warranty period" can be considered.

[0282] Step 5:

[0283] The server generates a response using the trained model. The server generates an accurate answer to the question using the trained model based on the user's prompt sentence and integrated data. The input is the prompt sentence and the integrated dataset, and the output is the generated response. The server edits the response to provide information optimized for the user's background.

[0284] Step 6:

[0285] Provide the response generated by the terminal to the user. The response from the server is conveyed to the user through the terminal. The input is the response data from the server, and the output is the provision of information to the end user. Through this process, the user can receive a quick and accurate answer.

[0286] (Application Example 1)

[0287] Next, Application Example 1 will be described. In the following description, the data processing device 12 is referred to as the "server", and the smart glasses 214 are referred to as the "terminal".

[0288] In a physical store, there is a problem that it is difficult to provide accurate information in real time in response to customer inquiries. This problem may lead to a decrease in customer satisfaction and ultimately result in a loss of sales opportunities. Furthermore, a general inquiry response system cannot provide an accurate response integrated with external information. The object of the present invention is to provide an effective method to solve this problem.

[0289] The specific processing by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0290] In this invention, the server includes means for extracting its own information and performing preprocessing, means for obtaining and integrating reliable external knowledge information, means for training a generation model, and means for performing real-time speech recognition and presenting the result to a visual display device. Thereby, it becomes possible to immediately provide accurate information in response to customer inquiries in a physical store.

[0291] "Own information" refers to data that is independently collected and managed, such as business-related data and customer information held by a specific organization.

[0292] "External knowledge information" refers to information from external sources such as industry standards and legal information obtained from reliable third parties or publicly available databases.

[0293] A "generative model" is an algorithmic system that uses machine learning techniques to generate information based on data such as text and audio.

[0294] "Training" is the process by which a model uses data to optimize its performance in order to improve its performance towards a defined objective.

[0295] "Real-time speech recognition" is a technology that converts input audio signals into text in real time.

[0296] A "visual display device" refers to hardware equipment, including displays and screens, used to visually present information to users.

[0297] The system implementing this invention aims to provide rapid and accurate information in response to customer inquiries. The system mainly consists of a server, a terminal including a visual display device worn by the user, and a generation model.

[0298] The server first extracts the company's own information and performs preprocessing such as data cleaning and formatting standardization. Next, it obtains external knowledge information from reliable data sources and integrates it with the company's information. This integrated data is used to train generative models, which undergo specialized training for specific customer responses.

[0299] The user wears a visual display device similar to smart glasses. The device enables voice input and uses speech recognition technology such as the Google Speech-to-Text API to transcribe customer inquiries into text in real time. The recognized text is sent to a server, where a trained generative model analyzes the data and generates the most appropriate response.

[0300] The generated response is displayed on the display of the visual display device, and the user conveys the necessary information to the customer based on it. As a specific example, when a customer asks, "Is this product heat-resistant?", the system quickly displays information regarding the heat resistance of the product and visually conveys it to the store clerk.

[0301] As an example of a prompt sentence, it can take a form that specifically indicates the user's question, such as "Product name: [product name], Attribute: Please tell me about the heat resistance performance." Using this prompt sentence, it is possible for the generative AI model to provide an accurate response.

[0302] The flow of the specific process in Application Example 1 will be described using FIG. 12.

[0303] Step 1:

[0304] The server retrieves its own information from the database and performs cleaning and format unification on that information. The input is raw company information, and the output is data in a unified format. As specific operations, it deletes duplicate data and converts data in different formats into a unified format.

[0305] Step 2:

[0306] The server collects external knowledge information from reliable data sources and integrates it with the preprocessed company information. The input is external knowledge information and the preprocessed company information, and the output is an integrated dataset. At this stage, the relevance of the information is confirmed, and the selection and integration of necessary data are performed.

[0307] Step 3:

[0308] The server trains the generative AI model using the integrated dataset. The input is the integrated dataset, and the output is a trained generative model. The generative model uses machine learning algorithms to learn the patterns of the dataset.

[0309] Step 4:

[0310] The user receives voice input from customers through the device's visual display. The input is voice data, and the output is a text-based query. Specifically, the device uses the Google Speech-to-Text API to convert the voice to text in real time.

[0311] Step 5:

[0312] The server receives a text-based query and generates the optimal response using a trained generative AI model. The input is the query text, and the output is the generated response. The generative model leverages advanced natural language processing to generate appropriate information for the query.

[0313] Step 6:

[0314] The terminal displays the generated response on a visual display device, and the user refers to it to convey information to the customer. The input is the generated response, and the output is the visually presented information. Specifically, text is displayed on the screen, and the user confirms it.

[0315] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[0316] As an embodiment of the present invention, we propose the construction of a system that enables companies to recognize user emotions and respond accordingly in customer service. This system incorporates an emotion engine in addition to conventional generative AI models to simultaneously analyze user queries and the emotions behind them, thereby achieving more appropriate and satisfying responses.

[0317] First, the server trains a generative model using internal data and reliable external data. Furthermore, an emotion engine is developed to analyze emotions from user text input and voice data. This emotion engine uses natural language processing technology to determine the user's emotions (e.g., joy, anger, sadness) in real time.

[0318] The terminal receives queries from the user and extracts sentiment data as part of the response. This sentiment data is sent to the server and analyzed by a pre-trained model. This allows the server to generate the optimal response while taking the user's emotions into consideration.

[0319] As a concrete example, consider a customer support scenario. When a user contacts support while feeling dissatisfied with a product defect, the emotion engine identifies emotions such as "dissatisfaction" or "anger." This allows the server to generate a more careful and sympathetic response. For example, the response might be, "We apologize for the inconvenience. Here are some possible solutions..."

[0320] In this way, this system can improve customer satisfaction by providing customized responses tailored to each user's emotions, going beyond mere information provision. This approach allows companies to offer more effective customer support, potentially contributing to increased brand loyalty.

[0321] The following describes the processing flow.

[0322] Step 1:

[0323] Users enter questions and opinions through contact forms or voice devices. Both text and voice data are accepted as input.

[0324] Step 2:

[0325] The terminal analyzes the user input it receives and extracts both text and audio data. In the case of audio input, speech recognition is performed and the data is converted into text format.

[0326] Step 3:

[0327] The device sends text data to the emotion engine. The emotion engine uses natural language processing techniques to identify emotions from the user's text. In this process, emotion labels such as "joy," "anger," and "sadness" are assigned.

[0328] Step 4:

[0329] The terminal sends a user query containing sentiment data, which is the result of the analysis, to the server. The server integrates the data based on the received sentiment labels and query content.

[0330] Step 5:

[0331] The server uses integrated data to run a generative AI model and generate the optimal response. During this process, it adjusts the wording and tone as needed based on emotion labels.

[0332] Step 6:

[0333] The server sends the generated response back to the terminal.

[0334] Step 7:

[0335] The device displays a response to the user. This response takes the user's emotions into consideration and includes appropriate tone and information.

[0336] Step 8:

[0337] Users provide feedback on the response. This feedback is based on factors such as the speed and accuracy of the response and overall satisfaction.

[0338] Step 9:

[0339] The server collects and analyzes user feedback. This data is used to train future models, contributing to continuous system improvement.

[0340] (Example 2)

[0341] Next, we will describe Example 2. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal".

[0342] In modern business customer service, effectively understanding user emotions and providing appropriate responses is crucial. However, traditional systems have limitations in their ability to analyze user emotions accurately in real time and generate optimal responses. This results in decreased customer satisfaction and hinders the improvement of brand loyalty.

[0343] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[0344] In this invention, the server includes means for extracting and preprocessing proprietary data, means for acquiring and integrating reliable external knowledge data, means for training a generative model, means for analyzing sentiment in response to user queries, means for generating responses based on the analyzed sentiment information, and means for continuously improving the model. This enables companies to understand user sentiment and generate optimal responses based on it, thereby improving customer satisfaction and strengthening brand loyalty.

[0345] "Company data" refers to all information owned and collected by a company in the course of its operations, including past interactions with customers and documents generated within the company.

[0346] "External knowledge data" refers to reliable information obtained from outside the company and used in its activities, including publicly available databases and information provided by third-party organizations.

[0347] A "generative model" is a form of artificial intelligence technology that is trained on collected data and has the ability to generate appropriate responses to user input.

[0348] "Sentiment analysis" refers to the process of identifying and analyzing emotions from user input, and is a technology that is performed in real time using natural language processing techniques.

[0349] "Response generation" is the process of constructing and providing an appropriate response based on the user's query and the emotions behind it.

[0350] "Model improvement" refers to the ongoing process of tuning and optimizing generative models and sentiment analysis engines to improve their performance and accuracy.

[0351] "Feedback information" refers to information based on reactions and evaluations obtained from users and the environment, and is data used to improve and optimize the system.

[0352] "Specialized data" refers to information carefully selected according to specific industries or applications, and is data used to maximize the use of its characteristics in training generative models.

[0353] This invention provides a system for companies to recognize user emotions in customer service and provide appropriate responses. This system integrates a generative AI model and an emotion analysis engine to analyze user queries and provide the optimal response.

[0354] First, the server collects internal data as well as reliable external knowledge data. This data is used to train generative AI models and sentiment analysis engines. The generative AI models can generate optimal responses while leveraging natural language processing techniques to discern user emotions and intentions.

[0355] Specific software includes natural language processing libraries and machine learning frameworks, and these tools are used to build and continuously improve generative AI models and sentiment analysis engines.

[0356] The terminal is a device that constantly receives queries from users, analyzing user text and voice data in real time to detect emotional data. By introducing an emotional analysis engine, it accurately determines the user's emotions such as "joy," "anger," and "sadness" and sends this information to the server.

[0357] As a concrete example, consider a case where a user inquires about a product defect. In this case, the device identifies the user's feelings of dissatisfaction and transmits that data to the server. Based on this emotional data, the server uses a generative AI model to generate a response such as, "We apologize for the inconvenience. To help resolve your issue, please try the following steps."

[0358] An example of a prompt to input into a generative AI model might be: "If the user's emotion is 'anger,' generate a careful and sympathetic response."

[0359] This system allows companies to comprehensively understand user emotions and improve the quality of their responses, ultimately leading to a significant increase in customer satisfaction.

[0360] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0361] Step 1:

[0362] The server collects and preprocesses internal data and reliable external knowledge data. Input data includes raw data, logs, and information extracted from external databases. This data is cleansed, normalized, and processed into a format suitable for training generative AI models. The output is data in a format usable as a training dataset.

[0363] Step 2:

[0364] The server trains a generative AI model and an emotion analysis engine. The input is the training dataset formatted in step 1. The model is trained using machine learning algorithms to learn responses to user input. The output is the trained generative AI model and emotion analysis engine.

[0365] Step 3:

[0366] The terminal receives queries from the user. The input is text or audio data provided by the user. The terminal extracts the necessary information from this query and converts it into a standardized input data format through text analysis or speech recognition. The output is the parsed text data.

[0367] Step 4:

[0368] The device analyzes emotions using the parsed text data. The input is the text data obtained in step 3. It runs the emotion analysis engine and identifies the user's emotional state (e.g., joy, anger, sadness). The output is a dataset containing the detected emotions.

[0369] Step 5:

[0370] The server receives sentiment analysis results and generates a response using a generative AI model. The input consists of user query data and sentiment data. The generative AI model determines the optimal response to the input and constructs appropriate phrasing and information based on the prompt. The output is the response sentence to be provided to the user.

[0371] Step 6:

[0372] The terminal provides the user with the response from the server. The input is the response statement generated in step 5. The terminal presents this response to the user visually or aurally. The output is the response information received by the user.

[0373] (Application Example 2)

[0374] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as the "server," and the smart glasses 214 will be referred to as the "terminal."

[0375] Traditional content distribution services have not adequately captured users' emotions through responses and content recommendations, making it a challenge to improve user satisfaction. In particular, there is a need for a means to extract and provide appropriate content based on the emotions users experience while viewing.

[0376] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[0377] In this invention, the server includes means for recognizing and analyzing the user's emotions, means for providing appropriate responses and recommendations based on the analyzed user's emotions, and means for constructing a set of learning data using industry-specific specialized data. This makes it possible to optimize the user's viewing experience based on emotions and improve satisfaction.

[0378] "Company information" refers to internal data and knowledge owned by a company, and is generally obtained from databases and documents that exist within the organization.

[0379] "External knowledge" refers to reliable information and data obtained from outside the company, and is based on third-party organizations, publicly available statistics, research data, etc.

[0380] A "generative model" refers to an artificial intelligence algorithm trained to generate appropriate responses or outputs in response to user inquiries or other inputs.

[0381] "User" refers to an individual or group that operates or uses the system and is interested in its output.

[0382] "Means for recognizing and analyzing emotions" refers to technologies that analyze a user's emotions from facial expressions, voice, or text data and identify their emotional state.

[0383] "Content recommendation" refers to a method of presenting optimal information and entertainment based on the recognized user's state and preferences.

[0384] A "server" refers to a central processing unit that handles data processing, algorithm execution, and communication with users over a network.

[0385] "Optimizing the viewing experience" refers to making adjustments and providing content tailored to the user's situation in order to increase their satisfaction and interest when consuming media content.

[0386] The system for carrying out the present invention includes three main elements: a server, a terminal, and a user.

[0387] The server first acquires its own information and reliable knowledge data obtained from external sources, and then integrates them. At this time, it uses the integrated dataset to train a generative AI model. The generative AI model is an algorithm for generating appropriate responses to user queries. The server also maintains an emotion engine that recognizes and analyzes user emotions in real time. This engine utilizes machine learning libraries such as TensorFlow and works in conjunction with natural language processing techniques to analyze emotional information.

[0388] The terminal functions as an interface with the user and sends user input to the server. The terminal is integrated into a smartphone, smart glasses, or similar device and utilizes a camera and microphone to collect the user's facial expressions and voice in real time for emotion analysis.

[0389] Users can feel that the content and information they receive through this system are most appropriate to their current emotions. Recommendations generated by the server based on emotion analysis present the most suitable content according to the user's state at that time. For example, when a user is smiling, humorous movies or TV shows can be recommended.

[0390] For example, if the emotion engine detects a user smiling while watching enjoyable content, a prompt message such as "Recommend a movie to watch next that is optimized for the current emotional tone" is input to the AI ​​model, which then provides a list of specific movies that are appropriate for that situation.

[0391] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0392] Step 1:

[0393] The device collects the user's facial expressions and voice. Using its built-in camera and microphone, the device records the user's facial expressions and voice tone in real time. This video and audio data is input to the device and sent to a server for emotion analysis.

[0394] Step 2:

[0395] The server analyzes the received video and audio data using an emotion engine. The emotion engine uses natural language processing techniques and machine learning algorithms to identify the user's emotional state (e.g., joy, sadness, surprise). Specifically, it quantifies the emotional tone based on the analysis results. This analysis result becomes the output, and the process proceeds to the next processing step.

[0396] Step 3:

[0397] The server inputs prompt sentences into a generating AI model based on the analyzed sentiment data, and then generates responses and content recommendations. For example, a prompt sentence such as "The user is happy, so suggest some interesting and fun movies" might be generated. The generating AI model processes this prompt and outputs the most suitable list of movies.

[0398] Step 4:

[0399] The server sends the generated response and content recommendations to the terminal. The terminal presents the received information to the user. As output from the terminal, the user can view, select, and watch the movie list and recommended content displayed on the screen.

[0400] Step 5:

[0401] The system collects user feedback and sends it to the server. The device records user feedback such as viewing time for selected content and post-viewing survey results, and sends this data to the server. This feedback is used to adjust the recommendation algorithm for future content.

[0402] The specific processing unit 290 transmits the result of the specific processing to the smart glasses 214. In the smart glasses 214, the control unit 46A causes the speaker 240 to output the result of the specific processing. The microphone 238 acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[0403] Data generation model 58 is a type of so-called generative AI (Artificial Intelligence). One example of data generation model 58 is ChatGPT (Internet Search).<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0404] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and the specific processing may also be performed by the smart glasses 214.

[0405] [Third Embodiment]

[0406] Figure 5 shows an example of the configuration of the data processing system 310 according to the third embodiment.

[0407] As shown in Figure 5, the data processing system 310 includes a data processing device 12 and a headset terminal 314. An example of the data processing device 12 is a server.

[0408] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0409] The headset terminal 314 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication interface 44, and a display 343. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, camera 42, and display 343 are also connected to the bus 52.

[0410] The microphone 238 receives voice signals from the user 20 and receives instructions from the user 20. The microphone 238 captures the voice signals from the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[0411] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the area around the user 20 (for example, an imaging range defined by a field of view equivalent to the width of a typical healthy person's field of vision).

[0412] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[0413] Figure 6 shows an example of the main functions of the data processing device 12 and the headset terminal 314. As shown in Figure 6, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[0414] The specific processing program 56 is an example of a "program" relating to the technology of this disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0415] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0416] In the headset terminal 314, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[0417] Next, the specific processing performed by the specific processing unit 290 of the data processing device 12 will be described. In the following description, the data processing device 12 will be referred to as the "server" and the headset terminal 314 will be referred to as the "terminal".

[0418] One embodiment of this invention is a system that enables companies and organizations to use customized generative AI models specifically for their own use. The basic components of this system and their operation are described below.

[0419] The central element of this system is the process of integrating internal data with external knowledge data to effectively train generative models. The server's first task is to collect data related to the company's operations and customer interactions and preprocess it appropriately. Preprocessing includes data cleaning and formatting standardization. This makes it easier for the model to learn efficiently.

[0420] Next, the server acquires knowledge data from trusted external data sources and integrates it with the company's own data. This includes legal information, industry standards, and general knowledge. Using this constructed dataset, the server trains a generative model. During training, the model is optimized by referring to feedback, particularly to reduce misinformation and improve the accuracy of responses.

[0421] The terminal accepts queries from users. However, since each user has different needs, the server selects information relevant to the user's specific background and edits the response accordingly during the response generation stage. Through this process, the system enables fast and accurate support.

[0422] For example, if a company receives an inquiry about a new product, the user enters this information into the system. The query received by the terminal is sent to a server, where a trained model analyzes the information by combining the company's data with external knowledge data. As a result, an accurate answer regarding the product's characteristics and warranty is generated and sent back to the user via the terminal.

[0423] This process allows companies to reduce the risk of misinformation, increase customer satisfaction, and efficiently manage costs.

[0424] The following describes the processing flow.

[0425] Step 1:

[0426] The server extracts company data from the database. This process includes past customer inquiry data and product-related documents.

[0427] Step 2:

[0428] The server preprocesses the extracted data. Specifically, this involves cleaning the data, imputing missing values, and standardizing the data format. This ensures data consistency.

[0429] Step 3:

[0430] The server connects to reliable external data sources and retrieves the necessary knowledge data. This data includes laws, industry standards, and general knowledge, which are then integrated as pre-cleaned information.

[0431] Step 4:

[0432] The server integrates processed internal data with external data and begins training the generative AI model. At this stage, the model is optimized using supervised learning and reinforcement learning techniques.

[0433] Step 5:

[0434] The server evaluates the training results based on validation data and checks the model's accuracy and performance. If insufficient, it adjusts the training parameters and runs the training again.

[0435] Step 6:

[0436] Users enter their questions through a contact form or chat interface.

[0437] Step 7:

[0438] The terminal receives the user's query and sends it to the server.

[0439] Step 8:

[0440] The server parses the received queries and prepares the data to be input into the model in the most optimal format. At this time, relevant company data and knowledge data are selected and used.

[0441] Step 9:

[0442] The server uses a generated AI model to produce a response from the prepared data. This response is structured in a format that matches the information requested by the user.

[0443] Step 10:

[0444] The server sends the generated response back to the terminal.

[0445] Step 11:

[0446] The terminal displays a response to the user. The user is presented with information that is easy to understand, accurate, and timely.

[0447] Step 12:

[0448] The user provides feedback on the response they received via the terminal.

[0449] Step 13:

[0450] The server receives and analyzes user feedback, accumulating it as data that helps improve the model's accuracy. Once sufficient feedback has been collected, the model is retrained to further improve the system's accuracy.

[0451] (Example 1)

[0452] Next, we will describe Example 1. In the following description, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[0453] The goal of using generative AI models is to effectively integrate company-specific data with reliable external data to provide fast, accurate, and optimized responses for each user. A challenge with conventional technologies is that such data integration and individualized responses are not performed efficiently.

[0454] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0455] In this invention, the server includes means for collecting and preprocessing data, means for acquiring and integrating external information, and means for training a generative model. This makes it possible to provide fast and accurate responses to user inquiries while maintaining data consistency.

[0456] "Means of collecting and pre-processing data" refers to methods used by a server to gather necessary information and prepare it in a format that can be analyzed.

[0457] "Methods for acquiring and integrating external information" refers to methods of extracting information from reliable external sources and transforming it into a format that can be used together with one's own company's information.

[0458] "Means for training generative models" are methods for training a model using collected and pre-processed data to improve the accuracy of its responses.

[0459] "Data consistency" means that information obtained from different sources is not contradictory and is in a coherent state.

[0460] "Providing a fast and accurate response" refers to the ability to generate and deliver reliable answers to user inquiries in the shortest possible time.

[0461] To implement this invention, a server plays a central role. The server first collects internal company data using a data management system. Database management software such as PostgreSQL or MySQL is used in this process. The collected data is preprocessed using a Python script to impute missing data and remove outliers.

[0462] Next, the server obtains external information via the internet, extracting necessary data from reliable public APIs and databases. This data is then integrated with the internal data using ETL tools. Using the integrated dataset, the server trains a generative AI model using deep learning libraries such as TensorFlow and PyTorch. During this training process, the model is optimized based on past feedback.

[0463] The terminal receives prompt messages from the user and sends them to the server. For example, if the prompt message is "Please tell me more about the environmental impact of the new product," the server uses a trained model to analyze internal data and external information to generate the optimal response. As a result, detailed and accurate information is quickly provided to the user via the terminal.

[0464] The information generated in this way allows companies to provide highly accurate support to users while improving the overall efficiency of their business processes.

[0465] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0466] Step 1:

[0467] The server collects internal data. The server uses a database management system (PostgreSQL or MySQL) to retrieve business data from the company. The input consists of business records and customer interaction data, and the output is a database containing this data. This information is preprocessed using a Python script to handle missing or outlier values.

[0468] Step 2:

[0469] The server retrieves and integrates external information. The server obtains legal information and general knowledge from trusted public APIs and external databases. The input is the endpoints of the external APIs and databases, and the output is the retrieved external knowledge data. This data is integrated with internal data using ETL tools to create a consistent dataset.

[0470] Step 3:

[0471] The server trains the generative AI model. The server uses a unified dataset to train the model using deep learning libraries such as TensorFlow and PyTorch. The input is the unified dataset, and the output is the trained generative AI model. During model training, parameters are optimized by referencing past feedback to improve the model's accuracy.

[0472] Step 4:

[0473] The terminal accepts prompt messages from the user. The user enters queries through the terminal's interface and sends them to the server. The input is the prompt message entered by the user, and the output is the query sent to the server. For example, a specific question such as "Please tell me about the product's warranty period" is possible.

[0474] Step 5:

[0475] The server generates responses using a trained model. Based on the user's prompt and integrated data, the server uses its trained model to generate accurate answers to questions. The input is the prompt and integrated dataset, and the output is the generated response. The server edits the response to provide information optimized for the user's background.

[0476] Step 6:

[0477] The terminal provides the user with the generated response. The response from the server is transmitted to the user through the terminal. The input is the response data from the server, and the output is the information provided to the end user. This process allows the user to receive quick and accurate answers.

[0478] (Application Example 1)

[0479] Next, we will explain Application Example 1. In the following explanation, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[0480] In physical stores, there is a challenge in providing accurate information to customers in real time in response to their inquiries. This challenge can lead to decreased customer satisfaction and, consequently, lost sales opportunities. Furthermore, general inquiry response systems cannot provide accurate responses that integrate with external information. The objective of this invention is to provide an effective method for solving this problem.

[0481] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0482] In this invention, the server includes means for extracting and pre-processing company information, means for acquiring and integrating reliable external knowledge information, means for training a generative model, and means for performing speech recognition in real time and presenting the results on a visual display device. This makes it possible to provide accurate information immediately in response to customer inquiries in physical stores.

[0483] "Company information" refers to data that is independently collected and managed by a specific organization, such as business-related data and customer information.

[0484] "External knowledge information" refers to information from external sources, such as industry standards and legal information, obtained from reliable third parties or publicly available databases.

[0485] A "generative model" is an algorithmic system that uses machine learning techniques to generate information based on data such as text and audio.

[0486] "Training" is the process by which a model uses data to optimize its performance in order to improve its performance towards a defined objective.

[0487] "Real-time speech recognition" is a technology that converts input audio signals into text in real time.

[0488] A "visual display device" refers to hardware equipment, including displays and screens, used to visually present information to users.

[0489] The system implementing this invention aims to provide rapid and accurate information in response to customer inquiries. The system mainly consists of a server, a terminal including a visual display device worn by the user, and a generation model.

[0490] The server first extracts the company's own information and performs preprocessing such as data cleaning and formatting standardization. Next, it obtains external knowledge information from reliable data sources and integrates it with the company's information. This integrated data is used to train generative models, which undergo specialized training for specific customer responses.

[0491] The user wears a visual display device similar to smart glasses. The device enables voice input and uses speech recognition technology such as the Google Speech-to-Text API to transcribe customer inquiries into text in real time. The recognized text is sent to a server, where a trained generative model analyzes the data and generates the most appropriate response.

[0492] The generated response is displayed on the visual display device, and the user uses it to communicate the necessary information to the customer. For example, if a customer asks, "Is this product heat resistant?", the system quickly displays information about the product's heat resistance and communicates it visually to the store clerk.

[0493] An example of a prompt is one that specifically states the user's question, such as "Product name: Product name, Attribute: Please tell me about the heat resistance performance." Using this prompt, the generative AI model can provide an accurate response.

[0494] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0495] Step 1:

[0496] The server retrieves company information from a database, cleans that information, and standardizes its format. The input is raw company information, and the output is data in a standardized format. Specifically, it removes duplicate data and converts data in different formats to a unified format.

[0497] Step 2:

[0498] The server collects external knowledge information from reliable data sources and integrates it with pre-processed internal information. The input consists of external knowledge information and pre-processed internal information, and the output is an integrated dataset. At this stage, the relevance of the information is verified, and the necessary data is selected and integrated.

[0499] Step 3:

[0500] The server trains a generative AI model using an integrated dataset. The input is the integrated dataset, and the output is the trained generative model. The generative model uses machine learning algorithms to learn patterns from the dataset.

[0501] Step 4:

[0502] The user receives voice input from customers through the device's visual display. The input is voice data, and the output is a text-based query. Specifically, the device uses the Google Speech-to-Text API to convert the voice to text in real time.

[0503] Step 5:

[0504] The server receives a text-based query and generates the optimal response using a trained generative AI model. The input is the query text, and the output is the generated response. The generative model leverages advanced natural language processing to generate appropriate information for the query.

[0505] Step 6:

[0506] The terminal displays the generated response on a visual display device, and the user refers to it to convey information to the customer. The input is the generated response, and the output is the visually presented information. Specifically, text is displayed on the screen, and the user confirms it.

[0507] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[0508] As an embodiment of the present invention, we propose the construction of a system that enables companies to recognize user emotions and respond accordingly in customer service. This system incorporates an emotion engine in addition to conventional generative AI models to simultaneously analyze user queries and the emotions behind them, thereby achieving more appropriate and satisfying responses.

[0509] First, the server trains a generative model using internal data and reliable external data. Furthermore, an emotion engine is developed to analyze emotions from user text input and voice data. This emotion engine uses natural language processing technology to determine the user's emotions (e.g., joy, anger, sadness) in real time.

[0510] The terminal receives queries from the user and extracts sentiment data as part of the response. This sentiment data is sent to the server and analyzed by a pre-trained model. This allows the server to generate the optimal response while taking the user's emotions into consideration.

[0511] As a concrete example, consider a customer support scenario. When a user contacts support while feeling dissatisfied with a product defect, the emotion engine identifies emotions such as "dissatisfaction" or "anger." This allows the server to generate a more careful and sympathetic response. For example, the response might be, "We apologize for the inconvenience. Here are some possible solutions..."

[0512] In this way, this system can improve customer satisfaction by providing customized responses tailored to each user's emotions, going beyond mere information provision. This approach allows companies to offer more effective customer support, potentially contributing to increased brand loyalty.

[0513] The following describes the processing flow.

[0514] Step 1:

[0515] Users enter questions and opinions through contact forms or voice devices. Both text and voice data are accepted as input.

[0516] Step 2:

[0517] The terminal analyzes the user input it receives and extracts both text and audio data. In the case of audio input, speech recognition is performed and the data is converted into text format.

[0518] Step 3:

[0519] The device sends text data to the emotion engine. The emotion engine uses natural language processing techniques to identify emotions from the user's text. In this process, emotion labels such as "joy," "anger," and "sadness" are assigned.

[0520] Step 4:

[0521] The terminal sends a user query containing sentiment data, which is the result of the analysis, to the server. The server integrates the data based on the received sentiment labels and query content.

[0522] Step 5:

[0523] The server uses integrated data to run a generative AI model and generate the optimal response. During this process, it adjusts the wording and tone as needed based on emotion labels.

[0524] Step 6:

[0525] The server sends the generated response back to the terminal.

[0526] Step 7:

[0527] The device displays a response to the user. This response takes the user's emotions into consideration and includes appropriate tone and information.

[0528] Step 8:

[0529] Users provide feedback on the response. This feedback is based on factors such as the speed and accuracy of the response and overall satisfaction.

[0530] Step 9:

[0531] The server collects and analyzes user feedback. This data is used to train future models, contributing to continuous system improvement.

[0532] (Example 2)

[0533] Next, we will describe Example 2. In the following description, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[0534] In modern business customer service, effectively understanding user emotions and providing appropriate responses is crucial. However, traditional systems have limitations in their ability to analyze user emotions accurately in real time and generate optimal responses. This results in decreased customer satisfaction and hinders the improvement of brand loyalty.

[0535] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[0536] In this invention, the server includes means for extracting and preprocessing proprietary data, means for acquiring and integrating reliable external knowledge data, means for training a generative model, means for analyzing sentiment in response to user queries, means for generating responses based on the analyzed sentiment information, and means for continuously improving the model. This enables companies to understand user sentiment and generate optimal responses based on it, thereby improving customer satisfaction and strengthening brand loyalty.

[0537] "Company data" refers to all information owned and collected by a company in the course of its operations, including past interactions with customers and documents generated within the company.

[0538] "External knowledge data" refers to reliable information obtained from outside the company and used in its activities, including publicly available databases and information provided by third-party organizations.

[0539] A "generative model" is a form of artificial intelligence technology that is trained on collected data and has the ability to generate appropriate responses to user input.

[0540] "Sentiment analysis" refers to the process of identifying and analyzing emotions from user input, and is a technology that is performed in real time using natural language processing techniques.

[0541] "Response generation" is the process of constructing and providing an appropriate response based on the user's query and the emotions behind it.

[0542] "Model improvement" refers to the ongoing process of tuning and optimizing generative models and sentiment analysis engines to improve their performance and accuracy.

[0543] "Feedback information" refers to information based on reactions and evaluations obtained from users and the environment, and is data used to improve and optimize the system.

[0544] "Specialized data" refers to information carefully selected according to specific industries or applications, and is data used to maximize the use of its characteristics in training generative models.

[0545] This invention provides a system for companies to recognize user emotions in customer service and provide appropriate responses. This system integrates a generative AI model and an emotion analysis engine to analyze user queries and provide the optimal response.

[0546] First, the server collects internal data as well as reliable external knowledge data. This data is used to train generative AI models and sentiment analysis engines. The generative AI models can generate optimal responses while leveraging natural language processing techniques to discern user emotions and intentions.

[0547] Specific software includes natural language processing libraries and machine learning frameworks, and these tools are used to build and continuously improve generative AI models and sentiment analysis engines.

[0548] The terminal is a device that constantly receives queries from users, analyzing user text and voice data in real time to detect emotional data. By introducing an emotional analysis engine, it accurately determines the user's emotions such as "joy," "anger," and "sadness" and sends this information to the server.

[0549] As a concrete example, consider a case where a user inquires about a product defect. In this case, the device identifies the user's feelings of dissatisfaction and transmits that data to the server. Based on this emotional data, the server uses a generative AI model to generate a response such as, "We apologize for the inconvenience. To help resolve your issue, please try the following steps."

[0550] An example of a prompt to input into a generative AI model might be: "If the user's emotion is 'anger,' generate a careful and sympathetic response."

[0551] This system allows companies to comprehensively understand user emotions and improve the quality of their responses, ultimately leading to a significant increase in customer satisfaction.

[0552] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0553] Step 1:

[0554] The server collects and preprocesses internal data and reliable external knowledge data. Input data includes raw data, logs, and information extracted from external databases. This data is cleansed, normalized, and processed into a format suitable for training generative AI models. The output is data in a format usable as a training dataset.

[0555] Step 2:

[0556] The server trains a generative AI model and an emotion analysis engine. The input is the training dataset formatted in step 1. The model is trained using machine learning algorithms to learn responses to user input. The output is the trained generative AI model and emotion analysis engine.

[0557] Step 3:

[0558] The terminal receives queries from the user. The input is text or audio data provided by the user. The terminal extracts the necessary information from this query and converts it into a standardized input data format through text analysis or speech recognition. The output is the parsed text data.

[0559] Step 4:

[0560] The device analyzes emotions using the parsed text data. The input is the text data obtained in step 3. It runs the emotion analysis engine and identifies the user's emotional state (e.g., joy, anger, sadness). The output is a dataset containing the detected emotions.

[0561] Step 5:

[0562] The server receives sentiment analysis results and generates a response using a generative AI model. The input consists of user query data and sentiment data. The generative AI model determines the optimal response to the input and constructs appropriate phrasing and information based on the prompt. The output is the response sentence to be provided to the user.

[0563] Step 6:

[0564] The terminal provides the user with the response from the server. The input is the response statement generated in step 5. The terminal presents this response to the user visually or aurally. The output is the response information received by the user.

[0565] (Application Example 2)

[0566] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[0567] Traditional content distribution services have not adequately captured users' emotions through responses and content recommendations, making it a challenge to improve user satisfaction. In particular, there is a need for a means to extract and provide appropriate content based on the emotions users experience while viewing.

[0568] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[0569] In this invention, the server includes means for recognizing and analyzing the user's emotions, means for providing appropriate responses and recommendations based on the analyzed user's emotions, and means for constructing a set of learning data using industry-specific specialized data. This makes it possible to optimize the user's viewing experience based on emotions and improve satisfaction.

[0570] "Company information" refers to internal data and knowledge owned by a company, and is generally obtained from databases and documents that exist within the organization.

[0571] "External knowledge" refers to reliable information and data obtained from outside the company, and is based on third-party organizations, publicly available statistics, research data, etc.

[0572] A "generative model" refers to an artificial intelligence algorithm trained to generate appropriate responses or outputs in response to user inquiries or other inputs.

[0573] "User" refers to an individual or group that operates or uses the system and is interested in its output.

[0574] "Means for recognizing and analyzing emotions" refers to technologies that analyze a user's emotions from facial expressions, voice, or text data and identify their emotional state.

[0575] "Content recommendation" refers to a method of presenting optimal information and entertainment based on the recognized user's state and preferences.

[0576] A "server" refers to a central processing unit that handles data processing, algorithm execution, and communication with users over a network.

[0577] "Optimizing the viewing experience" refers to making adjustments and providing content tailored to the user's situation in order to increase their satisfaction and interest when consuming media content.

[0578] The system for carrying out the present invention includes three main elements: a server, a terminal, and a user.

[0579] The server first acquires its own information and reliable knowledge data obtained from external sources, and then integrates them. At this time, it uses the integrated dataset to train a generative AI model. The generative AI model is an algorithm for generating appropriate responses to user queries. The server also maintains an emotion engine that recognizes and analyzes user emotions in real time. This engine utilizes machine learning libraries such as TensorFlow and works in conjunction with natural language processing techniques to analyze emotional information.

[0580] The terminal functions as an interface with the user and sends user input to the server. The terminal is integrated into a smartphone, smart glasses, or similar device and utilizes a camera and microphone to collect the user's facial expressions and voice in real time for emotion analysis.

[0581] Users can feel that the content and information they receive through this system are most appropriate to their current emotions. Recommendations generated by the server based on emotion analysis present the most suitable content according to the user's state at that time. For example, when a user is smiling, humorous movies or TV shows can be recommended.

[0582] For example, if the emotion engine detects a user smiling while watching enjoyable content, a prompt message such as "Recommend a movie to watch next that is optimized for the current emotional tone" is input to the AI ​​model, which then provides a list of specific movies that are appropriate for that situation.

[0583] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0584] Step 1:

[0585] The device collects the user's facial expressions and voice. Using its built-in camera and microphone, the device records the user's facial expressions and voice tone in real time. This video and audio data is input to the device and sent to a server for emotion analysis.

[0586] Step 2:

[0587] The server analyzes the received video and audio data using an emotion engine. The emotion engine uses natural language processing techniques and machine learning algorithms to identify the user's emotional state (e.g., joy, sadness, surprise). Specifically, it quantifies the emotional tone based on the analysis results. This analysis result becomes the output, and the process proceeds to the next processing step.

[0588] Step 3:

[0589] The server inputs prompt sentences into a generating AI model based on the analyzed sentiment data, and then generates responses and content recommendations. For example, a prompt sentence such as "The user is happy, so suggest some interesting and fun movies" might be generated. The generating AI model processes this prompt and outputs the most suitable list of movies.

[0590] Step 4:

[0591] The server sends the generated response and content recommendations to the terminal. The terminal presents the received information to the user. As output from the terminal, the user can view, select, and watch the movie list and recommended content displayed on the screen.

[0592] Step 5:

[0593] The system collects user feedback and sends it to the server. The device records user feedback such as viewing time for selected content and post-viewing survey results, and sends this data to the server. This feedback is used to adjust the recommendation algorithm for future content.

[0594] The specific processing unit 290 transmits the result of the specific processing to the headset terminal 314. In the headset terminal 314, the control unit 46A causes the speaker 240 and display 343 to output the result of the specific processing. The microphone 238 acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[0595] Data generation model 58 is a type of so-called generative AI (Artificial Intelligence). One example of data generation model 58 is ChatGPT (Internet search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0596] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and specific processing may also be performed by the headset terminal 314.

[0597] [Fourth Embodiment]

[0598] Figure 7 shows an example of the configuration of the data processing system 410 according to the fourth embodiment.

[0599] As shown in Figure 7, the data processing system 410 includes a data processing device 12 and a robot 414. An example of the data processing device 12 is a server.

[0600] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0601] The robot 414 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication interface 44, and a controlled object 443. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, camera 42, and controlled object 443 are also connected to the bus 52.

[0602] The microphone 238 receives voice signals from the user 20 and receives instructions from the user 20. The microphone 238 captures the voice signals from the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[0603] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the area around the user 20 (for example, an imaging range defined by a field of view equivalent to the width of a typical healthy person's field of vision).

[0604] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[0605] The controlled object 443 includes a display device, LEDs in the eyes, and motors that drive the arms, hands, and feet. The posture and gestures of the robot 414 are controlled by controlling the motors of the arms, hands, and feet. Some of the robot 414's emotions can be expressed by controlling these motors. Furthermore, the robot 414's facial expressions can also be expressed by controlling the illumination state of the LEDs in its eyes.

[0606] Figure 8 shows an example of the main functions of the data processing device 12 and the robot 414. As shown in Figure 8, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[0607] The specific processing program 56 is an example of a "program" relating to the technology of this disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0608] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0609] In robot 414, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[0610] Next, the specific processing performed by the specific processing unit 290 of the data processing device 12 will be described. In the following description, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[0611] One embodiment of this invention is a system that enables companies and organizations to use customized generative AI models specifically for their own use. The basic components of this system and their operation are described below.

[0612] The central element of this system is the process of integrating internal data with external knowledge data to effectively train generative models. The server's first task is to collect data related to the company's operations and customer interactions and preprocess it appropriately. Preprocessing includes data cleaning and formatting standardization. This makes it easier for the model to learn efficiently.

[0613] Next, the server acquires knowledge data from trusted external data sources and integrates it with the company's own data. This includes legal information, industry standards, and general knowledge. Using this constructed dataset, the server trains a generative model. During training, the model is optimized by referring to feedback, particularly to reduce misinformation and improve the accuracy of responses.

[0614] The terminal accepts queries from users. However, since each user has different needs, the server selects information relevant to the user's specific background and edits the response accordingly during the response generation stage. Through this process, the system enables fast and accurate support.

[0615] For example, if a company receives an inquiry about a new product, the user enters this information into the system. The query received by the terminal is sent to a server, where a trained model analyzes the information by combining the company's data with external knowledge data. As a result, an accurate answer regarding the product's characteristics and warranty is generated and sent back to the user via the terminal.

[0616] This process allows companies to reduce the risk of misinformation, increase customer satisfaction, and efficiently manage costs.

[0617] The following describes the processing flow.

[0618] Step 1:

[0619] The server extracts company data from the database. This process includes past customer inquiry data and product-related documents.

[0620] Step 2:

[0621] The server preprocesses the extracted data. Specifically, this involves cleaning the data, imputing missing values, and standardizing the data format. This ensures data consistency.

[0622] Step 3:

[0623] The server connects to reliable external data sources and retrieves the necessary knowledge data. This data includes laws, industry standards, and general knowledge, which are then integrated as pre-cleaned information.

[0624] Step 4:

[0625] The server integrates processed internal data with external data and begins training the generative AI model. At this stage, the model is optimized using supervised learning and reinforcement learning techniques.

[0626] Step 5:

[0627] The server evaluates the training results based on validation data and checks the model's accuracy and performance. If insufficient, it adjusts the training parameters and runs the training again.

[0628] Step 6:

[0629] Users enter their questions through a contact form or chat interface.

[0630] Step 7:

[0631] The terminal receives the user's query and sends it to the server.

[0632] Step 8:

[0633] The server parses the received queries and prepares the data to be input into the model in the most optimal format. At this time, relevant company data and knowledge data are selected and used.

[0634] Step 9:

[0635] The server uses a generated AI model to produce a response from the prepared data. This response is structured in a format that matches the information requested by the user.

[0636] Step 10:

[0637] The server sends the generated response back to the terminal.

[0638] Step 11:

[0639] The terminal displays a response to the user. The user is presented with information that is easy to understand, accurate, and timely.

[0640] Step 12:

[0641] The user provides feedback on the response they received via the terminal.

[0642] Step 13:

[0643] The server receives and analyzes user feedback, accumulating it as data that helps improve the model's accuracy. Once sufficient feedback has been collected, the model is retrained to further improve the system's accuracy.

[0644] (Example 1)

[0645] Next, we will describe Example 1. In the following description, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[0646] The goal of using generative AI models is to effectively integrate company-specific data with reliable external data to provide fast, accurate, and optimized responses for each user. A challenge with conventional technologies is that such data integration and individualized responses are not performed efficiently.

[0647] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0648] In this invention, the server includes means for collecting and preprocessing data, means for acquiring and integrating external information, and means for training a generative model. This makes it possible to provide fast and accurate responses to user inquiries while maintaining data consistency.

[0649] "Means of collecting and pre-processing data" refers to methods used by a server to gather necessary information and prepare it in a format that can be analyzed.

[0650] "Methods for acquiring and integrating external information" refers to methods of extracting information from reliable external sources and transforming it into a format that can be used together with one's own company's information.

[0651] "Means for training generative models" are methods for training a model using collected and pre-processed data to improve the accuracy of its responses.

[0652] "Data consistency" means that information obtained from different sources is not contradictory and is in a coherent state.

[0653] "Providing a fast and accurate response" refers to the ability to generate and deliver reliable answers to user inquiries in the shortest possible time.

[0654] To implement this invention, a server plays a central role. The server first collects internal company data using a data management system. Database management software such as PostgreSQL or MySQL is used in this process. The collected data is preprocessed using a Python script to impute missing data and remove outliers.

[0655] Next, the server obtains external information via the internet, extracting necessary data from reliable public APIs and databases. This data is then integrated with the internal data using ETL tools. Using the integrated dataset, the server trains a generative AI model using deep learning libraries such as TensorFlow and PyTorch. During this training process, the model is optimized based on past feedback.

[0656] The terminal receives prompt messages from the user and sends them to the server. For example, if the prompt message is "Please tell me more about the environmental impact of the new product," the server uses a trained model to analyze internal data and external information to generate the optimal response. As a result, detailed and accurate information is quickly provided to the user via the terminal.

[0657] The information generated in this way allows companies to provide highly accurate support to users while improving the overall efficiency of their business processes.

[0658] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0659] Step 1:

[0660] The server collects internal data. The server uses a database management system (PostgreSQL or MySQL) to retrieve business data from the company. The input consists of business records and customer interaction data, and the output is a database containing this data. This information is preprocessed using a Python script to handle missing or outlier values.

[0661] Step 2:

[0662] The server retrieves and integrates external information. The server obtains legal information and general knowledge from trusted public APIs and external databases. The input is the endpoints of the external APIs and databases, and the output is the retrieved external knowledge data. This data is integrated with internal data using ETL tools to create a consistent dataset.

[0663] Step 3:

[0664] The server trains the generative AI model. The server uses a unified dataset to train the model using deep learning libraries such as TensorFlow and PyTorch. The input is the unified dataset, and the output is the trained generative AI model. During model training, parameters are optimized by referencing past feedback to improve the model's accuracy.

[0665] Step 4:

[0666] The terminal accepts prompt messages from the user. The user enters queries through the terminal's interface and sends them to the server. The input is the prompt message entered by the user, and the output is the query sent to the server. For example, a specific question such as "Please tell me about the product's warranty period" is possible.

[0667] Step 5:

[0668] The server generates responses using a trained model. Based on the user's prompt and integrated data, the server uses its trained model to generate accurate answers to questions. The input is the prompt and integrated dataset, and the output is the generated response. The server edits the response to provide information optimized for the user's background.

[0669] Step 6:

[0670] The terminal provides the user with the generated response. The response from the server is transmitted to the user through the terminal. The input is the response data from the server, and the output is the information provided to the end user. This process allows the user to receive quick and accurate answers.

[0671] (Application Example 1)

[0672] Next, we will explain Application Example 1. In the following explanation, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[0673] In physical stores, there is a challenge in providing accurate information to customers in real time in response to their inquiries. This challenge can lead to decreased customer satisfaction and, consequently, lost sales opportunities. Furthermore, general inquiry response systems cannot provide accurate responses that integrate with external information. The objective of this invention is to provide an effective method for solving this problem.

[0674] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0675] In this invention, the server includes means for extracting and pre-processing company information, means for acquiring and integrating reliable external knowledge information, means for training a generative model, and means for performing speech recognition in real time and presenting the results on a visual display device. This makes it possible to provide accurate information immediately in response to customer inquiries in physical stores.

[0676] "Company information" refers to data that is independently collected and managed by a specific organization, such as business-related data and customer information.

[0677] "External knowledge information" refers to information from external sources, such as industry standards and legal information, obtained from reliable third parties or publicly available databases.

[0678] A "generative model" is an algorithmic system that uses machine learning techniques to generate information based on data such as text and audio.

[0679] "Training" is the process by which a model uses data to optimize its performance in order to improve its performance towards a defined objective.

[0680] "Real-time speech recognition" is a technology that converts input audio signals into text in real time.

[0681] A "visual display device" refers to hardware equipment, including displays and screens, used to visually present information to users.

[0682] The system implementing this invention aims to provide rapid and accurate information in response to customer inquiries. The system mainly consists of a server, a terminal including a visual display device worn by the user, and a generation model.

[0683] The server first extracts the company's own information and performs preprocessing such as data cleaning and formatting standardization. Next, it obtains external knowledge information from reliable data sources and integrates it with the company's information. This integrated data is used to train generative models, which undergo specialized training for specific customer responses.

[0684] The user wears a visual display device similar to smart glasses. The device enables voice input and uses speech recognition technology such as the Google Speech-to-Text API to transcribe customer inquiries into text in real time. The recognized text is sent to a server, where a trained generative model analyzes the data and generates the most appropriate response.

[0685] The generated response is displayed on the visual display device, and the user uses it to communicate the necessary information to the customer. For example, if a customer asks, "Is this product heat resistant?", the system quickly displays information about the product's heat resistance and communicates it visually to the store clerk.

[0686] An example of a prompt is one that specifically states the user's question, such as "Product name: Product name, Attribute: Please tell me about the heat resistance performance." Using this prompt, the generative AI model can provide an accurate response.

[0687] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0688] Step 1:

[0689] The server retrieves company information from a database, cleans that information, and standardizes its format. The input is raw company information, and the output is data in a standardized format. Specifically, it removes duplicate data and converts data in different formats to a unified format.

[0690] Step 2:

[0691] The server collects external knowledge information from reliable data sources and integrates it with pre-processed internal information. The input consists of external knowledge information and pre-processed internal information, and the output is an integrated dataset. At this stage, the relevance of the information is verified, and the necessary data is selected and integrated.

[0692] Step 3:

[0693] The server trains a generative AI model using an integrated dataset. The input is the integrated dataset, and the output is the trained generative model. The generative model uses machine learning algorithms to learn patterns from the dataset.

[0694] Step 4:

[0695] The user receives voice input from customers through the device's visual display. The input is voice data, and the output is a text-based query. Specifically, the device uses the Google Speech-to-Text API to convert the voice to text in real time.

[0696] Step 5:

[0697] The server receives a text-based query and generates the optimal response using a trained generative AI model. The input is the query text, and the output is the generated response. The generative model leverages advanced natural language processing to generate appropriate information for the query.

[0698] Step 6:

[0699] The terminal displays the generated response on a visual display device, and the user refers to it to convey information to the customer. The input is the generated response, and the output is the visually presented information. Specifically, text is displayed on the screen, and the user confirms it.

[0700] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[0701] As an embodiment of the present invention, we propose the construction of a system that enables companies to recognize user emotions and respond accordingly in customer service. This system incorporates an emotion engine in addition to conventional generative AI models to simultaneously analyze user queries and the emotions behind them, thereby achieving more appropriate and satisfying responses.

[0702] First, the server trains a generative model using internal data and reliable external data. Furthermore, an emotion engine is developed to analyze emotions from user text input and voice data. This emotion engine uses natural language processing technology to determine the user's emotions (e.g., joy, anger, sadness) in real time.

[0703] The terminal receives queries from the user and extracts sentiment data as part of the response. This sentiment data is sent to the server and analyzed by a pre-trained model. This allows the server to generate the optimal response while taking the user's emotions into consideration.

[0704] As a concrete example, consider a customer support scenario. When a user contacts support while feeling dissatisfied with a product defect, the emotion engine identifies emotions such as "dissatisfaction" or "anger." This allows the server to generate a more careful and sympathetic response. For example, the response might be, "We apologize for the inconvenience. Here are some possible solutions..."

[0705] In this way, this system can improve customer satisfaction by providing customized responses tailored to each user's emotions, going beyond mere information provision. This approach allows companies to offer more effective customer support, potentially contributing to increased brand loyalty.

[0706] The following describes the processing flow.

[0707] Step 1:

[0708] Users enter questions and opinions through contact forms or voice devices. Both text and voice data are accepted as input.

[0709] Step 2:

[0710] The terminal analyzes the user input it receives and extracts both text and audio data. In the case of audio input, speech recognition is performed and the data is converted into text format.

[0711] Step 3:

[0712] The device sends text data to the emotion engine. The emotion engine uses natural language processing techniques to identify emotions from the user's text. In this process, emotion labels such as "joy," "anger," and "sadness" are assigned.

[0713] Step 4:

[0714] The terminal sends a user query containing sentiment data, which is the result of the analysis, to the server. The server integrates the data based on the received sentiment labels and query content.

[0715] Step 5:

[0716] The server uses integrated data to run a generative AI model and generate the optimal response. During this process, it adjusts the wording and tone as needed based on emotion labels.

[0717] Step 6:

[0718] The server sends the generated response back to the terminal.

[0719] Step 7:

[0720] The device displays a response to the user. This response takes the user's emotions into consideration and includes appropriate tone and information.

[0721] Step 8:

[0722] Users provide feedback on the response. This feedback is based on factors such as the speed and accuracy of the response and overall satisfaction.

[0723] Step 9:

[0724] The server collects and analyzes user feedback. This data is used to train future models, contributing to continuous system improvement.

[0725] (Example 2)

[0726] Next, we will describe Example 2. In the following description, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[0727] In modern business customer service, effectively understanding user emotions and providing appropriate responses is crucial. However, traditional systems have limitations in their ability to analyze user emotions accurately in real time and generate optimal responses. This results in decreased customer satisfaction and hinders the improvement of brand loyalty.

[0728] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[0729] In this invention, the server includes means for extracting and preprocessing proprietary data, means for acquiring and integrating reliable external knowledge data, means for training a generative model, means for analyzing sentiment in response to user queries, means for generating responses based on the analyzed sentiment information, and means for continuously improving the model. This enables companies to understand user sentiment and generate optimal responses based on it, thereby improving customer satisfaction and strengthening brand loyalty.

[0730] "Company data" refers to all information owned and collected by a company in the course of its operations, including past interactions with customers and documents generated within the company.

[0731] "External knowledge data" refers to reliable information obtained from outside the company and used in its activities, including publicly available databases and information provided by third-party organizations.

[0732] A "generative model" is a form of artificial intelligence technology that is trained on collected data and has the ability to generate appropriate responses to user input.

[0733] "Sentiment analysis" refers to the process of identifying and analyzing emotions from user input, and is a technology that is performed in real time using natural language processing techniques.

[0734] "Response generation" is the process of constructing and providing an appropriate response based on the user's query and the emotions behind it.

[0735] "Model improvement" refers to the ongoing process of tuning and optimizing generative models and sentiment analysis engines to improve their performance and accuracy.

[0736] "Feedback information" refers to information based on reactions and evaluations obtained from users and the environment, and is data used to improve and optimize the system.

[0737] "Specialized data" refers to information carefully selected according to specific industries or applications, and is data used to maximize the use of its characteristics in training generative models.

[0738] This invention provides a system for companies to recognize user emotions in customer service and provide appropriate responses. This system integrates a generative AI model and an emotion analysis engine to analyze user queries and provide the optimal response.

[0739] First, the server collects internal data as well as reliable external knowledge data. This data is used to train generative AI models and sentiment analysis engines. The generative AI models can generate optimal responses while leveraging natural language processing techniques to discern user emotions and intentions.

[0740] Specific software includes natural language processing libraries and machine learning frameworks, and these tools are used to build and continuously improve generative AI models and sentiment analysis engines.

[0741] The terminal is a device that constantly receives queries from users, analyzing user text and voice data in real time to detect emotional data. By introducing an emotional analysis engine, it accurately determines the user's emotions such as "joy," "anger," and "sadness" and sends this information to the server.

[0742] As a concrete example, consider a case where a user inquires about a product defect. In this case, the device identifies the user's feelings of dissatisfaction and transmits that data to the server. Based on this emotional data, the server uses a generative AI model to generate a response such as, "We apologize for the inconvenience. To help resolve your issue, please try the following steps."

[0743] An example of a prompt to input into a generative AI model might be: "If the user's emotion is 'anger,' generate a careful and sympathetic response."

[0744] This system allows companies to comprehensively understand user emotions and improve the quality of their responses, ultimately leading to a significant increase in customer satisfaction.

[0745] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0746] Step 1:

[0747] The server collects and preprocesses internal data and reliable external knowledge data. Input data includes raw data, logs, and information extracted from external databases. This data is cleansed, normalized, and processed into a format suitable for training generative AI models. The output is data in a format usable as a training dataset.

[0748] Step 2:

[0749] The server trains a generative AI model and an emotion analysis engine. The input is the training dataset formatted in step 1. The model is trained using machine learning algorithms to learn responses to user input. The output is the trained generative AI model and emotion analysis engine.

[0750] Step 3:

[0751] The terminal receives queries from the user. The input is text or audio data provided by the user. The terminal extracts the necessary information from this query and converts it into a standardized input data format through text analysis or speech recognition. The output is the parsed text data.

[0752] Step 4:

[0753] The device analyzes emotions using the parsed text data. The input is the text data obtained in step 3. It runs the emotion analysis engine and identifies the user's emotional state (e.g., joy, anger, sadness). The output is a dataset containing the detected emotions.

[0754] Step 5:

[0755] The server receives sentiment analysis results and generates a response using a generative AI model. The input consists of user query data and sentiment data. The generative AI model determines the optimal response to the input and constructs appropriate phrasing and information based on the prompt. The output is the response sentence to be provided to the user.

[0756] Step 6:

[0757] The terminal provides the user with the response from the server. The input is the response statement generated in step 5. The terminal presents this response to the user visually or aurally. The output is the response information received by the user.

[0758] (Application Example 2)

[0759] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[0760] Traditional content distribution services have not adequately captured users' emotions through responses and content recommendations, making it a challenge to improve user satisfaction. In particular, there is a need for a means to extract and provide appropriate content based on the emotions users experience while viewing.

[0761] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[0762] In this invention, the server includes means for recognizing and analyzing the user's emotions, means for providing appropriate responses and recommendations based on the analyzed user's emotions, and means for constructing a set of learning data using industry-specific specialized data. This makes it possible to optimize the user's viewing experience based on emotions and improve satisfaction.

[0763] "Company information" refers to internal data and knowledge owned by a company, and is generally obtained from databases and documents that exist within the organization.

[0764] "External knowledge" refers to reliable information and data obtained from outside the company, and is based on third-party organizations, publicly available statistics, research data, etc.

[0765] A "generative model" refers to an artificial intelligence algorithm trained to generate appropriate responses or outputs in response to user inquiries or other inputs.

[0766] "User" refers to an individual or group that operates or uses the system and is interested in its output.

[0767] "Means for recognizing and analyzing emotions" refers to technologies that analyze a user's emotions from facial expressions, voice, or text data and identify their emotional state.

[0768] "Content recommendation" refers to a method of presenting optimal information and entertainment based on the recognized user's state and preferences.

[0769] A "server" refers to a central processing unit that handles data processing, algorithm execution, and communication with users over a network.

[0770] "Optimizing the viewing experience" refers to making adjustments and providing content tailored to the user's situation in order to increase their satisfaction and interest when consuming media content.

[0771] The system for carrying out the present invention includes three main elements: a server, a terminal, and a user.

[0772] The server first acquires its own information and reliable knowledge data obtained from external sources, and then integrates them. At this time, it uses the integrated dataset to train a generative AI model. The generative AI model is an algorithm for generating appropriate responses to user queries. The server also maintains an emotion engine that recognizes and analyzes user emotions in real time. This engine utilizes machine learning libraries such as TensorFlow and works in conjunction with natural language processing techniques to analyze emotional information.

[0773] The terminal functions as an interface with the user and sends user input to the server. The terminal is integrated into a smartphone, smart glasses, or similar device and utilizes a camera and microphone to collect the user's facial expressions and voice in real time for emotion analysis.

[0774] Users can feel that the content and information they receive through this system are most appropriate to their current emotions. Recommendations generated by the server based on emotion analysis present the most suitable content according to the user's state at that time. For example, when a user is smiling, humorous movies or TV shows can be recommended.

[0775] For example, if the emotion engine detects a user smiling while watching enjoyable content, a prompt message such as "Recommend a movie to watch next that is optimized for the current emotional tone" is input to the AI ​​model, which then provides a list of specific movies that are appropriate for that situation.

[0776] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0777] Step 1:

[0778] The device collects the user's facial expressions and voice. Using its built-in camera and microphone, the device records the user's facial expressions and voice tone in real time. This video and audio data is input to the device and sent to a server for emotion analysis.

[0779] Step 2:

[0780] The server analyzes the received video and audio data using an emotion engine. The emotion engine uses natural language processing techniques and machine learning algorithms to identify the user's emotional state (e.g., joy, sadness, surprise). Specifically, it quantifies the emotional tone based on the analysis results. This analysis result becomes the output, and the process proceeds to the next processing step.

[0781] Step 3:

[0782] The server inputs prompt sentences into a generating AI model based on the analyzed sentiment data, and then generates responses and content recommendations. For example, a prompt sentence such as "The user is happy, so suggest some interesting and fun movies" might be generated. The generating AI model processes this prompt and outputs the most suitable list of movies.

[0783] Step 4:

[0784] The server sends the generated response and content recommendations to the terminal. The terminal presents the received information to the user. As output from the terminal, the user can view, select, and watch the movie list and recommended content displayed on the screen.

[0785] Step 5:

[0786] The system collects user feedback and sends it to the server. The device records user feedback such as viewing time for selected content and post-viewing survey results, and sends this data to the server. This feedback is used to adjust the recommendation algorithm for future content.

[0787] The specific processing unit 290 transmits the result of the specific processing to the robot 414. In the robot 414, the control unit 46A causes the speaker 240 and the controlled object 443 to output the result of the specific processing. The microphone 238 acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[0788] Data generation model 58 is a type of so-called generative AI (Artificial Intelligence). One example of data generation model 58 is ChatGPT (Internet search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0789] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and the specific processing may also be performed by the robot 414.

[0790] Furthermore, the emotion identification model 59, acting as an emotion engine, may determine the user's emotion according to a specific mapping. Specifically, the emotion identification model 59 may determine the user's emotion according to a specific mapping, which is an emotion map (see Figure 9). Similarly, the emotion identification model 59 may also determine the robot's emotion, and the identification processing unit 290 may perform identification processing using the robot's emotion.

[0791] Figure 9 shows an emotion map 400 in which multiple emotions are mapped. In the emotion map 400, emotions are arranged in concentric circles radiating from the center. The closer to the center of the concentric circles, the more primitive the emotions are located. Further out of the concentric circles, emotions representing states and actions arising from mental states are located. Emotion is a concept that includes feelings and mental states. On the left side of the concentric circles, emotions that are generally generated from reactions occurring in the brain are located. On the right side of the concentric circles, emotions that are generally induced by situational judgment are located. Above and below the concentric circles, emotions that are generally generated from reactions occurring in the brain and induced by situational judgment are located. In addition, the emotion of "pleasure" is located on the upper side of the concentric circles, and the emotion of "displeasure" is located on the lower side. Thus, in the emotion map 400, multiple emotions are mapped based on the structure in which emotions arise, and emotions that are likely to occur simultaneously are mapped close together.

[0792] These emotions are distributed at the 3 o'clock position on the Emotion Map 400, and usually fluctuate between feelings of security and anxiety. In the right half of the Emotion Map 400, situational awareness takes precedence over internal feelings, resulting in a calm impression.

[0793] The inside of the Emotion Map 400 represents inner thoughts, while the outside represents actions. Therefore, the further you go from the outside of the Emotion Map 400, the more visible (expressed in actions) your emotions become.

[0794] Here, human emotions are based on various balances, such as posture and blood sugar levels. When these balances deviate from the ideal, it results in discomfort, and when they approach the ideal, it results in pleasure. Similarly, in robots, cars, motorcycles, etc., emotions can be created based on various balances, such as posture and battery level. When these balances deviate from the ideal, it results in discomfort, and when they approach the ideal, it results in pleasure. The emotion map can be generated, for example, based on Dr. Mitsuyoshi's emotion map (Research on a system for analyzing brain physiological signals of speech emotion recognition and emotion, Tokushima University, doctoral dissertation: https: / / ci.nii.ac.jp / naid / 500000375379). The left half of the emotion map contains emotions belonging to a region called "response," where sensation is dominant. The right half of the emotion map contains emotions belonging to a region called "situation," where situational awareness is dominant.

[0795] The emotion map defines two emotions that promote learning. One is the emotion around the middle of the negative "repentance" and "reflection" on the situation side. In other words, it is when the robot experiences negative emotions such as "I never want to feel this way again" or "I don't want to be scolded again." The other is the emotion around the positive "desire" on the reaction side. In other words, it is when the robot has positive feelings such as "I want more" or "I want to know more."

[0796] The emotion identification model 59 inputs user input into a pre-trained neural network, obtains emotion values ​​representing each emotion shown in the emotion map 400, and determines the user's emotion. This neural network is pre-trained based on multiple training data sets, which are combinations of user input and emotion values ​​representing each emotion shown in the emotion map 400. Furthermore, this neural network is trained so that emotions located close together have similar values, as shown in the emotion map 900 in Figure 10. Figure 10 shows an example where multiple emotions such as "reassured," "calm," and "confident" have similar emotion values.

[0797] The above description primarily focuses on the functions of the data processing device 12 in relation to this disclosure. However, the system related to this disclosure is not necessarily implemented on a server. The system related to this disclosure may be implemented as a general information processing system. This disclosure may be implemented, for example, as a software program that runs on a personal computer or as an application that runs on a smartphone. The method related to this disclosure may be provided to users in SaaS (Software as a Service) format.

[0798] In the above embodiment, an example was given in which a specific process is performed by a single computer 22. However, the technology of this disclosure is not limited thereto, and a distributed processing of the specific process may be performed by multiple computers, including computer 22. For example, a data generation model 58 may be provided in an external device of the data processing device 12, and the external device may generate data according to the input data.

[0799] In the above embodiment, an example was given in which the specific processing program 56 is stored in the storage 32, but the technology of this disclosure is not limited thereto. For example, the specific processing program 56 may be stored in a portable, computer-readable, non-temporary storage medium such as a USB (Universal Serial Bus) memory. The specific processing program 56 stored in the non-temporary storage medium is installed in the computer 22 of the data processing device 12. The processor 28 executes specific processing according to the specific processing program 56.

[0800] 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.

[0801] Furthermore, it is not necessary to store the entirety of the specific processing program 56 in a storage device such as a server connected to the data processing device 12 via the network 54, or to store the entirety of the specific processing program 56 in the storage 32; it is acceptable to store only a portion of the specific processing program 56.

[0802] The following types of processors can be used as hardware resources to perform specific processing. Examples of processors include a CPU, a general-purpose processor that functions as a hardware resource to perform specific processing by executing software, i.e., a program. Other examples of processors include dedicated electrical circuits, such as FPGAs (Field-Programmable Gate Arrays), PLDs (Programmable Logic Devices), or ASICs (Application Specific Integrated Circuits), which have circuit configurations specifically designed to perform specific processing. All of these processors have built-in or connected memory, and all of them perform specific processing by using memory.

[0803] The hardware resource that performs a specific process may consist of one of these various processors, or it may consist of a combination of two or more processors of the same or different types (for example, a combination of multiple FPGAs, or a combination of a CPU and an FPGA). Alternatively, the hardware resource that performs a specific process may consist of a single processor.

[0804] Examples of configurations using a single processor include, firstly, a configuration in which one or more CPUs and software are combined to form a single processor, and this processor functions as a hardware resource that performs a specific process. Secondly, there is a configuration using a processor that realizes the functions of the entire system, including multiple hardware resources that perform a specific process, on a single IC chip, as exemplified by SoCs (System-on-a-chip). In this way, a specific process is realized using one or more of the above types of processors as hardware resources.

[0805] Furthermore, the hardware structure of these various processors can more specifically utilize electrical circuits that combine circuit elements such as semiconductor devices. Also, the specific processing described above is merely an example. Therefore, it goes without saying that unnecessary steps can be deleted, new steps added, or the processing order rearranged, as long as it does not deviate from the main purpose.

[0806] The descriptions and illustrations presented above are detailed explanations of the technical aspects of this disclosure and are merely examples of the technical aspects. For example, the above descriptions of the structure, function, operation, and effect are examples of the structure, function, operation, and effect of the technical aspects of this disclosure. Therefore, it goes without saying that you may delete unnecessary parts, add new elements, or replace elements in the descriptions and illustrations presented above, as long as you do not deviate from the essence of the technical aspects of this disclosure. Furthermore, in order to avoid confusion and facilitate understanding of the technical aspects of this disclosure, explanations of common technical knowledge and the like that do not require special explanation to enable the implementation of the technical aspects of this disclosure have been omitted from the descriptions and illustrations presented above.

[0807] All documents, patent applications, and technical standards described herein are incorporated by reference to the same extent as if each individual document, patent application, and technical standard were specifically and individually noted to be incorporated by reference.

[0808] The following is further disclosed regarding the embodiments described above.

[0809] (Claim 1)

[0810] A method for extracting and pre-processing company data,

[0811] Means for acquiring and integrating reliable external knowledge data,

[0812] Methods for training generative models,

[0813] A means of generating a response to a query from a user,

[0814] A system that includes means for continuously improving the model.

[0815] (Claim 2)

[0816] The system according to claim 1, further comprising means for adjusting the generative model using the user feedback information.

[0817] (Claim 3)

[0818] The system according to claim 1, comprising means for constructing a training dataset using industry-specific specialized data.

[0819] "Example 1"

[0820] (Claim 1)

[0821] Means for collecting and preprocessing data,

[0822] Means for acquiring and integrating external information,

[0823] Means for training generative models,

[0824] A means of generating a response to a user inquiry,

[0825] A system that includes means for optimizing a model through operations.

[0826] (Claim 2)

[0827] The system according to claim 1, comprising means for improving the generative model using feedback information.

[0828] (Claim 3)

[0829] The system according to claim 1, comprising means for creating a dataset using specialized data relevant to a specific industry.

[0830] "Application Example 1"

[0831] (Claim 1)

[0832] A means of extracting and pre-processing company information,

[0833] Means for acquiring and integrating reliable external knowledge information,

[0834] Means for training generative models,

[0835] A means of creating a response to an inquiry from a user,

[0836] A means for performing speech recognition in real time and displaying the results on a visual display device,

[0837] A means of constructing a training dataset using information specific to the economic field,

[0838] A system that includes means for continuously improving the model.

[0839] (Claim 2)

[0840] The system according to claim 1, which adjusts the generation model using the response information from the aforementioned user.

[0841] (Claim 3)

[0842] The system according to claim 1, which displays response information to a user through a visual presentation device.

[0843] "Example 2 of combining an emotion engine"

[0844] (Claim 1)

[0845] A method for extracting and pre-processing company data,

[0846] Means for acquiring and integrating reliable external knowledge data,

[0847] Methods for training generative models,

[0848] A means of analyzing sentiment in response to user queries,

[0849] Means for generating a response based on analyzed emotional information,

[0850] A system that includes means for continuously improving the model.

[0851] (Claim 2)

[0852] The system according to claim 1, comprising means for adjusting a generative model and an emotion analysis engine using user feedback information.

[0853] (Claim 3)

[0854] The system according to claim 1, comprising means for constructing a training dataset using industry-specific specialized data.

[0855] "Application example 2 when combining with an emotional engine"

[0856] (Claim 1)

[0857] A means of extracting and pre-processing company information,

[0858] Means for acquiring and integrating reliable external knowledge,

[0859] Methods for training generative models,

[0860] A means of generating a response to an inquiry from a user,

[0861] A means of recognizing and analyzing the emotions of users,

[0862] A means of providing appropriate responses and recommendations based on analyzed user emotions,

[0863] A system that includes means for continuously improving the model.

[0864] (Claim 2)

[0865] The system according to claim 1, further comprising means for adjusting the generative model using the aforementioned user feedback information.

[0866] (Claim 3)

[0867] A method for constructing a training data set using specialized data for each industry,

[0868] The system according to claim 1, including means for providing content recommendations based on user sentiment. [Explanation of Symbols]

[0869] 10, 210, 310, 410 Data Processing Systems 12 Data Processing Devices 14 Smart Devices 214 Smart Glasses 314 Headset-type terminal 414 Robots< / url:> < / url:> < / url:> < / url:>

Claims

1. A method for extracting and pre-processing company data, Means for acquiring and integrating reliable external knowledge data, Methods for training generative models, A means of generating a response to a query from a user, A system that includes means for continuously improving the model.

2. The system according to claim 1, further comprising means for adjusting the generative model using the user feedback information.

3. The system according to claim 1, comprising means for constructing a training dataset using industry-specific specialized data.

Images