system

Abstract

We provide the system. [Solution] A means of analyzing queries received from users and extracting keywords within those queries, A means for selecting a language model with the relevant specialization from among multiple language models based on the extracted keywords, A means for generating a response to the query using a selected language model, A means for integrating responses obtained from multiple language models and selecting the most appropriate response, Means for providing the selected response to the user, A system that includes this.

Description

Technical Field

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

Background Art

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

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] With the improvement of the performance of large language models (LLMs), there is a concern about the increase in their power consumption. As it is, a large amount of power is required to process specific tasks, and it is inevitable to increase the environmental load and operation cost. Also, in a structure that depends on a single model, the accuracy of responses may not always be sufficient in a specific field. Therefore, a new method that generates high-precision responses while consuming low power is required.

Means for Solving the Problems

[0005] To solve the above-mentioned problems, the present invention provides a means for analyzing a query received from a user and extracting keywords. Next, it provides a means for selecting a model with the appropriate expertise from multiple language models based on the extracted keywords. Using this selected model, it generates a response to the query, and further provides a means for integrating the responses obtained from multiple models and selecting the most appropriate response. Finally, by providing the selected response to the user, it achieves low power consumption and high-precision response generation.

[0006] A "query" is a question or request entered by a user into a system, and the system is the entity that responds to it.

[0007] "Keywords" are important vocabulary words extracted from a query, and they serve as the criteria for selecting relevant knowledge and models.

[0008] A "language model" is a type of artificial intelligence model used as a means of natural language processing. It generates and interprets text by learning the statistical properties of language based on a large amount of text data.

[0009] A "specialized field" refers to a specific area in which each language model possesses specialized knowledge, and this serves as a selection criterion when generating the most appropriate response to a query.

[0010] A "response" refers to the solution or information that the system generates in response to a query and provides to the user.

[0011] "Integration" is the process of comparing responses obtained from multiple language models and ultimately combining them into a single response.

[0012] "Selection" is the act of choosing the most appropriate response from among several responses obtained as part of the integration process. [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, the numbered 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, the numbered 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, the numbered 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, and the like.

[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] This invention provides a system that receives queries from users and generates highly accurate responses with low power consumption by efficiently combining multiple language models. The following shows the processing flow based on an embodiment of the present invention.

[0035] First, the user inputs a specific question or problem through the terminal. For example, the user might ask, "Tell me about the 18th-century French education system." The terminal then sends this query to the server.

[0036] Next, the server receives this query and extracts keywords. In this step, it identifies important vocabulary such as "18th century," "France," and "education system." This process reveals basic information about the content of the query.

[0037] Subsequently, the server selects a language model with relevant expertise from among several language models based on the extracted keywords. For example, a model specializing in history and education might be selected.

[0038] The selected model generates responses based on its respective area of ​​expertise. For example, it might generate information such as, "In 18th-century France, education was primarily conducted by religious institutions."

[0039] The server integrates the generated responses and selects the most appropriate one. This process narrows down the multiple responses to those that are most relevant and contain useful information.

[0040] Finally, the server provides the user with the selected response. The user can obtain the most accurate and relevant information through their terminal.

[0041] Thus, the present invention efficiently utilizes multiple language models to quickly and effectively answer user inquiries, thereby minimizing power consumption while providing highly accurate information.

[0042] The following describes the processing flow.

[0043] Step 1:

[0044] The terminal receives a query from the user. The user enters a specific question into the terminal, such as "Tell me about the 18th-century French education system." The terminal prepares to send this input to the server.

[0045] Step 2:

[0046] The server receives queries sent from terminals. To analyze the queries, it interprets their content using natural language processing techniques and extracts keywords. For example, words like "18th century," "France," and "education system" might be identified.

[0047] Step 3:

[0048] The server selects language models with relevant expertise based on keywords. This includes models related to history and models specializing in education. The server makes this selection by referring to the specialized domains of each model.

[0049] Step 4:

[0050] The server passes queries to each selected language model and instructs them to generate responses. Each model generates a response from its area of ​​expertise based on the given query. For example, it might generate information such as, "In 18th-century France, the education system was mainly run by religious institutions."

[0051] Step 5:

[0052] The server receives multiple generated responses. It evaluates these responses and performs integration to select the optimal response. The server considers the relevance and accuracy of each response to find the response that contains the most useful information to provide to the user.

[0053] Step 6:

[0054] The server sends a unified, optimal response back to the terminal. The terminal displays this response to the user. The user can obtain detailed and accurate information through the terminal and resolve any questions regarding the query.

[0055] (Example 1)

[0056] 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."

[0057] A challenge exists in efficiently utilizing multiple natural language models to provide rapid and highly accurate responses to user information requests while minimizing power consumption. To address this challenge, technology is needed to appropriately select and integrate highly relevant information and provide it to the user.

[0058] 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.

[0059] In this invention, the server includes means for analyzing user queries received by an information terminal and extracting important words within the query; means for selecting a natural language model with a predetermined knowledge domain from among a plurality of natural language models based on the extracted important words; and means for generating a response to the query using the selected natural language model. This makes it possible to provide the user with highly relevant responses quickly and accurately.

[0060] An "information terminal" is an electronic device used by users to input queries and receive the results.

[0061] A "user query" refers to a question or request for information entered by a user through an information terminal.

[0062] "Key words" are words or phrases that contain essential information and are particularly relevant in meaning within a user query.

[0063] A "natural language model" is an algorithm or mathematical model trained for natural language processing, which understands and generates text data related to a specific knowledge domain.

[0064] A "knowledge domain" is a concept that represents the scope of a specific area of ​​expertise or content, and indicates the area of ​​expertise that a natural language model possesses.

[0065] "Numerical data" refers to information that indicates the tuning parameters and weights that a natural language model has optimized for a specific knowledge domain.

[0066] This invention begins with a process in which a user inputs a specific question or problem using an information terminal. The information terminal is often an electronic device such as a personal computer or smartphone. A concrete example would be a user inputting, "Tell me about the 18th-century French education system."

[0067] The information terminal sends the user query entered to the server. This transmission uses communication methods such as HTTP requests over the internet. The server parses the received query and extracts important words from its text. For this parsing, it is suitable to use libraries that provide natural language processing technology (e.g., NLTK or spaCy).

[0068] Based on the extracted key words, the server selects the appropriate model from among several natural language models. These natural language models possess numerical data relevant to their knowledge domain and can utilize existing technologies such as BERT and GPT-3®. The selected model performs data processing related to the specific knowledge domain and generates a response.

[0069] The generated responses are integrated within the server, and the most relevant response is selected from among them. Scoring algorithms and filtering techniques can be applied to this integration and selection process.

[0070] Ultimately, the server sends the selected response back to the information terminal, providing it to the user. Through this response, the user can obtain the requested information. For example, using the prompt "Explain the promotion of art during the Renaissance," it is possible to obtain relevant historical information.

[0071] In this way, this invention provides a system that efficiently reduces power consumption while responding to user requirements with high precision and speed.

[0072] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0073] Step 1:

[0074] The user enters a query using an information terminal. This query is entered as text data indicating the content of the inquiry. For example, a prompt such as "Tell me about the 18th-century French education system" might be entered. This input data is sent from the information terminal to the server.

[0075] Step 2:

[0076] The server analyzes the query received from the user and extracts important words from the text. By analyzing the input prompt using natural language processing techniques and extracting key keywords such as "18th century," "France," and "education system," the server understands the core of the query. The result of this processing is output as a list of extracted keywords.

[0077] Step 3:

[0078] The server selects an appropriate model from among several natural language models based on the extracted keywords. The selection criterion is whether the knowledge domain related to the keywords matches the model's area of ​​expertise. For example, a model capable of handling data related to history and education might be selected. This selection result is output as identification information for the selected natural language model.

[0079] Step 4:

[0080] Using the selected natural language model, the server generates a response to the query. In this step, keyword-based information processing and data calculations are performed using the numerical data held by the model. The generated response is output as text data that answers the query. A concrete example of a response would include the information, "In 18th-century France, education was primarily provided by religious institutions."

[0081] Step 5:

[0082] The server integrates responses from multiple models and selects the most relevant response. During the integration process, the relevance and confidence level of each response are evaluated, and a scoring algorithm is used to select the optimal response. The output of this step is the selected single response text.

[0083] Step 6:

[0084] The server sends the selected response back to the information terminal. This response data is displayed to the user, allowing them to quickly obtain the most useful information. This ensures that the user receives clear and accurate answers.

[0085] (Application Example 1)

[0086] 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."

[0087] A challenge in virtual stores is that users cannot obtain product information smoothly and accurately. When such information acquisition is inefficient, the user experience deteriorates, and store usage is not encouraged.

[0088] 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.

[0089] In this invention, the server includes a device for analyzing queries received from a user and extracting keywords within those queries; a device for selecting a natural language processing model with the relevant expertise from among a plurality of natural language processing models based on the extracted keywords; and a device for generating a response to the query using the selected natural language processing model. This enables users to efficiently and accurately obtain product information.

[0090] A "user" is someone who uses a system or application.

[0091] A "query" refers to information that a user gives to a system, such as a question or request.

[0092] "Keywords" are important terms extracted from within a query.

[0093] A "natural language processing model" is an algorithm or model designed to understand and process natural language.

[0094] A "specialized field" refers to a specific area of ​​knowledge or topic that a language processing model specializes in addressing.

[0095] A "response" is the answer or information generated in response to a query.

[0096] A "device" is a part of a machine or software designed to perform a specific function.

[0097] "Integration" is the process of combining multiple pieces of information or responses into one.

[0098] "Selection" is the act of choosing the most appropriate option from multiple choices.

[0099] A "virtual store" is a store that sells products and provides information online.

[0100] This section describes embodiments for carrying out the invention. This invention realizes a system for efficiently providing product information in a virtual store. Queries are received from users via a terminal. These queries are mainly questions or requests related to product information.

[0101] The server first parses the received query and extracts keywords. This process utilizes natural language processing libraries such as spaCy and NLTK. Based on the extracted keywords, the server uses machine learning frameworks such as TENSORFLOW® and PyTorch to select the appropriate model from among multiple natural language processing models. Each natural language processing model is specialized in a particular field, allowing it to generate responses suitable for the relevant product information.

[0102] After generating responses, the server integrates multiple responses and selects the most relevant and accurate response. This integrated response is then provided to the user again via the terminal. The cloud server achieves high processing speed and availability by using services such as AWS® EC2.

[0103] For example, if a user enters a query such as "What is the material of this product?", the system will select an appropriate natural language processing model and provide accurate material information. An example of a prompt would be: "User question: 'Please tell me the user reviews for this product.' Output: Collect review information and generate a response listing the key points." This allows users to efficiently obtain product information within the virtual store.

[0104] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0105] Step 1:

[0106] The user enters a query about a product using a terminal. This query is sent to the server. The input might be something like, "What is the material of this product?" The terminal forwards the query to the server in text format.

[0107] Step 2:

[0108] The server parses the received query and extracts keywords using natural language processing libraries (e.g., spaCy, NLTK). The input is the user's query, and the output is a list of keywords such as "product" and "material." The server applies text processing techniques to identify the key points of the information.

[0109] Step 3:

[0110] The server selects an appropriate natural language processing model based on the extracted keywords. This is done using a machine learning framework (e.g., TensorFlow, PyTorch). The input is a list of keywords, and the output is the selected model. The server then deploys a model selection algorithm to determine the most relevant model.

[0111] Step 4:

[0112] The server generates a response using a selected natural language processing model. The model processes the query and related information to produce an answer. The input is the selected model and information related to the query, and the output is the generated response. The server utilizes the model's inference capabilities to generate accurate information.

[0113] Step 5:

[0114] The system integrates responses generated by multiple natural language processing models to select the most appropriate response. The input consists of multiple responses, and the output is the most relevant response. The server uses an algorithm that compares the responses using evaluation criteria and selects the best one.

[0115] Step 6:

[0116] The server sends the selected response back to the terminal. The user receives the response through the terminal and can learn detailed information about the product. The input is the selected response, and the output is the information displayed on the user's terminal. The terminal visually presents the information and provides the user with useful data.

[0117] 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.

[0118] This invention provides a system that recognizes user emotions and adjusts responses accordingly. The system aims to generate highly accurate and user-friendly responses by utilizing multiple language models based on queries received from the user.

[0119] First, the user enters a query using a terminal. For example, they might enter a query that includes an emotion, such as "I feel angry about recent news." The terminal then sends this query to the server.

[0120] The server parses the received query. It extracts keywords using natural language processing techniques and analyzes the user's emotions through an emotion engine. In this step, the polarity of the emotions contained in the query, whether positive or negative, is determined.

[0121] Next, the server selects relevant models from multiple language models based on the extracted keywords. These selected models then generate responses to the query based on their respective areas of expertise.

[0122] The generated response is adjusted based on the emotional information detected by the emotion engine. For example, if the user is experiencing negative emotions, the response will be adjusted to be more polite and reassuring.

[0123] The server then integrates all the generated responses and selects the most appropriate one. This response selection process takes into account the accuracy, relevance, and emotional impact of the information.

[0124] Finally, the server provides the selected response to the terminal and displays it to the user. This allows the user to receive a response that takes their emotions into consideration, resulting in a better user experience.

[0125] Thus, by combining the system of the present invention with an emotion engine, it is possible to address not only technical accuracy but also the emotional needs of the user.

[0126] The following describes the processing flow.

[0127] Step 1:

[0128] The terminal receives a query from the user. The user enters a message expressing emotion, such as "I feel angry about the recent news." This input is then ready to be sent from the terminal to the server.

[0129] Step 2:

[0130] The server receives a query from the terminal. To process the query, the server uses natural language processing techniques to analyze it and extract important keywords. Simultaneously, it uses an emotion engine to analyze the user's emotions contained in the query. This analysis determines whether the query contains negative emotions.

[0131] Step 3:

[0132] Based on the analysis results, the server selects a language model with the appropriate expertise. In this case, models related to news and sentiment analysis are chosen. The selected model generates responses to the input query based on its respective area of ​​expertise.

[0133] Step 4:

[0134] The server adjusts the content of the generated response based on emotional information analyzed by the emotion engine. If a negative emotion is detected, the response is modified to be more considerate and encouraging, taking care to soothe the user's mood.

[0135] Step 5:

[0136] The server integrates the coordinated responses and selects the most appropriate one. This response selection process takes into account the high relevance and accuracy of the information, as well as consideration for the user's emotional needs.

[0137] Step 6:

[0138] The server sends the selected response to the terminal. The terminal displays this response to the user. As a result, the user receives accurate information that takes their emotional state into account, leading to a sense of reassurance and understanding.

[0139] (Example 2)

[0140] 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".

[0141] Conventional information response systems have struggled to improve the accuracy and relevance of responses to user queries, and also have difficulty generating responses that take user emotions into consideration. As a result, they have been unable to improve the quality of the user experience, and have a particular challenge in providing appropriate responses to users with negative emotions.

[0142] 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.

[0143] In this invention, the server includes means for analyzing queries received from a user and extracting keywords; means for selecting a language model with the relevant expertise from among multiple language models based on the extracted keywords; means for generating a response to the query using the selected language model; means for adjusting the generated response based on the user's emotional information; means for integrating the adjusted responses and selecting the most appropriate response; and means for providing the selected response to the user. This makes it possible to provide appropriate and considerate responses that take into account the user's emotions.

[0144] A "query" is information that represents a question or request that a user enters into a system.

[0145] A "keyword" is a phrase that indicates the main theme or topic extracted from a query.

[0146] A "language model" is an algorithm or data set used to process natural language and generate responses based on a specific area of ​​expertise.

[0147] "Emotional information" refers to information that indicates the polarity and characteristics of emotions included in user queries.

[0148] "Response adjustment" refers to the process of flexibly modifying the content of a generated response, taking into account the user's feelings.

[0149] "Response integration" is the process of combining multiple responses generated from different language models and selecting the single most appropriate response.

[0150] A "server" is a computing device or system that receives queries from users, performs the necessary processing, and generates responses.

[0151] This invention relates to an interactive system that generates appropriate responses while taking into account the user's emotions. This system primarily uses a terminal, a server, a language model, and an emotion engine.

[0152] Users use a terminal to enter queries into the system. These queries can include requests for information or expressions of opinion, and may contain emotional nuances. For example, consider a case where a user enters the query, "I feel angry about the recent news."

[0153] The terminal's role is to send the input query to the server. The server uses natural language processing libraries (e.g., spaCy or NLTK) to extract keywords in order to process the received query. It also uses sentiment engines (e.g., TextBlob or VADER) to analyze the sentiment information contained in the query and determine the polarity of the sentiment.

[0154] Based on this information, the server selects a model with the appropriate expertise from among multiple generative AI models. For example, for news-related queries, a model specializing in news will be selected. This process can utilize general generative AI technologies such as OpenAI.

[0155] The selected model leverages the query content and relevant expertise to generate an appropriate response. This response is refined within the server, and may be modified to take into account the emotional information provided by the user. For example, a query indicating negative emotions may be adjusted to include more reassuring and polite language in the response.

[0156] Finally, the server integrates the generated responses and selects the most appropriate one. This selection process takes into account the accuracy, relevance, and emotional considerations of the information. The selected response is sent to the terminal and displayed to the user.

[0157] This system configuration allows users to receive responses that take their emotions into account, resulting in a higher user experience. A concrete example of a prompt would be, "How should you respond if the user is angry about recent news?"

[0158] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0159] Step 1:

[0160] The user enters a query using a terminal. The entered query includes the user's information requests and emotions. For example, they might enter, "I feel angry about recent news." The terminal sends this query to the server.

[0161] Step 2:

[0162] The server parses the queries received from the terminal. First, it extracts keywords from the input query using a natural language processing library (e.g., spaCy or NLTK). This process identifies the theme or topic of the query. The extracted keywords are then output.

[0163] Step 3:

[0164] The server uses a sentiment engine (e.g., TextBlob or VADER) to analyze the sentiment contained in the query. This analysis determines the sentiment polarity (positive, negative, or neutral) of the query. The sentiment polarity information is obtained as output.

[0165] Step 4:

[0166] The server selects the most relevant generative AI model from multiple models based on the extracted keywords and sentiment polarity. For example, if keywords related to news are included, it will select a generative AI model specialized in news. This results in the output of a highly relevant generative AI model.

[0167] Step 5:

[0168] The server generates responses to queries using the selected generative AI model. This process leverages the model's expertise to produce highly accurate text output. The general content of the generated response is largely determined by the model.

[0169] Step 6:

[0170] The server adjusts the generated response based on the emotional information obtained by the emotion engine. If a negative emotion is detected, the response is modified to be more polite and considerate. The output is an emotionally conscious, adjusted response.

[0171] Step 7:

[0172] The server integrates multiple pre-tuned responses and selects the most appropriate one. This selection takes into account the accuracy, relevance, and emotional relevance of the response. The single best response is chosen as the output.

[0173] Step 8:

[0174] The server sends the selected response to the terminal. The terminal displays this response to the user, providing an appropriate answer that takes emotions into consideration. Through this process, the user receives an answer that reflects their emotions.

[0175] (Application Example 2)

[0176] 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".

[0177] In recent years, content delivery services have seen a growing demand for personalized content recommendations based on user emotions to improve the user experience. However, conventional technologies struggle to accurately analyze user emotions and provide appropriate responses and content immediately. This is a reason why user expectations are not being met, impacting service usage frequency and satisfaction.

[0178] 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.

[0179] In this invention, the server includes means for analyzing a query received from a user and extracting keywords and emotional states within the query; means for selecting a natural language processing model with the relevant expertise from among a plurality of natural language processing models based on the extracted keywords and emotional states; and means for generating a response to the query using the selected natural language processing model and adjusting the response content according to the user's emotional state. This enables the recommendation of personalized content that takes the user's emotions into consideration.

[0180] A "query" is information or a question that a user gives to a system, and it serves as the starting point for the system to analyze it and generate a response.

[0181] "Keywords" are important words or phrases extracted from a query, and they form the basis for the system to generate information and responses.

[0182] "Emotional state" refers to the psychological and emotional state of a user, analyzed from their queries and their actions before and after those queries, and is used when adjusting the content of the response.

[0183] A "natural language processing model" is an algorithmic method or system for understanding human language and generating responses, and it specializes in providing responses in various fields of expertise.

[0184] A "specialized natural language processing model" refers to a natural language processing model that specializes in processing information in a specific knowledge domain or topic, and has the ability to generate appropriate responses to relevant queries.

[0185] "Adjusting response content" refers to the process of modifying the tone and content of responses generated based on the user's emotional state, and delivering them in a way that is optimal for the user.

[0186] "Personalized content" refers to content that is individually generated and recommended based on a user's emotional state and past behavior, with the aim of improving the user experience.

[0187] The system for implementing this invention begins with a query input from the user via a terminal. The query received by the terminal from the user is transmitted to a server via a communication network. The server first analyzes the query and has the function of extracting keywords and sentiment states. Natural language processing technology is used for this, and software such as Python and TensorFlow is applied. The analysis of sentiment states is performed based on the text data contained in the user's query.

[0188] The server then selects the most suitable model from among several natural language processing models based on the extracted keywords and sentiment states. This process utilizes natural language processing models specialized in various fields, enabling sophisticated response generation. Generative AI models such as GPT are used in this process.

[0189] Using the selected model, the server generates a response to the query and further adjusts the response content considering the user's emotional state. This creates a thoughtful and appropriate response that takes the user's emotions into consideration. As a result, the responses and content provided by the system are personalized and tailored to the user's preferences and emotions.

[0190] Finally, the integrated response is returned to the terminal and presented to the user. This allows users to receive content in a way that takes their emotional state into consideration, improving the quality of the service and the user experience. A specific example of its use is when a user reacts to "anxious news": the server recommends reassuring content to support and improve the user's emotional state.

[0191] Example of a prompt:

[0192] "When users are feeling anxious, create content that provides reassurance. For example, think about lighthearted news that can refresh their mood or calming videos."

[0193] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0194] Step 1:

[0195] The user enters a query into the terminal. The entered query is sent from the terminal to the server in natural language text format. The input at this stage is the user's text data, and no specific information processing or analysis is performed.

[0196] Step 2:

[0197] The server parses the received query. The input is text data sent by the user. The server runs a natural language processing algorithm using Python to perform data analysis to extract keywords and sentiment states from the query. The output is the parsed keyword and sentiment state information.

[0198] Step 3:

[0199] The server selects an appropriate natural language processing model based on the extracted keywords and sentiment states. The input is the keywords and sentiment states obtained from step 2, and a comparison operation is performed to select a model with the corresponding area of ​​expertise. The output is the identification information of the selected model.

[0200] Step 4:

[0201] The server generates responses to queries using a selected natural language processing model. The input consists of the selected model and the user's query, and the generating AI model works to create an appropriate response. During this process, the response content is also adjusted according to the emotional state. The output is the adjusted response data.

[0202] Step 5:

[0203] When multiple responses are generated, the server integrates them and selects the most appropriate response. The input consists of multiple adjusted responses, and the server performs an optimization evaluation based on content, relevance, etc., to select the response deemed most appropriate. The output is the selected optimal response.

[0204] Step 6:

[0205] The selected response is sent to the terminal and displayed to the user. The input is the optimal response received from step 5, and the output is the response information presented to the user. This allows the user to receive a personalized response and obtain emotionally sensitive information.

[0206] 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.

[0207] 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.

[0208] 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.

[0209] [Second Embodiment]

[0210] Figure 3 shows an example of the configuration of the data processing system 210 according to the second embodiment.

[0211] 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.

[0212] 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).

[0213] 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.

[0214] 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.

[0215] 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).

[0216] 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.

[0217] 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.

[0218] 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.

[0219] 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.

[0220] 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.

[0221] 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".

[0222] This invention provides a system that receives queries from users and generates highly accurate responses with low power consumption by efficiently combining multiple language models. The following shows the processing flow based on an embodiment of the present invention.

[0223] First, the user inputs a specific question or problem through the terminal. For example, the user might ask, "Tell me about the 18th-century French education system." The terminal then sends this query to the server.

[0224] Next, the server receives this query and extracts keywords. In this step, it identifies important vocabulary such as "18th century," "France," and "education system." This process reveals basic information about the content of the query.

[0225] Subsequently, the server selects a language model with relevant expertise from among several language models based on the extracted keywords. For example, a model specializing in history and education might be selected.

[0226] The selected model generates responses based on its respective area of ​​expertise. For example, it might generate information such as, "In 18th-century France, education was primarily conducted by religious institutions."

[0227] The server integrates the generated responses and selects the most appropriate one. This process narrows down the multiple responses to those that are most relevant and contain useful information.

[0228] Finally, the server provides the user with the selected response. The user can obtain the most accurate and relevant information through their terminal.

[0229] Thus, the present invention efficiently utilizes multiple language models to quickly and effectively answer user inquiries, thereby minimizing power consumption while providing highly accurate information.

[0230] The following describes the processing flow.

[0231] Step 1:

[0232] The terminal receives a query from the user. The user enters a specific question into the terminal, such as "Tell me about the 18th-century French education system." The terminal prepares to send this input to the server.

[0233] Step 2:

[0234] The server receives queries sent from terminals. To analyze the queries, it interprets their content using natural language processing techniques and extracts keywords. For example, words like "18th century," "France," and "education system" might be identified.

[0235] Step 3:

[0236] The server selects language models with relevant expertise based on keywords. This includes models related to history and models specializing in education. The server makes this selection by referring to the specialized domains of each model.

[0237] Step 4:

[0238] The server passes queries to each selected language model and instructs them to generate responses. Each model generates a response from its area of ​​expertise based on the given query. For example, it might generate information such as, "In 18th-century France, the education system was mainly run by religious institutions."

[0239] Step 5:

[0240] The server receives multiple generated responses. It evaluates these responses and performs integration to select the optimal response. The server considers the relevance and accuracy of each response to find the response that contains the most useful information to provide to the user.

[0241] Step 6:

[0242] The server sends a unified, optimal response back to the terminal. The terminal displays this response to the user. The user can obtain detailed and accurate information through the terminal and resolve any questions regarding the query.

[0243] (Example 1)

[0244] 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."

[0245] A challenge exists in efficiently utilizing multiple natural language models to provide rapid and highly accurate responses to user information requests while minimizing power consumption. To address this challenge, technology is needed to appropriately select and integrate highly relevant information and provide it to the user.

[0246] 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.

[0247] In this invention, the server includes means for analyzing user queries received by an information terminal and extracting important words within the query; means for selecting a natural language model with a predetermined knowledge domain from among a plurality of natural language models based on the extracted important words; and means for generating a response to the query using the selected natural language model. This makes it possible to provide the user with highly relevant responses quickly and accurately.

[0248] An "information terminal" is an electronic device used by users to input queries and receive the results.

[0249] A "user query" refers to a question or request for information entered by a user through an information terminal.

[0250] "Key words" are words or phrases that contain essential information and are particularly relevant in meaning within a user query.

[0251] A "natural language model" is an algorithm or mathematical model trained for natural language processing, which understands and generates text data related to a specific knowledge domain.

[0252] A "knowledge domain" is a concept that represents the scope of a specific area of ​​expertise or content, and indicates the area of ​​expertise that a natural language model possesses.

[0253] "Numerical data" refers to information that indicates the tuning parameters and weights that a natural language model has optimized for a specific knowledge domain.

[0254] This invention begins with a process in which a user inputs a specific question or problem using an information terminal. The information terminal is often an electronic device such as a personal computer or smartphone. A concrete example would be a user inputting, "Tell me about the 18th-century French education system."

[0255] The information terminal sends the user query entered to the server. This transmission uses communication methods such as HTTP requests over the internet. The server parses the received query and extracts important words from its text. For this parsing, it is suitable to use libraries that provide natural language processing technology (e.g., NLTK or spaCy).

[0256] Based on the extracted key words, the server selects the appropriate model from among several natural language models. These natural language models possess numerical data relevant to their knowledge domain and can leverage existing technologies such as BERT and GPT-3. The selected model performs data processing related to the specific knowledge domain and generates a response.

[0257] The generated responses are integrated within the server, and the most relevant response is selected from among them. Scoring algorithms and filtering techniques can be applied to this integration and selection process.

[0258] Ultimately, the server sends the selected response back to the information terminal, providing it to the user. Through this response, the user can obtain the requested information. For example, using the prompt "Explain the promotion of art during the Renaissance," it is possible to obtain relevant historical information.

[0259] In this way, this invention provides a system that efficiently reduces power consumption while responding to user requirements with high precision and speed.

[0260] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0261] Step 1:

[0262] The user enters a query using an information terminal. This query is entered as text data indicating the content of the inquiry. For example, a prompt such as "Tell me about the 18th-century French education system" might be entered. This input data is sent from the information terminal to the server.

[0263] Step 2:

[0264] The server analyzes the query received from the user and extracts important words from the text. By analyzing the input prompt using natural language processing techniques and extracting key keywords such as "18th century," "France," and "education system," the server understands the core of the query. The result of this processing is output as a list of extracted keywords.

[0265] Step 3:

[0266] The server selects an appropriate model from among several natural language models based on the extracted keywords. The selection criterion is whether the knowledge domain related to the keywords matches the model's area of ​​expertise. For example, a model capable of handling data related to history and education might be selected. This selection result is output as identification information for the selected natural language model.

[0267] Step 4:

[0268] Using the selected natural language model, the server generates a response to the query. In this step, keyword-based information processing and data calculations are performed using the numerical data held by the model. The generated response is output as text data that answers the query. A concrete example of a response would include the information, "In 18th-century France, education was primarily provided by religious institutions."

[0269] Step 5:

[0270] The server integrates responses from multiple models and selects the most relevant response. During the integration process, the relevance and confidence level of each response are evaluated, and a scoring algorithm is used to select the optimal response. The output of this step is the selected single response text.

[0271] Step 6:

[0272] The server sends the selected response back to the information terminal. This response data is displayed to the user, allowing them to quickly obtain the most useful information. This ensures that the user receives clear and accurate answers.

[0273] (Application Example 1)

[0274] Next, we will explain Application Example 1. In the following explanation, the data processing device 12 will be referred to as the "server," and the smart glasses 214 will be referred to as the "terminal."

[0275] A challenge in virtual stores is that users cannot obtain product information smoothly and accurately. When such information acquisition is inefficient, the user experience deteriorates, and store usage is not encouraged.

[0276] 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.

[0277] In this invention, the server includes: a device that analyzes a query received from a user and extracts keywords in the query; a device that selects, based on the extracted keywords, a natural language processing model having a corresponding specialized field from among a plurality of natural language processing models; and a device that generates a response to the query using the selected natural language processing model. Thereby, the user can efficiently and accurately obtain product information.

[0278] A "user" is a person who uses a system or an application.

[0279] A "query" is information that represents a question or a request given from a user to a system.

[0280] A "keyword" is an important phrase extracted from within a query.

[0281] A "natural language processing model" is an algorithm or a model designed to understand and process natural language.

[0282] A "specialized field" is a specific knowledge area or theme that a language processing model specializes in handling.

[0283] A "response" is an answer or information generated in response to a query.

[0284] A "device" is a part of a machine or software designed to perform a specific function.

[0285] "Integration" is a process of combining multiple pieces of information or responses into one.

[0286] "Selection" is an act of selecting the most appropriate one from a plurality of options.

[0287] A "virtual store" is a store that sells products and provides information online.

[0288] This section describes embodiments for carrying out the invention. This invention realizes a system for efficiently providing product information in a virtual store. Queries are received from users via a terminal. These queries are mainly questions or requests related to product information.

[0289] The server first parses the received query and extracts keywords. This process utilizes natural language processing libraries such as spaCy and NLTK. Based on the extracted keywords, the server leverages machine learning frameworks such as TensorFlow and PyTorch to select the appropriate model from among multiple natural language processing models. Each natural language processing model is specialized in a particular field, allowing it to generate responses suitable for the relevant product information.

[0290] After generating responses, the server integrates multiple responses and selects the most relevant and accurate response. This integrated response is then provided to the user again via the terminal. Cloud servers, such as AWS EC2, achieve high processing speed and availability.

[0291] For example, if a user enters a query such as "What is the material of this product?", the system will select an appropriate natural language processing model and provide accurate material information. An example of a prompt would be: "User question: 'Please tell me the user reviews for this product.' Output: Collect review information and generate a response listing the key points." This allows users to efficiently obtain product information within the virtual store.

[0292] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0293] Step 1:

[0294] The user enters a query about a product using a terminal. This query is sent to the server. The input might be something like, "What is the material of this product?" The terminal forwards the query to the server in text format.

[0295] Step 2:

[0296] The server parses the received query and extracts keywords using natural language processing libraries (e.g., spaCy, NLTK). The input is the user's query, and the output is a list of keywords such as "product" and "material." The server applies text processing techniques to identify the key points of the information.

[0297] Step 3:

[0298] The server selects an appropriate natural language processing model based on the extracted keywords. This is done using a machine learning framework (e.g., TensorFlow, PyTorch). The input is a list of keywords, and the output is the selected model. The server then deploys a model selection algorithm to determine the most relevant model.

[0299] Step 4:

[0300] The server generates a response using a selected natural language processing model. The model processes the query and related information to produce an answer. The input is the selected model and information related to the query, and the output is the generated response. The server utilizes the model's inference capabilities to generate accurate information.

[0301] Step 5:

[0302] The system integrates responses generated by multiple natural language processing models to select the most appropriate response. The input consists of multiple responses, and the output is the most relevant response. The server uses an algorithm that compares the responses using evaluation criteria and selects the best one.

[0303] Step 6:

[0304] The server returns the selected response to the terminal. The user can receive the response through the terminal and obtain detailed information about the product. The input is the selected response, and the output is the information displayed on the user's terminal. The terminal visually presents the information and provides useful data to the user.

[0305] Furthermore, an emotion engine for estimating the user's emotion may be combined. That is, the specific processing unit 290 may estimate the user's emotion using the emotion identification model 59 and perform specific processing using the user's emotion.

[0306] The present invention provides a system that recognizes the user's emotion and adjusts the response based on it. This system aims to generate a highly accurate and user - considerate response by utilizing a plurality of language models based on the query received from the user.

[0307] First, the user inputs a query using the terminal. For example, the user may input a query containing an emotion such as "I feel angry about the recent news". The terminal transmits this query to the server.

[0308] The server analyzes the received query. It extracts keywords using natural language processing technology and analyzes the user's emotion through the emotion engine. In this step, the polarity of the emotion contained in the query, whether it is positive or negative, is determined.

[0309] Next, the server selects relevant models from a plurality of language models based on the extracted keywords. These selected models generate responses to the query based on their respective specialized fields.

[0310] The generated response is adjusted in content based on the emotion information detected by the emotion engine. For example, when the user has a negative emotion, the response is adjusted to be more polite and reassuring.

[0311] The server then integrates all the generated responses and selects the most appropriate one. This response selection process takes into account the accuracy, relevance, and emotional impact of the information.

[0312] Finally, the server provides the selected response to the terminal and displays it to the user. This allows the user to receive a response that takes their emotions into consideration, resulting in a better user experience.

[0313] Thus, by combining the system of the present invention with an emotion engine, it is possible to address not only technical accuracy but also the emotional needs of the user.

[0314] The following describes the processing flow.

[0315] Step 1:

[0316] The terminal receives a query from the user. The user enters a message expressing emotion, such as "I feel angry about the recent news." This input is then ready to be sent from the terminal to the server.

[0317] Step 2:

[0318] The server receives a query from the terminal. To process the query, the server uses natural language processing techniques to analyze it and extract important keywords. Simultaneously, it uses an emotion engine to analyze the user's emotions contained in the query. This analysis determines whether the query contains negative emotions.

[0319] Step 3:

[0320] Based on the analysis results, the server selects a language model with the appropriate expertise. In this case, models related to news and sentiment analysis are chosen. The selected model generates responses to the input query based on its respective area of ​​expertise.

[0321] Step 4:

[0322] The server adjusts the content of the generated response based on emotional information analyzed by the emotion engine. If a negative emotion is detected, the response is modified to be more considerate and encouraging, taking care to soothe the user's mood.

[0323] Step 5:

[0324] The server integrates the coordinated responses and selects the most appropriate one. This response selection process takes into account the high relevance and accuracy of the information, as well as consideration for the user's emotional needs.

[0325] Step 6:

[0326] The server sends the selected response to the terminal. The terminal displays this response to the user. As a result, the user receives accurate information that takes their emotional state into account, leading to a sense of reassurance and understanding.

[0327] (Example 2)

[0328] 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".

[0329] Conventional information response systems have struggled to improve the accuracy and relevance of responses to user queries, and also have difficulty generating responses that take user emotions into consideration. As a result, they have been unable to improve the quality of the user experience, and have a particular challenge in providing appropriate responses to users with negative emotions.

[0330] 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.

[0331] In this invention, the server includes means for analyzing queries received from a user and extracting keywords; means for selecting a language model with the relevant expertise from among multiple language models based on the extracted keywords; means for generating a response to the query using the selected language model; means for adjusting the generated response based on the user's emotional information; means for integrating the adjusted responses and selecting the most appropriate response; and means for providing the selected response to the user. This makes it possible to provide appropriate and considerate responses that take into account the user's emotions.

[0332] A "query" is information that represents a question or request that a user enters into a system.

[0333] A "keyword" is a phrase that indicates the main theme or topic extracted from a query.

[0334] A "language model" is an algorithm or data set used to process natural language and generate responses based on a specific area of ​​expertise.

[0335] "Emotional information" refers to information that indicates the polarity and characteristics of emotions included in user queries.

[0336] "Response adjustment" refers to the process of flexibly modifying the content of a generated response, taking into account the user's feelings.

[0337] "Response integration" is the process of combining multiple responses generated from different language models and selecting the single most appropriate response.

[0338] A "server" is a computing device or system that receives queries from users, performs the necessary processing, and generates responses.

[0339] This invention relates to an interactive system that generates appropriate responses while taking into account the user's emotions. This system primarily uses a terminal, a server, a language model, and an emotion engine.

[0340] Users use a terminal to enter queries into the system. These queries can include requests for information or expressions of opinion, and may contain emotional nuances. For example, consider a case where a user enters the query, "I feel angry about the recent news."

[0341] The terminal's role is to send the input query to the server. The server uses natural language processing libraries (e.g., spaCy or NLTK) to extract keywords in order to process the received query. It also uses sentiment engines (e.g., TextBlob or VADER) to analyze the sentiment information contained in the query and determine the polarity of the sentiment.

[0342] Based on this information, the server selects a model with the appropriate expertise from among multiple generative AI models. For example, for news-related queries, a model specializing in news will be selected. This process can utilize general generative AI technologies such as OpenAI.

[0343] The selected model leverages the query content and relevant expertise to generate an appropriate response. This response is refined within the server, and may be modified to take into account the emotional information provided by the user. For example, a query indicating negative emotions may be adjusted to include more reassuring and polite language in the response.

[0344] Finally, the server integrates the generated responses and selects the most appropriate one. This selection process takes into account the accuracy, relevance, and emotional considerations of the information. The selected response is sent to the terminal and displayed to the user.

[0345] This system configuration allows users to receive responses that take their emotions into account, resulting in a higher user experience. A concrete example of a prompt would be, "How should you respond if the user is angry about recent news?"

[0346] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0347] Step 1:

[0348] The user enters a query using a terminal. The entered query includes the user's information requests and emotions. For example, they might enter, "I feel angry about recent news." The terminal sends this query to the server.

[0349] Step 2:

[0350] The server parses the queries received from the terminal. First, it extracts keywords from the input query using a natural language processing library (e.g., spaCy or NLTK). This process identifies the theme or topic of the query. The extracted keywords are then output.

[0351] Step 3:

[0352] The server uses a sentiment engine (e.g., TextBlob or VADER) to analyze the sentiment contained in the query. This analysis determines the sentiment polarity (positive, negative, or neutral) of the query. The sentiment polarity information is obtained as output.

[0353] Step 4:

[0354] The server selects the most relevant generative AI model from multiple models based on the extracted keywords and sentiment polarity. For example, if keywords related to news are included, it will select a generative AI model specialized in news. This results in the output of a highly relevant generative AI model.

[0355] Step 5:

[0356] The server generates responses to queries using the selected generative AI model. This process leverages the model's expertise to produce highly accurate text output. The general content of the generated response is largely determined by the model.

[0357] Step 6:

[0358] The server adjusts the generated response based on the emotional information obtained by the emotion engine. If a negative emotion is detected, the response is modified to be more polite and considerate. The output is an emotionally conscious, adjusted response.

[0359] Step 7:

[0360] The server integrates multiple pre-tuned responses and selects the most appropriate one. This selection takes into account the accuracy, relevance, and emotional relevance of the response. The single best response is chosen as the output.

[0361] Step 8:

[0362] The server sends the selected response to the terminal. The terminal displays this response to the user, providing an appropriate answer that takes emotions into consideration. Through this process, the user receives an answer that reflects their emotions.

[0363] (Application Example 2)

[0364] 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."

[0365] In recent years, content delivery services have seen a growing demand for personalized content recommendations based on user emotions to improve the user experience. However, conventional technologies struggle to accurately analyze user emotions and provide appropriate responses and content immediately. This is a reason why user expectations are not being met, impacting service usage frequency and satisfaction.

[0366] 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.

[0367] In this invention, the server includes means for analyzing a query received from a user and extracting keywords and emotional states within the query; means for selecting a natural language processing model with the relevant expertise from among a plurality of natural language processing models based on the extracted keywords and emotional states; and means for generating a response to the query using the selected natural language processing model and adjusting the response content according to the user's emotional state. This enables the recommendation of personalized content that takes the user's emotions into consideration.

[0368] A "query" is information or a question that a user gives to a system, and it serves as the starting point for the system to analyze it and generate a response.

[0369] "Keywords" are important words or phrases extracted from a query, and they form the basis for the system to generate information and responses.

[0370] "Emotional state" refers to the psychological and emotional state of a user, analyzed from their queries and their actions before and after those queries, and is used when adjusting the content of the response.

[0371] A "natural language processing model" is an algorithmic method or system for understanding human language and generating responses, and it specializes in providing responses in various fields of expertise.

[0372] A "specialized natural language processing model" refers to a natural language processing model that specializes in processing information in a particular knowledge domain or topic, and has the ability to generate appropriate responses to relevant queries.

[0373] "Adjusting response content" refers to the process of modifying the tone and content of responses generated based on the user's emotional state, and delivering them in a way that is optimal for the user.

[0374] "Personalized content" refers to content that is individually generated and recommended based on a user's emotional state and past behavior, with the aim of improving the user experience.

[0375] The system for implementing this invention begins with a query input from the user via a terminal. The query received by the terminal from the user is transmitted to a server via a communication network. The server first analyzes the query and has the function of extracting keywords and sentiment states. Natural language processing technology is used for this, and software such as Python and TensorFlow is applied. The analysis of sentiment states is performed based on the text data contained in the user's query.

[0376] The server then selects the most suitable model from among several natural language processing models based on the extracted keywords and sentiment states. This process utilizes natural language processing models specialized in various fields, enabling sophisticated response generation. Generative AI models such as GPT are used in this process.

[0377] Using the selected model, the server generates a response to the query and further adjusts the response content considering the user's emotional state. This creates a thoughtful and appropriate response that takes the user's emotions into consideration. As a result, the responses and content provided by the system are personalized and tailored to the user's preferences and emotions.

[0378] Finally, the integrated response is returned to the terminal and presented to the user. This allows users to receive content in a way that takes their emotional state into consideration, improving the quality of the service and the user experience. A specific example of its use is when a user reacts to "anxious news": the server recommends reassuring content to support and improve the user's emotional state.

[0379] Example of a prompt:

[0380] "When users are feeling anxious, create content that provides reassurance. For example, think about lighthearted news that can refresh their mood or calming videos."

[0381] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0382] Step 1:

[0383] The user enters a query into the terminal. The entered query is sent from the terminal to the server in natural language text format. The input at this stage is the user's text data, and no specific information processing or analysis is performed.

[0384] Step 2:

[0385] The server parses the received query. The input is text data sent by the user. The server runs a natural language processing algorithm using Python to perform data analysis to extract keywords and sentiment states from the query. The output is the parsed keyword and sentiment state information.

[0386] Step 3:

[0387] The server selects an appropriate natural language processing model based on the extracted keywords and sentiment states. The input is the keywords and sentiment states obtained from step 2, and a comparison operation is performed to select a model with the corresponding area of ​​expertise. The output is the identification information of the selected model.

[0388] Step 4:

[0389] The server generates responses to queries using a selected natural language processing model. The input consists of the selected model and the user's query, and the generating AI model works to create an appropriate response. During this process, the response content is also adjusted according to the emotional state. The output is the adjusted response data.

[0390] Step 5:

[0391] When multiple responses are generated, the server integrates them and selects the most appropriate response. The input consists of multiple adjusted responses, and the server performs an optimization evaluation based on content, relevance, etc., to select the response deemed most appropriate. The output is the selected optimal response.

[0392] Step 6:

[0393] The selected response is sent to the terminal and displayed to the user. The input is the optimal response received from step 5, and the output is the response information presented to the user. This allows the user to receive a personalized response and obtain emotionally sensitive information.

[0394] 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.

[0395] 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.

[0396] 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.

[0397] [Third Embodiment]

[0398] Figure 5 shows an example of the configuration of the data processing system 310 according to the third embodiment.

[0399] 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.

[0400] 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).

[0401] 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.

[0402] 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.

[0403] 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).

[0404] 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.

[0405] 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.

[0406] 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.

[0407] 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.

[0408] 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.

[0409] 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".

[0410] This invention provides a system that receives queries from users and generates highly accurate responses with low power consumption by efficiently combining multiple language models. The following shows the processing flow based on an embodiment of the present invention.

[0411] First, the user inputs a specific question or problem through the terminal. For example, the user might ask, "Tell me about the 18th-century French education system." The terminal then sends this query to the server.

[0412] Next, the server receives this query and extracts keywords. In this step, it identifies important vocabulary such as "18th century," "France," and "education system." This process reveals basic information about the content of the query.

[0413] Subsequently, the server selects a language model with relevant expertise from among several language models based on the extracted keywords. For example, a model specializing in history and education might be selected.

[0414] The selected model generates responses based on its respective area of ​​expertise. For example, it might generate information such as, "In 18th-century France, education was primarily conducted by religious institutions."

[0415] The server integrates the generated responses and selects the most appropriate one. This process narrows down the multiple responses to those that are most relevant and contain useful information.

[0416] Finally, the server provides the user with the selected response. The user can obtain the most accurate and relevant information through their terminal.

[0417] Thus, the present invention efficiently utilizes multiple language models to quickly and effectively answer user inquiries, thereby minimizing power consumption while providing highly accurate information.

[0418] The following describes the processing flow.

[0419] Step 1:

[0420] The terminal receives a query from the user. The user enters a specific question into the terminal, such as "Tell me about the 18th-century French education system." The terminal prepares to send this input to the server.

[0421] Step 2:

[0422] The server receives queries sent from terminals. To analyze the queries, it interprets their content using natural language processing techniques and extracts keywords. For example, words like "18th century," "France," and "education system" might be identified.

[0423] Step 3:

[0424] The server selects language models with relevant expertise based on keywords. This includes models related to history and models specializing in education. The server makes this selection by referring to the specialized domains of each model.

[0425] Step 4:

[0426] The server passes queries to each selected language model and instructs them to generate responses. Each model generates a response from its area of ​​expertise based on the given query. For example, it might generate information such as, "In 18th-century France, the education system was mainly run by religious institutions."

[0427] Step 5:

[0428] The server receives multiple generated responses. It evaluates these responses and performs integration to select the optimal response. The server considers the relevance and accuracy of each response to find the response that contains the most useful information to provide to the user.

[0429] Step 6:

[0430] The server sends a unified, optimal response back to the terminal. The terminal displays this response to the user. The user can obtain detailed and accurate information through the terminal and resolve any questions regarding the query.

[0431] (Example 1)

[0432] 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."

[0433] A challenge exists in efficiently utilizing multiple natural language models to provide rapid and highly accurate responses to user information requests while minimizing power consumption. To address this challenge, technology is needed to appropriately select and integrate highly relevant information and provide it to the user.

[0434] 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.

[0435] In this invention, the server includes means for analyzing user queries received by an information terminal and extracting important words within the query; means for selecting a natural language model with a predetermined knowledge domain from among a plurality of natural language models based on the extracted important words; and means for generating a response to the query using the selected natural language model. This makes it possible to provide the user with highly relevant responses quickly and accurately.

[0436] An "information terminal" is an electronic device used by users to input queries and receive the results.

[0437] A "user query" refers to a question or request for information entered by a user through an information terminal.

[0438] "Key words" are words or phrases that contain essential information and are particularly relevant in meaning within a user query.

[0439] A "natural language model" is an algorithm or mathematical model trained for natural language processing, which understands and generates text data related to a specific knowledge domain.

[0440] A "knowledge domain" is a concept that represents the scope of a specific area of ​​expertise or content, and indicates the area of ​​expertise that a natural language model possesses.

[0441] "Numerical data" refers to information that indicates the tuning parameters and weights that a natural language model has optimized for a specific knowledge domain.

[0442] This invention begins with a process in which a user inputs a specific question or problem using an information terminal. The information terminal is often an electronic device such as a personal computer or smartphone. A concrete example would be a user inputting, "Tell me about the 18th-century French education system."

[0443] The information terminal sends the user query entered to the server. This transmission uses communication methods such as HTTP requests over the internet. The server parses the received query and extracts important words from its text. For this parsing, it is suitable to use libraries that provide natural language processing technology (e.g., NLTK or spaCy).

[0444] Based on the extracted key words, the server selects the appropriate model from among several natural language models. These natural language models possess numerical data relevant to their knowledge domain and can leverage existing technologies such as BERT and GPT-3. The selected model performs data processing related to the specific knowledge domain and generates a response.

[0445] The generated responses are integrated within the server, and the most relevant response is selected from among them. Scoring algorithms and filtering techniques can be applied to this integration and selection process.

[0446] Ultimately, the server sends the selected response back to the information terminal, providing it to the user. Through this response, the user can obtain the requested information. For example, using the prompt "Explain the promotion of art during the Renaissance," it is possible to obtain relevant historical information.

[0447] In this way, this invention provides a system that efficiently reduces power consumption while responding to user requirements with high precision and speed.

[0448] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0449] Step 1:

[0450] The user enters a query using an information terminal. This query is entered as text data indicating the content of the inquiry. For example, a prompt such as "Tell me about the 18th-century French education system" might be entered. This input data is sent from the information terminal to the server.

[0451] Step 2:

[0452] The server analyzes the query received from the user and extracts important words from the text. By analyzing the input prompt using natural language processing techniques and extracting key keywords such as "18th century," "France," and "education system," the server understands the core of the query. The result of this processing is output as a list of extracted keywords.

[0453] Step 3:

[0454] The server selects an appropriate model from among several natural language models based on the extracted keywords. The selection criterion is whether the knowledge domain related to the keywords matches the model's area of ​​expertise. For example, a model capable of handling data related to history and education might be selected. This selection result is output as identification information for the selected natural language model.

[0455] Step 4:

[0456] Using the selected natural language model, the server generates a response to the query. In this step, keyword-based information processing and data calculations are performed using the numerical data held by the model. The generated response is output as text data that answers the query. A concrete example of a response would include the information, "In 18th-century France, education was primarily provided by religious institutions."

[0457] Step 5:

[0458] The server integrates responses from multiple models and selects the most relevant response. During the integration process, the relevance and confidence level of each response are evaluated, and a scoring algorithm is used to select the optimal response. The output of this step is the selected single response text.

[0459] Step 6:

[0460] The server sends the selected response back to the information terminal. This response data is displayed to the user, allowing them to quickly obtain the most useful information. This ensures that the user receives clear and accurate answers.

[0461] (Application Example 1)

[0462] 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."

[0463] A challenge in virtual stores is that users cannot obtain product information smoothly and accurately. When such information acquisition is inefficient, the user experience deteriorates, and store usage is not encouraged.

[0464] 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.

[0465] In this invention, the server includes a device for analyzing queries received from a user and extracting keywords within those queries; a device for selecting a natural language processing model with the relevant expertise from among a plurality of natural language processing models based on the extracted keywords; and a device for generating a response to the query using the selected natural language processing model. This enables users to efficiently and accurately obtain product information.

[0466] A "user" is someone who uses a system or application.

[0467] A "query" refers to information that a user gives to a system, such as a question or request.

[0468] "Keywords" are important terms extracted from within a query.

[0469] A "natural language processing model" is an algorithm or model designed to understand and process natural language.

[0470] A "specialized field" refers to a specific area of ​​knowledge or topic that a language processing model specializes in addressing.

[0471] A "response" is the answer or information generated in response to a query.

[0472] A "device" is a part of a machine or software designed to perform a specific function.

[0473] "Integration" is the process of combining multiple pieces of information or responses into one.

[0474] "Selection" is the act of choosing the most appropriate option from multiple choices.

[0475] A "virtual store" is a store that sells products and provides information online.

[0476] This section describes embodiments for carrying out the invention. This invention realizes a system for efficiently providing product information in a virtual store. Queries are received from users via a terminal. These queries are mainly questions or requests related to product information.

[0477] The server first parses the received query and extracts keywords. This process utilizes natural language processing libraries such as spaCy and NLTK. Based on the extracted keywords, the server leverages machine learning frameworks such as TensorFlow and PyTorch to select the appropriate model from among multiple natural language processing models. Each natural language processing model is specialized in a particular field, allowing it to generate responses suitable for the relevant product information.

[0478] After generating responses, the server integrates multiple responses and selects the most relevant and accurate response. This integrated response is then provided to the user again via the terminal. Cloud servers, such as AWS EC2, achieve high processing speed and availability.

[0479] For example, if a user enters a query such as "What is the material of this product?", the system will select an appropriate natural language processing model and provide accurate material information. An example of a prompt would be: "User question: 'Please tell me the user reviews for this product.' Output: Collect review information and generate a response listing the key points." This allows users to efficiently obtain product information within the virtual store.

[0480] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0481] Step 1:

[0482] The user enters a query about a product using a terminal. This query is sent to the server. The input might be something like, "What is the material of this product?" The terminal forwards the query to the server in text format.

[0483] Step 2:

[0484] The server parses the received query and extracts keywords using natural language processing libraries (e.g., spaCy, NLTK). The input is the user's query, and the output is a list of keywords such as "product" and "material." The server applies text processing techniques to identify the key points of the information.

[0485] Step 3:

[0486] The server selects an appropriate natural language processing model based on the extracted keywords. This is done using a machine learning framework (e.g., TensorFlow, PyTorch). The input is a list of keywords, and the output is the selected model. The server then deploys a model selection algorithm to determine the most relevant model.

[0487] Step 4:

[0488] The server generates a response using a selected natural language processing model. The model processes the query and related information to produce an answer. The input is the selected model and information related to the query, and the output is the generated response. The server utilizes the model's inference capabilities to generate accurate information.

[0489] Step 5:

[0490] The system integrates responses generated by multiple natural language processing models to select the most appropriate response. The input consists of multiple responses, and the output is the most relevant response. The server uses an algorithm that compares the responses using evaluation criteria and selects the best one.

[0491] Step 6:

[0492] The server sends the selected response back to the terminal. The user receives the response through the terminal and can learn detailed information about the product. The input is the selected response, and the output is the information displayed on the user's terminal. The terminal visually presents the information and provides the user with useful data.

[0493] 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.

[0494] This invention provides a system that recognizes user emotions and adjusts responses accordingly. The system aims to generate highly accurate and user-friendly responses by utilizing multiple language models based on queries received from the user.

[0495] First, the user enters a query using a terminal. For example, they might enter a query that includes an emotion, such as "I feel angry about recent news." The terminal then sends this query to the server.

[0496] The server parses the received query. It extracts keywords using natural language processing techniques and analyzes the user's emotions through an emotion engine. In this step, the polarity of the emotions contained in the query, whether positive or negative, is determined.

[0497] Next, the server selects relevant models from multiple language models based on the extracted keywords. These selected models then generate responses to the query based on their respective areas of expertise.

[0498] The generated response is adjusted based on the emotional information detected by the emotion engine. For example, if the user is experiencing negative emotions, the response will be adjusted to be more polite and reassuring.

[0499] The server then integrates all the generated responses and selects the most appropriate one. This response selection process takes into account the accuracy, relevance, and emotional impact of the information.

[0500] Finally, the server provides the selected response to the terminal and displays it to the user. This allows the user to receive a response that takes their emotions into consideration, resulting in a better user experience.

[0501] Thus, by combining the system of the present invention with an emotion engine, it is possible to address not only technical accuracy but also the emotional needs of the user.

[0502] The following describes the processing flow.

[0503] Step 1:

[0504] The terminal receives a query from the user. The user enters a message expressing emotion, such as "I feel angry about the recent news." This input is then ready to be sent from the terminal to the server.

[0505] Step 2:

[0506] The server receives a query from the terminal. To process the query, the server uses natural language processing techniques to analyze it and extract important keywords. Simultaneously, it uses an emotion engine to analyze the user's emotions contained in the query. This analysis determines whether the query contains negative emotions.

[0507] Step 3:

[0508] Based on the analysis results, the server selects a language model with the appropriate expertise. In this case, models related to news and sentiment analysis are chosen. The selected model generates responses to the input query based on its respective area of ​​expertise.

[0509] Step 4:

[0510] The server adjusts the content of the generated response based on emotional information analyzed by the emotion engine. If a negative emotion is detected, the response is modified to be more considerate and encouraging, taking care to soothe the user's mood.

[0511] Step 5:

[0512] The server integrates the coordinated responses and selects the most appropriate one. This response selection process takes into account the high relevance and accuracy of the information, as well as consideration for the user's emotional needs.

[0513] Step 6:

[0514] The server sends the selected response to the terminal. The terminal displays this response to the user. As a result, the user receives accurate information that takes their emotional state into account, leading to a sense of reassurance and understanding.

[0515] (Example 2)

[0516] 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."

[0517] Conventional information response systems have struggled to improve the accuracy and relevance of responses to user queries, and also have difficulty generating responses that take user emotions into consideration. As a result, they have been unable to improve the quality of the user experience, and have a particular challenge in providing appropriate responses to users with negative emotions.

[0518] 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.

[0519] In this invention, the server includes means for analyzing queries received from a user and extracting keywords; means for selecting a language model with the relevant expertise from among multiple language models based on the extracted keywords; means for generating a response to the query using the selected language model; means for adjusting the generated response based on the user's emotional information; means for integrating the adjusted responses and selecting the most appropriate response; and means for providing the selected response to the user. This makes it possible to provide appropriate and considerate responses that take into account the user's emotions.

[0520] A "query" is information that represents a question or request that a user enters into a system.

[0521] A "keyword" is a phrase that indicates the main theme or topic extracted from a query.

[0522] A "language model" is an algorithm or data set used to process natural language and generate responses based on a specific area of ​​expertise.

[0523] "Emotional information" refers to information that indicates the polarity and characteristics of emotions included in user queries.

[0524] "Response adjustment" refers to the process of flexibly modifying the content of a generated response, taking into account the user's feelings.

[0525] "Response integration" is the process of combining multiple responses generated from different language models and selecting the single most appropriate response.

[0526] A "server" is a computing device or system that receives queries from users, performs the necessary processing, and generates responses.

[0527] This invention relates to an interactive system that generates appropriate responses while taking into account the user's emotions. This system primarily uses a terminal, a server, a language model, and an emotion engine.

[0528] Users use a terminal to enter queries into the system. These queries can include requests for information or expressions of opinion, and may contain emotional nuances. For example, consider a case where a user enters the query, "I feel angry about the recent news."

[0529] The terminal's role is to send the input query to the server. The server uses natural language processing libraries (e.g., spaCy or NLTK) to extract keywords in order to process the received query. It also uses sentiment engines (e.g., TextBlob or VADER) to analyze the sentiment information contained in the query and determine the polarity of the sentiment.

[0530] Based on this information, the server selects a model with the appropriate expertise from among multiple generative AI models. For example, for news-related queries, a model specializing in news will be selected. This process can utilize general generative AI technologies such as OpenAI.

[0531] The selected model leverages the query content and relevant expertise to generate an appropriate response. This response is refined within the server, and may be modified to take into account the emotional information provided by the user. For example, a query indicating negative emotions may be adjusted to include more reassuring and polite language in the response.

[0532] Finally, the server integrates the generated responses and selects the most appropriate one. This selection process takes into account the accuracy, relevance, and emotional considerations of the information. The selected response is sent to the terminal and displayed to the user.

[0533] This system configuration allows users to receive responses that take their emotions into account, resulting in a higher user experience. A concrete example of a prompt would be, "How should you respond if the user is angry about recent news?"

[0534] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0535] Step 1:

[0536] The user enters a query using a terminal. The entered query includes the user's information requests and emotions. For example, they might enter, "I feel angry about recent news." The terminal sends this query to the server.

[0537] Step 2:

[0538] The server parses the queries received from the terminal. First, it extracts keywords from the input query using a natural language processing library (e.g., spaCy or NLTK). This process identifies the theme or topic of the query. The extracted keywords are then output.

[0539] Step 3:

[0540] The server uses a sentiment engine (e.g., TextBlob or VADER) to analyze the sentiment contained in the query. This analysis determines the sentiment polarity (positive, negative, or neutral) of the query. The sentiment polarity information is obtained as output.

[0541] Step 4:

[0542] The server selects the most relevant generative AI model from multiple models based on the extracted keywords and sentiment polarity. For example, if keywords related to news are included, it will select a generative AI model specialized in news. This results in the output of a highly relevant generative AI model.

[0543] Step 5:

[0544] The server generates responses to queries using the selected generative AI model. This process leverages the model's expertise to produce highly accurate text output. The general content of the generated response is largely determined by the model.

[0545] Step 6:

[0546] The server adjusts the generated response based on the emotional information obtained by the emotion engine. If a negative emotion is detected, the response is modified to be more polite and considerate. The output is an emotionally conscious, adjusted response.

[0547] Step 7:

[0548] The server integrates multiple pre-tuned responses and selects the most appropriate one. This selection takes into account the accuracy, relevance, and emotional relevance of the response. The single best response is chosen as the output.

[0549] Step 8:

[0550] The server sends the selected response to the terminal. The terminal displays this response to the user, providing an appropriate answer that takes emotions into consideration. Through this process, the user can receive an answer that reflects their emotions.

[0551] (Application Example 2)

[0552] 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."

[0553] In recent years, content delivery services have seen a growing demand for personalized content recommendations based on user emotions to improve the user experience. However, conventional technologies struggle to accurately analyze user emotions and provide appropriate responses and content immediately. This is a reason why user expectations are not being met, impacting service usage frequency and satisfaction.

[0554] 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.

[0555] In this invention, the server includes means for analyzing a query received from a user and extracting keywords and emotional states within the query; means for selecting a natural language processing model with the relevant expertise from among a plurality of natural language processing models based on the extracted keywords and emotional states; and means for generating a response to the query using the selected natural language processing model and adjusting the response content according to the user's emotional state. This enables the recommendation of personalized content that takes the user's emotions into consideration.

[0556] A "query" is information or a question that a user gives to a system, and it serves as the starting point for the system to analyze it and generate a response.

[0557] "Keywords" are important words or phrases extracted from a query, and they form the basis for the system to generate information and responses.

[0558] "Emotional state" refers to the psychological and emotional state of a user, analyzed from their queries and their actions before and after those queries, and is used when adjusting the content of the response.

[0559] A "natural language processing model" is an algorithmic method or system for understanding human language and generating responses, and it specializes in providing responses in various fields of expertise.

[0560] A "specialized natural language processing model" refers to a natural language processing model that specializes in processing information in a specific knowledge domain or topic, and has the ability to generate appropriate responses to relevant queries.

[0561] "Adjusting response content" refers to the process of modifying the tone and content of responses generated based on the user's emotional state, and delivering them in a way that is optimal for the user.

[0562] "Personalized content" refers to content that is individually generated and recommended based on a user's emotional state and past behavior, with the aim of improving the user experience.

[0563] The system for implementing this invention begins with a query input from the user via a terminal. The query received by the terminal from the user is transmitted to a server via a communication network. The server first analyzes the query and has the function of extracting keywords and sentiment states. Natural language processing technology is used for this, and software such as Python and TensorFlow is applied. The analysis of sentiment states is performed based on the text data contained in the user's query.

[0564] The server then selects the most suitable model from among several natural language processing models based on the extracted keywords and sentiment states. This process utilizes natural language processing models specialized in various fields, enabling sophisticated response generation. Generative AI models such as GPT are used in this process.

[0565] Using the selected model, the server generates a response to the query and further adjusts the response content considering the user's emotional state. This creates a thoughtful and appropriate response that takes the user's emotions into consideration. As a result, the responses and content provided by the system are personalized and tailored to the user's preferences and emotions.

[0566] Finally, the integrated response is returned to the terminal and presented to the user. This allows users to receive content in a way that takes their emotional state into consideration, improving the quality of the service and the user experience. A specific example of its use is when a user reacts to "anxious news": the server recommends reassuring content to support and improve the user's emotional state.

[0567] Example of a prompt:

[0568] "When users are feeling anxious, create content that provides reassurance. For example, think about lighthearted news that can refresh their mood or calming videos."

[0569] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0570] Step 1:

[0571] The user enters a query into the terminal. The entered query is sent from the terminal to the server in natural language text format. The input at this stage is the user's text data, and no specific information processing or analysis is performed.

[0572] Step 2:

[0573] The server parses the received query. The input is text data sent by the user. The server runs a natural language processing algorithm using Python to perform data analysis to extract keywords and sentiment states from the query. The output is the parsed keyword and sentiment state information.

[0574] Step 3:

[0575] The server selects an appropriate natural language processing model based on the extracted keywords and sentiment states. The input is the keywords and sentiment states obtained from step 2, and a comparison operation is performed to select a model with the corresponding area of ​​expertise. The output is the identification information of the selected model.

[0576] Step 4:

[0577] The server generates responses to queries using a selected natural language processing model. The input consists of the selected model and the user's query, and the generating AI model works to create an appropriate response. During this process, the response content is also adjusted according to the emotional state. The output is the adjusted response data.

[0578] Step 5:

[0579] When multiple responses are generated, the server integrates them and selects the most appropriate response. The input consists of multiple adjusted responses, and the server performs an optimization evaluation based on content, relevance, etc., to select the response deemed most appropriate. The output is the selected optimal response.

[0580] Step 6:

[0581] The selected response is sent to the terminal and displayed to the user. The input is the optimal response received from step 5, and the output is the response information presented to the user. This allows the user to receive a personalized response and obtain emotionally sensitive information.

[0582] 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.

[0583] 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.

[0584] 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.

[0585] [Fourth Embodiment]

[0586] Figure 7 shows an example of the configuration of the data processing system 410 according to the fourth embodiment.

[0587] 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.

[0588] 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).

[0589] 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.

[0590] 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.

[0591] 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).

[0592] 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.

[0593] 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.

[0594] 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.

[0595] 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.

[0596] 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.

[0597] 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.

[0598] 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".

[0599] This invention provides a system that receives queries from users and generates highly accurate responses with low power consumption by efficiently combining multiple language models. The following shows the processing flow based on an embodiment of the present invention.

[0600] First, the user inputs a specific question or problem through the terminal. For example, the user might ask, "Tell me about the 18th-century French education system." The terminal then sends this query to the server.

[0601] Next, the server receives this query and extracts keywords. In this step, it identifies important vocabulary such as "18th century," "France," and "education system." This process reveals basic information about the content of the query.

[0602] Subsequently, the server selects a language model with relevant expertise from among several language models based on the extracted keywords. For example, a model specializing in history and education might be selected.

[0603] The selected model generates responses based on its respective area of ​​expertise. For example, it might generate information such as, "In 18th-century France, education was primarily conducted by religious institutions."

[0604] The server integrates the generated responses and selects the most appropriate one. This process narrows down the multiple responses to those that are most relevant and contain useful information.

[0605] Finally, the server provides the user with the selected response. The user can obtain the most accurate and relevant information through their terminal.

[0606] Thus, the present invention efficiently utilizes multiple language models to quickly and effectively answer user inquiries, thereby minimizing power consumption while providing highly accurate information.

[0607] The following describes the processing flow.

[0608] Step 1:

[0609] The terminal receives a query from the user. The user enters a specific question into the terminal, such as "Tell me about the 18th-century French education system." The terminal prepares to send this input to the server.

[0610] Step 2:

[0611] The server receives queries sent from terminals. To analyze the queries, it interprets their content using natural language processing techniques and extracts keywords. For example, words like "18th century," "France," and "education system" might be identified.

[0612] Step 3:

[0613] The server selects language models with relevant expertise based on keywords. This includes models related to history and models specializing in education. The server makes this selection by referring to the specialized domains of each model.

[0614] Step 4:

[0615] The server passes queries to each selected language model and instructs them to generate responses. Each model generates a response from its area of ​​expertise based on the given query. For example, it might generate information such as, "In 18th-century France, the education system was mainly run by religious institutions."

[0616] Step 5:

[0617] The server receives multiple generated responses. It evaluates these responses and performs integration to select the optimal response. The server considers the relevance and accuracy of each response to find the response that contains the most useful information to provide to the user.

[0618] Step 6:

[0619] The server sends a unified, optimal response back to the terminal. The terminal displays this response to the user. The user can obtain detailed and accurate information through the terminal and resolve any questions regarding the query.

[0620] (Example 1)

[0621] 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".

[0622] A challenge exists in efficiently utilizing multiple natural language models to provide rapid and highly accurate responses to user information requests while minimizing power consumption. To address this challenge, technology is needed to appropriately select and integrate highly relevant information and provide it to the user.

[0623] 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.

[0624] In this invention, the server includes means for analyzing user queries received by an information terminal and extracting important words within the query; means for selecting a natural language model with a predetermined knowledge domain from among a plurality of natural language models based on the extracted important words; and means for generating a response to the query using the selected natural language model. This makes it possible to provide the user with highly relevant responses quickly and accurately.

[0625] An "information terminal" is an electronic device used by users to input queries and receive the results.

[0626] A "user query" refers to a question or request for information entered by a user through an information terminal.

[0627] "Key words" are words or phrases that contain essential information and are particularly relevant in meaning within a user query.

[0628] A "natural language model" is an algorithm or mathematical model trained for natural language processing, which understands and generates text data related to a specific knowledge domain.

[0629] A "knowledge domain" is a concept that represents the scope of a specific area of ​​expertise or content, and indicates the area of ​​expertise that a natural language model possesses.

[0630] "Numerical data" refers to information that indicates the tuning parameters and weights that a natural language model has optimized for a specific knowledge domain.

[0631] This invention begins with a process in which a user inputs a specific question or problem using an information terminal. The information terminal is often an electronic device such as a personal computer or smartphone. A concrete example would be a user inputting, "Tell me about the 18th-century French education system."

[0632] The information terminal sends the user query entered to the server. This transmission uses communication methods such as HTTP requests over the internet. The server parses the received query and extracts important words from its text. For this parsing, it is suitable to use libraries that provide natural language processing technology (e.g., NLTK or spaCy).

[0633] Based on the extracted key words, the server selects the appropriate model from among several natural language models. These natural language models possess numerical data relevant to their knowledge domain and can leverage existing technologies such as BERT and GPT-3. The selected model performs data processing related to the specific knowledge domain and generates a response.

[0634] The generated responses are integrated within the server, and the most relevant response is selected from among them. Scoring algorithms and filtering techniques can be applied to this integration and selection process.

[0635] Ultimately, the server sends the selected response back to the information terminal, providing it to the user. Through this response, the user can obtain the requested information. For example, using the prompt "Explain the promotion of art during the Renaissance," it is possible to obtain relevant historical information.

[0636] In this way, this invention provides a system that efficiently reduces power consumption while responding to user requirements with high precision and speed.

[0637] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0638] Step 1:

[0639] The user enters a query using an information terminal. This query is entered as text data indicating the content of the inquiry. For example, a prompt such as "Tell me about the 18th-century French education system" might be entered. This input data is sent from the information terminal to the server.

[0640] Step 2:

[0641] The server analyzes the query received from the user and extracts important words from the text. By analyzing the input prompt using natural language processing techniques and extracting key keywords such as "18th century," "France," and "education system," the server understands the core of the query. The result of this processing is output as a list of extracted keywords.

[0642] Step 3:

[0643] The server selects an appropriate model from among several natural language models based on the extracted keywords. The selection criterion is whether the knowledge domain related to the keywords matches the model's area of ​​expertise. For example, a model capable of handling data related to history and education might be selected. This selection result is output as identification information for the selected natural language model.

[0644] Step 4:

[0645] Using the selected natural language model, the server generates a response to the query. In this step, keyword-based information processing and data calculations are performed using the numerical data held by the model. The generated response is output as text data that answers the query. A concrete example of a response would include the information, "In 18th-century France, education was primarily provided by religious institutions."

[0646] Step 5:

[0647] The server integrates responses from multiple models and selects the most relevant response. During the integration process, the relevance and confidence level of each response are evaluated, and a scoring algorithm is used to select the optimal response. The output of this step is the selected single response text.

[0648] Step 6:

[0649] The server sends the selected response back to the information terminal. This response data is displayed to the user, allowing them to quickly obtain the most useful information. This ensures that the user receives clear and accurate answers.

[0650] (Application Example 1)

[0651] 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".

[0652] A challenge in virtual stores is that users cannot obtain product information smoothly and accurately. When such information acquisition is inefficient, the user experience deteriorates, and store usage is not encouraged.

[0653] 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.

[0654] In this invention, the server includes a device for analyzing queries received from a user and extracting keywords within those queries; a device for selecting a natural language processing model with the relevant expertise from among a plurality of natural language processing models based on the extracted keywords; and a device for generating a response to the query using the selected natural language processing model. This enables users to efficiently and accurately obtain product information.

[0655] A "user" is someone who uses a system or application.

[0656] A "query" refers to information that a user gives to a system, such as a question or request.

[0657] "Keywords" are important terms extracted from within a query.

[0658] A "natural language processing model" is an algorithm or model designed to understand and process natural language.

[0659] A "specialized field" refers to a specific area of ​​knowledge or topic that a language processing model specializes in addressing.

[0660] A "response" is the answer or information generated in response to a query.

[0661] A "device" is a part of a machine or software designed to perform a specific function.

[0662] "Integration" is the process of combining multiple pieces of information or responses into one.

[0663] "Selection" is the act of choosing the most appropriate option from multiple choices.

[0664] A "virtual store" is a store that sells products and provides information online.

[0665] This section describes embodiments for carrying out the invention. This invention realizes a system for efficiently providing product information in a virtual store. Queries are received from users via a terminal. These queries are mainly questions or requests related to product information.

[0666] The server first parses the received query and extracts keywords. This process utilizes natural language processing libraries such as spaCy and NLTK. Based on the extracted keywords, the server leverages machine learning frameworks such as TensorFlow and PyTorch to select the appropriate model from among multiple natural language processing models. Each natural language processing model is specialized in a particular field, allowing it to generate responses suitable for the relevant product information.

[0667] After generating responses, the server integrates multiple responses and selects the most relevant and accurate response. This integrated response is then provided to the user again via the terminal. Cloud servers, such as AWS EC2, achieve high processing speed and availability.

[0668] For example, if a user enters a query such as "What is the material of this product?", the system will select an appropriate natural language processing model and provide accurate material information. An example of a prompt would be: "User question: 'Please tell me the user reviews for this product.' Output: Collect review information and generate a response listing the key points." This allows users to efficiently obtain product information within the virtual store.

[0669] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0670] Step 1:

[0671] The user enters a query about a product using a terminal. This query is sent to the server. The input might be something like, "What is the material of this product?" The terminal forwards the query to the server in text format.

[0672] Step 2:

[0673] The server parses the received query and extracts keywords using natural language processing libraries (e.g., spaCy, NLTK). The input is the user's query, and the output is a list of keywords such as "product" and "material." The server applies text processing techniques to identify the key points of the information.

[0674] Step 3:

[0675] The server selects an appropriate natural language processing model based on the extracted keywords. This is done using a machine learning framework (e.g., TensorFlow, PyTorch). The input is a list of keywords, and the output is the selected model. The server then deploys a model selection algorithm to determine the most relevant model.

[0676] Step 4:

[0677] The server generates a response using a selected natural language processing model. The model processes the query and related information to produce an answer. The input is the selected model and information related to the query, and the output is the generated response. The server utilizes the model's inference capabilities to generate accurate information.

[0678] Step 5:

[0679] The system integrates responses generated by multiple natural language processing models to select the most appropriate response. The input consists of multiple responses, and the output is the most relevant response. The server uses an algorithm that compares the responses using evaluation criteria and selects the best one.

[0680] Step 6:

[0681] The server sends the selected response back to the terminal. The user receives the response through the terminal and can learn detailed information about the product. The input is the selected response, and the output is the information displayed on the user's terminal. The terminal visually presents the information and provides the user with useful data.

[0682] 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.

[0683] This invention provides a system that recognizes user emotions and adjusts responses accordingly. The system aims to generate highly accurate and user-friendly responses by utilizing multiple language models based on queries received from the user.

[0684] First, the user enters a query using a terminal. For example, they might enter a query that includes an emotion, such as "I feel angry about recent news." The terminal then sends this query to the server.

[0685] The server parses the received query. It extracts keywords using natural language processing techniques and analyzes the user's emotions through an emotion engine. In this step, the polarity of the emotions contained in the query, whether positive or negative, is determined.

[0686] Next, the server selects relevant models from multiple language models based on the extracted keywords. These selected models then generate responses to the query based on their respective areas of expertise.

[0687] The generated response is adjusted based on the emotional information detected by the emotion engine. For example, if the user is experiencing negative emotions, the response will be adjusted to be more polite and reassuring.

[0688] The server then integrates all the generated responses and selects the most appropriate one. This response selection process takes into account the accuracy, relevance, and emotional impact of the information.

[0689] Finally, the server provides the selected response to the terminal and displays it to the user. This allows the user to receive a response that takes their emotions into consideration, resulting in a better user experience.

[0690] Thus, by combining the system of the present invention with an emotion engine, it is possible to address not only technical accuracy but also the emotional needs of the user.

[0691] The following describes the processing flow.

[0692] Step 1:

[0693] The terminal receives a query from the user. The user enters a message expressing emotion, such as "I feel angry about the recent news." This input is then ready to be sent from the terminal to the server.

[0694] Step 2:

[0695] The server receives a query from the terminal. To process the query, the server uses natural language processing techniques to analyze it and extract important keywords. Simultaneously, it uses an emotion engine to analyze the user's emotions contained in the query. This analysis determines whether the query contains negative emotions.

[0696] Step 3:

[0697] Based on the analysis results, the server selects a language model with the appropriate expertise. In this case, models related to news and sentiment analysis are chosen. The selected model generates responses to the input query based on its respective area of ​​expertise.

[0698] Step 4:

[0699] The server adjusts the content of the generated response based on emotional information analyzed by the emotion engine. If a negative emotion is detected, the response is modified to be more considerate and encouraging, taking care to soothe the user's mood.

[0700] Step 5:

[0701] The server integrates the coordinated responses and selects the most appropriate one. This response selection process takes into account the high relevance and accuracy of the information, as well as consideration for the user's emotional needs.

[0702] Step 6:

[0703] The server sends the selected response to the terminal. The terminal displays this response to the user. As a result, the user receives accurate information that takes their emotional state into account, leading to a sense of reassurance and understanding.

[0704] (Example 2)

[0705] 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".

[0706] Conventional information response systems have struggled to improve the accuracy and relevance of responses to user queries, and also have difficulty generating responses that take user emotions into consideration. As a result, they have been unable to improve the quality of the user experience, and have a particular challenge in providing appropriate responses to users with negative emotions.

[0707] 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.

[0708] In this invention, the server includes means for analyzing queries received from a user and extracting keywords; means for selecting a language model with the relevant expertise from among multiple language models based on the extracted keywords; means for generating a response to the query using the selected language model; means for adjusting the generated response based on the user's emotional information; means for integrating the adjusted responses and selecting the most appropriate response; and means for providing the selected response to the user. This makes it possible to provide appropriate and considerate responses that take into account the user's emotions.

[0709] A "query" is information that represents a question or request that a user enters into a system.

[0710] A "keyword" is a phrase that indicates the main theme or topic extracted from a query.

[0711] A "language model" is an algorithm or data set used to process natural language and generate responses based on a specific area of ​​expertise.

[0712] "Emotional information" refers to information that indicates the polarity and characteristics of emotions included in user queries.

[0713] "Response adjustment" refers to the process of flexibly modifying the content of a generated response, taking into account the user's feelings.

[0714] "Response integration" is the process of combining multiple responses generated from different language models and selecting the single most appropriate response.

[0715] A "server" is a computing device or system that receives queries from users, performs the necessary processing, and generates responses.

[0716] This invention relates to an interactive system that generates appropriate responses while taking into account the user's emotions. This system primarily uses a terminal, a server, a language model, and an emotion engine.

[0717] Users use a terminal to enter queries into the system. These queries can include requests for information or expressions of opinion, and may contain emotional nuances. For example, consider a case where a user enters the query, "I feel angry about the recent news."

[0718] The terminal's role is to send the input query to the server. The server uses natural language processing libraries (e.g., spaCy or NLTK) to extract keywords in order to process the received query. It also uses sentiment engines (e.g., TextBlob or VADER) to analyze the sentiment information contained in the query and determine the polarity of the sentiment.

[0719] Based on this information, the server selects a model with the appropriate expertise from among multiple generative AI models. For example, for news-related queries, a model specializing in news will be selected. This process can utilize general generative AI technologies such as OpenAI.

[0720] The selected model leverages the query content and relevant expertise to generate an appropriate response. This response is refined within the server, and may be modified to take into account the emotional information provided by the user. For example, a query indicating negative emotions may be adjusted to include more reassuring and polite language in the response.

[0721] Finally, the server integrates the generated responses and selects the most appropriate one. This selection process takes into account the accuracy, relevance, and emotional considerations of the information. The selected response is sent to the terminal and displayed to the user.

[0722] This system configuration allows users to receive responses that take their emotions into account, resulting in a higher user experience. A concrete example of a prompt would be, "How should you respond if the user is angry about recent news?"

[0723] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0724] Step 1:

[0725] The user enters a query using a terminal. The entered query includes the user's information requests and emotions. For example, they might enter, "I feel angry about recent news." The terminal sends this query to the server.

[0726] Step 2:

[0727] The server parses the queries received from the terminal. First, it extracts keywords from the input query using a natural language processing library (e.g., spaCy or NLTK). This process identifies the theme or topic of the query. The extracted keywords are then output.

[0728] Step 3:

[0729] The server uses a sentiment engine (e.g., TextBlob or VADER) to analyze the sentiment contained in the query. This analysis determines the sentiment polarity (positive, negative, or neutral) of the query. The sentiment polarity information is obtained as output.

[0730] Step 4:

[0731] The server selects the most relevant generative AI model from multiple models based on the extracted keywords and sentiment polarity. For example, if keywords related to news are included, it will select a generative AI model specialized in news. This results in the output of a highly relevant generative AI model.

[0732] Step 5:

[0733] The server generates responses to queries using the selected generative AI model. This process leverages the model's expertise to produce highly accurate text output. The general content of the generated response is largely determined by the model.

[0734] Step 6:

[0735] The server adjusts the generated response based on the emotional information obtained by the emotion engine. If a negative emotion is detected, the response is modified to be more polite and considerate. The output is an emotionally conscious, adjusted response.

[0736] Step 7:

[0737] The server integrates multiple pre-tuned responses and selects the most appropriate one. This selection takes into account the accuracy, relevance, and emotional relevance of the response. The single best response is chosen as the output.

[0738] Step 8:

[0739] The server sends the selected response to the terminal. The terminal displays this response to the user, providing an appropriate answer that takes emotions into consideration. Through this process, the user receives an answer that reflects their emotions.

[0740] (Application Example 2)

[0741] 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".

[0742] In recent years, content delivery services have seen a growing demand for personalized content recommendations based on user emotions to improve the user experience. However, conventional technologies struggle to accurately analyze user emotions and provide appropriate responses and content immediately. This is a reason why user expectations are not being met, impacting service usage frequency and satisfaction.

[0743] 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.

[0744] In this invention, the server includes means for analyzing a query received from a user and extracting keywords and emotional states within the query; means for selecting a natural language processing model with the relevant expertise from among a plurality of natural language processing models based on the extracted keywords and emotional states; and means for generating a response to the query using the selected natural language processing model and adjusting the response content according to the user's emotional state. This enables the recommendation of personalized content that takes the user's emotions into consideration.

[0745] A "query" is information or a question that a user gives to a system, and it serves as the starting point for the system to analyze it and generate a response.

[0746] "Keywords" are important words or phrases extracted from a query, and they form the basis for the system to generate information and responses.

[0747] "Emotional state" refers to the psychological and emotional state of a user, analyzed from their queries and their actions before and after those queries, and is used when adjusting the content of the response.

[0748] A "natural language processing model" is an algorithmic method or system for understanding human language and generating responses, and it specializes in providing responses in various fields of expertise.

[0749] A "specialized natural language processing model" refers to a natural language processing model that specializes in processing information in a specific knowledge domain or topic, and has the ability to generate appropriate responses to relevant queries.

[0750] "Adjusting response content" refers to the process of modifying the tone and content of responses generated based on the user's emotional state, and delivering them in a way that is optimal for the user.

[0751] "Personalized content" refers to content that is individually generated and recommended based on a user's emotional state and past behavior, with the aim of improving the user experience.

[0752] The system for implementing this invention begins with a query input from the user via a terminal. The query received by the terminal from the user is transmitted to a server via a communication network. The server first analyzes the query and has the function of extracting keywords and sentiment states. Natural language processing technology is used for this, and software such as Python and TensorFlow is applied. The analysis of sentiment states is performed based on the text data contained in the user's query.

[0753] The server then selects the most suitable model from among several natural language processing models based on the extracted keywords and sentiment states. This process utilizes natural language processing models specialized in various fields, enabling sophisticated response generation. Generative AI models such as GPT are used in this process.

[0754] Using the selected model, the server generates a response to the query and further adjusts the response content considering the user's emotional state. This creates a thoughtful and appropriate response that takes the user's emotions into consideration. As a result, the responses and content provided by the system are personalized and tailored to the user's preferences and emotions.

[0755] Finally, the integrated response is returned to the terminal and presented to the user. This allows users to receive content in a way that takes their emotional state into consideration, improving the quality of the service and the user experience. A specific example of its use is when a user reacts to "anxious news": the server recommends reassuring content to support and improve the user's emotional state.

[0756] Example of a prompt:

[0757] "When users are feeling anxious, create content that provides reassurance. For example, think about lighthearted news that can refresh their mood or calming videos."

[0758] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0759] Step 1:

[0760] The user enters a query into the terminal. The entered query is sent from the terminal to the server in natural language text format. The input at this stage is the user's text data, and no specific information processing or analysis is performed.

[0761] Step 2:

[0762] The server parses the received query. The input is text data sent by the user. The server runs a natural language processing algorithm using Python to perform data analysis to extract keywords and sentiment states from the query. The output is the parsed keyword and sentiment state information.

[0763] Step 3:

[0764] The server selects an appropriate natural language processing model based on the extracted keywords and sentiment states. The input is the keywords and sentiment states obtained from step 2, and a comparison operation is performed to select a model with the corresponding area of ​​expertise. The output is the identification information of the selected model.

[0765] Step 4:

[0766] The server generates responses to queries using a selected natural language processing model. The input consists of the selected model and the user's query, and the generating AI model works to create an appropriate response. During this process, the response content is also adjusted according to the emotional state. The output is the adjusted response data.

[0767] Step 5:

[0768] When multiple responses are generated, the server integrates them and selects the most appropriate response. The input consists of multiple adjusted responses, and the server performs an optimization evaluation based on content, relevance, etc., to select the response deemed most appropriate. The output is the selected optimal response.

[0769] Step 6:

[0770] The selected response is sent to the terminal and displayed to the user. The input is the optimal response received from step 5, and the output is the response information presented to the user. This allows the user to receive a personalized response and obtain emotionally sensitive information.

[0771] 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.

[0772] 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.

[0773] 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.

[0774] 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.

[0775] 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.

[0776] 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.

[0777] 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.

[0778] 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.

[0779] 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."

[0780] 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.

[0781] 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.

[0782] 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.

[0783] 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.

[0784] 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.

[0785] 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.

[0786] 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.

[0787] 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.

[0788] 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.

[0789] 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.

[0790] 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.

[0791] 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.

[0792] The following is further disclosed regarding the embodiments described above.

[0793] (Claim 1)

[0794] A means of analyzing queries received from users and extracting keywords within those queries,

[0795] A means for selecting a language model with the relevant specialization from among multiple language models based on the extracted keywords,

[0796] A means for generating a response to the query using a selected language model,

[0797] A means for integrating responses obtained from multiple language models and selecting the most appropriate response,

[0798] Means for providing the selected response to the user,

[0799] A system that includes this.

[0800] (Claim 2)

[0801] The system according to claim 1, characterized in that the multiple language models each have parameters specialized for different fields of expertise.

[0802] (Claim 3)

[0803] The system according to claim 1, characterized in that it utilizes natural language processing techniques to understand the user's intent during query analysis.

[0804] "Example 1"

[0805] (Claim 1)

[0806] A means for analyzing user queries received by an information terminal and extracting important words within those queries,

[0807] A means for selecting a natural language model with a predetermined knowledge domain from among multiple natural language models based on the extracted important words,

[0808] A means for generating a response to the query using a selected natural language model,

[0809] A means for integrating responses obtained from multiple natural language models and determining the most relevant response,

[0810] Means for sending the determined response to the user,

[0811] An information processing system that includes this.

[0812] (Claim 2)

[0813] The information processing system according to claim 1, characterized in that each of the multiple natural language models has numerical data specialized in a different knowledge domain.

[0814] (Claim 3)

[0815] The information processing system according to claim 1, characterized in that it uses natural language processing technology to understand the user's intent during query analysis.

[0816] "Application Example 1"

[0817] (Claim 1)

[0818] A device that analyzes queries received from users and extracts keywords from those queries,

[0819] A device that selects a natural language processing model with the relevant expertise from among multiple natural language processing models based on the extracted keywords,

[0820] A device that generates a response to the query using a selected natural language processing model,

[0821] A device that integrates responses obtained from multiple natural language processing models and selects the most appropriate response,

[0822] A device that provides the selected response to the user,

[0823] A device that utilizes responses generated by a natural language processing model for providing product information,

[0824] A system that includes this.

[0825] (Claim 2)

[0826] The system according to claim 1, characterized in that the multiple natural language processing models each have parameters specialized for different fields.

[0827] (Claim 3)

[0828] The system according to claim 1, characterized in that it utilizes natural language understanding technology to understand the user's intent during query analysis.

[0829] "Example 2 of combining an emotion engine"

[0830] (Claim 1)

[0831] A means of analyzing queries received from users and extracting keywords within those queries,

[0832] A means for selecting a language model with the relevant specialization from among multiple language models based on the extracted keywords,

[0833] A means for generating a response to the query using a selected language model,

[0834] A means for adjusting the generated response based on the user's emotional information,

[0835] A means for integrating the adjusted responses and selecting the most appropriate response,

[0836] Means for providing the selected response to the user,

[0837] A system that includes this.

[0838] (Claim 2)

[0839] The system according to claim 1, characterized in that the multiple language models each have parameters specialized in different fields and perform adjustments based on the emotional information.

[0840] (Claim 3)

[0841] The system according to claim 1, characterized in that it utilizes natural language processing technology and emotion detection technology to understand the user's intent and emotions during query analysis.

[0842] "Application example 2 when combining with an emotional engine"

[0843] (Claim 1)

[0844] A means for analyzing queries received from users and extracting keywords and emotional states within those queries,

[0845] A means for selecting a natural language processing model with expertise in the relevant field from among multiple natural language processing models based on the extracted keywords and emotional states,

[0846] A means for generating a response to the query using a selected natural language processing model and adjusting the response content according to the user's emotional state,

[0847] A means for integrating responses obtained from multiple natural language processing models and selecting the most appropriate response,

[0848] A means of providing the user with the selected response and recommending personalized content that takes the user's emotions into consideration,

[0849] A system that includes this.

[0850] (Claim 2)

[0851] The system according to claim 1, characterized in that the multiple natural language processing models each have parameters specialized for different fields.

[0852] (Claim 3)

[0853] The system according to claim 1, characterized in that, during query analysis, it utilizes natural language processing techniques to understand the user's intent and adjust the response based on their emotional state. [Explanation of Symbols]

[0854] 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 means of analyzing queries received from users and extracting keywords within those queries, A means for selecting a language model with the relevant specialization from among multiple language models based on the extracted keywords, A means for generating a response to the query using a selected language model, A means for integrating responses obtained from multiple language models and selecting the most appropriate response, Means for providing the selected response to the user, A system that includes this.

2. The system according to claim 1, characterized in that the plurality of language models each have parameters specialized for different fields of expertise.

3. The system according to claim 1, characterized in that it utilizes natural language processing technology to understand the user's intent during query analysis.

Images