System

Abstract

A system is provided.SOLUTION: A system comprising means for a user to input a question from a terminal, server means for receiving and analyzing the question transmitted from the terminal, means for transmitting the question received from the server means to a plurality of artificial intelligence models and acquiring an answer from each artificial intelligence model, means for evaluating and scoring the answer received from each of the artificial intelligence models, means for generating feedback for each artificial intelligence model based on the evaluation result, means for calculating an average score from the scores of the artificial intelligence models and determining the most reliable answer, and means for transmitting the determined answer to the terminal and presenting it to the user.SELECTED DRAWING: Figure 1

Description

[Technical Field]

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

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

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

[0004] Current generative AI models suffer from the problem that answers from a single model may lack accuracy and reliability. In particular, they may provide incorrect information due to factors such as an inability to fully understand the question, a lack of training data, or answers based on outdated data. This can lead to misunderstandings and a loss of trust among users. There is a need to solve these problems and improve the quality and reliability of answers. [Means for solving the problem]

[0005] In order to solve the above problems, the present invention provides the following means: A server means is provided for allowing a user to input a question from a terminal, and receiving and analyzing the question sent from the terminal. The server means includes means for sending the received question to a plurality of artificial intelligence models and obtaining an answer from each of the artificial intelligence models. The server means further includes means for evaluating and scoring the answers received from each artificial intelligence model, means for generating feedback for each artificial intelligence model based on the evaluation results, and means for calculating an average score from the scores of each artificial intelligence model to determine the most reliable answer. Finally, a system is provided that includes means for sending the determined answer to the terminal and presenting it to the user.

[0006] This system makes it possible to comprehensively evaluate the answers of multiple AI models and provide the most reliable answer to the user, thereby improving the quality and reliability of the answers of the generative AI model and providing accurate and reliable information to the user.

[0007] A "user" is an entity that uses the system to enter questions and receive answers.

[0008] A "terminal" is a device through which a user inputs questions and communicates with the server. Specifically, this applies to a PC, smartphone, etc.

[0009] A "server" is a device or system that receives questions from a terminal, analyzes them, and sends the questions to multiple artificial intelligence models to obtain answers.

[0010] An "artificial intelligence model" is a program or algorithm that is trained to generate answers to specific questions.

[0011] An "answer" is information that is generated by an artificial intelligence model in response to a question entered by a user and sent back to the server.

[0012] "Evaluation" is the act or process of quantifying and determining the accuracy, reliability, and currency of answers from an artificial intelligence model.

[0013] "Scoring" is the act of assigning a numerical score to each AI model's response based on evaluation criteria.

[0014] "Feedback" is information provided to an artificial intelligence model for improvement based on the evaluation results.

[0015] The "average score" is a numerical value obtained by averaging the scores obtained from multiple AI models and is used to determine the most reliable answer.

[0016] The "most reliable answer" is the answer obtained from multiple artificial intelligence models that is determined to have the highest reliability based on evaluation and scoring.

[0017] "System" refers to the technology, based on the present invention, consisting of the entire components and processes by which a user inputs a question, and a server receives, analyzes, and evaluates the question to provide the most reliable answer. [Brief explanation of the drawings]

[0018] [Figure 1] 1 is a conceptual diagram showing an example of the configuration of a data processing system according to a first embodiment. [Figure 2] 1 is a conceptual diagram showing an example of main functions of a data processing device and a smart device according to a first embodiment. [Figure 3] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a second embodiment. [Figure 4] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and smart glasses according to a second embodiment. [Figure 5] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a third embodiment. [Figure 6]FIG. 11 is a conceptual diagram showing an example of main functions of a data processing device and a headset-type terminal according to a third embodiment. [Figure 7] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a fourth embodiment. [Figure 8] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and a robot according to a fourth embodiment. [Figure 9] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 10] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 11] FIG. 3 is a sequence diagram showing a processing flow of the data processing system according to the first embodiment. [Figure 12] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 1. [Figure 13] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system according to the second embodiment when an emotion engine is combined. [Figure 14] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 2 when an emotion engine is combined. DETAILED DESCRIPTION OF THE INVENTION

[0019] An example of an embodiment of a system according to the technology of the present disclosure will be described below with reference to the accompanying drawings.

[0020] First, the terms used in the following description will be explained.

[0021] In the following embodiments, a coded processor (hereinafter simply referred to as a "processor") may be a single arithmetic device or a combination of multiple arithmetic devices. Furthermore, a processor may be a single type of arithmetic device or a combination of multiple types of arithmetic devices. Examples of arithmetic devices include a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), a GPGPU (General-Purpose computing on Graphics Processing Units), and an APU (Accelerated Processing Unit).

[0022] In the following embodiments, a coded RAM (Random Access Memory) is a memory in which information is temporarily stored and is used as a working memory by a processor.

[0023] In the following embodiments, the coded storage is one or more non-volatile storage devices that store various programs, various parameters, etc. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), magnetic disks (e.g., hard disks), and magnetic tapes.

[0024] In the following embodiments, a communication I / F (Interface) with a symbol is an interface including a communication processor, an antenna, etc. The communication I / F controls communication between multiple computers. Examples of communication standards applied to the communication I / F include wireless communication standards including 5G (5th Generation Mobile Communication System), Wi-Fi (registered trademark), Bluetooth (registered trademark), etc.

[0025] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B." In other words, "A and / or B" means that it may be only A, only B, or a combination of A and B. Furthermore, in this specification, the same concept as "A and / or B" is also applied when three or more things are expressed connected by "and / or."

[0026] [First embodiment]

[0027] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0028] 1, a data processing system 10 includes a data processing device 12 and a smart device 14. An example of the data processing device 12 is a server.

[0029] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0030] The smart device 14 includes a computer 36, a reception device 38, an output device 40, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The reception device 38, the output device 40, and the camera 42 are also connected to the bus 52.

[0031] The reception device 38 includes a touch panel 38A, a microphone 38B, and the like, and receives user input. The touch panel 38A detects contact with an indicator (for example, a pen or a finger) to receive user input by the touch of the indicator. The microphone 38B detects the user's voice to receive user input by voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the data indicating the user input.

[0032] The output device 40 includes a display 40A and a speaker 40B, and presents data to the user 20 by outputting the data in a form of expression that the user 20 can perceive (for example, audio and / or text). The display 40A displays visible information such as text and images in accordance with instructions from the processor 46. The speaker 40B outputs audio in accordance with instructions from the processor 46. The camera 42 is a compact digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0033] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 control the exchange of various information between the processor 46 and the processor 28 via the network 54.

[0034] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0035] 2, in the data processing device 12, a specific process is performed by the processor 28. A specific processing program 56 is stored in the storage 32. The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific process is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0036] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0037] In the smart device 14, the processor 46 performs the reception output process. The storage 50 stores a reception output program 60. The reception output program 60 is used in conjunction with the specific processing program 56 by the data processing system 10. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0038] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0039] The system according to the present invention uses multiple artificial intelligence models to generate answers to questions entered by a user, evaluates the answers, provides feedback, and provides the most reliable answer to the user. An embodiment of this system will be described in detail below.

[0040] User Input

[0041] A user enters a question into the device's interface. For example, the user might enter the question "How old is the universe?" When the user presses the submit button, the device sends the question data to the server. This transmission is typically done using an HTTP request, with the question data packaged in JSON format.

[0042] Receiving and analyzing

[0043] The server receives the question sent from the device and analyzes the question. It checks the format of the received data, confirms that it is in JSON format, and then parses it to extract the question. The extracted question is used in the next process.

[0044] Ask a question and get an answer

[0045] The server sends the received questions to multiple artificial intelligence models. For example, it sends a question to model A and model B via API and obtains answers from each model. At this time, the server uses the appropriate API endpoint to send to each model.

[0046] Evaluating and scoring answers

[0047] The server evaluates the answers received from each AI model. Evaluation criteria include accuracy, reliability, and recency. These criteria are applied to each answer, and a score is generated. For example, model A's answer "13.7 billion years" is given a score of 99% accuracy and 95% reliability, while model B's answer "13.8 billion years" is given a score of 98% accuracy and 90% reliability.

[0048] Generate feedback

[0049] The server generates feedback for each AI model based on the evaluation results. The feedback indicates specific areas for improvement for each model. For example, for Model A, the server generates feedback such as "The answer is accurate, but further improvement in reliability is needed."

[0050] Calculating the average score and determining the final answer

[0051] The server calculates the average score from the scores of each AI model and determines the most reliable answer. For example, if the answers are "13.7 billion years" and "13.8 billion years," the server calculates the median and average scores and determines the final answer as "13.75 billion years."

[0052] Presenting the results to the user

[0053] The server sends the final answer to the user's device. The device receives the result from the server and displays it to the user. The user can check the presented result and obtain the necessary information. For example, the result will be displayed to the user as "The age of the universe is approximately 13.75 billion years."

[0054] This system allows users to comprehensively evaluate the answers of multiple AI models to obtain the most reliable information, thereby improving the quality and reliability of the answers of the generative AI model and providing users with accurate and reliable information.

[0055] The processing flow will be explained below.

[0056] Step 1:

[0057] A user types a question into a terminal interface. For example, the user types the question "How old is the universe?"

[0058] Step 2:

[0059] When the user presses the submit button, the device sends the question data to the server, typically using an HTTP request, with the question data packaged in JSON format.

[0060] Step 3:

[0061] The server receives the question sent from the device, analyzes the question, checks the format of the received data, and parses the JSON data to extract the question.

[0062] Step 4:

[0063] The server sends the received questions to multiple AI models. For example, it sends a question to model A and model B via API and obtains answers from each model.

[0064] Step 5:

[0065] Each AI model generates an answer to the question and returns it to the server. For example, model A answers "13.7 billion years," and model B answers "13.8 billion years." The server receives these answers.

[0066] Step 6:

[0067] The server evaluates the answers received from each AI model, using criteria such as accuracy, reliability, and recency. Each answer is scored based on these criteria.

[0068] Step 7:

[0069] The server generates feedback for each model based on the scoring results, including specific suggestions for improvement, such as "data needs updating."

[0070] Step 8:

[0071] The server calculates the average score from each AI model's score and determines the most reliable answer. For example, it uses the median of "13.7 billion years" and "13.8 billion years" and determines the median of 13.75 billion years as the final answer.

[0072] Step 9:

[0073] The server then sends the final answer to the user's device, also using an HTTP request.

[0074] Step 10:

[0075] The device displays the final answer received from the server to the user, who then confirms the information displayed on the device screen. For example, the information might read, "The age of the universe is approximately 13.75 billion years."

[0076] Example 1

[0077] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0078] Conventional information retrieval systems rely on a single AI model to generate answers, which limits the accuracy and reliability of the answers. Furthermore, there is insufficient assurance of the quality of the information users receive, increasing the risk of providing incorrect information. This makes it difficult to quickly provide the most reliable information in response to user questions.

[0079] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0080] In this invention, the server includes a device for a user to input a question from a terminal, an information processing device that receives and analyzes the question sent from the terminal, a device that sends the question received by the information processing device to a plurality of artificial intelligence systems and obtains answers from each of the artificial intelligence systems, a device that evaluates and quantifies the answers received from each of the artificial intelligence systems, a device that generates feedback for each artificial intelligence system based on the evaluation results, a device that calculates an average value from the numerical values ​​of the artificial intelligence systems and determines the most reliable answer, and a device that sends the determined answer to the terminal and presents it to the user. This makes it possible to comprehensively evaluate the answers of a plurality of artificial intelligence models and quickly provide accurate and reliable information to the user.

[0081] "Terminal" refers to an electronic device through which a user can enter questions and view results.

[0082] The "information processing device" refers to a computer system that receives a question sent from a terminal and analyzes its content.

[0083] "Artificial intelligence system" refers to a collection of computational models or algorithms designed to generate answers to specific questions.

[0084] "Device" refers to a combination of hardware and software for achieving a particular function.

[0085] "Question" refers to text or other form of data that allows a user to enter information they would like to know.

[0086] "Analysis" refers to the process of breaking down received data and understanding its meaning.

[0087] "Answer" refers to the information generated by an artificial intelligence system in response to a question.

[0088] "Evaluation" refers to the process for determining the quality of answers received from an artificial intelligence system.

[0089] "Quantification" refers to expressing the evaluation results as quantitative values.

[0090] "Feedback" refers to information used to provide improvements and evaluation results to an artificial intelligence system.

[0091] The "average value" refers to the value obtained by dividing the sum of multiple numerical data by the number of data items.

[0092] A "trusted answer" refers to an answer that is determined to be the most trustworthy based on the evaluation criteria.

[0093] "Presenting" refers to displaying information to the user on a screen or the like.

[0094] The system according to the present invention uses multiple artificial intelligence models to generate answers to questions entered by a user, evaluates the answers, provides feedback, and provides the most reliable answers to the user. An embodiment of this system is described in detail below.

[0095] Hardware and Software Configuration

[0096] This system consists of a terminal used by the user, a server that receives and analyzes questions, multiple artificial intelligence models, and an evaluation system that evaluates and provides feedback on each answer.

[0097] Terminal

[0098] Users input questions using a device (e.g., smartphone, PC, tablet). The device uses a web browser or dedicated application as an interface to input, submit, and display the results of the user's question.

[0099] server

[0100] The server receives and analyzes questions sent by users. The server typically runs on a Linux (registered trademark) server or a cloud service (e.g., AWS (registered trademark) or Google (registered trademark) Cloud). The received question data is in JSON format, and the content of the question is analyzed and extracted.

[0101] Artificial Intelligence Model

[0102] The server sends the parsed questions to multiple artificial intelligence models (e.g., GPT-4®, BERT), which are connected through API endpoints and each model independently generates an answer based on the questions posed by the server.

[0103] Rating System

[0104] The server evaluates and quantifies the answers obtained from each AI model. The evaluation criteria are accuracy, reliability, and recency. Based on the evaluation results, feedback is generated for each AI model.

[0105] Specific examples

[0106] For example, if a user inputs the question "What is the radius of the Earth?", this question is processed as follows:

[0107] 1. The user types "What is the radius of the Earth?" into the terminal interface and presses the send button.

[0108] 2. The device sends the question data to the server in JSON format.

[0109] 3. The server receives the question data and parses and extracts the question content using a JSON parser.

[0110] 4. The server sends the extracted questions to multiple AI models. For example, it sends questions to Model A and Model B via API.

[0111] 5. Model A generates the answer "approximately 6371 km" and Model B generates the answer "approximately 6370 km".

[0112] 6. The server evaluates each model's answer and assigns Model A an accuracy of 98% and a confidence of 97%, and Model B an accuracy of 97% and a confidence of 95%.

[0113] 7. Based on the evaluation results, the server generates feedback for Model A saying, "The answer is accurate, but you need to further improve its reliability," and for Model B saying, "The accuracy is high, but you need to strengthen its reliability."

[0114] 8. The server calculates the average value of each model and determines that the radius of the Earth is approximately 6,370.5 km.

[0115] 9. The final answer is sent to the user's device and the user is shown "The radius of the Earth is approximately 6370.5 km."

[0116] Prompt Sentence Examples

[0117] An example of a prompt that a user might enter into a terminal is as follows:

[0118] User: What is the radius of the Earth?

[0119] This system allows users to comprehensively evaluate the answers of multiple AI models to obtain the most reliable information, improving the quality and reliability of the answers of the generative AI model and providing users with accurate and reliable information.

[0120] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0121] System program processing flow

[0122] Step 1: User Input

[0123] Contents:

[0124] The user enters a question in the terminal interface.

[0125] Specific behavior:

[0126] The user enters "What is the radius of the Earth?" in the text box and presses the submit button.

[0127] Input and Output:

[0128] Input: User's question (e.g., "What is the radius of the Earth?")

[0129] Data processing: Convert user input into JSON format

[0130] Output: Question data in JSON format

[0131] Step 2: Submit your question

[0132] Contents:

[0133] After the send button is pressed, the terminal transmits the user's question to the server.

[0134] Specific behavior:

[0135] The terminal processes the user's input and sends an HTTP request (POST) to the server, which contains the question data in JSON format.

[0136] Input and Output:

[0137] Input: Question data in JSON format

[0138] Data processing: None

[0139] Output: HTTP request sent (including question data in JSON format)

[0140] Step 3: Receiving and parsing the question

[0141] Contents:

[0142] The server receives the question sent from the terminal and analyzes its contents.

[0143] Specific behavior:

[0144] After receiving the HTTP request, the server extracts the question data in JSON format from the request, parses it, and extracts the question content.

[0145] Input and Output:

[0146] Input: HTTP request (including question data in JSON format)

[0147] Data processing: Extracting question content by parsing JSON

[0148] Output: Parsed question (e.g., "What is the radius of the Earth?")

[0149] Step 4: Sending questions to the AI ​​model

[0150] Contents:

[0151] The server sends the analyzed questions to multiple artificial intelligence models.

[0152] Specific behavior:

[0153] The server sends a query to a prepared API endpoint, for example, an HTTP POST request to model A and model B.

[0154] Input and Output:

[0155] Input: Parsed question

[0156] Data processing: Converting questions into a request format for each model

[0157] Output: Multiple HTTP requests (one for model A, one for model B)

[0158] Step 5: Getting the answer from the model

[0159] Contents:

[0160] The server receives responses from each artificial intelligence model.

[0161] Specific behavior:

[0162] Receive a response from the artificial intelligence model and extract the answer data.

[0163] Input and Output:

[0164] Input: HTTP response from each model

[0165] Data processing: Extract answer data from responses

[0166] Output: Extracted response data (e.g., Model A "approximately 6371 km", Model B "approximately 6370 km")

[0167] Step 6: Evaluate and quantify responses

[0168] Contents:

[0169] The server evaluates and quantifies the answers from each artificial intelligence model.

[0170] Specific behavior:

[0171] The server rates each answer on the criteria of accuracy, reliability, and currency, and assigns a numerical score for each criterion.

[0172] Input and Output:

[0173] Input: Response data from each model

[0174] Data processing: Assigning a numerical score based on evaluation criteria

[0175] Output: Evaluated numerical score (e.g., Model A: Accuracy 98%, Confidence 97%)

[0176] Step 7: Generate feedback

[0177] Contents:

[0178] The server generates feedback for each artificial intelligence model based on the evaluation results.

[0179] Specific behavior:

[0180] The evaluation results are analyzed and feedback messages are generated that specifically indicate areas for improvement for each model.

[0181] Input and Output:

[0182] Input: Evaluation result

[0183] Data processing: feedback message generation

[0184] Output: Feedback message (e.g., Model A "Reliability needs to be improved")

[0185] Step 8: Calculate the average and determine the final answer

[0186] Contents:

[0187] The server calculates an average score from each AI model and determines the most reliable answer.

[0188] Specific behavior:

[0189] The evaluation scores of each model are averaged to select the most reliable answer.

[0190] Input and Output:

[0191] Input: Rating score

[0192] Data processing: Calculating the average and determining the final answer

[0193] Output: Final answer (e.g. "The radius of the Earth is approximately 6370.5 km")

[0194] Step 9: Present the results to the user

[0195] Contents:

[0196] The server sends the determined final answer to the user's terminal, and the terminal displays the result to the user.

[0197] Specific behavior:

[0198] The server generates the final answer in JSON format and sends it as an HTTP response to the terminal, which receives this data and displays the result on the screen.

[0199] Input and Output:

[0200] Input: Final answer

[0201] Data processing: Converting the final answer to JSON and generating an HTTP response

[0202] Output: The result displayed on the user's screen (e.g., "The radius of the Earth is approximately 6370.5 km.")

[0203] (Application example 1)

[0204] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0205] In modern food delivery applications, when users ask questions about food or restaurants, it is difficult for them to determine whether the answers are reliable. Furthermore, answers provided by a single AI model often have limitations in accuracy and reliability. Therefore, there is a need for a system that uses multiple AI models to provide more reliable information.

[0206] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0207] In this invention, the server includes: means for a user to input a question from a terminal; information processing means for receiving and analyzing the question sent from the terminal; means for sending the question received by the information processing means to a plurality of artificial intelligence models and obtaining answers from each of the artificial intelligence models; means for evaluating and scoring the answers received from each of the artificial intelligence models; means for generating feedback for each artificial intelligence model based on the evaluation results; means for calculating an average score from the scores of the artificial intelligence models and determining the most reliable answer; means for sending the determined answer to the terminal and presenting it to the user; and means for adding a function to a food delivery application that generates answers to questions from users using a plurality of artificial intelligence models and provides the most reliable answer, thereby enabling users to obtain reliable answers.

[0208] A "terminal" is a computing device through which a user can enter questions and send and receive information.

[0209] The "information processing device means" is a computer server for receiving and analyzing questions sent from the terminals.

[0210] An "artificial intelligence model" is a system that uses pre-trained machine learning algorithms to generate answers to user questions.

[0211] "Means for evaluating and scoring" means a method or device for evaluating and assigning a score to the answers received from each artificial intelligence model based on criteria of accuracy, reliability, and currency.

[0212] The "means for generating feedback" refers to a method or device for creating feedback that specifically indicates areas for improvement based on the evaluation results of each artificial intelligence model.

[0213] A "food delivery application" is a mobile application that allows users to search for information about food and restaurants and place orders.

[0214] A "reliable answer" is information that is judged to have the highest reliability based on the evaluation results of multiple artificial intelligence models.

[0215] A food delivery application system based on this invention begins with a user inputting a question from a terminal. For example, the question might be, "What is the most popular menu item at this restaurant?" The terminal then sends the question to a server. The server receives and analyzes the question and sends it to multiple artificial intelligence models. Each artificial intelligence model generates an answer, which is then received by the server. The received answers are evaluated and scored based on accuracy, reliability, and recency.

[0216] Hardware and software used

[0217] 1. Hardware:

[0218] Device: The smartphone used by the user.

[0219] Server: Use a cloud-based server system (e.g., AWS, Google Cloud).

[0220] 2. Software:

[0221] Food delivery app: A mobile application compatible with ANDROID (registered trademark) and iOS.

[0222] API endpoints: APIs for accessing multiple artificial intelligence models (e.g., GPT-3®, etc.).

[0223] Assessment and feedback generation system: A software module for analyzing responses, scoring, and generating feedback.

[0224] The server generates feedback for each AI model based on the evaluation results. The feedback includes specific areas for improvement for each model. For example, the server might generate feedback such as, "The answer is accurate, but you need to improve its reliability."

[0225] The server then calculates an average score from each AI model's score and determines the most reliable answer. This reliable answer is sent to the user's device and presented to the user through the application. The user can then review the presented results and obtain the necessary information.

[0226] For example, if a user asks, "How many calories are in salmon sushi?" the system will proceed as follows:

[0227] Example prompt sentence:

[0228] "How many calories are in salmon sushi?"

[0229] In response to this question, multiple AI models generate answers such as "Salmon sushi has 250 kcal" and "Salmon sushi has 260 kcal." The server evaluates these answers and provides the most reliable answer, "Salmon sushi has 255 kcal," to the user.

[0230] This allows the user to quickly and accurately obtain highly reliable information.

[0231] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0232] Step 1:

[0233] The user enters a question from the terminal and submits it. The entered question is converted to JSON format and sent to the server as an HTTP request. The input of this process is the question text entered by the user, and the output is the JSON-formatted question data sent to the server.

[0234] Step 2:

[0235] The server receives the question sent from the device and analyzes the question content. It checks the format of the received data, confirms that it is in JSON format, and then parses it to extract the question content. The input of this process is the question data in JSON format, and the output is the analyzed question text.

[0236] Step 3:

[0237] The server sends the parsed question to multiple AI models and obtains an answer from each model. The appropriate API endpoint is used to send the question to each model. The input of this process is the parsed question text, and the output is the answer text from each AI model.

[0238] Step 4:

[0239] The server evaluates and scores the answers received from each AI model. Evaluation criteria include accuracy, reliability, and recency, and each answer is scored. The input to this process is the answer text from each model, and the output is the evaluation result and its respective score.

[0240] Step 5:

[0241] The server generates feedback for each AI model based on the evaluation results. The feedback specifically indicates areas for improvement for each model. The input to this process is the evaluation results and scores, and the output is a feedback message.

[0242] Step 6:

[0243] The server calculates the average score from the scores of each AI model and determines the most reliable answer. The input of this process is the score data, and the output is the most reliable answer text.

[0244] Step 7:

[0245] The server sends the determined most authoritative answer to the terminal and presents it to the user. The terminal receives the result from the server and displays it to the user. The input of this process is the most authoritative answer text, and the output is the answer message displayed on the user's terminal.

[0246] Furthermore, an emotion engine that estimates the user's emotion may be combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59 and perform identification processing using the user's emotion.

[0247] The system based on this invention uses multiple artificial intelligence models to generate answers to questions entered by a user, evaluates and provides feedback on the answers, and provides the user with the most reliable answer. Furthermore, by combining it with an emotion engine that recognizes the user's emotions, it is possible to provide a more appropriate answer that takes the user's emotions into consideration. An embodiment of this system will be described in detail below.

[0248] User Input and Emotion Recognition

[0249] A user inputs a question into the device's interface. For example, suppose the user inputs the question, "How old is the universe?" At this time, the device's emotion engine also recognizes the user's emotion. The emotion engine uses methods such as voice analysis, facial expression recognition, or text analysis to generate emotion data that the user indicates when inputting.

[0250] Submitting questions and emotion data

[0251] When the user presses the send button, the device sends the question data and emotion data to the server, typically using an HTTP request, with the question data and emotion data packaged in JSON format.

[0252] Receiving and analyzing

[0253] The server receives the question data and emotion data sent from the device and analyzes them. After checking the format of the received data and confirming that it is in JSON format, it parses it to extract the question content and emotion data.

[0254] Ask a question and get an answer

[0255] The server sends the received questions to multiple artificial intelligence models. For example, it sends a question to model A and model B via API and obtains answers from each model. At this time, the server uses the appropriate API endpoint to send to each model.

[0256] Evaluating and scoring answers

[0257] The server evaluates the answers received from each AI model. Evaluation criteria include accuracy, reliability, and recency. These criteria are applied to each answer, and a score is generated. Emotional data is also taken into account in the evaluation. For example, model A's answer "13.7 billion years" is given a score of 99% accuracy and 95% reliability, while model B's answer "13.8 billion years" is given a score of 98% accuracy and 90% reliability.

[0258] Generate feedback

[0259] The server generates feedback for each AI model based on the evaluation results. The feedback includes specific areas for improvement for each model, while also taking into account the user's emotional data. For example, if the user expresses anxiety, the server may generate feedback such as "your answer is accurate, but more information is needed."

[0260] Calculating the average score and determining the final answer

[0261] The server calculates the average score from each AI model and determines the most reliable answer. For example, the median of "13.7 billion years" and "13.8 billion years" is taken to determine 13.75 billion years. Taking into account emotional data, answers that are closest to the user's needs are also given more weight.

[0262] Presenting the results to the user

[0263] The server sends the final answer to the user's device. This transmission is also performed using an HTTP request. The device receives the result from the server and displays it to the user. The user can check the presented result and obtain the necessary information. For example, the result is displayed to the user as "The age of the universe is approximately 13.75 billion years."

[0264] This system allows users to comprehensively evaluate the answers of multiple AI models to obtain the most reliable information, while also providing optimal answers that take emotional data into account. This improves the quality and reliability of the generative AI model's answers, enabling the provision of accurate and reliable information to users.

[0265] The processing flow will be explained below.

[0266] System Embodiments

[0267] The system based on this invention uses multiple artificial intelligence models to generate answers to questions entered by the user, evaluates and provides feedback on the answers, and provides the user with the most reliable answer. Furthermore, by combining it with an emotion engine that recognizes the user's emotions, it is possible to provide a more appropriate answer that takes the user's emotions into consideration. This will be explained in detail in the following processing steps.

[0268] Step 1:

[0269] A user types a question into a terminal interface. For example, the user types the question "How old is the universe?"

[0270] Step 2:

[0271] When a user inputs a question, the device's emotion engine recognizes the user's emotion using voice analysis, facial expression recognition, text analysis, etc. For example, emotion data is generated from the user's tone of voice and facial expression.

[0272] Step 3:

[0273] When the user presses the send button, the device sends the question data and emotion data to the server, typically using an HTTP request, with the question data and emotion data packaged in JSON format.

[0274] Step 4:

[0275] The server receives the question data and emotion data sent from the device and analyzes them. After checking the format of the received data and confirming that it is in JSON format, it parses it to extract the question content and emotion data.

[0276] Step 5:

[0277] The server sends the received question to multiple artificial intelligence models. For example, it sends a question to model A and model B via API and obtains an answer from each model. The server uses the appropriate API endpoint to send to each model.

[0278] Step 6:

[0279] Each AI model generates an answer to the question and returns it to the server. For example, model A answers "13.7 billion years," and model B answers "13.8 billion years." The server receives these answers.

[0280] Step 7:

[0281] The server evaluates the answers received from each AI model. Evaluation criteria include accuracy, reliability, and recency. These criteria are applied to each answer, resulting in a score. For example, model A's answer "13.7 billion years" is given a score of 99% accuracy and 95% reliability, while model B's answer "13.8 billion years" is given a score of 98% accuracy and 90% reliability. Emotional data is also taken into account in this evaluation.

[0282] Step 8:

[0283] The server generates feedback for each model based on the scoring results. The feedback includes specific areas for improvement for each model, and also takes into account the user's emotional data. For example, if the user expresses anxiety, the server generates feedback such as "The answer is accurate, but further explanation is needed."

[0284] Step 9:

[0285] The server calculates the average score from each AI model's score and determines the most reliable answer. For example, the median of "13.7 billion years" and "13.8 billion years" is taken to determine 13.75 billion years. Taking into account the user's emotional data, the server selects an answer that is optimized for the user's emotional state.

[0286] Step 10:

[0287] The server then sends the final answer to the user's device, also using an HTTP request.

[0288] Step 11:

[0289] The device displays the final answer received from the server to the user, who then confirms the information displayed on the device screen. For example, the information might read, "The age of the universe is approximately 13.75 billion years."

[0290] This system allows users to comprehensively evaluate the answers of multiple AI models to obtain the most reliable information, while also providing optimal answers that take emotional data into account. This improves the quality and reliability of the generative AI model's answers, enabling the provision of accurate and reliable information to users.

[0291] Example 2

[0292] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0293] Currently, when a user inputs a question into an AI, they must compare and consider the answers of multiple AI models to obtain a reliable answer. However, doing this individually is technically time-consuming and ultimately inconvenient for users. Furthermore, if an answer is provided without taking the user's feelings into consideration, user satisfaction decreases. Furthermore, there is a lack of functionality to provide feedback on the obtained answers and use it to improve the model, which also poses the problem of delays in improving the model itself.

[0294] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0295] In this invention, the server includes: means for a user to input a question from a terminal; means for generating user emotion data using an emotion recognition engine of the terminal; means for receiving and analyzing the question and emotion data sent from the terminal; means for transmitting the question received by the server means to multiple artificial intelligence models and obtaining answers from each of the artificial intelligence models; means for evaluating and scoring the answers and emotion data received from each of the artificial intelligence models; means for generating feedback for each artificial intelligence model based on the evaluation results; means for calculating an average score from the scores of the artificial intelligence models and determining the most reliable answer; and means for transmitting the determined answer to the terminal and presenting it to the user. This allows the user to obtain a reliable answer and provides appropriate feedback according to the user's emotion, thereby improving satisfaction. Furthermore, since the artificial intelligence models are rapidly improved through feedback, the performance of the entire system can also be improved.

[0296] A "terminal" is a device where a user inputs a question and generates emotion data.

[0297] An "emotion recognition engine" is a technology installed in a device that analyzes a user's voice, facial expressions, text, etc. to generate emotional data.

[0298] "Emotion data" is information generated by an emotion recognition engine that expresses the emotions felt by the user when entering a question.

[0299] A "server" is a computer system that receives and analyzes question data and emotion data sent from a terminal.

[0300] An "artificial intelligence model" is a set of machine learning algorithms and / or deep learning models used to generate answers to specific questions.

[0301] An "API endpoint" is an interface for sending questions to an artificial intelligence model and obtaining answers.

[0302] "Scoring" is the process of assigning a score to the answer obtained from an artificial intelligence model based on evaluation criteria.

[0303] "Feedback" is information that provides improvements and additional information based on the evaluation results of the AI ​​model's answers.

[0304] The "average score" is the average score of the answers from multiple artificial intelligence models, and is the value used to determine the most reliable answer.

[0305] An "HTTP request" is one of the Internet protocols used to send and receive data.

[0306] The "JSON format" is a format for structuring and storing data, and is used to package question data and sentiment data.

[0307] The system based on this invention uses multiple artificial intelligence models to generate answers to questions entered by users, evaluates and provides feedback on the answers, and provides the most reliable answers to the users. Furthermore, by combining it with an emotion engine that recognizes the user's emotions, it is possible to provide more appropriate answers that take the user's emotions into consideration.

[0308] User Input and Emotion Recognition

[0309] A user inputs a question into the device's interface. For example, when a user inputs a question such as "How old is the universe?", the device's emotion recognition engine recognizes the user's emotions in real time. The emotion recognition engine generates emotion data using voice analysis, facial expression recognition, or text analysis technologies. Specifically, it captures the user's tone of voice and facial expressions and analyzes them to determine emotions such as "surprise" or "interest."

[0310] Submitting questions and emotion data

[0311] When the user presses the send button, the device sends the question data and emotion data to the server. This transmission is made via an HTTP request, and the data is packaged in JSON format. For example, the question data might include "How old is the universe?", and the emotion data might include "surprise."

[0312] Receiving and analyzing

[0313] The server receives the question data and emotion data sent from the device and analyzes them. First, it checks that the received data is in JSON format, then parses the data using a JSON library. This extracts the question content and emotion data, which are then stored in internal variables. For example, the question data is stored in the "question" variable, and the emotion data is stored in the "emotion" variable.

[0314] Ask a question and get an answer

[0315] The server sends the received question to multiple artificial intelligence models via API. For example, the server calls the API endpoints of Model A and Model B and sends the question to each model. This returns an answer from each model. For example, Model A might return the answer "13.7 billion years," and Model B might return the answer "13.8 billion years."

[0316] Evaluating and scoring answers

[0317] The server evaluates the answers received from each model, using criteria such as accuracy, reliability, and recency. For example, model A's answer "13.7 billion years" is given a score of 99% accuracy and 95% reliability. On the other hand, model B's answer "13.8 billion years" is given a score of 98% accuracy and 90% reliability. Emotional data is also taken into account in the evaluation.

[0318] Generate feedback

[0319] The server generates feedback for each model based on the evaluation results. The feedback includes suggestions for improvement and additional information for each model, and also takes into account emotional data. For example, if the user expresses surprise, the server generates feedback such as "The answer is accurate, but it would be better if you provided more detailed information."

[0320] Calculating the average score and determining the final answer

[0321] The server calculates the average score from each model's score and determines the most reliable answer. Specifically, it takes the median value of 13.7 billion years for Model A and 13.8 billion years for Model B as its answer, resulting in 13.75 billion years. It also takes into account sentiment data to provide the answer that best suits the user's needs.

[0322] Presenting the results to the user

[0323] The server sends the final answer to the user's device. This transmission also uses an HTTP request. The device receives the result from the server and displays it to the user. For example, it may say, "The age of the universe is approximately 13.75 billion years." The user can confirm this and obtain the information they need.

[0324] Prompt Sentence Examples

[0325] Here are some example prompts to input to a generative AI model:

[0326] User: How old is the universe?

[0327] Emotions: Interest, Surprise

[0328] As a result, the system of the present invention can provide high-quality information to users and improve user satisfaction by providing optimal answers that take emotion data into consideration.

[0329] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0330] Step 1:

[0331] A user inputs a question into the device's interface. For example, a user might input "How old is the universe?" This question input is received by the device as text data. Next, the device's emotion recognition engine generates the user's emotion data. The emotion recognition engine uses voice analysis, facial expression recognition, and text analysis techniques to analyze the user's emotion at the time of input. Specifically, the device's camera captures the user's facial expression and determines the emotion based on that data. The input at this time is the user's question text and real-time video data, and the output is emotion data such as "interest" or "surprise."

[0332] Step 2:

[0333] When the user presses the send button, the device sends the question data and emotion data to the server. The transmission is done using an HTTP request, and the data is packaged in JSON format. For example, the question data might include "How old is the universe?", and the emotion data might include "surprise." The input in this case is the text question and emotion data from the user, and the output is an HTTP request containing these.

[0334] Step 3:

[0335] The server receives the question data and emotion data sent from the device and analyzes them. First, it checks that the received data is in JSON format. Next, it uses a JSON library to parse the data, separates the question content and emotion data, and stores them in internal variables. For example, it stores the question data in the "question" variable and the emotion data in the "emotion" variable. The input in this case is the JSON data included in the HTTP request, and the output is the separated question text and emotion data.

[0336] Step 4:

[0337] The server sends the received question to multiple artificial intelligence models. This is done using an API, for example by calling the API endpoints of Model A and Model B to send the question. An answer is returned from each model. For example, Model A might answer "13.7 billion years," and Model B might answer "13.8 billion years." The input here is the question text, and the output is the answer from each model. Specifically, the server sends an asynchronous request to the model's API endpoint and waits for the response.

[0338] Step 5:

[0339] The server evaluates and scores the answers obtained from each model. Evaluation criteria include accuracy, reliability, and recency, and also considers emotional data. For example, model A's answer "13.7 billion years" is given a score of 99% accuracy and 95% reliability. On the other hand, model B's answer "13.8 billion years" is given a score of 98% accuracy and 90% reliability. The inputs are the answers and emotional data from each model, and the output is the evaluation results and scoring.

[0340] Step 6:

[0341] The server generates feedback for each model based on the evaluation results. The feedback includes improvements and additional information for each model, and also takes into account the user's emotional data. For example, if the user expresses surprise, the server generates feedback such as "The answer is accurate, but it would be better if you provided more detailed information." The inputs are the evaluation results and emotional data, and the output is feedback information.

[0342] Step 7:

[0343] The server calculates the average score based on the scores of each model and determines the most reliable answer. For example, take the median of 13.7 billion years for Model A and 13.8 billion years for Model B, and arrive at 13.75 billion years. The input is the scores of each model, and the output is the final reliable answer. Emotional data is also taken into account to provide the answer that best suits the user's needs.

[0344] Step 8:

[0345] The server sends the final answer to the user's device. This is also done using an HTTP request. The device receives the result and displays it to the user. For example, it might say, "The age of the universe is approximately 13.75 billion years." The input is the final answer and emotion data, and the output is the result displayed on the user's device.

[0346] (Application example 2)

[0347] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0348] Conventional answer generation systems using artificial intelligence often provide answers without considering the user's feelings, and there is a demand for a highly reliable system that can quickly respond to security-related questions and concerns. Therefore, a system that provides reliable answers that take the user's feelings into consideration is needed.

[0349] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes: means for a user to input a question from a terminal; means for receiving and analyzing the question sent from the terminal; means for sending the question received by the server means to a plurality of artificial intelligence models and obtaining answers from each of the artificial intelligence models; means for evaluating and scoring the answers received from each of the artificial intelligence models; means for generating feedback for each artificial intelligence model based on the evaluation results; means for calculating an average score from the scores of the artificial intelligence models and determining the most reliable answer; means for sending the determined answer to the terminal and presenting it to the user; emotion recognition means for recognizing the user's emotion and generating emotion data; and means for providing the most appropriate answer in consideration of the emotion data. This makes it possible to provide a reliable answer that takes the user's emotion into consideration.

[0350] "Means for users to input questions from a terminal" refers to an interface that allows users to input and send questions using a terminal such as a smartphone or computer.

[0351] "Server means for receiving and analyzing questions sent from terminals" refers to a server system that receives question data entered by users and analyzes the content of the data.

[0352] "Means for transmitting questions received by the server means to multiple artificial intelligence models and obtaining answers from each of the artificial intelligence models" refers to a function for transmitting questions received by the server to multiple artificial intelligence models and obtaining answers from each model.

[0353] "Means for evaluating and scoring the answers received from each of the AI ​​models" refers to a function for evaluating and scoring the answers obtained from each AI model based on their reliability and accuracy.

[0354] "Means for generating feedback for each artificial intelligence model based on the evaluation results" refers to a function for generating feedback including improvements for the artificial intelligence model based on the evaluation results.

[0355] "Means for calculating an average score from the scores of AI models and determining the most reliable answer" refers to a function for integrating the scores of each AI model to calculate an average score and determine the most reliable answer.

[0356] "Means for transmitting the determined answer to the terminal and presenting it to the user" refers to a function for transmitting the optimal answer determined by the server to the terminal and displaying it to the user.

[0357] "Emotion recognition means for recognizing a user's emotions and generating emotion data" refers to technology that recognizes emotions from the user's input, voice, facial expressions, etc., and generates that data.

[0358] "Means for providing the most appropriate answer taking into consideration emotional data" refers to a technology that provides the most appropriate answer based on the user's emotions, based on recognized emotional data.

[0359] "Means for inputting security questions and concerns" refers to an interface that allows a user to input security questions and concerns into a terminal.

[0360] "Means for recognizing anxiety and making an evaluation that places particular emphasis on reliability based on that emotion" refers to a function that uses emotion recognition technology to recognize the user's feelings of anxiety and makes an answer evaluation that places emphasis on reliability, taking those emotions into consideration.

[0361] A "database containing prompt sentences for an artificial intelligence model" refers to a database that stores prompt sentences used by an artificial intelligence model to generate answers.

[0362]

[0363] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be described in detail below with reference to the accompanying drawings.

[0364] The system based on the present invention provides fast and reliable answers to security-related questions and concerns. A user inputs a question from a terminal, and an emotion recognition means analyzes the user's emotions. The analysis results are sent to a server, where multiple artificial intelligence models generate answers to the question. The server evaluates each answer and presents the most reliable answer to the user.

[0365] The server is configured with various software and API endpoints. For example, emotion recognition uses a voice analysis API, facial expression recognition API, or text analysis API. As a specific example, the Emotion API (e.g., https: / / emotion-api.example.com / analyze) can be used to obtain emotional data from the user's input text. For multiple AI models, API endpoints for Model A and Model B (e.g., https: / / ai-model-a.example.com / answer and https: / / ai-model-b.example.com / answer) are used. Answers to questions are obtained through these APIs.

[0366] Next, we will explain the emotion recognition means and the artificial intelligence model acquisition means. When a user inputs a question on the terminal, the emotion recognition engine operates and analyzes either the text, voice, or facial expression to generate emotion data. If the user inputs a question such as, "I feel like there's been some suspicious activity around me recently. What should I do?", the emotion recognition engine will detect the user's anxiety. This emotion data and question text are then sent to the server.

[0367] The server sends the received question and emotion data to each AI model and obtains an answer. The answers received from each model are evaluated based on criteria such as accuracy, reliability, and recency. For example, Model A's answer "Contact the nearest police station" and Model B's answer "Install a security camera" are evaluated and a score is calculated for each. The user's emotion data is also taken into account. If anxiety is detected, reliability is emphasized.

[0368] The server then calculates the average score from each answer and determines the most reliable answer. The determined answer is sent from the server to the device and displayed to the user. This process allows the user to receive the optimal answer that takes emotions into account.

[0369] For example, the following prompt sentence is used:

[0370] "What should I do if I notice any suspicious activity in my neighborhood?"

[0371] "Please tell me what to do when you feel anxious."

[0372] By using the above means, the present invention can take into consideration the user's feelings and quickly provide the most reliable answers to security-related questions and concerns.

[0373] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0374] Step 1:

[0375] The user uses the terminal to input a question, for example, "I've noticed some suspicious activity around me recently. What should I do?" The input data is stored in text format.

[0376] Step 2:

[0377] The device sends the input question to the emotion recognition engine, which generates the user's emotion data using text analysis, voice analysis, or facial expression analysis. For example, it recognizes the emotion "anxious" from the user's text and generates emotion data. The output emotion data is in the format { "emotion": "anxious", "confidence": 0.85}.

[0378] Step 3:

[0379] The device sends the question data and emotion data to the server via an HTTP request. The data is packaged in JSON format. For example, the following data is sent: { "question": "I feel like there's been some suspicious activity around me recently. What should I do?", "emotion": "anxious", "confidence": 0.85}

[0380] Step 4:

[0381] The server receives the question data and emotion data sent from the device and parses it in JSON format. It extracts the received data and separates each field (question text, emotion, confidence level). This data is used in the next processing step.

[0382] Step 5:

[0383] The server sends the received question to multiple artificial intelligence models. Specifically, it sends the question to Model A and Model B via an API endpoint. The question data is sent via an HTTP request to each model, and the answer data is obtained. For example, if Model A says "Contact the nearby police station," Model B will respond with "Install a security camera."

[0384] Step 6:

[0385] The server evaluates the answers received from each AI model. The evaluation criteria are accuracy, reliability, and recency. Each answer is assigned a score based on these criteria. For example, the answer from model A may be assigned a score of 99% accuracy and 95% reliability, while the answer from model B may be assigned a score of 98% accuracy and 90% reliability.

[0386] Step 7:

[0387] The server generates an evaluation result by taking into account the emotional data. If the user expresses "anxiety," it places special emphasis on the reliability of the answer. For example, it adds a reliability score to Model A's answer "Contact the nearest police station" and evaluates it. The evaluation score is then assigned to each answer to generate the final score.

[0388] Step 8:

[0389] The server generates feedback for each AI model, specifically indicating areas for improvement for each model. For example, it may generate feedback such as "Accuracy is high, but the reliability of the answers has decreased."

[0390] Step 9:

[0391] The server calculates the average score from each AI model and determines the most reliable answer. For example, it combines the scores of "contact the nearest police station" and "install security cameras" and selects the most reliable answer based on the average score.

[0392] Step 10:

[0393] The server sends the final answer to the device again. The sending is done via an HTTP request, and the device presents the final answer received from the server to the user. For example, the answer "Contact the nearest police station" is displayed to the user.

[0394] This series of processes allows users to receive the most reliable answer to their input question, taking emotions into consideration.

[0395] The specific processing unit 290 transmits the result of the specific processing to the smart device 14. In the smart device 14, the control unit 46A causes the output device 40 to output the result of the specific processing. The microphone 38B acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0396] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (registered trademark) (Internet search engine).<URL: https: / / openai.com / blog / chatgpt> ), Gemini (registered trademark) (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0397] In the above embodiment, an example in which the specific process is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific process may be performed by the smart device 14.

[0398] [Second embodiment]

[0399] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0400] 3, the data processing system 210 includes the data processing device 12 and smart glasses 214. An example of the data processing device 12 is a server.

[0401] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0402] The smart glasses 214 include a computer 36, a microphone 238, a speaker 240, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, and the camera 42 are also connected to the bus 52.

[0403] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[0404] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[0405] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0406] Fig. 4 shows an example of the main functions of the data processing device 12 and the smart glasses 214. As shown in Fig. 4, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0407] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0408] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0409] In the smart glasses 214, the reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0410] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal."

[0411] The system according to the present invention uses multiple artificial intelligence models to generate answers to questions entered by a user, evaluates the answers, provides feedback, and provides the most reliable answer to the user. An embodiment of this system will be described in detail below.

[0412] User Input

[0413] A user enters a question into the device's interface. For example, the user might enter the question "How old is the universe?" When the user presses the submit button, the device sends the question data to the server. This transmission is typically done using an HTTP request, with the question data packaged in JSON format.

[0414] Receiving and analyzing

[0415] The server receives the question sent from the device and analyzes the question. It checks the format of the received data, confirms that it is in JSON format, and then parses it to extract the question. The extracted question is used in the next process.

[0416] Ask a question and get an answer

[0417] The server sends the received questions to multiple artificial intelligence models. For example, it sends a question to model A and model B via API and obtains answers from each model. At this time, the server uses the appropriate API endpoint to send to each model.

[0418] Evaluating and scoring answers

[0419] The server evaluates the answers received from each AI model. Evaluation criteria include accuracy, reliability, and recency. These criteria are applied to each answer, and a score is generated. For example, model A's answer "13.7 billion years" is given a score of 99% accuracy and 95% reliability, while model B's answer "13.8 billion years" is given a score of 98% accuracy and 90% reliability.

[0420] Generate feedback

[0421] The server generates feedback for each AI model based on the evaluation results. The feedback indicates specific areas for improvement for each model. For example, for Model A, the server generates feedback such as "The answer is accurate, but further improvement in reliability is needed."

[0422] Calculating the average score and determining the final answer

[0423] The server calculates the average score from the scores of each AI model and determines the most reliable answer. For example, if the answers are "13.7 billion years" and "13.8 billion years," the server calculates the median and average scores and determines the final answer as "13.75 billion years."

[0424] Presenting the results to the user

[0425] The server sends the final answer to the user's device. The device receives the result from the server and displays it to the user. The user can check the presented result and obtain the necessary information. For example, the result will be displayed to the user as "The age of the universe is approximately 13.75 billion years."

[0426] This system allows users to comprehensively evaluate the answers of multiple AI models to obtain the most reliable information, thereby improving the quality and reliability of the answers of the generative AI model and providing users with accurate and reliable information.

[0427] The processing flow will be explained below.

[0428] Step 1:

[0429] A user types a question into a terminal interface. For example, the user types the question "How old is the universe?"

[0430] Step 2:

[0431] When the user presses the submit button, the device sends the question data to the server, typically using an HTTP request, with the question data packaged in JSON format.

[0432] Step 3:

[0433] The server receives the question sent from the device, analyzes the question, checks the format of the received data, and parses the JSON data to extract the question.

[0434] Step 4:

[0435] The server sends the received questions to multiple AI models. For example, it sends a question to model A and model B via API and obtains answers from each model.

[0436] Step 5:

[0437] Each AI model generates an answer to the question and returns it to the server. For example, model A answers "13.7 billion years," and model B answers "13.8 billion years." The server receives these answers.

[0438] Step 6:

[0439] The server evaluates the answers received from each AI model, using criteria such as accuracy, reliability, and recency. Each answer is scored based on these criteria.

[0440] Step 7:

[0441] The server generates feedback for each model based on the scoring results, including specific suggestions for improvement, such as "data needs updating."

[0442] Step 8:

[0443] The server calculates the average score from each AI model's score and determines the most reliable answer. For example, it uses the median of "13.7 billion years" and "13.8 billion years" and determines the median of 13.75 billion years as the final answer.

[0444] Step 9:

[0445] The server then sends the final answer to the user's device, also using an HTTP request.

[0446] Step 10:

[0447] The device displays the final answer received from the server to the user, who then confirms the information displayed on the device screen. For example, the information might read, "The age of the universe is approximately 13.75 billion years."

[0448] Example 1

[0449] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0450] Conventional information retrieval systems rely on a single AI model to generate answers, which limits the accuracy and reliability of the answers. Furthermore, there is insufficient assurance of the quality of the information users receive, increasing the risk of providing incorrect information. This makes it difficult to quickly provide the most reliable information in response to user questions.

[0451] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0452] In this invention, the server includes a device for a user to input a question from a terminal, an information processing device that receives and analyzes the question sent from the terminal, a device that sends the question received by the information processing device to a plurality of artificial intelligence systems and obtains answers from each of the artificial intelligence systems, a device that evaluates and quantifies the answers received from each of the artificial intelligence systems, a device that generates feedback for each artificial intelligence system based on the evaluation results, a device that calculates an average value from the numerical values ​​of the artificial intelligence systems and determines the most reliable answer, and a device that sends the determined answer to the terminal and presents it to the user. This makes it possible to comprehensively evaluate the answers of a plurality of artificial intelligence models and quickly provide accurate and reliable information to the user.

[0453] "Terminal" refers to an electronic device through which a user can enter questions and view results.

[0454] The "information processing device" refers to a computer system that receives a question sent from a terminal and analyzes its content.

[0455] "Artificial intelligence system" refers to a collection of computational models or algorithms designed to generate answers to specific questions.

[0456] "Device" refers to a combination of hardware and software for achieving a particular function.

[0457] "Question" refers to text or other form of data that allows a user to enter information they would like to know.

[0458] "Analysis" refers to the process of breaking down received data and understanding its meaning.

[0459] "Answer" refers to the information generated by an artificial intelligence system in response to a question.

[0460] "Evaluation" refers to the process for determining the quality of answers received from an artificial intelligence system.

[0461] "Quantification" refers to expressing the evaluation results as quantitative values.

[0462] "Feedback" refers to information used to provide improvements and evaluation results to an artificial intelligence system.

[0463] The "average value" refers to the value obtained by dividing the sum of multiple numerical data by the number of data items.

[0464] A "trusted answer" refers to an answer that is determined to be the most trustworthy based on the evaluation criteria.

[0465] "Presenting" refers to displaying information to the user on a screen or the like.

[0466] The system according to the present invention uses multiple artificial intelligence models to generate answers to questions entered by a user, evaluates the answers, provides feedback, and provides the most reliable answers to the user. An embodiment of this system is described in detail below.

[0467] Hardware and Software Configuration

[0468] This system consists of a terminal used by the user, a server that receives and analyzes questions, multiple artificial intelligence models, and an evaluation system that evaluates and provides feedback on each answer.

[0469] Terminal

[0470] Users input questions using a device (e.g., smartphone, PC, tablet). The device uses a web browser or dedicated application as an interface to input, submit, and display the results of the user's question.

[0471] server

[0472] The server receives and analyzes questions sent by users. The server typically runs on a Linux server or a cloud service (e.g., AWS or Google Cloud). The received question data is in JSON format, and the question content is parsed and extracted.

[0473] Artificial Intelligence Model

[0474] The server sends the parsed questions to multiple artificial intelligence models (e.g., GPT-4, BERT), which are connected through API endpoints and each model independently generates an answer based on the questions posed by the server.

[0475] Rating System

[0476] The server evaluates and quantifies the answers obtained from each AI model. The evaluation criteria are accuracy, reliability, and recency. Based on the evaluation results, feedback is generated for each AI model.

[0477] Specific examples

[0478] For example, if a user inputs the question "What is the radius of the Earth?", this question is processed as follows:

[0479] 1. The user types "What is the radius of the Earth?" into the terminal interface and presses the send button.

[0480] 2. The device sends the question data to the server in JSON format.

[0481] 3. The server receives the question data and parses and extracts the question content using a JSON parser.

[0482] 4. The server sends the extracted questions to multiple AI models. For example, it sends questions to Model A and Model B via API.

[0483] 5. Model A generates the answer "approximately 6371 km" and Model B generates the answer "approximately 6370 km".

[0484] 6. The server evaluates each model's answer and assigns Model A an accuracy of 98% and a confidence of 97%, and Model B an accuracy of 97% and a confidence of 95%.

[0485] 7. Based on the evaluation results, the server generates feedback for Model A saying, "The answer is accurate, but you need to further improve its reliability," and for Model B saying, "The accuracy is high, but you need to strengthen its reliability."

[0486] 8. The server calculates the average value of each model and determines that the radius of the Earth is approximately 6,370.5 km.

[0487] 9. The final answer is sent to the user's device and the user is shown "The radius of the Earth is approximately 6370.5 km."

[0488] Prompt Sentence Examples

[0489] An example of a prompt that a user might enter into a terminal is as follows:

[0490] User: What is the radius of the Earth?

[0491] This system allows users to comprehensively evaluate the answers of multiple AI models to obtain the most reliable information, improving the quality and reliability of the answers of the generative AI model and providing users with accurate and reliable information.

[0492] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0493] System program processing flow

[0494] Step 1: User Input

[0495] Contents:

[0496] The user enters a question in the terminal interface.

[0497] Specific behavior:

[0498] The user enters "What is the radius of the Earth?" in the text box and presses the submit button.

[0499] Input and Output:

[0500] Input: User's question (e.g., "What is the radius of the Earth?")

[0501] Data processing: Convert user input into JSON format

[0502] Output: Question data in JSON format

[0503] Step 2: Submit your question

[0504] Contents:

[0505] After the send button is pressed, the terminal transmits the user's question to the server.

[0506] Specific behavior:

[0507] The terminal processes the user's input and sends an HTTP request (POST) to the server, which contains the question data in JSON format.

[0508] Input and Output:

[0509] Input: Question data in JSON format

[0510] Data processing: None

[0511] Output: HTTP request sent (including question data in JSON format)

[0512] Step 3: Receiving and parsing the question

[0513] Contents:

[0514] The server receives the question sent from the terminal and analyzes its contents.

[0515] Specific behavior:

[0516] After receiving the HTTP request, the server extracts the question data in JSON format from the request, parses it, and extracts the question content.

[0517] Input and Output:

[0518] Input: HTTP request (including question data in JSON format)

[0519] Data processing: Extracting question content by parsing JSON

[0520] Output: Parsed question (e.g., "What is the radius of the Earth?")

[0521] Step 4: Sending questions to the AI ​​model

[0522] Contents:

[0523] The server sends the analyzed questions to multiple artificial intelligence models.

[0524] Specific behavior:

[0525] The server sends a query to a prepared API endpoint, for example, an HTTP POST request to model A and model B.

[0526] Input and Output:

[0527] Input: Parsed question

[0528] Data processing: Converting questions into a request format for each model

[0529] Output: Multiple HTTP requests (one for model A, one for model B)

[0530] Step 5: Getting the answer from the model

[0531] Contents:

[0532] The server receives responses from each artificial intelligence model.

[0533] Specific behavior:

[0534] Receive a response from the artificial intelligence model and extract the answer data.

[0535] Input and Output:

[0536] Input: HTTP response from each model

[0537] Data processing: Extract answer data from responses

[0538] Output: Extracted response data (e.g., Model A "approximately 6371 km", Model B "approximately 6370 km")

[0539] Step 6: Evaluate and quantify responses

[0540] Contents:

[0541] The server evaluates and quantifies the answers from each artificial intelligence model.

[0542] Specific behavior:

[0543] The server rates each answer on the criteria of accuracy, reliability, and currency, and assigns a numerical score for each criterion.

[0544] Input and Output:

[0545] Input: Response data from each model

[0546] Data processing: Assigning a numerical score based on evaluation criteria

[0547] Output: Evaluated numerical score (e.g., Model A: Accuracy 98%, Confidence 97%)

[0548] Step 7: Generate feedback

[0549] Contents:

[0550] The server generates feedback for each artificial intelligence model based on the evaluation results.

[0551] Specific behavior:

[0552] The evaluation results are analyzed and feedback messages are generated that specifically indicate areas for improvement for each model.

[0553] Input and Output:

[0554] Input: Evaluation result

[0555] Data processing: feedback message generation

[0556] Output: Feedback message (e.g., Model A "Reliability needs to be improved")

[0557] Step 8: Calculate the average and determine the final answer

[0558] Contents:

[0559] The server calculates an average score from each AI model and determines the most reliable answer.

[0560] Specific behavior:

[0561] The evaluation scores of each model are averaged to select the most reliable answer.

[0562] Input and Output:

[0563] Input: Rating score

[0564] Data processing: Calculating the average and determining the final answer

[0565] Output: Final answer (e.g. "The radius of the Earth is approximately 6370.5 km")

[0566] Step 9: Present the results to the user

[0567] Contents:

[0568] The server sends the determined final answer to the user's terminal, and the terminal displays the result to the user.

[0569] Specific behavior:

[0570] The server generates the final answer in JSON format and sends it as an HTTP response to the terminal, which receives this data and displays the result on the screen.

[0571] Input and Output:

[0572] Input: Final answer

[0573] Data processing: Converting the final answer to JSON and generating an HTTP response

[0574] Output: The result displayed on the user's screen (e.g., "The radius of the Earth is approximately 6370.5 km.")

[0575] (Application example 1)

[0576] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0577] In modern food delivery applications, when users ask questions about food or restaurants, it is difficult for them to determine whether the answers are reliable. Furthermore, answers provided by a single AI model often have limitations in accuracy and reliability. Therefore, there is a need for a system that uses multiple AI models to provide more reliable information.

[0578] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0579] In this invention, the server includes: means for a user to input a question from a terminal; information processing means for receiving and analyzing the question sent from the terminal; means for sending the question received by the information processing means to a plurality of artificial intelligence models and obtaining answers from each of the artificial intelligence models; means for evaluating and scoring the answers received from each of the artificial intelligence models; means for generating feedback for each artificial intelligence model based on the evaluation results; means for calculating an average score from the scores of the artificial intelligence models and determining the most reliable answer; means for sending the determined answer to the terminal and presenting it to the user; and means for adding a function to a food delivery application that generates answers to questions from users using a plurality of artificial intelligence models and provides the most reliable answer, thereby enabling users to obtain reliable answers.

[0580] A "terminal" is a computing device through which a user can enter questions and send and receive information.

[0581] The "information processing device means" is a computer server for receiving and analyzing questions sent from the terminals.

[0582] An "artificial intelligence model" is a system that uses pre-trained machine learning algorithms to generate answers to user questions.

[0583] "Means for evaluating and scoring" means a method or device for evaluating and assigning a score to the answers received from each artificial intelligence model based on criteria of accuracy, reliability, and currency.

[0584] The "means for generating feedback" refers to a method or device for creating feedback that specifically indicates areas for improvement based on the evaluation results of each artificial intelligence model.

[0585] A "food delivery application" is a mobile application that allows users to search for information about food and restaurants and place orders.

[0586] A "reliable answer" is information that is judged to have the highest reliability based on the evaluation results of multiple artificial intelligence models.

[0587] A food delivery application system based on this invention begins with a user inputting a question from a terminal. For example, the question might be, "What is the most popular menu item at this restaurant?" The terminal then sends the question to a server. The server receives and analyzes the question and sends it to multiple artificial intelligence models. Each artificial intelligence model generates an answer, which is then received by the server. The received answers are evaluated and scored based on accuracy, reliability, and recency.

[0588] Hardware and software used

[0589] 1. Hardware:

[0590] Device: The smartphone used by the user.

[0591] Server: Use a cloud-based server system (e.g., AWS, Google Cloud).

[0592] 2. Software:

[0593] Food delivery app: A mobile application for Android and iOS.

[0594] API endpoints: APIs for accessing multiple artificial intelligence models (e.g., GPT-3).

[0595] Assessment and feedback generation system: A software module for analyzing responses, scoring, and generating feedback.

[0596] The server generates feedback for each AI model based on the evaluation results. The feedback includes specific areas for improvement for each model. For example, the server might generate feedback such as, "The answer is accurate, but you need to improve its reliability."

[0597] The server then calculates an average score from each AI model's score and determines the most reliable answer. This reliable answer is sent to the user's device and presented to the user through the application. The user can then review the presented results and obtain the necessary information.

[0598] For example, if a user asks, "How many calories are in salmon sushi?" the system will proceed as follows:

[0599] Example prompt sentence:

[0600] "How many calories are in salmon sushi?"

[0601] In response to this question, multiple AI models generate answers such as "Salmon sushi has 250 kcal" and "Salmon sushi has 260 kcal." The server evaluates these answers and provides the most reliable answer, "Salmon sushi has 255 kcal," to the user.

[0602] This allows the user to quickly and accurately obtain highly reliable information.

[0603] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0604] Step 1:

[0605] The user enters a question from the terminal and submits it. The entered question is converted to JSON format and sent to the server as an HTTP request. The input of this process is the question text entered by the user, and the output is the JSON-formatted question data sent to the server.

[0606] Step 2:

[0607] The server receives the question sent from the device and analyzes the question content. It checks the format of the received data, confirms that it is in JSON format, and then parses it to extract the question content. The input of this process is the question data in JSON format, and the output is the analyzed question text.

[0608] Step 3:

[0609] The server sends the parsed question to multiple AI models and obtains an answer from each model. The appropriate API endpoint is used to send the question to each model. The input of this process is the parsed question text, and the output is the answer text from each AI model.

[0610] Step 4:

[0611] The server evaluates and scores the answers received from each AI model. Evaluation criteria include accuracy, reliability, and recency, and each answer is scored. The input to this process is the answer text from each model, and the output is the evaluation result and its respective score.

[0612] Step 5:

[0613] The server generates feedback for each AI model based on the evaluation results. The feedback specifically indicates areas for improvement for each model. The input to this process is the evaluation results and scores, and the output is a feedback message.

[0614] Step 6:

[0615] The server calculates the average score from the scores of each AI model and determines the most reliable answer. The input of this process is the score data, and the output is the most reliable answer text.

[0616] Step 7:

[0617] The server sends the determined most authoritative answer to the terminal and presents it to the user. The terminal receives the result from the server and displays it to the user. The input of this process is the most authoritative answer text, and the output is the answer message displayed on the user's terminal.

[0618] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[0619] The system based on this invention uses multiple artificial intelligence models to generate answers to questions entered by a user, evaluates and provides feedback on the answers, and provides the user with the most reliable answer. Furthermore, by combining it with an emotion engine that recognizes the user's emotions, it is possible to provide a more appropriate answer that takes the user's emotions into consideration. An embodiment of this system will be described in detail below.

[0620] User Input and Emotion Recognition

[0621] A user inputs a question into the device's interface. For example, suppose the user inputs the question, "How old is the universe?" At this time, the device's emotion engine also recognizes the user's emotion. The emotion engine uses methods such as voice analysis, facial expression recognition, or text analysis to generate emotion data that the user indicates when inputting.

[0622] Submitting questions and emotion data

[0623] When the user presses the send button, the device sends the question data and emotion data to the server, typically using an HTTP request, with the question data and emotion data packaged in JSON format.

[0624] Receiving and analyzing

[0625] The server receives the question data and emotion data sent from the device and analyzes them. After checking the format of the received data and confirming that it is in JSON format, it parses it to extract the question content and emotion data.

[0626] Ask a question and get an answer

[0627] The server sends the received questions to multiple artificial intelligence models. For example, it sends a question to model A and model B via API and obtains answers from each model. At this time, the server uses the appropriate API endpoint to send to each model.

[0628] Evaluating and scoring answers

[0629] The server evaluates the answers received from each AI model. Evaluation criteria include accuracy, reliability, and recency. These criteria are applied to each answer, and a score is generated. Emotional data is also taken into account in the evaluation. For example, model A's answer "13.7 billion years" is given a score of 99% accuracy and 95% reliability, while model B's answer "13.8 billion years" is given a score of 98% accuracy and 90% reliability.

[0630] Generate feedback

[0631] The server generates feedback for each AI model based on the evaluation results. The feedback includes specific areas for improvement for each model, while also taking into account the user's emotional data. For example, if the user expresses anxiety, the server may generate feedback such as "your answer is accurate, but more information is needed."

[0632] Calculating the average score and determining the final answer

[0633] The server calculates the average score from each AI model and determines the most reliable answer. For example, the median of "13.7 billion years" and "13.8 billion years" is taken to determine 13.75 billion years. Taking into account emotional data, answers that are closest to the user's needs are also given more weight.

[0634] Presenting the results to the user

[0635] The server sends the final answer to the user's device. This transmission is also performed using an HTTP request. The device receives the result from the server and displays it to the user. The user can check the presented result and obtain the necessary information. For example, the result is displayed to the user as "The age of the universe is approximately 13.75 billion years."

[0636] This system allows users to comprehensively evaluate the answers of multiple AI models to obtain the most reliable information, while also providing optimal answers that take emotional data into account. This improves the quality and reliability of the generative AI model's answers, enabling the provision of accurate and reliable information to users.

[0637] The processing flow will be explained below.

[0638] System Embodiments

[0639] The system based on this invention uses multiple artificial intelligence models to generate answers to questions entered by the user, evaluates and provides feedback on the answers, and provides the user with the most reliable answer. Furthermore, by combining it with an emotion engine that recognizes the user's emotions, it is possible to provide a more appropriate answer that takes the user's emotions into consideration. This will be explained in detail in the following processing steps.

[0640] Step 1:

[0641] A user types a question into a terminal interface. For example, the user types the question "How old is the universe?"

[0642] Step 2:

[0643] When a user inputs a question, the device's emotion engine recognizes the user's emotion using voice analysis, facial expression recognition, text analysis, etc. For example, emotion data is generated from the user's tone of voice and facial expression.

[0644] Step 3:

[0645] When the user presses the send button, the device sends the question data and emotion data to the server, typically using an HTTP request, with the question data and emotion data packaged in JSON format.

[0646] Step 4:

[0647] The server receives the question data and emotion data sent from the device and analyzes them. After checking the format of the received data and confirming that it is in JSON format, it parses it to extract the question content and emotion data.

[0648] Step 5:

[0649] The server sends the received question to multiple artificial intelligence models. For example, it sends a question to model A and model B via API and obtains an answer from each model. The server uses the appropriate API endpoint to send to each model.

[0650] Step 6:

[0651] Each AI model generates an answer to the question and returns it to the server. For example, model A answers "13.7 billion years," and model B answers "13.8 billion years." The server receives these answers.

[0652] Step 7:

[0653] The server evaluates the answers received from each AI model. Evaluation criteria include accuracy, reliability, and recency. These criteria are applied to each answer, resulting in a score. For example, model A's answer "13.7 billion years" is given a score of 99% accuracy and 95% reliability, while model B's answer "13.8 billion years" is given a score of 98% accuracy and 90% reliability. Emotional data is also taken into account in this evaluation.

[0654] Step 8:

[0655] The server generates feedback for each model based on the scoring results. The feedback includes specific areas for improvement for each model, and also takes into account the user's emotional data. For example, if the user expresses anxiety, the server generates feedback such as "The answer is accurate, but further explanation is needed."

[0656] Step 9:

[0657] The server calculates the average score from each AI model's score and determines the most reliable answer. For example, the median of "13.7 billion years" and "13.8 billion years" is taken to determine 13.75 billion years. Taking into account the user's emotional data, the server selects an answer that is optimized for the user's emotional state.

[0658] Step 10:

[0659] The server then sends the final answer to the user's device, also using an HTTP request.

[0660] Step 11:

[0661] The device displays the final answer received from the server to the user, who then confirms the information displayed on the device screen. For example, the information might read, "The age of the universe is approximately 13.75 billion years."

[0662] This system allows users to comprehensively evaluate the answers of multiple AI models to obtain the most reliable information, while also providing optimal answers that take emotional data into account. This improves the quality and reliability of the generative AI model's answers, enabling the provision of accurate and reliable information to users.

[0663] Example 2

[0664] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0665] Currently, when a user inputs a question into an AI, they must compare and consider the answers of multiple AI models to obtain a reliable answer. However, doing this individually is technically time-consuming and ultimately inconvenient for users. Furthermore, if an answer is provided without taking the user's feelings into consideration, user satisfaction decreases. Furthermore, there is a lack of functionality to provide feedback on the obtained answers and use it to improve the model, which also poses the problem of delays in improving the model itself.

[0666] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0667] In this invention, the server includes: means for a user to input a question from a terminal; means for generating user emotion data using an emotion recognition engine of the terminal; means for receiving and analyzing the question and emotion data sent from the terminal; means for transmitting the question received by the server means to multiple artificial intelligence models and obtaining answers from each of the artificial intelligence models; means for evaluating and scoring the answers and emotion data received from each of the artificial intelligence models; means for generating feedback for each artificial intelligence model based on the evaluation results; means for calculating an average score from the scores of the artificial intelligence models and determining the most reliable answer; and means for transmitting the determined answer to the terminal and presenting it to the user. This allows the user to obtain a reliable answer and provides appropriate feedback according to the user's emotion, thereby improving satisfaction. Furthermore, since the artificial intelligence models are rapidly improved through feedback, the performance of the entire system can also be improved.

[0668] A "terminal" is a device where a user inputs a question and generates emotion data.

[0669] An "emotion recognition engine" is a technology installed in a device that analyzes a user's voice, facial expressions, text, etc. to generate emotional data.

[0670] "Emotion data" is information generated by an emotion recognition engine that expresses the emotions felt by the user when entering a question.

[0671] A "server" is a computer system that receives and analyzes question data and emotion data sent from a terminal.

[0672] An "artificial intelligence model" is a set of machine learning algorithms and / or deep learning models used to generate answers to specific questions.

[0673] An "API endpoint" is an interface for sending questions to an artificial intelligence model and obtaining answers.

[0674] "Scoring" is the process of assigning a score to the answer obtained from an artificial intelligence model based on evaluation criteria.

[0675] "Feedback" is information that provides improvements and additional information based on the evaluation results of the AI ​​model's answers.

[0676] The "average score" is the average score of the answers from multiple artificial intelligence models, and is the value used to determine the most reliable answer.

[0677] An "HTTP request" is one of the Internet protocols used to send and receive data.

[0678] The "JSON format" is a format for structuring and storing data, and is used to package question data and sentiment data.

[0679] The system based on this invention uses multiple artificial intelligence models to generate answers to questions entered by users, evaluates and provides feedback on the answers, and provides the most reliable answers to the users. Furthermore, by combining it with an emotion engine that recognizes the user's emotions, it is possible to provide more appropriate answers that take the user's emotions into consideration.

[0680] User Input and Emotion Recognition

[0681] A user inputs a question into the device's interface. For example, when a user inputs a question such as "How old is the universe?", the device's emotion recognition engine recognizes the user's emotions in real time. The emotion recognition engine generates emotion data using voice analysis, facial expression recognition, or text analysis technologies. Specifically, it captures the user's tone of voice and facial expressions and analyzes them to determine emotions such as "surprise" or "interest."

[0682] Submitting questions and emotion data

[0683] When the user presses the send button, the device sends the question data and emotion data to the server. This transmission is made via an HTTP request, and the data is packaged in JSON format. For example, the question data might include "How old is the universe?", and the emotion data might include "surprise."

[0684] Receiving and analyzing

[0685] The server receives the question data and emotion data sent from the device and analyzes them. First, it checks that the received data is in JSON format, then parses the data using a JSON library. This extracts the question content and emotion data, which are then stored in internal variables. For example, the question data is stored in the "question" variable, and the emotion data is stored in the "emotion" variable.

[0686] Ask a question and get an answer

[0687] The server sends the received question to multiple artificial intelligence models via API. For example, the server calls the API endpoints of Model A and Model B and sends the question to each model. This returns an answer from each model. For example, Model A might return the answer "13.7 billion years," and Model B might return the answer "13.8 billion years."

[0688] Evaluating and scoring answers

[0689] The server evaluates the answers received from each model, using criteria such as accuracy, reliability, and recency. For example, model A's answer "13.7 billion years" is given a score of 99% accuracy and 95% reliability. On the other hand, model B's answer "13.8 billion years" is given a score of 98% accuracy and 90% reliability. Emotional data is also taken into account in the evaluation.

[0690] Generate feedback

[0691] The server generates feedback for each model based on the evaluation results. The feedback includes suggestions for improvement and additional information for each model, and also takes into account emotional data. For example, if the user expresses surprise, the server generates feedback such as "The answer is accurate, but it would be better if you provided more detailed information."

[0692] Calculating the average score and determining the final answer

[0693] The server calculates the average score from each model's score and determines the most reliable answer. Specifically, it takes the median value of 13.7 billion years for Model A and 13.8 billion years for Model B as its answer, resulting in 13.75 billion years. It also takes into account sentiment data to provide the answer that best suits the user's needs.

[0694] Presenting the results to the user

[0695] The server sends the final answer to the user's device. This transmission also uses an HTTP request. The device receives the result from the server and displays it to the user. For example, it may say, "The age of the universe is approximately 13.75 billion years." The user can confirm this and obtain the information they need.

[0696] Prompt Sentence Examples

[0697] Here are some example prompts to input to a generative AI model:

[0698] User: How old is the universe?

[0699] Emotions: Interest, Surprise

[0700] As a result, the system of the present invention can provide high-quality information to users and improve user satisfaction by providing optimal answers that take emotion data into consideration.

[0701] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0702] Step 1:

[0703] A user inputs a question into the device's interface. For example, a user might input "How old is the universe?" This question input is received by the device as text data. Next, the device's emotion recognition engine generates the user's emotion data. The emotion recognition engine uses voice analysis, facial expression recognition, and text analysis techniques to analyze the user's emotion at the time of input. Specifically, the device's camera captures the user's facial expression and determines the emotion based on that data. The input at this time is the user's question text and real-time video data, and the output is emotion data such as "interest" or "surprise."

[0704] Step 2:

[0705] When the user presses the send button, the device sends the question data and emotion data to the server. The transmission is done using an HTTP request, and the data is packaged in JSON format. For example, the question data might include "How old is the universe?", and the emotion data might include "surprise." The input in this case is the text question and emotion data from the user, and the output is an HTTP request containing these.

[0706] Step 3:

[0707] The server receives the question data and emotion data sent from the device and analyzes them. First, it checks that the received data is in JSON format. Next, it uses a JSON library to parse the data, separates the question content and emotion data, and stores them in internal variables. For example, it stores the question data in the "question" variable and the emotion data in the "emotion" variable. The input in this case is the JSON data included in the HTTP request, and the output is the separated question text and emotion data.

[0708] Step 4:

[0709] The server sends the received question to multiple artificial intelligence models. This is done using an API, for example by calling the API endpoints of Model A and Model B to send the question. An answer is returned from each model. For example, Model A might answer "13.7 billion years," and Model B might answer "13.8 billion years." The input here is the question text, and the output is the answer from each model. Specifically, the server sends an asynchronous request to the model's API endpoint and waits for the response.

[0710] Step 5:

[0711] The server evaluates and scores the answers obtained from each model. Evaluation criteria include accuracy, reliability, and recency, and also considers emotional data. For example, model A's answer "13.7 billion years" is given a score of 99% accuracy and 95% reliability. On the other hand, model B's answer "13.8 billion years" is given a score of 98% accuracy and 90% reliability. The inputs are the answers and emotional data from each model, and the output is the evaluation results and scoring.

[0712] Step 6:

[0713] The server generates feedback for each model based on the evaluation results. The feedback includes improvements and additional information for each model, and also takes into account the user's emotional data. For example, if the user expresses surprise, the server generates feedback such as "The answer is accurate, but it would be better if you provided more detailed information." The inputs are the evaluation results and emotional data, and the output is feedback information.

[0714] Step 7:

[0715] The server calculates the average score based on the scores of each model and determines the most reliable answer. For example, take the median of 13.7 billion years for Model A and 13.8 billion years for Model B, and arrive at 13.75 billion years. The input is the scores of each model, and the output is the final reliable answer. Emotional data is also taken into account to provide the answer that best suits the user's needs.

[0716] Step 8:

[0717] The server sends the final answer to the user's device. This is also done using an HTTP request. The device receives the result and displays it to the user. For example, it might say, "The age of the universe is approximately 13.75 billion years." The input is the final answer and emotion data, and the output is the result displayed on the user's device.

[0718] (Application example 2)

[0719] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0720] Conventional answer generation systems using artificial intelligence often provide answers without considering the user's feelings, and there is a demand for a highly reliable system that can quickly respond to security-related questions and concerns. Therefore, a system that provides reliable answers that take the user's feelings into consideration is needed.

[0721] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes: means for a user to input a question from a terminal; means for receiving and analyzing the question sent from the terminal; means for sending the question received by the server means to a plurality of artificial intelligence models and obtaining answers from each of the artificial intelligence models; means for evaluating and scoring the answers received from each of the artificial intelligence models; means for generating feedback for each artificial intelligence model based on the evaluation results; means for calculating an average score from the scores of the artificial intelligence models and determining the most reliable answer; means for sending the determined answer to the terminal and presenting it to the user; emotion recognition means for recognizing the user's emotion and generating emotion data; and means for providing the most appropriate answer in consideration of the emotion data. This makes it possible to provide a reliable answer that takes the user's emotion into consideration.

[0722] "Means for users to input questions from a terminal" refers to an interface that allows users to input and send questions using a terminal such as a smartphone or computer.

[0723] "Server means for receiving and analyzing questions sent from terminals" refers to a server system that receives question data entered by users and analyzes the content of the data.

[0724] "Means for transmitting questions received by the server means to multiple artificial intelligence models and obtaining answers from each of the artificial intelligence models" refers to a function for transmitting questions received by the server to multiple artificial intelligence models and obtaining answers from each model.

[0725] "Means for evaluating and scoring the answers received from each of the AI ​​models" refers to a function for evaluating and scoring the answers obtained from each AI model based on their reliability and accuracy.

[0726] "Means for generating feedback for each artificial intelligence model based on the evaluation results" refers to a function for generating feedback including improvements for the artificial intelligence model based on the evaluation results.

[0727] "Means for calculating an average score from the scores of AI models and determining the most reliable answer" refers to a function for integrating the scores of each AI model to calculate an average score and determine the most reliable answer.

[0728] "Means for transmitting the determined answer to the terminal and presenting it to the user" refers to a function for transmitting the optimal answer determined by the server to the terminal and displaying it to the user.

[0729] "Emotion recognition means for recognizing a user's emotions and generating emotion data" refers to technology that recognizes emotions from the user's input, voice, facial expressions, etc., and generates that data.

[0730] "Means for providing the most appropriate answer taking into consideration emotional data" refers to a technology that provides the most appropriate answer based on the user's emotions, based on recognized emotional data.

[0731] "Means for inputting security questions and concerns" refers to an interface that allows a user to input security questions and concerns into a terminal.

[0732] "Means for recognizing anxiety and making an evaluation that places particular emphasis on reliability based on that emotion" refers to a function that uses emotion recognition technology to recognize the user's feelings of anxiety and makes an answer evaluation that places emphasis on reliability, taking those emotions into consideration.

[0733] A "database containing prompt sentences for an artificial intelligence model" refers to a database that stores prompt sentences used by an artificial intelligence model to generate answers.

[0734]

[0735] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be described in detail below with reference to the accompanying drawings.

[0736] The system based on the present invention provides fast and reliable answers to security-related questions and concerns. A user inputs a question from a terminal, and an emotion recognition means analyzes the user's emotions. The analysis results are sent to a server, where multiple artificial intelligence models generate answers to the question. The server evaluates each answer and presents the most reliable answer to the user.

[0737] The server is configured with various software and API endpoints. For example, emotion recognition uses a voice analysis API, facial expression recognition API, or text analysis API. As a specific example, the Emotion API (e.g., https: / / emotion-api.example.com / analyze) can be used to obtain emotional data from the user's input text. For multiple AI models, API endpoints for Model A and Model B (e.g., https: / / ai-model-a.example.com / answer and https: / / ai-model-b.example.com / answer) are used. Answers to questions are obtained through these APIs.

[0738] Next, we will explain the emotion recognition means and the artificial intelligence model acquisition means. When a user inputs a question on the terminal, the emotion recognition engine operates and analyzes either the text, voice, or facial expression to generate emotion data. If the user inputs a question such as, "I feel like there's been some suspicious activity around me recently. What should I do?", the emotion recognition engine will detect the user's anxiety. This emotion data and question text are then sent to the server.

[0739] The server sends the received question and emotion data to each AI model and obtains an answer. The answers received from each model are evaluated based on criteria such as accuracy, reliability, and recency. For example, Model A's answer "Contact the nearest police station" and Model B's answer "Install a security camera" are evaluated and a score is calculated for each. The user's emotion data is also taken into account. If anxiety is detected, reliability is emphasized.

[0740] The server then calculates the average score from each answer and determines the most reliable answer. The determined answer is sent from the server to the device and displayed to the user. This process allows the user to receive the optimal answer that takes emotions into account.

[0741] For example, the following prompt sentence is used:

[0742] "What should I do if I notice any suspicious activity in my neighborhood?"

[0743] "Please tell me what to do when you feel anxious."

[0744] By using the above means, the present invention can take into consideration the user's feelings and quickly provide the most reliable answers to security-related questions and concerns.

[0745] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0746] Step 1:

[0747] The user uses the terminal to input a question, for example, "I've noticed some suspicious activity around me recently. What should I do?" The input data is stored in text format.

[0748] Step 2:

[0749] The device sends the input question to the emotion recognition engine, which generates the user's emotion data using text analysis, voice analysis, or facial expression analysis. For example, it recognizes the emotion "anxious" from the user's text and generates emotion data. The output emotion data is in the format { "emotion": "anxious", "confidence": 0.85}.

[0750] Step 3:

[0751] The device sends the question data and emotion data to the server via an HTTP request. The data is packaged in JSON format. For example, the following data is sent: { "question": "I feel like there's been some suspicious activity around me recently. What should I do?", "emotion": "anxious", "confidence": 0.85}

[0752] Step 4:

[0753] The server receives the question data and emotion data sent from the device and parses it in JSON format. It extracts the received data and separates each field (question text, emotion, confidence level). This data is used in the next processing step.

[0754] Step 5:

[0755] The server sends the received question to multiple artificial intelligence models. Specifically, it sends the question to Model A and Model B via an API endpoint. The question data is sent via an HTTP request to each model, and the answer data is obtained. For example, if Model A says "Contact the nearby police station," Model B will respond with "Install a security camera."

[0756] Step 6:

[0757] The server evaluates the answers received from each AI model. The evaluation criteria are accuracy, reliability, and recency. Each answer is assigned a score based on these criteria. For example, the answer from model A may be assigned a score of 99% accuracy and 95% reliability, while the answer from model B may be assigned a score of 98% accuracy and 90% reliability.

[0758] Step 7:

[0759] The server generates an evaluation result by taking into account the emotional data. If the user expresses "anxiety," it places special emphasis on the reliability of the answer. For example, it adds a reliability score to Model A's answer "Contact the nearest police station" and evaluates it. The evaluation score is then assigned to each answer to generate the final score.

[0760] Step 8:

[0761] The server generates feedback for each AI model, specifically indicating areas for improvement for each model. For example, it may generate feedback such as "Accuracy is high, but the reliability of the answers has decreased."

[0762] Step 9:

[0763] The server calculates the average score from each AI model and determines the most reliable answer. For example, it combines the scores of "contact the nearest police station" and "install security cameras" and selects the most reliable answer based on the average score.

[0764] Step 10:

[0765] The server sends the final answer to the device again. The sending is done via an HTTP request, and the device presents the final answer received from the server to the user. For example, the answer "Contact the nearest police station" is displayed to the user.

[0766] This series of processes allows users to receive the most reliable answer to their input question, taking emotions into consideration.

[0767] The specific processing unit 290 transmits the result of the specific processing to the smart glasses 214. In the smart glasses 214, the control unit 46A causes the speaker 240 to output the result of the specific processing. The microphone 238 acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0768] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0769] In the above embodiment, an example in which the specific processing is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the smart glasses 214.

[0770] [Third embodiment]

[0771] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[0772] 5, the data processing system 310 includes the data processing device 12 and a headset terminal 314. An example of the data processing device 12 is a server.

[0773] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0774] The headset type terminal 314 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a display 343. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the display 343 are also connected to the bus 52.

[0775] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[0776] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[0777] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0778] Fig. 6 shows an example of the main functions of the data processing device 12 and the headset type terminal 314. As shown in Fig. 6, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0779] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0780] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0781] In the headset type terminal 314, a reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0782] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as the "server" and the headset type terminal 314 will be referred to as the "terminal."

[0783] The system according to the present invention uses multiple artificial intelligence models to generate answers to questions entered by a user, evaluates the answers, provides feedback, and provides the most reliable answer to the user. An embodiment of this system will be described in detail below.

[0784] User Input

[0785] A user enters a question into the device's interface. For example, the user might enter the question "How old is the universe?" When the user presses the submit button, the device sends the question data to the server. This transmission is typically done using an HTTP request, with the question data packaged in JSON format.

[0786] Receiving and analyzing

[0787] The server receives the question sent from the device and analyzes the question. It checks the format of the received data, confirms that it is in JSON format, and then parses it to extract the question. The extracted question is used in the next process.

[0788] Ask a question and get an answer

[0789] The server sends the received questions to multiple artificial intelligence models. For example, it sends a question to model A and model B via API and obtains answers from each model. At this time, the server uses the appropriate API endpoint to send to each model.

[0790] Evaluating and scoring answers

[0791] The server evaluates the answers received from each AI model. Evaluation criteria include accuracy, reliability, and recency. These criteria are applied to each answer, and a score is generated. For example, model A's answer "13.7 billion years" is given a score of 99% accuracy and 95% reliability, while model B's answer "13.8 billion years" is given a score of 98% accuracy and 90% reliability.

[0792] Generate feedback

[0793] The server generates feedback for each AI model based on the evaluation results. The feedback indicates specific areas for improvement for each model. For example, for Model A, the server generates feedback such as "The answer is accurate, but further improvement in reliability is needed."

[0794] Calculating the average score and determining the final answer

[0795] The server calculates the average score from the scores of each AI model and determines the most reliable answer. For example, if the answers are "13.7 billion years" and "13.8 billion years," the server calculates the median and average scores and determines the final answer as "13.75 billion years."

[0796] Presenting the results to the user

[0797] The server sends the final answer to the user's device. The device receives the result from the server and displays it to the user. The user can check the presented result and obtain the necessary information. For example, the result will be displayed to the user as "The age of the universe is approximately 13.75 billion years."

[0798] This system allows users to comprehensively evaluate the answers of multiple AI models to obtain the most reliable information, thereby improving the quality and reliability of the answers of the generative AI model and providing users with accurate and reliable information.

[0799] The processing flow will be explained below.

[0800] Step 1:

[0801] A user types a question into a terminal interface. For example, the user types the question "How old is the universe?"

[0802] Step 2:

[0803] When the user presses the submit button, the device sends the question data to the server, typically using an HTTP request, with the question data packaged in JSON format.

[0804] Step 3:

[0805] The server receives the question sent from the device, analyzes the question, checks the format of the received data, and parses the JSON data to extract the question.

[0806] Step 4:

[0807] The server sends the received questions to multiple AI models. For example, it sends a question to model A and model B via API and obtains answers from each model.

[0808] Step 5:

[0809] Each AI model generates an answer to the question and returns it to the server. For example, model A answers "13.7 billion years," and model B answers "13.8 billion years." The server receives these answers.

[0810] Step 6:

[0811] The server evaluates the answers received from each AI model, using criteria such as accuracy, reliability, and recency. Each answer is scored based on these criteria.

[0812] Step 7:

[0813] The server generates feedback for each model based on the scoring results, including specific suggestions for improvement, such as "data needs updating."

[0814] Step 8:

[0815] The server calculates the average score from each AI model's score and determines the most reliable answer. For example, it uses the median of "13.7 billion years" and "13.8 billion years" and determines the median of 13.75 billion years as the final answer.

[0816] Step 9:

[0817] The server then sends the final answer to the user's device, also using an HTTP request.

[0818] Step 10:

[0819] The device displays the final answer received from the server to the user, who then confirms the information displayed on the device screen. For example, the information might read, "The age of the universe is approximately 13.75 billion years."

[0820] Example 1

[0821] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[0822] Conventional information retrieval systems rely on a single AI model to generate answers, which limits the accuracy and reliability of the answers. Furthermore, there is insufficient assurance of the quality of the information users receive, increasing the risk of providing incorrect information. This makes it difficult to quickly provide the most reliable information in response to user questions.

[0823] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0824] In this invention, the server includes a device for a user to input a question from a terminal, an information processing device that receives and analyzes the question sent from the terminal, a device that sends the question received by the information processing device to a plurality of artificial intelligence systems and obtains answers from each of the artificial intelligence systems, a device that evaluates and quantifies the answers received from each of the artificial intelligence systems, a device that generates feedback for each artificial intelligence system based on the evaluation results, a device that calculates an average value from the numerical values ​​of the artificial intelligence systems and determines the most reliable answer, and a device that sends the determined answer to the terminal and presents it to the user. This makes it possible to comprehensively evaluate the answers of a plurality of artificial intelligence models and quickly provide accurate and reliable information to the user.

[0825] "Terminal" refers to an electronic device through which a user can enter questions and view results.

[0826] The "information processing device" refers to a computer system that receives a question sent from a terminal and analyzes its content.

[0827] "Artificial intelligence system" refers to a collection of computational models or algorithms designed to generate answers to specific questions.

[0828] "Device" refers to a combination of hardware and software for achieving a particular function.

[0829] "Question" refers to text or other form of data that allows a user to enter information they would like to know.

[0830] "Analysis" refers to the process of breaking down received data and understanding its meaning.

[0831] "Answer" refers to the information generated by an artificial intelligence system in response to a question.

[0832] "Evaluation" refers to the process for determining the quality of answers received from an artificial intelligence system.

[0833] "Quantification" refers to expressing the evaluation results as quantitative values.

[0834] "Feedback" refers to information used to provide improvements and evaluation results to an artificial intelligence system.

[0835] The "average value" refers to the value obtained by dividing the sum of multiple numerical data by the number of data items.

[0836] A "trusted answer" refers to an answer that is determined to be the most trustworthy based on the evaluation criteria.

[0837] "Presenting" refers to displaying information to the user on a screen or the like.

[0838] The system according to the present invention uses multiple artificial intelligence models to generate answers to questions entered by a user, evaluates the answers, provides feedback, and provides the most reliable answers to the user. An embodiment of this system is described in detail below.

[0839] Hardware and Software Configuration

[0840] This system consists of a terminal used by the user, a server that receives and analyzes questions, multiple artificial intelligence models, and an evaluation system that evaluates and provides feedback on each answer.

[0841] Terminal

[0842] Users input questions using a device (e.g., smartphone, PC, tablet). The device uses a web browser or dedicated application as an interface to input, submit, and display the results of the user's question.

[0843] server

[0844] The server receives and analyzes questions sent by users. The server typically runs on a Linux server or a cloud service (e.g., AWS or Google Cloud). The received question data is in JSON format, and the question content is parsed and extracted.

[0845] Artificial Intelligence Model

[0846] The server sends the parsed questions to multiple artificial intelligence models (e.g., GPT-4, BERT), which are connected through API endpoints and each model independently generates an answer based on the questions posed by the server.

[0847] Rating System

[0848] The server evaluates and quantifies the answers obtained from each AI model. The evaluation criteria are accuracy, reliability, and recency. Based on the evaluation results, feedback is generated for each AI model.

[0849] Specific examples

[0850] For example, if a user inputs the question "What is the radius of the Earth?", this question is processed as follows:

[0851] 1. The user types "What is the radius of the Earth?" into the terminal interface and presses the send button.

[0852] 2. The device sends the question data to the server in JSON format.

[0853] 3. The server receives the question data and parses and extracts the question content using a JSON parser.

[0854] 4. The server sends the extracted questions to multiple AI models. For example, it sends questions to Model A and Model B via API.

[0855] 5. Model A generates the answer "approximately 6371 km" and Model B generates the answer "approximately 6370 km".

[0856] 6. The server evaluates each model's answer and assigns Model A an accuracy of 98% and a confidence of 97%, and Model B an accuracy of 97% and a confidence of 95%.

[0857] 7. Based on the evaluation results, the server generates feedback for Model A saying, "The answer is accurate, but you need to further improve its reliability," and for Model B saying, "The accuracy is high, but you need to strengthen its reliability."

[0858] 8. The server calculates the average value of each model and determines that the radius of the Earth is approximately 6,370.5 km.

[0859] 9. The final answer is sent to the user's device and the user is shown "The radius of the Earth is approximately 6370.5 km."

[0860] Prompt Sentence Examples

[0861] An example of a prompt that a user might enter into a terminal is as follows:

[0862] User: What is the radius of the Earth?

[0863] This system allows users to comprehensively evaluate the answers of multiple AI models to obtain the most reliable information, improving the quality and reliability of the answers of the generative AI model and providing users with accurate and reliable information.

[0864] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0865] System program processing flow

[0866] Step 1: User Input

[0867] Contents:

[0868] The user enters a question in the terminal interface.

[0869] Specific behavior:

[0870] The user enters "What is the radius of the Earth?" in the text box and presses the submit button.

[0871] Input and Output:

[0872] Input: User's question (e.g., "What is the radius of the Earth?")

[0873] Data processing: Convert user input into JSON format

[0874] Output: Question data in JSON format

[0875] Step 2: Submit your question

[0876] Contents:

[0877] After the send button is pressed, the terminal transmits the user's question to the server.

[0878] Specific behavior:

[0879] The terminal processes the user's input and sends an HTTP request (POST) to the server, which contains the question data in JSON format.

[0880] Input and Output:

[0881] Input: Question data in JSON format

[0882] Data processing: None

[0883] Output: HTTP request sent (including question data in JSON format)

[0884] Step 3: Receiving and parsing the question

[0885] Contents:

[0886] The server receives the question sent from the terminal and analyzes its contents.

[0887] Specific behavior:

[0888] After receiving the HTTP request, the server extracts the question data in JSON format from the request, parses it, and extracts the question content.

[0889] Input and Output:

[0890] Input: HTTP request (including question data in JSON format)

[0891] Data processing: Extracting question content by parsing JSON

[0892] Output: Parsed question (e.g., "What is the radius of the Earth?")

[0893] Step 4: Sending questions to the AI ​​model

[0894] Contents:

[0895] The server sends the analyzed questions to multiple artificial intelligence models.

[0896] Specific behavior:

[0897] The server sends a query to a prepared API endpoint, for example, an HTTP POST request to model A and model B.

[0898] Input and Output:

[0899] Input: Parsed question

[0900] Data processing: Converting questions into a request format for each model

[0901] Output: Multiple HTTP requests (one for model A, one for model B)

[0902] Step 5: Getting the answer from the model

[0903] Contents:

[0904] The server receives responses from each artificial intelligence model.

[0905] Specific behavior:

[0906] Receive a response from the artificial intelligence model and extract the answer data.

[0907] Input and Output:

[0908] Input: HTTP response from each model

[0909] Data processing: Extract answer data from responses

[0910] Output: Extracted response data (e.g., Model A "approximately 6371 km", Model B "approximately 6370 km")

[0911] Step 6: Evaluate and quantify responses

[0912] Contents:

[0913] The server evaluates and quantifies the answers from each artificial intelligence model.

[0914] Specific behavior:

[0915] The server rates each answer on the criteria of accuracy, reliability, and currency, and assigns a numerical score for each criterion.

[0916] Input and Output:

[0917] Input: Response data from each model

[0918] Data processing: Assigning a numerical score based on evaluation criteria

[0919] Output: Evaluated numerical score (e.g., Model A: Accuracy 98%, Confidence 97%)

[0920] Step 7: Generate feedback

[0921] Contents:

[0922] The server generates feedback for each artificial intelligence model based on the evaluation results.

[0923] Specific behavior:

[0924] The evaluation results are analyzed and feedback messages are generated that specifically indicate areas for improvement for each model.

[0925] Input and Output:

[0926] Input: Evaluation result

[0927] Data processing: feedback message generation

[0928] Output: Feedback message (e.g., Model A "Reliability needs to be improved")

[0929] Step 8: Calculate the average and determine the final answer

[0930] Contents:

[0931] The server calculates an average score from each AI model and determines the most reliable answer.

[0932] Specific behavior:

[0933] The evaluation scores of each model are averaged to select the most reliable answer.

[0934] Input and Output:

[0935] Input: Rating score

[0936] Data processing: Calculating the average and determining the final answer

[0937] Output: Final answer (e.g. "The radius of the Earth is approximately 6370.5 km")

[0938] Step 9: Present the results to the user

[0939] Contents:

[0940] The server sends the determined final answer to the user's terminal, and the terminal displays the result to the user.

[0941] Specific behavior:

[0942] The server generates the final answer in JSON format and sends it as an HTTP response to the terminal, which receives this data and displays the result on the screen.

[0943] Input and Output:

[0944] Input: Final answer

[0945] Data processing: Converting the final answer to JSON and generating an HTTP response

[0946] Output: The result displayed on the user's screen (e.g., "The radius of the Earth is approximately 6370.5 km.")

[0947] (Application example 1)

[0948] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[0949] In modern food delivery applications, when users ask questions about food or restaurants, it is difficult for them to determine whether the answers are reliable. Furthermore, answers provided by a single AI model often have limitations in accuracy and reliability. Therefore, there is a need for a system that uses multiple AI models to provide more reliable information.

[0950] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0951] In this invention, the server includes: means for a user to input a question from a terminal; information processing means for receiving and analyzing the question sent from the terminal; means for sending the question received by the information processing means to a plurality of artificial intelligence models and obtaining answers from each of the artificial intelligence models; means for evaluating and scoring the answers received from each of the artificial intelligence models; means for generating feedback for each artificial intelligence model based on the evaluation results; means for calculating an average score from the scores of the artificial intelligence models and determining the most reliable answer; means for sending the determined answer to the terminal and presenting it to the user; and means for adding a function to a food delivery application that generates answers to questions from users using a plurality of artificial intelligence models and provides the most reliable answer, thereby enabling users to obtain reliable answers.

[0952] A "terminal" is a computing device through which a user can enter questions and send and receive information.

[0953] The "information processing device means" is a computer server for receiving and analyzing questions sent from the terminals.

[0954] An "artificial intelligence model" is a system that uses pre-trained machine learning algorithms to generate answers to user questions.

[0955] "Means for evaluating and scoring" means a method or device for evaluating and assigning a score to the answers received from each artificial intelligence model based on criteria of accuracy, reliability, and currency.

[0956] The "means for generating feedback" refers to a method or device for creating feedback that specifically indicates areas for improvement based on the evaluation results of each artificial intelligence model.

[0957] A "food delivery application" is a mobile application that allows users to search for information about food and restaurants and place orders.

[0958] A "reliable answer" is information that is judged to have the highest reliability based on the evaluation results of multiple artificial intelligence models.

[0959] A food delivery application system based on this invention begins with a user inputting a question from a terminal. For example, the question might be, "What is the most popular menu item at this restaurant?" The terminal then sends the question to a server. The server receives and analyzes the question and sends it to multiple artificial intelligence models. Each artificial intelligence model generates an answer, which is then received by the server. The received answers are evaluated and scored based on accuracy, reliability, and recency.

[0960] Hardware and software used

[0961] 1. Hardware:

[0962] Device: The smartphone used by the user.

[0963] Server: Use a cloud-based server system (e.g., AWS, Google Cloud).

[0964] 2. Software:

[0965] Food delivery app: A mobile application for Android and iOS.

[0966] API endpoints: APIs for accessing multiple artificial intelligence models (e.g., GPT-3).

[0967] Assessment and feedback generation system: A software module for analyzing responses, scoring, and generating feedback.

[0968] The server generates feedback for each AI model based on the evaluation results. The feedback includes specific areas for improvement for each model. For example, the server might generate feedback such as, "The answer is accurate, but you need to improve its reliability."

[0969] The server then calculates an average score from each AI model's score and determines the most reliable answer. This reliable answer is sent to the user's device and presented to the user through the application. The user can then review the presented results and obtain the necessary information.

[0970] For example, if a user asks, "How many calories are in salmon sushi?" the system will proceed as follows:

[0971] Example prompt sentence:

[0972] "How many calories are in salmon sushi?"

[0973] In response to this question, multiple AI models generate answers such as "Salmon sushi has 250 kcal" and "Salmon sushi has 260 kcal." The server evaluates these answers and provides the most reliable answer, "Salmon sushi has 255 kcal," to the user.

[0974] This allows the user to quickly and accurately obtain highly reliable information.

[0975] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0976] Step 1:

[0977] The user enters a question from the terminal and submits it. The entered question is converted to JSON format and sent to the server as an HTTP request. The input of this process is the question text entered by the user, and the output is the JSON-formatted question data sent to the server.

[0978] Step 2:

[0979] The server receives the question sent from the device and analyzes the question content. It checks the format of the received data, confirms that it is in JSON format, and then parses it to extract the question content. The input of this process is the question data in JSON format, and the output is the analyzed question text.

[0980] Step 3:

[0981] The server sends the parsed question to multiple AI models and obtains an answer from each model. The appropriate API endpoint is used to send the question to each model. The input of this process is the parsed question text, and the output is the answer text from each AI model.

[0982] Step 4:

[0983] The server evaluates and scores the answers received from each AI model. Evaluation criteria include accuracy, reliability, and recency, and each answer is scored. The input to this process is the answer text from each model, and the output is the evaluation result and its respective score.

[0984] Step 5:

[0985] The server generates feedback for each AI model based on the evaluation results. The feedback specifically indicates areas for improvement for each model. The input to this process is the evaluation results and scores, and the output is a feedback message.

[0986] Step 6:

[0987] The server calculates the average score from the scores of each AI model and determines the most reliable answer. The input of this process is the score data, and the output is the most reliable answer text.

[0988] Step 7:

[0989] The server sends the determined most authoritative answer to the terminal and presents it to the user. The terminal receives the result from the server and displays it to the user. The input of this process is the most authoritative answer text, and the output is the answer message displayed on the user's terminal.

[0990] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[0991] The system based on this invention uses multiple artificial intelligence models to generate answers to questions entered by a user, evaluates and provides feedback on the answers, and provides the user with the most reliable answer. Furthermore, by combining it with an emotion engine that recognizes the user's emotions, it is possible to provide a more appropriate answer that takes the user's emotions into consideration. An embodiment of this system will be described in detail below.

[0992] User Input and Emotion Recognition

[0993] A user inputs a question into the device's interface. For example, suppose the user inputs the question, "How old is the universe?" At this time, the device's emotion engine also recognizes the user's emotion. The emotion engine uses methods such as voice analysis, facial expression recognition, or text analysis to generate emotion data that the user indicates when inputting.

[0994] Submitting questions and emotion data

[0995] When the user presses the send button, the device sends the question data and emotion data to the server, typically using an HTTP request, with the question data and emotion data packaged in JSON format.

[0996] Receiving and analyzing

[0997] The server receives the question data and emotion data sent from the device and analyzes them. After checking the format of the received data and confirming that it is in JSON format, it parses it to extract the question content and emotion data.

[0998] Ask a question and get an answer

[0999] The server sends the received questions to multiple artificial intelligence models. For example, it sends a question to model A and model B via API and obtains answers from each model. At this time, the server uses the appropriate API endpoint to send to each model.

[1000] Evaluating and scoring answers

[1001] The server evaluates the answers received from each AI model. Evaluation criteria include accuracy, reliability, and recency. These criteria are applied to each answer, and a score is generated. Emotional data is also taken into account in the evaluation. For example, model A's answer "13.7 billion years" is given a score of 99% accuracy and 95% reliability, while model B's answer "13.8 billion years" is given a score of 98% accuracy and 90% reliability.

[1002] Generate feedback

[1003] The server generates feedback for each AI model based on the evaluation results. The feedback includes specific areas for improvement for each model, while also taking into account the user's emotional data. For example, if the user expresses anxiety, the server may generate feedback such as "your answer is accurate, but more information is needed."

[1004] Calculating the average score and determining the final answer

[1005] The server calculates the average score from each AI model and determines the most reliable answer. For example, the median of "13.7 billion years" and "13.8 billion years" is taken to determine 13.75 billion years. Taking into account emotional data, answers that are closest to the user's needs are also given more weight.

[1006] Presenting the results to the user

[1007] The server sends the final answer to the user's device. This transmission is also performed using an HTTP request. The device receives the result from the server and displays it to the user. The user can check the presented result and obtain the necessary information. For example, the result is displayed to the user as "The age of the universe is approximately 13.75 billion years."

[1008] This system allows users to comprehensively evaluate the answers of multiple AI models to obtain the most reliable information, while also providing optimal answers that take emotional data into account. This improves the quality and reliability of the generative AI model's answers, enabling the provision of accurate and reliable information to users.

[1009] The processing flow will be explained below.

[1010] System Embodiments

[1011] The system based on this invention uses multiple artificial intelligence models to generate answers to questions entered by the user, evaluates and provides feedback on the answers, and provides the user with the most reliable answer. Furthermore, by combining it with an emotion engine that recognizes the user's emotions, it is possible to provide a more appropriate answer that takes the user's emotions into consideration. This will be explained in detail in the following processing steps.

[1012] Step 1:

[1013] A user types a question into a terminal interface. For example, the user types the question "How old is the universe?"

[1014] Step 2:

[1015] When a user inputs a question, the device's emotion engine recognizes the user's emotion using voice analysis, facial expression recognition, text analysis, etc. For example, emotion data is generated from the user's tone of voice and facial expression.

[1016] Step 3:

[1017] When the user presses the send button, the device sends the question data and emotion data to the server, typically using an HTTP request, with the question data and emotion data packaged in JSON format.

[1018] Step 4:

[1019] The server receives the question data and emotion data sent from the device and analyzes them. After checking the format of the received data and confirming that it is in JSON format, it parses it to extract the question content and emotion data.

[1020] Step 5:

[1021] The server sends the received question to multiple artificial intelligence models. For example, it sends a question to model A and model B via API and obtains an answer from each model. The server uses the appropriate API endpoint to send to each model.

[1022] Step 6:

[1023] Each AI model generates an answer to the question and returns it to the server. For example, model A answers "13.7 billion years," and model B answers "13.8 billion years." The server receives these answers.

[1024] Step 7:

[1025] The server evaluates the answers received from each AI model. Evaluation criteria include accuracy, reliability, and recency. These criteria are applied to each answer, resulting in a score. For example, model A's answer "13.7 billion years" is given a score of 99% accuracy and 95% reliability, while model B's answer "13.8 billion years" is given a score of 98% accuracy and 90% reliability. Emotional data is also taken into account in this evaluation.

[1026] Step 8:

[1027] The server generates feedback for each model based on the scoring results. The feedback includes specific areas for improvement for each model, and also takes into account the user's emotional data. For example, if the user expresses anxiety, the server generates feedback such as "The answer is accurate, but further explanation is needed."

[1028] Step 9:

[1029] The server calculates the average score from each AI model's score and determines the most reliable answer. For example, the median of "13.7 billion years" and "13.8 billion years" is taken to determine 13.75 billion years. Taking into account the user's emotional data, the server selects an answer that is optimized for the user's emotional state.

[1030] Step 10:

[1031] The server then sends the final answer to the user's device, also using an HTTP request.

[1032] Step 11:

[1033] The device displays the final answer received from the server to the user, who then confirms the information displayed on the device screen. For example, the information might read, "The age of the universe is approximately 13.75 billion years."

[1034] This system allows users to comprehensively evaluate the answers of multiple AI models to obtain the most reliable information, while also providing optimal answers that take emotional data into account. This improves the quality and reliability of the generative AI model's answers, enabling the provision of accurate and reliable information to users.

[1035] Example 2

[1036] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1037] Currently, when a user inputs a question into an AI, they must compare and consider the answers of multiple AI models to obtain a reliable answer. However, doing this individually is technically time-consuming and ultimately inconvenient for users. Furthermore, if an answer is provided without taking the user's feelings into consideration, user satisfaction decreases. Furthermore, there is a lack of functionality to provide feedback on the obtained answers and use it to improve the model, which also poses the problem of delays in improving the model itself.

[1038] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[1039] In this invention, the server includes: means for a user to input a question from a terminal; means for generating user emotion data using an emotion recognition engine of the terminal; means for receiving and analyzing the question and emotion data sent from the terminal; means for transmitting the question received by the server means to multiple artificial intelligence models and obtaining answers from each of the artificial intelligence models; means for evaluating and scoring the answers and emotion data received from each of the artificial intelligence models; means for generating feedback for each artificial intelligence model based on the evaluation results; means for calculating an average score from the scores of the artificial intelligence models and determining the most reliable answer; and means for transmitting the determined answer to the terminal and presenting it to the user. This allows the user to obtain a reliable answer and provides appropriate feedback according to the user's emotion, thereby improving satisfaction. Furthermore, since the artificial intelligence models are rapidly improved through feedback, the performance of the entire system can also be improved.

[1040] A "terminal" is a device where a user inputs a question and generates emotion data.

[1041] An "emotion recognition engine" is a technology installed in a device that analyzes a user's voice, facial expressions, text, etc. to generate emotional data.

[1042] "Emotion data" is information generated by an emotion recognition engine that expresses the emotions felt by the user when entering a question.

[1043] A "server" is a computer system that receives and analyzes question data and emotion data sent from a terminal.

[1044] An "artificial intelligence model" is a set of machine learning algorithms and / or deep learning models used to generate answers to specific questions.

[1045] An "API endpoint" is an interface for sending questions to an artificial intelligence model and obtaining answers.

[1046] "Scoring" is the process of assigning a score to the answer obtained from an artificial intelligence model based on evaluation criteria.

[1047] "Feedback" is information that provides improvements and additional information based on the evaluation results of the AI ​​model's answers.

[1048] The "average score" is the average score of the answers from multiple artificial intelligence models, and is the value used to determine the most reliable answer.

[1049] An "HTTP request" is one of the Internet protocols used to send and receive data.

[1050] The "JSON format" is a format for structuring and storing data, and is used to package question data and sentiment data.

[1051] The system based on this invention uses multiple artificial intelligence models to generate answers to questions entered by users, evaluates and provides feedback on the answers, and provides the most reliable answers to the users. Furthermore, by combining it with an emotion engine that recognizes the user's emotions, it is possible to provide more appropriate answers that take the user's emotions into consideration.

[1052] User Input and Emotion Recognition

[1053] A user inputs a question into the device's interface. For example, when a user inputs a question such as "How old is the universe?", the device's emotion recognition engine recognizes the user's emotions in real time. The emotion recognition engine generates emotion data using voice analysis, facial expression recognition, or text analysis technologies. Specifically, it captures the user's tone of voice and facial expressions and analyzes them to determine emotions such as "surprise" or "interest."

[1054] Submitting questions and emotion data

[1055] When the user presses the send button, the device sends the question data and emotion data to the server. This transmission is made via an HTTP request, and the data is packaged in JSON format. For example, the question data might include "How old is the universe?", and the emotion data might include "surprise."

[1056] Receiving and analyzing

[1057] The server receives the question data and emotion data sent from the device and analyzes them. First, it checks that the received data is in JSON format, then parses the data using a JSON library. This extracts the question content and emotion data, which are then stored in internal variables. For example, the question data is stored in the "question" variable, and the emotion data is stored in the "emotion" variable.

[1058] Ask a question and get an answer

[1059] The server sends the received question to multiple artificial intelligence models via API. For example, the server calls the API endpoints of Model A and Model B and sends the question to each model. This returns an answer from each model. For example, Model A might return the answer "13.7 billion years," and Model B might return the answer "13.8 billion years."

[1060] Evaluating and scoring answers

[1061] The server evaluates the answers received from each model, using criteria such as accuracy, reliability, and recency. For example, model A's answer "13.7 billion years" is given a score of 99% accuracy and 95% reliability. On the other hand, model B's answer "13.8 billion years" is given a score of 98% accuracy and 90% reliability. Emotional data is also taken into account in the evaluation.

[1062] Generate feedback

[1063] The server generates feedback for each model based on the evaluation results. The feedback includes suggestions for improvement and additional information for each model, and also takes into account emotional data. For example, if the user expresses surprise, the server generates feedback such as "The answer is accurate, but it would be better if you provided more detailed information."

[1064] Calculating the average score and determining the final answer

[1065] The server calculates the average score from each model's score and determines the most reliable answer. Specifically, it takes the median value of 13.7 billion years for Model A and 13.8 billion years for Model B as its answer, resulting in 13.75 billion years. It also takes into account sentiment data to provide the answer that best suits the user's needs.

[1066] Presenting the results to the user

[1067] The server sends the final answer to the user's device. This transmission also uses an HTTP request. The device receives the result from the server and displays it to the user. For example, it may say, "The age of the universe is approximately 13.75 billion years." The user can confirm this and obtain the information they need.

[1068] Prompt Sentence Examples

[1069] Here are some example prompts to input to a generative AI model:

[1070] User: How old is the universe?

[1071] Emotions: Interest, Surprise

[1072] As a result, the system of the present invention can provide high-quality information to users and improve user satisfaction by providing optimal answers that take emotion data into consideration.

[1073] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1074] Step 1:

[1075] A user inputs a question into the device's interface. For example, a user might input "How old is the universe?" This question input is received by the device as text data. Next, the device's emotion recognition engine generates the user's emotion data. The emotion recognition engine uses voice analysis, facial expression recognition, and text analysis techniques to analyze the user's emotion at the time of input. Specifically, the device's camera captures the user's facial expression and determines the emotion based on that data. The input at this time is the user's question text and real-time video data, and the output is emotion data such as "interest" or "surprise."

[1076] Step 2:

[1077] When the user presses the send button, the device sends the question data and emotion data to the server. The transmission is done using an HTTP request, and the data is packaged in JSON format. For example, the question data might include "How old is the universe?", and the emotion data might include "surprise." The input in this case is the text question and emotion data from the user, and the output is an HTTP request containing these.

[1078] Step 3:

[1079] The server receives the question data and emotion data sent from the device and analyzes them. First, it checks that the received data is in JSON format. Next, it uses a JSON library to parse the data, separates the question content and emotion data, and stores them in internal variables. For example, it stores the question data in the "question" variable and the emotion data in the "emotion" variable. The input in this case is the JSON data included in the HTTP request, and the output is the separated question text and emotion data.

[1080] Step 4:

[1081] The server sends the received question to multiple artificial intelligence models. This is done using an API, for example by calling the API endpoints of Model A and Model B to send the question. An answer is returned from each model. For example, Model A might answer "13.7 billion years," and Model B might answer "13.8 billion years." The input here is the question text, and the output is the answer from each model. Specifically, the server sends an asynchronous request to the model's API endpoint and waits for the response.

[1082] Step 5:

[1083] The server evaluates and scores the answers obtained from each model. Evaluation criteria include accuracy, reliability, and recency, and also considers emotional data. For example, model A's answer "13.7 billion years" is given a score of 99% accuracy and 95% reliability. On the other hand, model B's answer "13.8 billion years" is given a score of 98% accuracy and 90% reliability. The inputs are the answers and emotional data from each model, and the output is the evaluation results and scoring.

[1084] Step 6:

[1085] The server generates feedback for each model based on the evaluation results. The feedback includes improvements and additional information for each model, and also takes into account the user's emotional data. For example, if the user expresses surprise, the server generates feedback such as "The answer is accurate, but it would be better if you provided more detailed information." The inputs are the evaluation results and emotional data, and the output is feedback information.

[1086] Step 7:

[1087] The server calculates the average score based on the scores of each model and determines the most reliable answer. For example, take the median of 13.7 billion years for Model A and 13.8 billion years for Model B, and arrive at 13.75 billion years. The input is the scores of each model, and the output is the final reliable answer. Emotional data is also taken into account to provide the answer that best suits the user's needs.

[1088] Step 8:

[1089] The server sends the final answer to the user's device. This is also done using an HTTP request. The device receives the result and displays it to the user. For example, it might say, "The age of the universe is approximately 13.75 billion years." The input is the final answer and emotion data, and the output is the result displayed on the user's device.

[1090] (Application example 2)

[1091] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1092] Conventional answer generation systems using artificial intelligence often provide answers without considering the user's feelings, and there is a demand for a highly reliable system that can quickly respond to security-related questions and concerns. Therefore, a system that provides reliable answers that take the user's feelings into consideration is needed.

[1093] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes: means for a user to input a question from a terminal; means for receiving and analyzing the question sent from the terminal; means for sending the question received by the server means to a plurality of artificial intelligence models and obtaining answers from each of the artificial intelligence models; means for evaluating and scoring the answers received from each of the artificial intelligence models; means for generating feedback for each artificial intelligence model based on the evaluation results; means for calculating an average score from the scores of the artificial intelligence models and determining the most reliable answer; means for sending the determined answer to the terminal and presenting it to the user; emotion recognition means for recognizing the user's emotion and generating emotion data; and means for providing the most appropriate answer in consideration of the emotion data. This makes it possible to provide a reliable answer that takes the user's emotion into consideration.

[1094] "Means for users to input questions from a terminal" refers to an interface that allows users to input and send questions using a terminal such as a smartphone or computer.

[1095] "Server means for receiving and analyzing questions sent from terminals" refers to a server system that receives question data entered by users and analyzes the content of the data.

[1096] "Means for transmitting questions received by the server means to multiple artificial intelligence models and obtaining answers from each of the artificial intelligence models" refers to a function for transmitting questions received by the server to multiple artificial intelligence models and obtaining answers from each model.

[1097] "Means for evaluating and scoring the answers received from each of the AI ​​models" refers to a function for evaluating and scoring the answers obtained from each AI model based on their reliability and accuracy.

[1098] "Means for generating feedback for each artificial intelligence model based on the evaluation results" refers to a function for generating feedback including improvements for the artificial intelligence model based on the evaluation results.

[1099] "Means for calculating an average score from the scores of AI models and determining the most reliable answer" refers to a function for integrating the scores of each AI model to calculate an average score and determine the most reliable answer.

[1100] "Means for transmitting the determined answer to the terminal and presenting it to the user" refers to a function for transmitting the optimal answer determined by the server to the terminal and displaying it to the user.

[1101] "Emotion recognition means for recognizing a user's emotions and generating emotion data" refers to technology that recognizes emotions from the user's input, voice, facial expressions, etc., and generates that data.

[1102] "Means for providing the most appropriate answer taking into consideration emotional data" refers to a technology that provides the most appropriate answer based on the user's emotions, based on recognized emotional data.

[1103] "Means for inputting security questions and concerns" refers to an interface that allows a user to input security questions and concerns into a terminal.

[1104] "Means for recognizing anxiety and making an evaluation that places particular emphasis on reliability based on that emotion" refers to a function that uses emotion recognition technology to recognize the user's feelings of anxiety and makes an answer evaluation that places emphasis on reliability, taking those emotions into consideration.

[1105] A "database containing prompt sentences for an artificial intelligence model" refers to a database that stores prompt sentences used by an artificial intelligence model to generate answers.

[1106]

[1107] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be described in detail below with reference to the accompanying drawings.

[1108] The system based on the present invention provides fast and reliable answers to security-related questions and concerns. A user inputs a question from a terminal, and an emotion recognition means analyzes the user's emotions. The analysis results are sent to a server, where multiple artificial intelligence models generate answers to the question. The server evaluates each answer and presents the most reliable answer to the user.

[1109] The server is configured with various software and API endpoints. For example, emotion recognition uses a voice analysis API, facial expression recognition API, or text analysis API. As a specific example, the Emotion API (e.g., https: / / emotion-api.example.com / analyze) can be used to obtain emotional data from the user's input text. For multiple AI models, API endpoints for Model A and Model B (e.g., https: / / ai-model-a.example.com / answer and https: / / ai-model-b.example.com / answer) are used. Answers to questions are obtained through these APIs.

[1110] Next, we will explain the emotion recognition means and the artificial intelligence model acquisition means. When a user inputs a question on the terminal, the emotion recognition engine operates and analyzes either the text, voice, or facial expression to generate emotion data. If the user inputs a question such as, "I feel like there's been some suspicious activity around me recently. What should I do?", the emotion recognition engine will detect the user's anxiety. This emotion data and question text are then sent to the server.

[1111] The server sends the received question and emotion data to each AI model and obtains an answer. The answers received from each model are evaluated based on criteria such as accuracy, reliability, and recency. For example, Model A's answer "Contact the nearest police station" and Model B's answer "Install a security camera" are evaluated and a score is calculated for each. The user's emotion data is also taken into account. If anxiety is detected, reliability is emphasized.

[1112] The server then calculates the average score from each answer and determines the most reliable answer. The determined answer is sent from the server to the device and displayed to the user. This process allows the user to receive the optimal answer that takes emotions into account.

[1113] For example, the following prompt sentence is used:

[1114] "What should I do if I notice any suspicious activity in my neighborhood?"

[1115] "Please tell me what to do when you feel anxious."

[1116] By using the above means, the present invention can take into consideration the user's feelings and quickly provide the most reliable answers to security-related questions and concerns.

[1117] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1118] Step 1:

[1119] The user uses the terminal to input a question, for example, "I've noticed some suspicious activity around me recently. What should I do?" The input data is stored in text format.

[1120] Step 2:

[1121] The device sends the input question to the emotion recognition engine, which generates the user's emotion data using text analysis, voice analysis, or facial expression analysis. For example, it recognizes the emotion "anxious" from the user's text and generates emotion data. The output emotion data is in the format { "emotion": "anxious", "confidence": 0.85}.

[1122] Step 3:

[1123] The device sends the question data and emotion data to the server via an HTTP request. The data is packaged in JSON format. For example, the following data is sent: { "question": "I feel like there's been some suspicious activity around me recently. What should I do?", "emotion": "anxious", "confidence": 0.85}

[1124] Step 4:

[1125] The server receives the question data and emotion data sent from the device and parses it in JSON format. It extracts the received data and separates each field (question text, emotion, confidence level). This data is used in the next processing step.

[1126] Step 5:

[1127] The server sends the received question to multiple artificial intelligence models. Specifically, it sends the question to Model A and Model B via an API endpoint. The question data is sent via an HTTP request to each model, and the answer data is obtained. For example, if Model A says "Contact the nearby police station," Model B will respond with "Install a security camera."

[1128] Step 6:

[1129] The server evaluates the answers received from each AI model. The evaluation criteria are accuracy, reliability, and recency. Each answer is assigned a score based on these criteria. For example, the answer from model A may be assigned a score of 99% accuracy and 95% reliability, while the answer from model B may be assigned a score of 98% accuracy and 90% reliability.

[1130] Step 7:

[1131] The server generates an evaluation result by taking into account the emotional data. If the user expresses "anxiety," it places special emphasis on the reliability of the answer. For example, it adds a reliability score to Model A's answer "Contact the nearest police station" and evaluates it. The evaluation score is then assigned to each answer to generate the final score.

[1132] Step 8:

[1133] The server generates feedback for each AI model, specifically indicating areas for improvement for each model. For example, it may generate feedback such as "Accuracy is high, but the reliability of the answers has decreased."

[1134] Step 9:

[1135] The server calculates the average score from each AI model and determines the most reliable answer. For example, it combines the scores of "contact the nearest police station" and "install security cameras" and selects the most reliable answer based on the average score.

[1136] Step 10:

[1137] The server sends the final answer to the device again. The sending is done via an HTTP request, and the device presents the final answer received from the server to the user. For example, the answer "Contact the nearest police station" is displayed to the user.

[1138] This series of processes allows users to receive the most reliable answer to their input question, taking emotions into consideration.

[1139] The specific processing unit 290 transmits the result of the specific processing to the headset type terminal 314. In the headset type terminal 314, the control unit 46A causes the speaker 240 and the display 343 to output the result of the specific processing. The microphone 238 acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[1140] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1141] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the headset type terminal 314.

[1142] [Fourth embodiment]

[1143] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1144] 7, a data processing system 410 includes a data processing device 12 and a robot 414. An example of the data processing device 12 is a server.

[1145] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[1146] The robot 414 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a control target 443. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the control target 443 are also connected to the bus 52.

[1147] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[1148] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[1149] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[1150] The control object 443 includes a display device, LEDs in the eyes, and motors for driving the arms, hands, and feet. The posture and gestures of the robot 414 are controlled by controlling the motors of the arms, hands, and feet. Some of the emotions of the robot 414 can be expressed by controlling these motors. In addition, the facial expressions of the robot 414 can also be expressed by controlling the light emission state of the LEDs in the eyes of the robot 414.

[1151] Fig. 8 shows an example of the main functions of the data processing device 12 and the robot 414. As shown in Fig. 8, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[1152] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[1153] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[1154] In the robot 414, the processor 46 performs the reception output process. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[1155] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1156] The system according to the present invention uses multiple artificial intelligence models to generate answers to questions entered by a user, evaluates the answers, provides feedback, and provides the most reliable answer to the user. An embodiment of this system will be described in detail below.

[1157] User Input

[1158] A user enters a question into the device's interface. For example, the user might enter the question "How old is the universe?" When the user presses the submit button, the device sends the question data to the server. This transmission is typically done using an HTTP request, with the question data packaged in JSON format.

[1159] Receiving and analyzing

[1160] The server receives the question sent from the device and analyzes the question. It checks the format of the received data, confirms that it is in JSON format, and then parses it to extract the question. The extracted question is used in the next process.

[1161] Ask a question and get an answer

[1162] The server sends the received questions to multiple artificial intelligence models. For example, it sends a question to model A and model B via API and obtains answers from each model. At this time, the server uses the appropriate API endpoint to send to each model.

[1163] Evaluating and scoring answers

[1164] The server evaluates the answers received from each AI model. Evaluation criteria include accuracy, reliability, and recency. These criteria are applied to each answer, and a score is generated. For example, model A's answer "13.7 billion years" is given a score of 99% accuracy and 95% reliability, while model B's answer "13.8 billion years" is given a score of 98% accuracy and 90% reliability.

[1165] Generate feedback

[1166] The server generates feedback for each AI model based on the evaluation results. The feedback indicates specific areas for improvement for each model. For example, for Model A, the server generates feedback such as "The answer is accurate, but further improvement in reliability is needed."

[1167] Calculating the average score and determining the final answer

[1168] The server calculates the average score from the scores of each AI model and determines the most reliable answer. For example, if the answers are "13.7 billion years" and "13.8 billion years," the server calculates the median and average scores and determines the final answer as "13.75 billion years."

[1169] Presenting the results to the user

[1170] The server sends the final answer to the user's device. The device receives the result from the server and displays it to the user. The user can check the presented result and obtain the necessary information. For example, the result will be displayed to the user as "The age of the universe is approximately 13.75 billion years."

[1171] This system allows users to comprehensively evaluate the answers of multiple AI models to obtain the most reliable information, thereby improving the quality and reliability of the answers of the generative AI model and providing users with accurate and reliable information.

[1172] The processing flow will be explained below.

[1173] Step 1:

[1174] A user types a question into a terminal interface. For example, the user types the question "How old is the universe?"

[1175] Step 2:

[1176] When the user presses the submit button, the device sends the question data to the server, typically using an HTTP request, with the question data packaged in JSON format.

[1177] Step 3:

[1178] The server receives the question sent from the device, analyzes the question, checks the format of the received data, and parses the JSON data to extract the question.

[1179] Step 4:

[1180] The server sends the received questions to multiple AI models. For example, it sends a question to model A and model B via API and obtains answers from each model.

[1181] Step 5:

[1182] Each AI model generates an answer to the question and returns it to the server. For example, model A answers "13.7 billion years," and model B answers "13.8 billion years." The server receives these answers.

[1183] Step 6:

[1184] The server evaluates the answers received from each AI model, using criteria such as accuracy, reliability, and recency. Each answer is scored based on these criteria.

[1185] Step 7:

[1186] The server generates feedback for each model based on the scoring results, including specific suggestions for improvement, such as "data needs updating."

[1187] Step 8:

[1188] The server calculates the average score from each AI model's score and determines the most reliable answer. For example, it uses the median of "13.7 billion years" and "13.8 billion years" and determines the median of 13.75 billion years as the final answer.

[1189] Step 9:

[1190] The server then sends the final answer to the user's device, also using an HTTP request.

[1191] Step 10:

[1192] The device displays the final answer received from the server to the user, who then confirms the information displayed on the device screen. For example, the information might read, "The age of the universe is approximately 13.75 billion years."

[1193] Example 1

[1194] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1195] Conventional information retrieval systems rely on a single AI model to generate answers, which limits the accuracy and reliability of the answers. Furthermore, there is insufficient assurance of the quality of the information users receive, increasing the risk of providing incorrect information. This makes it difficult to quickly provide the most reliable information in response to user questions.

[1196] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[1197] In this invention, the server includes a device for a user to input a question from a terminal, an information processing device that receives and analyzes the question sent from the terminal, a device that sends the question received by the information processing device to a plurality of artificial intelligence systems and obtains answers from each of the artificial intelligence systems, a device that evaluates and quantifies the answers received from each of the artificial intelligence systems, a device that generates feedback for each artificial intelligence system based on the evaluation results, a device that calculates an average value from the numerical values ​​of the artificial intelligence systems and determines the most reliable answer, and a device that sends the determined answer to the terminal and presents it to the user. This makes it possible to comprehensively evaluate the answers of a plurality of artificial intelligence models and quickly provide accurate and reliable information to the user.

[1198] "Terminal" refers to an electronic device through which a user can enter questions and view results.

[1199] The "information processing device" refers to a computer system that receives a question sent from a terminal and analyzes its content.

[1200] "Artificial intelligence system" refers to a collection of computational models or algorithms designed to generate answers to specific questions.

[1201] "Device" refers to a combination of hardware and software for achieving a particular function.

[1202] "Question" refers to text or other form of data that allows a user to enter information they would like to know.

[1203] "Analysis" refers to the process of breaking down received data and understanding its meaning.

[1204] "Answer" refers to the information generated by an artificial intelligence system in response to a question.

[1205] "Evaluation" refers to the process for determining the quality of answers received from an artificial intelligence system.

[1206] "Quantification" refers to expressing the evaluation results as quantitative values.

[1207] "Feedback" refers to information used to provide improvements and evaluation results to an artificial intelligence system.

[1208] The "average value" refers to the value obtained by dividing the sum of multiple numerical data by the number of data items.

[1209] A "trusted answer" refers to an answer that is determined to be the most trustworthy based on the evaluation criteria.

[1210] "Presenting" refers to displaying information to the user on a screen or the like.

[1211] The system according to the present invention uses multiple artificial intelligence models to generate answers to questions entered by a user, evaluates the answers, provides feedback, and provides the most reliable answers to the user. An embodiment of this system is described in detail below.

[1212] Hardware and Software Configuration

[1213] This system consists of a terminal used by the user, a server that receives and analyzes questions, multiple artificial intelligence models, and an evaluation system that evaluates and provides feedback on each answer.

[1214] Terminal

[1215] Users input questions using a device (e.g., smartphone, PC, tablet). The device uses a web browser or dedicated application as an interface to input, submit, and display the results of the user's question.

[1216] server

[1217] The server receives and analyzes questions sent by users. The server typically runs on a Linux server or a cloud service (e.g., AWS or Google Cloud). The received question data is in JSON format, and the question content is parsed and extracted.

[1218] Artificial Intelligence Model

[1219] The server sends the parsed questions to multiple artificial intelligence models (e.g., GPT-4, BERT), which are connected through API endpoints and each model independently generates an answer based on the questions posed by the server.

[1220] Rating System

[1221] The server evaluates and quantifies the answers obtained from each AI model. The evaluation criteria are accuracy, reliability, and recency. Based on the evaluation results, feedback is generated for each AI model.

[1222] Specific examples

[1223] For example, if a user inputs the question "What is the radius of the Earth?", this question is processed as follows:

[1224] 1. The user types "What is the radius of the Earth?" into the terminal interface and presses the send button.

[1225] 2. The device sends the question data to the server in JSON format.

[1226] 3. The server receives the question data and parses and extracts the question content using a JSON parser.

[1227] 4. The server sends the extracted questions to multiple AI models. For example, it sends questions to Model A and Model B via API.

[1228] 5. Model A generates the answer "approximately 6371 km" and Model B generates the answer "approximately 6370 km".

[1229] 6. The server evaluates each model's answer and assigns Model A an accuracy of 98% and a confidence of 97%, and Model B an accuracy of 97% and a confidence of 95%.

[1230] 7. Based on the evaluation results, the server generates feedback for Model A saying, "The answer is accurate, but you need to further improve its reliability," and for Model B saying, "The accuracy is high, but you need to strengthen its reliability."

[1231] 8. The server calculates the average value of each model and determines that the radius of the Earth is approximately 6,370.5 km.

[1232] 9. The final answer is sent to the user's device and the user is shown "The radius of the Earth is approximately 6370.5 km."

[1233] Prompt Sentence Examples

[1234] An example of a prompt that a user might enter into a terminal is as follows:

[1235] User: What is the radius of the Earth?

[1236] This system allows users to comprehensively evaluate the answers of multiple AI models to obtain the most reliable information, improving the quality and reliability of the answers of the generative AI model and providing users with accurate and reliable information.

[1237] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1238] System program processing flow

[1239] Step 1: User Input

[1240] Contents:

[1241] The user enters a question in the terminal interface.

[1242] Specific behavior:

[1243] The user enters "What is the radius of the Earth?" in the text box and presses the submit button.

[1244] Input and Output:

[1245] Input: User's question (e.g., "What is the radius of the Earth?")

[1246] Data processing: Convert user input into JSON format

[1247] Output: Question data in JSON format

[1248] Step 2: Submit your question

[1249] Contents:

[1250] After the send button is pressed, the terminal transmits the user's question to the server.

[1251] Specific behavior:

[1252] The terminal processes the user's input and sends an HTTP request (POST) to the server, which contains the question data in JSON format.

[1253] Input and Output:

[1254] Input: Question data in JSON format

[1255] Data processing: None

[1256] Output: HTTP request sent (including question data in JSON format)

[1257] Step 3: Receiving and parsing the question

[1258] Contents:

[1259] The server receives the question sent from the terminal and analyzes its contents.

[1260] Specific behavior:

[1261] After receiving the HTTP request, the server extracts the question data in JSON format from the request, parses it, and extracts the question content.

[1262] Input and Output:

[1263] Input: HTTP request (including question data in JSON format)

[1264] Data processing: Extracting question content by parsing JSON

[1265] Output: Parsed question (e.g., "What is the radius of the Earth?")

[1266] Step 4: Sending questions to the AI ​​model

[1267] Contents:

[1268] The server sends the analyzed questions to multiple artificial intelligence models.

[1269] Specific behavior:

[1270] The server sends a query to a prepared API endpoint, for example, an HTTP POST request to model A and model B.

[1271] Input and Output:

[1272] Input: Parsed question

[1273] Data processing: Converting questions into a request format for each model

[1274] Output: Multiple HTTP requests (one for model A, one for model B)

[1275] Step 5: Getting the answer from the model

[1276] Contents:

[1277] The server receives responses from each artificial intelligence model.

[1278] Specific behavior:

[1279] Receive a response from the artificial intelligence model and extract the answer data.

[1280] Input and Output:

[1281] Input: HTTP response from each model

[1282] Data processing: Extract answer data from responses

[1283] Output: Extracted response data (e.g., Model A "approximately 6371 km", Model B "approximately 6370 km")

[1284] Step 6: Evaluate and quantify responses

[1285] Contents:

[1286] The server evaluates and quantifies the answers from each artificial intelligence model.

[1287] Specific behavior:

[1288] The server rates each answer on the criteria of accuracy, reliability, and currency, and assigns a numerical score for each criterion.

[1289] Input and Output:

[1290] Input: Response data from each model

[1291] Data processing: Assigning a numerical score based on evaluation criteria

[1292] Output: Evaluated numerical score (e.g., Model A: Accuracy 98%, Confidence 97%)

[1293] Step 7: Generate feedback

[1294] Contents:

[1295] The server generates feedback for each artificial intelligence model based on the evaluation results.

[1296] Specific behavior:

[1297] The evaluation results are analyzed and feedback messages are generated that specifically indicate areas for improvement for each model.

[1298] Input and Output:

[1299] Input: Evaluation result

[1300] Data processing: feedback message generation

[1301] Output: Feedback message (e.g., Model A "Reliability needs to be improved")

[1302] Step 8: Calculate the average and determine the final answer

[1303] Contents:

[1304] The server calculates an average score from each AI model and determines the most reliable answer.

[1305] Specific behavior:

[1306] The evaluation scores of each model are averaged to select the most reliable answer.

[1307] Input and Output:

[1308] Input: Rating score

[1309] Data processing: Calculating the average and determining the final answer

[1310] Output: Final answer (e.g. "The radius of the Earth is approximately 6370.5 km")

[1311] Step 9: Present the results to the user

[1312] Contents:

[1313] The server sends the determined final answer to the user's terminal, and the terminal displays the result to the user.

[1314] Specific behavior:

[1315] The server generates the final answer in JSON format and sends it as an HTTP response to the terminal, which receives this data and displays the result on the screen.

[1316] Input and Output:

[1317] Input: Final answer

[1318] Data processing: Converting the final answer to JSON and generating an HTTP response

[1319] Output: The result displayed on the user's screen (e.g., "The radius of the Earth is approximately 6370.5 km.")

[1320] (Application example 1)

[1321] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1322] In modern food delivery applications, when users ask questions about food or restaurants, it is difficult for them to determine whether the answers are reliable. Furthermore, answers provided by a single AI model often have limitations in accuracy and reliability. Therefore, there is a need for a system that uses multiple AI models to provide more reliable information.

[1323] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[1324] In this invention, the server includes: means for a user to input a question from a terminal; information processing means for receiving and analyzing the question sent from the terminal; means for sending the question received by the information processing means to a plurality of artificial intelligence models and obtaining answers from each of the artificial intelligence models; means for evaluating and scoring the answers received from each of the artificial intelligence models; means for generating feedback for each artificial intelligence model based on the evaluation results; means for calculating an average score from the scores of the artificial intelligence models and determining the most reliable answer; means for sending the determined answer to the terminal and presenting it to the user; and means for adding a function to a food delivery application that generates answers to questions from users using a plurality of artificial intelligence models and provides the most reliable answer, thereby enabling users to obtain reliable answers.

[1325] A "terminal" is a computing device through which a user can enter questions and send and receive information.

[1326] The "information processing device means" is a computer server for receiving and analyzing questions sent from the terminals.

[1327] An "artificial intelligence model" is a system that uses pre-trained machine learning algorithms to generate answers to user questions.

[1328] "Means for evaluating and scoring" means a method or device for evaluating and assigning a score to the answers received from each artificial intelligence model based on criteria of accuracy, reliability, and currency.

[1329] The "means for generating feedback" refers to a method or device for creating feedback that specifically indicates areas for improvement based on the evaluation results of each artificial intelligence model.

[1330] A "food delivery application" is a mobile application that allows users to search for information about food and restaurants and place orders.

[1331] A "reliable answer" is information that is judged to have the highest reliability based on the evaluation results of multiple artificial intelligence models.

[1332] A food delivery application system based on this invention begins with a user inputting a question from a terminal. For example, the question might be, "What is the most popular menu item at this restaurant?" The terminal then sends the question to a server. The server receives and analyzes the question and sends it to multiple artificial intelligence models. Each artificial intelligence model generates an answer, which is then received by the server. The received answers are evaluated and scored based on accuracy, reliability, and recency.

[1333] Hardware and software used

[1334] 1. Hardware:

[1335] Device: The smartphone used by the user.

[1336] Server: Use a cloud-based server system (e.g., AWS, Google Cloud).

[1337] 2. Software:

[1338] Food delivery app: A mobile application for Android and iOS.

[1339] API endpoints: APIs for accessing multiple artificial intelligence models (e.g., GPT-3).

[1340] Assessment and feedback generation system: A software module for analyzing responses, scoring, and generating feedback.

[1341] The server generates feedback for each AI model based on the evaluation results. The feedback includes specific areas for improvement for each model. For example, the server might generate feedback such as, "The answer is accurate, but you need to improve its reliability."

[1342] The server then calculates an average score from each AI model's score and determines the most reliable answer. This reliable answer is sent to the user's device and presented to the user through the application. The user can then review the presented results and obtain the necessary information.

[1343] For example, if a user asks, "How many calories are in salmon sushi?" the system will proceed as follows:

[1344] Example prompt sentence:

[1345] "How many calories are in salmon sushi?"

[1346] In response to this question, multiple AI models generate answers such as "Salmon sushi has 250 kcal" and "Salmon sushi has 260 kcal." The server evaluates these answers and provides the most reliable answer, "Salmon sushi has 255 kcal," to the user.

[1347] This allows the user to quickly and accurately obtain highly reliable information.

[1348] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1349] Step 1:

[1350] The user enters a question from the terminal and submits it. The entered question is converted to JSON format and sent to the server as an HTTP request. The input of this process is the question text entered by the user, and the output is the JSON-formatted question data sent to the server.

[1351] Step 2:

[1352] The server receives the question sent from the device and analyzes the question content. It checks the format of the received data, confirms that it is in JSON format, and then parses it to extract the question content. The input of this process is the question data in JSON format, and the output is the analyzed question text.

[1353] Step 3:

[1354] The server sends the parsed question to multiple AI models and obtains an answer from each model. The appropriate API endpoint is used to send the question to each model. The input of this process is the parsed question text, and the output is the answer text from each AI model.

[1355] Step 4:

[1356] The server evaluates and scores the answers received from each AI model. Evaluation criteria include accuracy, reliability, and recency, and each answer is scored. The input to this process is the answer text from each model, and the output is the evaluation result and its respective score.

[1357] Step 5:

[1358] The server generates feedback for each AI model based on the evaluation results. The feedback specifically indicates areas for improvement for each model. The input to this process is the evaluation results and scores, and the output is a feedback message.

[1359] Step 6:

[1360] The server calculates the average score from the scores of each AI model and determines the most reliable answer. The input of this process is the score data, and the output is the most reliable answer text.

[1361] Step 7:

[1362] The server sends the determined most authoritative answer to the terminal and presents it to the user. The terminal receives the result from the server and displays it to the user. The input of this process is the most authoritative answer text, and the output is the answer message displayed on the user's terminal.

[1363] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[1364] The system based on this invention uses multiple artificial intelligence models to generate answers to questions entered by a user, evaluates and provides feedback on the answers, and provides the user with the most reliable answer. Furthermore, by combining it with an emotion engine that recognizes the user's emotions, it is possible to provide a more appropriate answer that takes the user's emotions into consideration. An embodiment of this system will be described in detail below.

[1365] User Input and Emotion Recognition

[1366] A user inputs a question into the device's interface. For example, suppose the user inputs the question, "How old is the universe?" At this time, the device's emotion engine also recognizes the user's emotion. The emotion engine uses methods such as voice analysis, facial expression recognition, or text analysis to generate emotion data that the user indicates when inputting.

[1367] Submitting questions and emotion data

[1368] When the user presses the send button, the device sends the question data and emotion data to the server, typically using an HTTP request, with the question data and emotion data packaged in JSON format.

[1369] Receiving and analyzing

[1370] The server receives the question data and emotion data sent from the device and analyzes them. After checking the format of the received data and confirming that it is in JSON format, it parses it to extract the question content and emotion data.

[1371] Ask a question and get an answer

[1372] The server sends the received questions to multiple artificial intelligence models. For example, it sends a question to model A and model B via API and obtains answers from each model. At this time, the server uses the appropriate API endpoint to send to each model.

[1373] Evaluating and scoring answers

[1374] The server evaluates the answers received from each AI model. Evaluation criteria include accuracy, reliability, and recency. These criteria are applied to each answer, and a score is generated. Emotional data is also taken into account in the evaluation. For example, model A's answer "13.7 billion years" is given a score of 99% accuracy and 95% reliability, while model B's answer "13.8 billion years" is given a score of 98% accuracy and 90% reliability.

[1375] Generate feedback

[1376] The server generates feedback for each AI model based on the evaluation results. The feedback includes specific areas for improvement for each model, while also taking into account the user's emotional data. For example, if the user expresses anxiety, the server may generate feedback such as "your answer is accurate, but more information is needed."

[1377] Calculating the average score and determining the final answer

[1378] The server calculates the average score from each AI model and determines the most reliable answer. For example, the median of "13.7 billion years" and "13.8 billion years" is taken to determine 13.75 billion years. Taking into account emotional data, answers that are closest to the user's needs are also given more weight.

[1379] Presenting the results to the user

[1380] The server sends the final answer to the user's device. This transmission is also performed using an HTTP request. The device receives the result from the server and displays it to the user. The user can check the presented result and obtain the necessary information. For example, the result is displayed to the user as "The age of the universe is approximately 13.75 billion years."

[1381] This system allows users to comprehensively evaluate the answers of multiple AI models to obtain the most reliable information, while also providing optimal answers that take emotional data into account. This improves the quality and reliability of the generative AI model's answers, enabling the provision of accurate and reliable information to users.

[1382] The processing flow will be explained below.

[1383] System Embodiments

[1384] The system based on this invention uses multiple artificial intelligence models to generate answers to questions entered by the user, evaluates and provides feedback on the answers, and provides the user with the most reliable answer. Furthermore, by combining it with an emotion engine that recognizes the user's emotions, it is possible to provide a more appropriate answer that takes the user's emotions into consideration. This will be explained in detail in the following processing steps.

[1385] Step 1:

[1386] A user types a question into a terminal interface. For example, the user types the question "How old is the universe?"

[1387] Step 2:

[1388] When a user inputs a question, the device's emotion engine recognizes the user's emotion using voice analysis, facial expression recognition, text analysis, etc. For example, emotion data is generated from the user's tone of voice and facial expression.

[1389] Step 3:

[1390] When the user presses the send button, the device sends the question data and emotion data to the server, typically using an HTTP request, with the question data and emotion data packaged in JSON format.

[1391] Step 4:

[1392] The server receives the question data and emotion data sent from the device and analyzes them. After checking the format of the received data and confirming that it is in JSON format, it parses it to extract the question content and emotion data.

[1393] Step 5:

[1394] The server sends the received question to multiple artificial intelligence models. For example, it sends a question to model A and model B via API and obtains an answer from each model. The server uses the appropriate API endpoint to send to each model.

[1395] Step 6:

[1396] Each AI model generates an answer to the question and returns it to the server. For example, model A answers "13.7 billion years," and model B answers "13.8 billion years." The server receives these answers.

[1397] Step 7:

[1398] The server evaluates the answers received from each AI model. Evaluation criteria include accuracy, reliability, and recency. These criteria are applied to each answer, resulting in a score. For example, model A's answer "13.7 billion years" is given a score of 99% accuracy and 95% reliability, while model B's answer "13.8 billion years" is given a score of 98% accuracy and 90% reliability. Emotional data is also taken into account in this evaluation.

[1399] Step 8:

[1400] The server generates feedback for each model based on the scoring results. The feedback includes specific areas for improvement for each model, and also takes into account the user's emotional data. For example, if the user expresses anxiety, the server generates feedback such as "The answer is accurate, but further explanation is needed."

[1401] Step 9:

[1402] The server calculates the average score from each AI model's score and determines the most reliable answer. For example, the median of "13.7 billion years" and "13.8 billion years" is taken to determine 13.75 billion years. Taking into account the user's emotional data, the server selects an answer that is optimized for the user's emotional state.

[1403] Step 10:

[1404] The server then sends the final answer to the user's device, also using an HTTP request.

[1405] Step 11:

[1406] The device displays the final answer received from the server to the user, who then confirms the information displayed on the device screen. For example, the information might read, "The age of the universe is approximately 13.75 billion years."

[1407] This system allows users to comprehensively evaluate the answers of multiple AI models to obtain the most reliable information, while also providing optimal answers that take emotional data into account. This improves the quality and reliability of the generative AI model's answers, enabling the provision of accurate and reliable information to users.

[1408] Example 2

[1409] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1410] Currently, when a user inputs a question into an AI, they must compare and consider the answers of multiple AI models to obtain a reliable answer. However, doing this individually is technically time-consuming and ultimately inconvenient for users. Furthermore, if an answer is provided without taking the user's feelings into consideration, user satisfaction decreases. Furthermore, there is a lack of functionality to provide feedback on the obtained answers and use it to improve the model, which also poses the problem of delays in improving the model itself.

[1411] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[1412] In this invention, the server includes: means for a user to input a question from a terminal; means for generating user emotion data using an emotion recognition engine of the terminal; means for receiving and analyzing the question and emotion data sent from the terminal; means for transmitting the question received by the server means to multiple artificial intelligence models and obtaining answers from each of the artificial intelligence models; means for evaluating and scoring the answers and emotion data received from each of the artificial intelligence models; means for generating feedback for each artificial intelligence model based on the evaluation results; means for calculating an average score from the scores of the artificial intelligence models and determining the most reliable answer; and means for transmitting the determined answer to the terminal and presenting it to the user. This allows the user to obtain a reliable answer and provides appropriate feedback according to the user's emotion, thereby improving satisfaction. Furthermore, since the artificial intelligence models are rapidly improved through feedback, the performance of the entire system can also be improved.

[1413] A "terminal" is a device where a user inputs a question and generates emotion data.

[1414] An "emotion recognition engine" is a technology installed in a device that analyzes a user's voice, facial expressions, text, etc. to generate emotional data.

[1415] "Emotion data" is information generated by an emotion recognition engine that expresses the emotions felt by the user when entering a question.

[1416] A "server" is a computer system that receives and analyzes question data and emotion data sent from a terminal.

[1417] An "artificial intelligence model" is a set of machine learning algorithms and / or deep learning models used to generate answers to specific questions.

[1418] An "API endpoint" is an interface for sending questions to an artificial intelligence model and obtaining answers.

[1419] "Scoring" is the process of assigning a score to the answer obtained from an artificial intelligence model based on evaluation criteria.

[1420] "Feedback" is information that provides improvements and additional information based on the evaluation results of the AI ​​model's answers.

[1421] The "average score" is the average score of the answers from multiple artificial intelligence models, and is the value used to determine the most reliable answer.

[1422] An "HTTP request" is one of the Internet protocols used to send and receive data.

[1423] The "JSON format" is a format for structuring and storing data, and is used to package question data and sentiment data.

[1424] The system based on this invention uses multiple artificial intelligence models to generate answers to questions entered by users, evaluates and provides feedback on the answers, and provides the most reliable answers to the users. Furthermore, by combining it with an emotion engine that recognizes the user's emotions, it is possible to provide more appropriate answers that take the user's emotions into consideration.

[1425] User Input and Emotion Recognition

[1426] A user inputs a question into the device's interface. For example, when a user inputs a question such as "How old is the universe?", the device's emotion recognition engine recognizes the user's emotions in real time. The emotion recognition engine generates emotion data using voice analysis, facial expression recognition, or text analysis technologies. Specifically, it captures the user's tone of voice and facial expressions and analyzes them to determine emotions such as "surprise" or "interest."

[1427] Submitting questions and emotion data

[1428] When the user presses the send button, the device sends the question data and emotion data to the server. This transmission is made via an HTTP request, and the data is packaged in JSON format. For example, the question data might include "How old is the universe?", and the emotion data might include "surprise."

[1429] Receiving and analyzing

[1430] The server receives the question data and emotion data sent from the device and analyzes them. First, it checks that the received data is in JSON format, then parses the data using a JSON library. This extracts the question content and emotion data, which are then stored in internal variables. For example, the question data is stored in the "question" variable, and the emotion data is stored in the "emotion" variable.

[1431] Ask a question and get an answer

[1432] The server sends the received question to multiple artificial intelligence models via API. For example, the server calls the API endpoints of Model A and Model B and sends the question to each model. This returns an answer from each model. For example, Model A might return the answer "13.7 billion years," and Model B might return the answer "13.8 billion years."

[1433] Evaluating and scoring answers

[1434] The server evaluates the answers received from each model, using criteria such as accuracy, reliability, and recency. For example, model A's answer "13.7 billion years" is given a score of 99% accuracy and 95% reliability. On the other hand, model B's answer "13.8 billion years" is given a score of 98% accuracy and 90% reliability. Emotional data is also taken into account in the evaluation.

[1435] Generate feedback

[1436] The server generates feedback for each model based on the evaluation results. The feedback includes suggestions for improvement and additional information for each model, and also takes into account emotional data. For example, if the user expresses surprise, the server generates feedback such as "The answer is accurate, but it would be better if you provided more detailed information."

[1437] Calculating the average score and determining the final answer

[1438] The server calculates the average score from each model's score and determines the most reliable answer. Specifically, it takes the median value of 13.7 billion years for Model A and 13.8 billion years for Model B as its answer, resulting in 13.75 billion years. It also takes into account sentiment data to provide the answer that best suits the user's needs.

[1439] Presenting the results to the user

[1440] The server sends the final answer to the user's device. This transmission also uses an HTTP request. The device receives the result from the server and displays it to the user. For example, it may say, "The age of the universe is approximately 13.75 billion years." The user can confirm this and obtain the information they need.

[1441] Prompt Sentence Examples

[1442] Here are some example prompts to input to a generative AI model:

[1443] User: How old is the universe?

[1444] Emotions: Interest, Surprise

[1445] As a result, the system of the present invention can provide high-quality information to users and improve user satisfaction by providing optimal answers that take emotion data into consideration.

[1446] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1447] Step 1:

[1448] A user inputs a question into the device's interface. For example, a user might input "How old is the universe?" This question input is received by the device as text data. Next, the device's emotion recognition engine generates the user's emotion data. The emotion recognition engine uses voice analysis, facial expression recognition, and text analysis techniques to analyze the user's emotion at the time of input. Specifically, the device's camera captures the user's facial expression and determines the emotion based on that data. The input at this time is the user's question text and real-time video data, and the output is emotion data such as "interest" or "surprise."

[1449] Step 2:

[1450] When the user presses the send button, the device sends the question data and emotion data to the server. The transmission is done using an HTTP request, and the data is packaged in JSON format. For example, the question data might include "How old is the universe?", and the emotion data might include "surprise." The input in this case is the text question and emotion data from the user, and the output is an HTTP request containing these.

[1451] Step 3:

[1452] The server receives the question data and emotion data sent from the device and analyzes them. First, it checks that the received data is in JSON format. Next, it uses a JSON library to parse the data, separates the question content and emotion data, and stores them in internal variables. For example, it stores the question data in the "question" variable and the emotion data in the "emotion" variable. The input in this case is the JSON data included in the HTTP request, and the output is the separated question text and emotion data.

[1453] Step 4:

[1454] The server sends the received question to multiple artificial intelligence models. This is done using an API, for example by calling the API endpoints of Model A and Model B to send the question. An answer is returned from each model. For example, Model A might answer "13.7 billion years," and Model B might answer "13.8 billion years." The input here is the question text, and the output is the answer from each model. Specifically, the server sends an asynchronous request to the model's API endpoint and waits for the response.

[1455] Step 5:

[1456] The server evaluates and scores the answers obtained from each model. Evaluation criteria include accuracy, reliability, and recency, and also considers emotional data. For example, model A's answer "13.7 billion years" is given a score of 99% accuracy and 95% reliability. On the other hand, model B's answer "13.8 billion years" is given a score of 98% accuracy and 90% reliability. The inputs are the answers and emotional data from each model, and the output is the evaluation results and scoring.

[1457] Step 6:

[1458] The server generates feedback for each model based on the evaluation results. The feedback includes improvements and additional information for each model, and also takes into account the user's emotional data. For example, if the user expresses surprise, the server generates feedback such as "The answer is accurate, but it would be better if you provided more detailed information." The inputs are the evaluation results and emotional data, and the output is feedback information.

[1459] Step 7:

[1460] The server calculates the average score based on the scores of each model and determines the most reliable answer. For example, take the median of 13.7 billion years for Model A and 13.8 billion years for Model B, and arrive at 13.75 billion years. The input is the scores of each model, and the output is the final reliable answer. Emotional data is also taken into account to provide the answer that best suits the user's needs.

[1461] Step 8:

[1462] The server sends the final answer to the user's device. This is also done using an HTTP request. The device receives the result and displays it to the user. For example, it might say, "The age of the universe is approximately 13.75 billion years." The input is the final answer and emotion data, and the output is the result displayed on the user's device.

[1463] (Application example 2)

[1464] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1465] Conventional answer generation systems using artificial intelligence often provide answers without considering the user's feelings, and there is a demand for a highly reliable system that can quickly respond to security-related questions and concerns. Therefore, a system that provides reliable answers that take the user's feelings into consideration is needed.

[1466] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes: means for a user to input a question from a terminal; means for receiving and analyzing the question sent from the terminal; means for sending the question received by the server means to a plurality of artificial intelligence models and obtaining answers from each of the artificial intelligence models; means for evaluating and scoring the answers received from each of the artificial intelligence models; means for generating feedback for each artificial intelligence model based on the evaluation results; means for calculating an average score from the scores of the artificial intelligence models and determining the most reliable answer; means for sending the determined answer to the terminal and presenting it to the user; emotion recognition means for recognizing the user's emotion and generating emotion data; and means for providing the most appropriate answer in consideration of the emotion data. This makes it possible to provide a reliable answer that takes the user's emotion into consideration.

[1467] "Means for users to input questions from a terminal" refers to an interface that allows users to input and send questions using a terminal such as a smartphone or computer.

[1468] "Server means for receiving and analyzing questions sent from terminals" refers to a server system that receives question data entered by users and analyzes the content of the data.

[1469] "Means for transmitting questions received by the server means to multiple artificial intelligence models and obtaining answers from each of the artificial intelligence models" refers to a function for transmitting questions received by the server to multiple artificial intelligence models and obtaining answers from each model.

[1470] "Means for evaluating and scoring the answers received from each of the AI ​​models" refers to a function for evaluating and scoring the answers obtained from each AI model based on their reliability and accuracy.

[1471] "Means for generating feedback for each artificial intelligence model based on the evaluation results" refers to a function for generating feedback including improvements for the artificial intelligence model based on the evaluation results.

[1472] "Means for calculating an average score from the scores of AI models and determining the most reliable answer" refers to a function for integrating the scores of each AI model to calculate an average score and determine the most reliable answer.

[1473] "Means for transmitting the determined answer to the terminal and presenting it to the user" refers to a function for transmitting the optimal answer determined by the server to the terminal and displaying it to the user.

[1474] "Emotion recognition means for recognizing a user's emotions and generating emotion data" refers to technology that recognizes emotions from the user's input, voice, facial expressions, etc., and generates that data.

[1475] "Means for providing the most appropriate answer taking into consideration emotional data" refers to a technology that provides the most appropriate answer based on the user's emotions, based on recognized emotional data.

[1476] "Means for inputting security questions and concerns" refers to an interface that allows a user to input security questions and concerns into a terminal.

[1477] "Means for recognizing anxiety and making an evaluation that places particular emphasis on reliability based on that emotion" refers to a function that uses emotion recognition technology to recognize the user's feelings of anxiety and makes an answer evaluation that places emphasis on reliability, taking those emotions into consideration.

[1478] A "database containing prompt sentences for an artificial intelligence model" refers to a database that stores prompt sentences used by an artificial intelligence model to generate answers.

[1479]

[1480] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be described in detail below with reference to the accompanying drawings.

[1481] The system based on the present invention provides fast and reliable answers to security-related questions and concerns. A user inputs a question from a terminal, and an emotion recognition means analyzes the user's emotions. The analysis results are sent to a server, where multiple artificial intelligence models generate answers to the question. The server evaluates each answer and presents the most reliable answer to the user.

[1482] The server is configured with various software and API endpoints. For example, emotion recognition uses a voice analysis API, facial expression recognition API, or text analysis API. As a specific example, the Emotion API (e.g., https: / / emotion-api.example.com / analyze) can be used to obtain emotional data from the user's input text. For multiple AI models, API endpoints for Model A and Model B (e.g., https: / / ai-model-a.example.com / answer and https: / / ai-model-b.example.com / answer) are used. Answers to questions are obtained through these APIs.

[1483] Next, we will explain the emotion recognition means and the artificial intelligence model acquisition means. When a user inputs a question on the terminal, the emotion recognition engine operates and analyzes either the text, voice, or facial expression to generate emotion data. If the user inputs a question such as, "I feel like there's been some suspicious activity around me recently. What should I do?", the emotion recognition engine will detect the user's anxiety. This emotion data and question text are then sent to the server.

[1484] The server sends the received question and emotion data to each AI model and obtains an answer. The answers received from each model are evaluated based on criteria such as accuracy, reliability, and recency. For example, Model A's answer "Contact the nearest police station" and Model B's answer "Install a security camera" are evaluated and a score is calculated for each. The user's emotion data is also taken into account. If anxiety is detected, reliability is emphasized.

[1485] The server then calculates the average score from each answer and determines the most reliable answer. The determined answer is sent from the server to the device and displayed to the user. This process allows the user to receive the optimal answer that takes emotions into account.

[1486] For example, the following prompt sentence is used:

[1487] "What should I do if I notice any suspicious activity in my neighborhood?"

[1488] "Please tell me what to do when you feel anxious."

[1489] By using the above means, the present invention can take into consideration the user's feelings and quickly provide the most reliable answers to security-related questions and concerns.

[1490] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1491] Step 1:

[1492] The user uses the terminal to input a question, for example, "I've noticed some suspicious activity around me recently. What should I do?" The input data is stored in text format.

[1493] Step 2:

[1494] The device sends the input question to the emotion recognition engine, which generates the user's emotion data using text analysis, voice analysis, or facial expression analysis. For example, it recognizes the emotion "anxious" from the user's text and generates emotion data. The output emotion data is in the format { "emotion": "anxious", "confidence": 0.85}.

[1495] Step 3:

[1496] The device sends the question data and emotion data to the server via an HTTP request. The data is packaged in JSON format. For example, the following data is sent: { "question": "I feel like there's been some suspicious activity around me recently. What should I do?", "emotion": "anxious", "confidence": 0.85}

[1497] Step 4:

[1498] The server receives the question data and emotion data sent from the device and parses it in JSON format. It extracts the received data and separates each field (question text, emotion, confidence level). This data is used in the next processing step.

[1499] Step 5:

[1500] The server sends the received question to multiple artificial intelligence models. Specifically, it sends the question to Model A and Model B via an API endpoint. The question data is sent via an HTTP request to each model, and the answer data is obtained. For example, if Model A says "Contact the nearby police station," Model B will respond with "Install a security camera."

[1501] Step 6:

[1502] The server evaluates the answers received from each AI model. The evaluation criteria are accuracy, reliability, and recency. Each answer is assigned a score based on these criteria. For example, the answer from model A may be assigned a score of 99% accuracy and 95% reliability, while the answer from model B may be assigned a score of 98% accuracy and 90% reliability.

[1503] Step 7:

[1504] The server generates an evaluation result by taking into account the emotional data. If the user expresses "anxiety," it places special emphasis on the reliability of the answer. For example, it adds a reliability score to Model A's answer "Contact the nearest police station" and evaluates it. The evaluation score is then assigned to each answer to generate the final score.

[1505] Step 8:

[1506] The server generates feedback for each AI model, specifically indicating areas for improvement for each model. For example, it may generate feedback such as "Accuracy is high, but the reliability of the answers has decreased."

[1507] Step 9:

[1508] The server calculates the average score from each AI model and determines the most reliable answer. For example, it combines the scores of "contact the nearest police station" and "install security cameras" and selects the most reliable answer based on the average score.

[1509] Step 10:

[1510] The server sends the final answer to the device again. The sending is done via an HTTP request, and the device presents the final answer received from the server to the user. For example, the answer "Contact the nearest police station" is displayed to the user.

[1511] This series of processes allows users to receive the most reliable answer to their input question, taking emotions into consideration.

[1512] The specific processing unit 290 transmits the result of the specific processing to the robot 414. In the robot 414, the control unit 46A causes the speaker 240 and the control target 443 to output the result of the specific processing. The microphone 238 acquires voice indicating a user input regarding the result of the specific processing. The control unit 46A transmits voice data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the voice data.

[1513] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1514] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the robot 414.

[1515] The emotion identification model 59 as an emotion engine may determine the user's emotion according to a specific mapping. Specifically, the emotion identification model 59 may determine the user's emotion according to an emotion map (see FIG. 9), which is a specific mapping. Similarly, the emotion identification model 59 may determine the robot's emotion, and the identification processing unit 290 may perform identification processing using the robot's emotion.

[1516] FIG. 9 is a diagram illustrating an emotion map 400 on which multiple emotions are mapped. In the emotion map 400, emotions are arranged in concentric circles radiating from the center. Emotions closer to the center of the concentric circles are more primitive. Emotions representing states and actions arising from a state of mind are arranged on the outer edges of the concentric circles. The concept of emotion includes both affect and mental states. Emotions generally generated from reactions occurring in the brain are arranged on the left side of the concentric circles. Emotions generally induced by situational judgment are arranged on the right side of the concentric circles. Emotions generally generated from reactions occurring in the brain and induced by situational judgment are arranged on the upper and lower sides of the concentric circles. Furthermore, the emotion of "pleasure" is arranged on the upper side of the concentric circles, and the emotion of "discomfort" is arranged on the lower side. In this way, in the emotion map 400, multiple emotions are mapped based on the structure by which emotions are generated, and emotions that tend to occur simultaneously are mapped close to each other.

[1517] These emotions are distributed in the 3 o'clock direction on emotion map 400, and typically fluctuate between relief and anxiety. In the right half of emotion map 400, situational awareness dominates over internal sensations, resulting in a sense of calm.

[1518] The inside of emotion map 400 represents what is going on in the mind, and the outside of emotion map 400 represents behavior, so the further you go outside emotion map 400, the more visible the emotions become (the more they are expressed in behavior).

[1519] Human emotions are based on various balances, such as posture and blood sugar levels. When these balances deviate from the ideal, a state of discomfort is indicated, and when they approach the ideal, a state of pleasure is indicated. Emotions can also be created for robots, automobiles, and motorcycles, based on various balances, such as posture and remaining battery life. When these balances deviate from the ideal, a state of discomfort is indicated, and when they approach the ideal, a state of pleasure is indicated. An emotion map can be generated, for example, based on Dr. Mitsuyoshi's emotion map (Research on Voice Emotion Recognition and Emotional Brain Physiological Signal Analysis Systems, Tokushima University, Doctoral Dissertation: https: / / ci.nii.ac.jp / naid / 500000375379). The left half of the emotion map lists emotions belonging to the "reaction" domain, where sensation is dominant. The right half of the emotion map lists emotions belonging to the "situation" domain, where situational awareness is dominant.

[1520] The emotion map defines two emotions that promote learning. One is a negative emotion on the situation side, around the middle of "repentance" or "reflection." In other words, this occurs when the robot experiences negative emotions such as "I never want to feel this way again" or "I don't want to be scolded again." The other is a positive emotion on the response side, around "desire." In other words, this occurs when the robot experiences positive feelings such as "I want more" or "I want to know more."

[1521] The emotion identification model 59 inputs user input into a pre-trained neural network, obtains emotion values ​​indicating each emotion shown in the emotion map 400, and determines the user's emotion. This neural network is pre-trained based on multiple pieces of training data that are combinations of user input and emotion values ​​indicating each emotion shown in the emotion map 400. Furthermore, this neural network is trained so that emotions that are located close to each other have similar values, as in the emotion map 900 shown in FIG. 10. FIG. 10 shows an example in which multiple emotions, "relieved," "calm," and "reassuring," have similar emotion values.

[1522] The system according to the present disclosure has been described above mainly with respect to the functions of the data processing device 12, but the system according to the present disclosure is not necessarily implemented on a server. The system according to the present disclosure may be implemented as a general information processing system. The present disclosure may be implemented, for example, as a software program running on a personal computer or an application running on a smartphone, etc. The method according to the present disclosure may be provided to users in the form of SaaS (Software as a Service).

[1523] In the above embodiment, an example was given in which the specific processing is performed by one computer 22, but the technology of the present disclosure is not limited to this, and the specific processing may be distributed and performed by a plurality of computers including the computer 22. For example, the data generation model 58 may be provided in an external device of the data processing device 12, and data may be generated in the external device in accordance with input data.

[1524] In the above embodiment, an example in which the specific processing program 56 is stored in the storage 32 has been described, but the technology of the present disclosure is not limited to this. For example, the specific processing program 56 may be stored in a portable, computer-readable, non-transitory storage medium such as a USB (Universal Serial Bus) memory. The specific processing program 56 stored in the non-transitory storage medium is installed in the computer 22 of the data processing device 12. The processor 28 executes the specific processing in accordance with the specific processing program 56.

[1525] 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.

[1526] It is not necessary to store all of the specific processing program 56 in a storage device such as a server connected to the data processing device 12 via the network 54, or to store all of the specific processing program 56 in the storage 32; only a portion of the specific processing program 56 may be stored.

[1527] The hardware resource for executing a specific process can be any of the following processors: An example of a processor is a CPU, which is a general-purpose processor that functions as a hardware resource for executing a specific process by executing software, i.e., a program. Another example of a processor is a dedicated electrical circuit, such as an FPGA (Field-Programmable Gate Array), a PLD (Programmable Logic Device), or an ASIC (Application Specific Integrated Circuit), which is a processor with a circuit configuration designed specifically for executing a specific process. Each processor has built-in or connected memory, and each processor uses the memory to execute the specific process.

[1528] The hardware resource that executes the specific processing may be configured with one of these various processors, or may be configured with a combination of two or more processors of the same or different types (for example, a combination of multiple FPGAs, or a combination of a CPU and an FPGA). Also, the hardware resource that executes the specific processing may be a single processor.

[1529] As an example of a system configured with a single processor, first, one processor is configured by combining one or more CPUs and software, and this processor functions as a hardware resource that executes a specific process. Second, there is a system that uses a processor that realizes the functions of an entire system including multiple hardware resources that execute a specific process on a single IC chip, as typified by SoC (System-on-a-chip). In this way, a specific process is realized using one or more of the above-mentioned various processors as hardware resources.

[1530] Furthermore, the hardware structure of these various processors can be, more specifically, an electric circuit that combines circuit elements such as semiconductor devices. The specific processing described above is merely an example. Therefore, it goes without saying that unnecessary steps may be deleted, new steps may be added, or the processing order may be rearranged, without departing from the spirit of the invention.

[1531] The above-described description and illustrations are a detailed explanation of the parts related to the technology of the present disclosure and are merely an example of the technology of the present disclosure. For example, the above description of the configuration, functions, actions, and effects is an explanation of an example of the configuration, functions, actions, and effects of the parts related to the technology of the present disclosure. Therefore, it goes without saying that unnecessary parts may be deleted, new elements may be added, or replacements may be made to the above-described description and illustrations within the scope of the gist of the technology of the present disclosure. Furthermore, to avoid confusion and facilitate understanding of the parts related to the technology of the present disclosure, the above-described description and illustrations omit explanations of common technical knowledge that do not require particular explanation to enable the implementation of the technology of the present disclosure.

[1532] All publications, patent applications, and technical standards mentioned in this specification are herein incorporated by reference to the same extent as if each individual publication, patent application, or technical standard was specifically and individually indicated to be incorporated by reference.

[1533] The following is further disclosed regarding the above embodiment.

[1534] (Claim 1)

[1535] A means for a user to input a question from a terminal;

[1536] a server means for receiving and analyzing questions sent from the terminal;

[1537] means for transmitting the question received by the server means to a plurality of artificial intelligence models and obtaining answers from each of the artificial intelligence models;

[1538] means for evaluating and scoring answers received from each of said artificial intelligence models;

[1539] a means for generating feedback for each artificial intelligence model based on the evaluation results;

[1540] means for calculating an average score from the scores of the artificial intelligence model and determining the most reliable answer;

[1541] means for transmitting the determined answer to a terminal and presenting it to a user;

[1542] A system including:

[1543] (Claim 2)

[1544] 10. The system of claim 1, further comprising means for evaluating the answers of the artificial intelligence model based on accuracy, reliability, and currency.

[1545] (Claim 3)

[1546] 2. The system of claim 1, wherein the feedback generating means generates feedback that specifically indicates areas for improvement for each artificial intelligence model.

[1547] "Example 1"

[1548] (Claim 1)

[1549] a device for allowing a user to input a question from a terminal;

[1550] an information processing device that receives and analyzes the question sent from the terminal;

[1551] a device for transmitting the question received by the information processing device to a plurality of artificial intelligence systems and obtaining answers from each of the artificial intelligence systems;

[1552] a device for evaluating and quantifying responses received from each of said artificial intelligence systems;

[1553] a device for generating feedback for each AI system based on the evaluation results;

[1554] a device for calculating an average value from the values ​​of the artificial intelligence system and determining the most reliable answer;

[1555] a device for transmitting the determined answer to a terminal and presenting the answer to a user;

[1556] A system including:

[1557] (Claim 2)

[1558] 10. The system of claim 1, further comprising a device for evaluating the answers of the artificial intelligence system based on accuracy, reliability, and currency.

[1559] (Claim 3)

[1560] 2. The system of claim 1, wherein the feedback generator generates feedback that specifically indicates areas for improvement for each artificial intelligence system.

[1561] "Application Example 1"

[1562] (Claim 1)

[1563] A means for a user to input a question from a terminal;

[1564] an information processing device that receives and analyzes questions sent from the terminal;

[1565] means for transmitting the question received by the information processing device means to a plurality of artificial intelligence models and obtaining answers from each of the artificial intelligence models;

[1566] means for evaluating and scoring answers received from each of said artificial intelligence models;

[1567] a means for generating feedback for each artificial intelligence model based on the evaluation results;

[1568] means for calculating an average score from the scores of the artificial intelligence model and determining the most reliable answer;

[1569] means for transmitting the determined answer to a terminal and presenting it to a user;

[1570] A means for adding a function to a food delivery application that generates answers to questions from users using multiple artificial intelligence models and provides the most reliable answers;

[1571] A system including:

[1572] (Claim 2)

[1573] 10. The system of claim 1, further comprising means for evaluating the answers of the artificial intelligence model based on accuracy, reliability, and currency.

[1574] (Claim 3)

[1575] 2. The system of claim 1, wherein the feedback generating means generates feedback that specifically indicates areas for improvement for each artificial intelligence model.

[1576] "Example 2: Combining Emotion Engines"

[1577] (Claim 1)

[1578] A means for a user to input a question from a terminal;

[1579] means for generating user emotion data by an emotion recognition engine of the terminal;

[1580] a server means for receiving and analyzing questions and emotion data transmitted from the terminal;

[1581] means for transmitting the question received by the server means to a plurality of artificial intelligence models and obtaining answers from each of the artificial intelligence models;

[1582] means for evaluating and scoring the responses and sentiment data received from each of said artificial intelligence models;

[1583] a means for generating feedback for each artificial intelligence model based on the evaluation results;

[1584] means for calculating an average score from the scores of the artificial intelligence model and determining the most reliable answer;

[1585] means for transmitting the determined answer to a terminal and presenting it to a user;

[1586] A system including:

[1587] (Claim 2)

[1588] 10. The system of claim 1, further comprising means for evaluating the answers of the artificial intelligence model based on accuracy, reliability, and currency.

[1589] (Claim 3)

[1590] 2. The system of claim 1, wherein the feedback generating means generates feedback that specifically indicates areas for improvement for each artificial intelligence model.

[1591] "Application example 2 when combining emotion engines"

[1592] (Claim 1)

[1593] A means for a user to input a question from a terminal;

[1594] a server means for receiving and analyzing questions sent from the terminal;

[1595] means for transmitting the question received by the server means to a plurality of artificial intelligence models and obtaining answers from each of the artificial intelligence models;

[1596] means for evaluating and scoring answers received from each of said artificial intelligence models;

[1597] a means for generating feedback for each artificial intelligence model based on the evaluation results;

[1598] means for calculating an average score from the scores of the artificial intelligence model and determining the most reliable answer;

[1599] means for transmitting the determined answer to a terminal and presenting it to a user;

[1600] emotion recognition means for recognizing an emotion of a user and generating emotion data;

[1601] means for providing the most appropriate answer in consideration of said emotion data;

[1602] A system including:

[1603] (Claim 2)

[1604] 10. The system of claim 1, further comprising means for evaluating the answers of the artificial intelligence model based on accuracy, reliability, and currency.

[1605] (Claim 3)

[1606] 2. The system of claim 1, wherein the feedback generating means generates feedback that specifically indicates areas for improvement for each artificial intelligence model.

[1607] (Claim 4)

[1608] 2. The system according to claim 1, wherein when a user inputs a question or anxieties about security, the emotion recognition means recognizes the user's anxieties and performs an evaluation that places particular emphasis on trustworthiness based on the emotions.

[1609] (Claim 5)

[1610] 10. The system of claim 1, further comprising means for selecting answers to security questions and concerns from a database containing prompts from an artificial intelligence model.

[1611] [Explanation of symbols]

[1612] 10, 210, 310, 410 Data Processing Systems 12 Data Processing Device 14 Smart Devices 214 Smart Glasses 314 Headset-type terminal 414 Robot< / url:> < / url:> < / url:> < / url:>

Claims

1. A means for a user to input a question from a terminal; a server means for receiving and analyzing questions sent from the terminal; means for transmitting the question received by the server means to a plurality of artificial intelligence models and obtaining answers from each of the artificial intelligence models; means for evaluating and scoring answers received from each of said artificial intelligence models; a means for generating feedback for each artificial intelligence model based on the evaluation results; means for calculating an average score from the scores of the artificial intelligence model and determining the most reliable answer; means for transmitting the determined answer to a terminal and presenting it to a user; A system including:

2. The system of claim 1 , further comprising means for evaluating the answers of the artificial intelligence model based on accuracy, reliability, and currency.

3. 2. The system of claim 1, wherein the feedback generating means generates feedback that specifically indicates improvements to each artificial intelligence model.

Images