system

Abstract

Provide a system. An input means for a user to input a question or problem; an answer generation means for generating an answer using a generation AI based on the question or problem sent from the input means; evaluation means for evaluating the reliability of the answer generated by the answer generation means; a recommendation means for recommending an additional verification method based on the likelihood evaluated by the evaluation means; a display means for displaying to a user the answer by the answer generating means, the evaluation of the likelihood by the evaluating means, and the verification method by the recommending means; A system including:

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] Answers provided by generative AI may not always be accurate, making it difficult for users to judge their reliability. Therefore, there is a need for a system that appropriately recommends methods and resources for users to verify the reliability of generated answers. The present invention aims to solve this problem by providing a system that enables users to efficiently judge the reliability of generative AI answers. [Means for solving the problem]

[0005] The present invention is a system including an input means for a user to input a question or problem, an answer generation means for generating an answer using a generation AI based on the question or problem sent from the input means, an evaluation means for evaluating the reliability of the answer generated by the answer generation means, a recommendation means for recommending an additional verification method based on the reliability evaluated by the evaluation means, and a display means for displaying to the user the answer generated by the answer generation means, the reliability evaluation by the evaluation means, and the verification method recommended by the recommendation means. This allows the user to check the reliability of the generated answer and, if necessary, perform additional verification using a reliable external database or resource.

[0006] An "input means" is a device or interface through which a user inputs a question or problem in text form.

[0007] An "answer generation means" is a device or system that uses a generation AI to create answers to users' questions or problems.

[0008] The "evaluation means" is a device or system that evaluates the reliability of the generated answer and calculates the reliability as a score.

[0009] The "recommendation means" is a device or system that suggests an appropriate verification method to the user based on the likelihood score obtained by the evaluation means.

[0010] The "display means" is a device or interface that visually displays to the user the answer created by the answer creation means, the evaluation of the likelihood by the evaluation means, and the verification method by the recommendation means.

[0011] "Generative AI" is an artificial intelligence model that automatically generates answers in natural language based on user input.

[0012] A "question" is text data that a user inputs to express the information they want to know or the problem they want to solve.

[0013] "Confidence" is a measure of the likelihood that the generated answer is accurate, usually expressed as a percentage.

[0014] "External Databases" are external information sources or resources used to provide reliable information.

[0015] "Resources" are sources of information such as documents, databases, and expert articles that can be used to verify the reliability of information. [Brief explanation of the drawings]

[0016] [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

[0017] 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.

[0018] First, the terms used in the following description will be explained.

[0019] 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).

[0020] 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.

[0021] 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.

[0022] 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.

[0023] 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."

[0024] [First embodiment]

[0025] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0026] 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.

[0027] 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).

[0028] 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.

[0029] 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.

[0030] 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.

[0031] 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.

[0032] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0033] 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.

[0034] 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.

[0035] 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.

[0036] 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."

[0037] The system of the present invention executes a series of processes in which a user inputs a question or problem, and a generation AI generates an answer based on that information, evaluates its reliability, and recommends an appropriate verification method.

[0038] User Input

[0039] A user uses a terminal to input a question or problem in text format. For example, the user might input "What is the diameter of the Earth?" This input is sent from the terminal to a server. This process is performed by sending the text data entered by the user to the server as an HTTP request.

[0040] Server-side answer generation

[0041] 1. Request received:

[0042] The server receives an HTTP request sent from a user terminal and extracts the question text. For example, if the request data is "What is the diameter of the Earth?", the server analyzes it and extracts it.

[0043] 2. Enter a question into the generative AI model:

[0044] The extracted question text is input into a generative AI model, which generates an answer based on this question. For example, the generated answer might be "The diameter of the Earth is approximately 12,742 km."

[0045] 3. Rate the confidence of your answer:

[0046] The server uses a model to calculate the likelihood of the generated answer, for example rating it as having a likelihood score of 80%.

[0047] Verification method recommendations

[0048] 1. Recommendation generation based on evaluation results:

[0049] The server determines the verification method to recommend to the user based on the reliability score. For example, if the reliability score is less than 90%, the server will recommend specific verification methods such as "check the NASA database" or "read the Encyclopaedia Britannica article."

[0050] 2. Result format:

[0051] Finally, we format the data to include the answer, a confidence score, and a recommended verification method. For example, the formatted result might look like this:

[0052] Answer: "The diameter of the Earth is approximately 12,742 km."

[0053] Probability: 80%

[0054] Recommended verification method:

[0055] Check the NASA database

[0056] Read the Encyclopaedia Britannica article

[0057] Output to the user

[0058] The server sends the formatted results as an HTTP response to the user's device, which then displays the received data in an easy-to-understand format for the user, who can see the generated answer, its accuracy, and the recommended verification method.

[0059] For example, the user receives the following result:

[0060] Answer: "The diameter of the Earth is approximately 12,742 km."

[0061] Probability: 80%

[0062] Recommended verification method:

[0063] Check the NASA database

[0064] Read the Encyclopaedia Britannica article

[0065] In this way, the system of the present invention helps users efficiently determine the accuracy of answers provided by the generating AI and, if necessary, verify them with more reliable sources of information.

[0066] The processing flow will be explained below.

[0067] Step 1:

[0068] The user enters a question or problem into the terminal. The user enters the question in text format using the touchscreen or keyboard and presses the submit button. At this time, the user's input is sent to the server as an HTTP POST request.

[0069] Step 2:

[0070] The terminal sends the user's input to the server. The terminal sends the text data entered by the user to the server as an HTTP request. For example, if the request data is "What is the diameter of the Earth?", this is sent to the server.

[0071] Step 3:

[0072] The server receives the request. The server receives the HTTP request sent from the device, analyzes the request data, and extracts the question text. Specifically, it extracts it using a method such as request.get("question").

[0073] Step 4:

[0074] The server inputs a question into the generative AI model. The extracted question text is input into the generative AI to generate an answer. If the generative AI generates an answer such as "The diameter of the Earth is approximately 12,742 km," this is accepted.

[0075] Step 5:

[0076] The server evaluates the reliability of the answer. A reliability evaluation model is used to evaluate the reliability of the generated answer. Specifically, the generated answer is input into the evaluation model and a calculated score is obtained. For example, suppose the evaluation result is 80%.

[0077] Step 6:

[0078] The server generates recommendations based on the evaluation results. The server recommends appropriate verification methods to the user based on the reliability score. If the score is less than 90%, the server recommends viewing reliable external databases or articles. For example, the server generates specific recommendations such as "Check the NASA database" or "Read the Encyclopaedia Britannica article."

[0079] Step 7:

[0080] The server formats the results. The server formats the generated answer, confidence score, and recommended verification method together, for example in JSON format like this:

[0081] {

[0082] "answer": "The diameter of the Earth is approximately 12742 km",

[0083] "certainty_score": 80,

[0084] "recommendations": ["Check out the NASA database", "Read the Encyclopaedia Britannica article"]

[0085] }

[0086] Step 8:

[0087] The server sends the formatted results to the terminal. The server sends the formatted data to the user terminal as an HTTP response.

[0088] Step 9:

[0089] The terminal displays the result to the user. The terminal analyzes the received response data and displays it in an easy-to-understand manner for the user. For example, the user receives the result as follows:

[0090] Answer: "The diameter of the Earth is approximately 12,742 km."

[0091] Probability: 80%

[0092] Recommended verification method:

[0093] Check the NASA database

[0094] Read the Encyclopaedia Britannica article

[0095] The user can then perform further verification based on this.

[0096] Example 1

[0097] 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."

[0098] In conventional question-answering systems, users have little means to judge the accuracy and reliability of the generated answers, which creates the risk of receiving incorrect information. Furthermore, users must independently verify the authenticity of the generated answers, which is time-consuming and laborious. Furthermore, due to the lack of functionality to evaluate the reliability of information and recommend appropriate verification methods, users have difficulty finding additional reliable external information sources.

[0099] 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.

[0100] In this invention, the server includes input means for a user to input a question or problem, answer generation means for receiving the question or problem sent from the input means, inputting an inquiry generated based on the question or problem into a generative AI model, and generating an answer from the model, evaluation means for evaluating the reliability of the answer generated by the answer generation means, recommendation means for recommending an additional verification method based on the reliability evaluated by the evaluation means, and display means for formatting and displaying to the user the answer generated by the answer generation means, the reliability evaluation by the evaluation means, and the verification method recommended by the recommendation means. This saves the user the trouble of verifying the reliability of the generated answer, and enables additional verification using a reliable information source.

[0101] "Input means" refers to the means by which a user inputs questions or problems to the system.

[0102] An "answer generation means" is a means for generating a corresponding answer using a generative AI model based on a question or problem from a user.

[0103] The "evaluation means" is a means for calculating and evaluating the likelihood of the generated answer.

[0104] The "recommendation means" is a means for recommending an additional verification method to the user based on the likelihood evaluated by the evaluation means.

[0105] The "display means" is a means for formatting the answer by the answer generation means, the evaluation of the likelihood by the evaluation means, and the verification method by the recommendation means, and displaying them in an easy-to-see manner for the user.

[0106] A "generative AI model" is an artificial intelligence model used to generate answers in natural language based on a user's question or problem.

[0107] "Likelihood" is an evaluation value that indicates the accuracy and reliability of the answer provided by the generative AI model.

[0108] "Validation methods" are additional steps or reference sources used to confirm the accuracy of the answers provided by a generative AI model.

[0109] "Network connection" refers to the communication environment for accessing generative AI models and external information sources.

[0110] This invention is a system in which a user inputs a question or problem, a generative AI model generates an answer based on that information, evaluates its reliability, and recommends an appropriate verification method. The system is composed of a server, terminals, and a communication network.

[0111] System Configuration

[0112] The system mainly consists of the following components:

[0113] 1. Input medium: The interface through which a user enters a question or problem. This can include a web browser or a mobile app.

[0114] 2. Answer generation means: Receives questions or problems from users and generates answers using a generative AI model. The generative AI model can be, for example, a general artificial intelligence model that excels at text generation.

[0115] 3. Evaluation method: The likelihood of the generated answer is calculated and evaluated. Here, a probabilistic evaluation algorithm is used.

[0116] 4. Recommendation: Based on the reliability assessment results, the system recommends additional verification methods to the user. These recommendations are based on databases and reliable external information sources.

[0117] 5. Display: The answers, the confidence scores, and the recommended verification methods are formatted and displayed to the user in an easy-to-read format on the device screen.

[0118] Specific examples

[0119] A user uses a terminal to input a question such as: "What is the diameter of the Earth?" This input is sent to the server as an HTTP request.

[0120] Server Processing

[0121] The server receives an HTTP request sent from the user's device and extracts the question text from it. The extracted question is input into a generative AI model, which then generates an answer to the question. For example, the generated answer might be, "The diameter of the Earth is approximately 12,742 km."

[0122] The generated answer is fed into an algorithm to calculate its likelihood. The algorithm rates the answer as 80% reliable. Based on this rating, it recommends additional verification methods. Because the score is below 90%, it recommends things like "checking the NASA database" or "reading the Encyclopaedia Britannica article."

[0123] Finally, the answer, confidence score, and recommended verification method are formatted into a text like this:

[0124] Answer: "The diameter of the Earth is approximately 12,742 km."

[0125] Probability: 80%

[0126] Recommended verification method:

[0127] Check the NASA database

[0128] Read the Encyclopaedia Britannica article

[0129] What the user sees

[0130] The server sends the formatted results as an HTTP response to the user's device, which receives it and displays it in an easy-to-understand manner for the user.

[0131] In this way, the system of the present invention helps users efficiently determine the accuracy of answers provided by the generating AI and, if necessary, verify them with reliable sources of information.

[0132] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0133] System program processing flow

[0134] Step 1: User Input

[0135] Input: The user types a question or problem into the terminal.

[0136] Output: Text data of the question or problem entered.

[0137] The user inputs a question or problem through the device's interface. For example, they might type, "What is the diameter of the Earth?" The input data is sent to the server as an HTTP request. When the user types a question into the input form and presses the "Submit" button, the device converts the text data into JSON format and sends it to the server.

[0138] Step 2: Receiving the request

[0139] Input: HTTP request sent from the user's device.

[0140] Output: The extracted question text.

[0141] The server receives the HTTP request sent from the user's device and extracts the question text. It then analyzes the JSON format data from the received HTTP request to obtain the specific question. For example, the extracted question might be "What is the diameter of the Earth?"

[0142] Step 3: Input to the generative AI model

[0143] Input: The extracted question text.

[0144] Output: The answer text returned by the generative AI model.

[0145] The server inputs the extracted question text into the generative AI model, which then generates an answer based on this question. Specifically, the prompt "What is the diameter of the Earth?" is input to the generative AI model, and the model responds with an answer such as "The diameter of the Earth is approximately 12,742 km."

[0146] Step 4: Evaluate the likelihood of your answer

[0147] Input: The answer text returned by the generative AI model.

[0148] Output: A likelihood score.

[0149] The server evaluates the reliability of the generated answer. The generated answer is input into the evaluation model and the reliability is calculated. For example, the reliability score is calculated as 80%. Specifically, the evaluation is performed using an algorithm that calculates a reliability score.

[0150] Step 5: Recommend a verification method

[0151] Input: A likelihood score.

[0152] Output: A list of recommended validation methods.

[0153] The server determines the verification method to recommend to the user based on the reliability score. For example, if the score is less than 90%, specific verification methods such as "check the NASA database" or "read the Encyclopaedia Britannica article" are recommended. This is done based on databases and reliable sources.

[0154] Step 6: Formatting the results

[0155] Input: Answer text, confidence score, and a list of recommended verification methods.

[0156] Output: Formatted result data.

[0157] The server formats the answer, confidence score, and recommended verification method, perhaps in a format like JSON or plain text, for example:

[0158] Answer: "The diameter of the Earth is approximately 12,742 km."

[0159] Probability: 80%

[0160] Recommended verification method:

[0161] Check the NASA database

[0162] Read the Encyclopaedia Britannica article

[0163] Step 7: Output to the User

[0164] Input: Formatted result data.

[0165] Output: The result that is displayed to the user.

[0166] The server sends the formatted results as an HTTP response to the user's device. The device then displays the received data in an easy-to-understand format for the user. Specifically, the results are displayed on the screen of a web browser or mobile app. The user can view the generated answer, its accuracy, and the recommended verification method.

[0167] (Application example 1)

[0168] 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."

[0169] In today's world, phishing emails are becoming increasingly sophisticated, making them extremely difficult to identify. Conventional phishing email countermeasures require users to check the emails themselves, which often leads to oversights or incorrect judgments. Furthermore, judgments may be made based on unreliable information. Therefore, there is a need for a system that can quickly and accurately evaluate the reliability of emails received by users and recommend necessary verification methods.

[0170] 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.

[0171] In this invention, the server includes: input means for a user to input a question or problem; answer generation means for generating an answer using a generation AI based on the question or problem sent from the input means; evaluation means for evaluating the reliability of the answer generated by the answer generation means; recommendation means for recommending an additional verification method based on the reliability evaluated by the evaluation means; display means for displaying to the user the answer generated by the answer generation means, the reliability evaluation by the evaluation means, and the verification method recommended by the recommendation means; reliability evaluation means for evaluating the reliability of email text; and warning means for notifying the user that the reliability evaluation means has determined that the email is likely to be a phishing email. This allows the user to quickly determine the reliability of received emails and prevent damage from phishing emails.

[0172] "User" means any person or entity that utilizes the System to enter questions or problems and receive results.

[0173] "Input means" refers to a device or interface that allows a user to input a question or problem in text form.

[0174] An "answer generation means" is a process or system for generating answers using a generation AI based on questions or problems entered by a user.

[0175] An "evaluation method" is a process or system for evaluating the reliability of answers created by generative AI.

[0176] A "recommendation means" is a process or system for recommending an appropriate verification method to a user based on the assessed likelihood.

[0177] A "display means" is a device or interface for visually displaying to a user the generated answer, the likelihood assessment, and the recommended verification method.

[0178] A "trustworthiness assessment means" is a process or system for assessing the trustworthiness of the text of a received email and determining whether it is a phishing email.

[0179] The "warning means" is a process or system for notifying the user when the reliability evaluation means determines that the email is highly likely to be a phishing email.

[0180] A "phishing email" is a fraudulent email sent to steal personal or confidential information from users for malicious purposes.

[0181] This paper describes how to build a system that can quickly and accurately evaluate the reliability of emails received by users and warn them of phishing emails as necessary. This system utilizes a generative AI model to evaluate the reliability of emails and recommend appropriate measures to users.

[0182] The server includes an input means for a user to input a question or problem, an answer generation means for generating an answer using a generation AI based on the question or problem sent from the input means, an evaluation means for evaluating the reliability of the answer generated by the answer generation means, and a recommendation means for recommending an additional verification method based on the reliability evaluated by the evaluation means.The server further includes a display means for displaying to the user the answer generated by the answer generation means, the reliability evaluation by the evaluation means, and the verification method recommended by the recommendation means, a reliability evaluation means for evaluating the reliability of the email text, and a warning means for notifying the user that the email may be a phishing email based on the reliability evaluation means.

[0183] Specifically, the user inputs the text of the email they received and sends it to the server. This input is done via the user's device (smartphone, tablet, PC, etc.). In this process, the email body is input in text format and sent to the server as an HTTP request. For example, the user inputs the following email text:

[0184] "The email said, 'Your account has been locked due to a security issue. Please click the link to verify your account immediately.'"

[0185] The server receives the data sent from the user's device and evaluates the reliability of the email using a generative AI model, such as one from Hugging Face's Transformers library. This model determines the likelihood that the email is a phishing email. For example, it generates a response such as "The email is likely to be phishing."

[0186] The server then recommends appropriate verification methods to the user based on the assessed trustworthiness score. For emails with low trustworthiness, additional verification methods may be recommended. For example, a recommendation such as "Please check official security guidelines" may be given.

[0187] Finally, this information is sent to the user's terminal, where the user can check the results in text format, making it easy for the user to quickly determine the authenticity of the received email and perform further checks if necessary.

[0188] This system allows users to confirm the authenticity of emails in a reliable way, even in an age where it is difficult to distinguish phishing emails.When commercializing the product, it is recommended that devices be standard smartphones, tablets, or PCs, and that software utilize Python or a framework that can handle HTTP requests.

[0189] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0190] Step 1: User enters email text

[0191] The user inputs the text of the received email into the terminal and sends it to the server. The input data could be, for example, the text "Your account has been locked due to a security issue. Please click the link immediately to confirm." This input is sent to the server as an HTTP request.

[0192] Step 2: The server receives the input

[0193] The server receives the HTTP request sent from the user terminal and extracts the email text. For example, it extracts the text "Your account has been locked due to a security issue. Please click the link immediately to confirm." from the request data.

[0194] Step 3: Generative AI reliability assessment

[0195] The server inputs the extracted email text into a generative AI model to evaluate its reliability. The generative AI model uses, for example, Hugging Face's Transformers library to determine whether the email is a phishing email. The model generates an answer such as "This is likely a phishing email." This process uses natural language processing techniques to analyze the input data and calculate a reliability score.

[0196] Step 4: Evaluate the reliability score

[0197] The server analyzes the reliability score provided by the generative AI model and evaluates its reliability. For example, if the reliability score is evaluated as 80%, the next processing step will proceed based on that score. A score analysis algorithm is used for this data processing.

[0198] Step 5: Recommend a verification method

[0199] The server recommends appropriate verification methods to the user based on the reliability score. For example, if the reliability score is low, the server provides a recommendation to "check official security guidelines." This process links the results of the reliability assessment to the user's countermeasure actions.

[0200] Step 6: Formatting the results

[0201] The server then combines the generated answers, confidence ratings, and recommended verification methods into a single format, such as the following:

[0202] "It's likely a phishing email."

[0203] Reliability score: 80%

[0204] Recommended verification: Check official security guidelines

[0205] Step 7: Displaying results to the user

[0206] The device receives the formatted results sent from the server and displays them to the user, who can view the generated answer, its confidence rating, and recommended verification methods, for example, displayed on a smartphone screen.

[0207] Each processing step has been explained in detail above. This system makes it easy for users to quickly determine the authenticity of received emails and to verify them as needed.

[0208] 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.

[0209] The system of the present invention uses generative AI to provide answers to questions or problems entered by users, evaluate their accuracy, and recommend appropriate verification methods. It also incorporates an emotion engine that recognizes emotions from the user's input, adjusts answers based on those emotions, and suggests verification methods.

[0210] User Input

[0211] A user uses a terminal to input a question or problem in text format. For example, a user might type, "The weather has been bad lately. What will happen this weekend?" This input is sent from the terminal to the server. It is sent as an HTTP POST request.

[0212] emotion recognition

[0213] 1. Request received:

[0214] The server receives the HTTP request sent from the terminal, analyzes the request data, and extracts the question text.

[0215] 2. Emotion recognition using emotion engine:

[0216] The extracted text is input into the emotion engine, which analyzes the user's emotions. For example, a question like "The weather has been bad lately. What will happen this weekend?" can be used to identify "negative emotions."

[0217] Answer generation and likelihood assessment

[0218] 1. Enter a question into the generative AI model:

[0219] After recognizing the emotion, the question text is fed into a generative AI model to generate an answer. For example, the generative AI might respond, "It's likely to be sunny over the weekend."

[0220] 2. Adjusting your answers:

[0221] It adjusts the tone and wording of responses based on emotion recognition. If a negative emotion is detected, it will adjust the response, for example, "Don't worry, it's likely to be sunny over the weekend."

[0222] 3. Assessing likelihood:

[0223] The server uses a model to assess the likelihood of the generated answer and calculates a confidence score. For example, suppose the assessment result is 85%.

[0224] Verification method recommendations

[0225] 1. Recommendation generation based on evaluation results:

[0226] Based on the reliability score, the server recommends appropriate verification methods to the user. If the score is below 90%, it suggests checking reliable external databases or expert articles. For example, it recommends "checking weather forecasting services" or "referring to the official website of the Meteorological Agency."

[0227] 2. Result format:

[0228] The server formats the generated answer, the confidence score, and the recommended verification method, for example in JSON format as follows:

[0229] {

[0230] "answer": "Don't worry, it's likely to be sunny over the weekend",

[0231] "certainty_score": 85,

[0232] "recommendations": ["Check the weather forecast service", "See the official Meteorological Agency website"]

[0233] }

[0234] Output to the user

[0235] The server sends the formatted results as an HTTP response to the user's device, which then parses the response data and displays it in a user-friendly format. The user can see the generated answer, its accuracy, and the recommended verification method.

[0236] For example, the user receives the following result:

[0237] Answer: "Don't worry, it's likely to be sunny over the weekend."

[0238] Probability: 85%

[0239] Recommended verification method:

[0240] Check the weather forecast service

[0241] Visit the official Meteorological Bureau website

[0242] In this way, the system of the present invention helps users efficiently judge the accuracy of the answers generated by the AI ​​and, if necessary, allows them to conduct additional verification using more reliable external sources. It also improves the user experience by adjusting the answers to take into account the user's feelings.

[0243] The processing flow will be explained below.

[0244] Step 1:

[0245] The user types a question or problem into the device. The user uses a keyboard or touchscreen to type the question in text format and clicks a submit button. For example, the user might type, "The weather has been bad lately. What's it going to be like this weekend?"

[0246] Step 2:

[0247] The device sends the user's input to the server. The device sends the entered text data to the server as an HTTP POST request. The data sent is in the format {"question": "The weather has been bad lately. What will happen this weekend?"}

[0248] Step 3:

[0249] The server receives the request. The server receives the HTTP request, analyzes its contents, and extracts the question text. Specifically, the question is extracted using request.get("question") .

[0250] Step 4:

[0251] The server inputs the question text into the emotion engine. The server then passes the extracted question text to the emotion engine, which analyzes the user's emotion. For example, the emotion engine recognizes "negative emotion."

[0252] Step 5:

[0253] The server inputs a question into the generative AI model. Based on the results of analysis by the emotion engine, the question text is input into the generative AI model to generate an answer. For example, the generative AI generates the answer "It's likely to be sunny over the weekend."

[0254] Step 6:

[0255] The server adjusts the tone of the generated answer. The tone and expression of the answer are adjusted appropriately depending on the user's emotions. For example, if a negative emotion is detected, the server adjusts the answer to "Don't worry. It's likely to be sunny over the weekend."

[0256] Step 7:

[0257] The server evaluates the reliability of the generated answer. The server inputs the answer into a reliability evaluation model and obtains a reliability score. For example, suppose the evaluation result is 85%.

[0258] Step 8:

[0259] The server generates recommendations based on the evaluation results. If the reliability score is less than 90%, it recommends using a more reliable external database or resource. For example, it suggests "checking the weather forecast service" or "referring to the official website of the Japan Meteorological Agency."

[0260] Step 9:

[0261] The server formats the results: the generated answer, the confidence score, and the recommended verification method in JSON format, for example:

[0262] {

[0263] "answer": "Don't worry, it's likely to be sunny over the weekend",

[0264] "certainty_score": 85,

[0265] "recommendations": ["Check the weather forecast service", "See the official Meteorological Agency website"]

[0266] }

[0267] Step 10:

[0268] The server sends the formatted results to the terminal. The server sends the formatted data to the user terminal as an HTTP response.

[0269] Step 11:

[0270] The terminal displays the result to the user. The terminal analyzes the received response data and displays it in an easy-to-understand manner for the user. For example, the user receives the result as follows:

[0271] Answer: "Don't worry, it's likely to be sunny over the weekend."

[0272] Probability: 85%

[0273] Recommended verification method:

[0274] Check the weather forecast service

[0275] Visit the official Meteorological Bureau website

[0276] Users can then use this information to verify the information with reliable external sources.

[0277] Example 2

[0278] 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."

[0279] Conventional question-answering systems lacked a means for evaluating the reliability of generated answers or a mechanism for adjusting answers based on user emotions. This made it difficult for users to judge the reliability of generated answers, potentially leading to actions based on unreliable information. Furthermore, there were concerns that one-sided answers that did not take the user's emotions into consideration could detract from the user experience. In response to this, the present invention aims to provide more reliable information and improve the user experience by combining reliability evaluation and emotion recognition in an answer generation system using a generative AI model.

[0280] 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.

[0281] In this invention, the server includes: input means for a user to input a question or problem; answer generation means for generating an answer using a generation AI based on the question or problem transmitted from the input means; evaluation means for evaluating the reliability of the answer generated by the answer generation means; recommendation means for recommending an additional verification method based on the reliability evaluated by the evaluation means; emotion recognition means for recognizing emotions in the user's input text and adjusting the answer based on the result; and display means for displaying to the user the answer generated by the answer generation means, the reliability evaluation by the evaluation means, and the verification method recommended by the recommendation means. This makes it possible to evaluate the reliability of the generated answer, provide the user with reliable information, and provide answers that take the user's emotions into consideration, thereby improving the user experience.

[0282] An "input means" is a device or software interface through which a user inputs a question or problem in text form.

[0283] An "answer generation means" is a device or software function that uses generation AI to generate appropriate answers based on questions or problems obtained from an input means.

[0284] The "evaluation means" is a device or software function that evaluates the likelihood of the answer generated by the answer generation means and calculates the reliability as a score.

[0285] The "recommendation means" is a device or software function that recommends additional verification methods to the user based on the likelihood evaluated by the evaluation means.

[0286] An "emotion recognizer" is a device or software function that recognizes the emotion of a user's input text and adjusts the response based on the results.

[0287] The "display means" is a device or software interface that displays to the user the answer from the answer generation means, the evaluation of the likelihood from the evaluation means, and the verification method from the recommendation means.

[0288] MODE FOR CARRYING OUT THE INVENTION

[0289] The system of the present invention uses generative AI to provide answers to questions or problems entered by users, evaluate their accuracy, and recommend appropriate verification methods. This system also incorporates an emotion engine that recognizes the user's emotions and adjusts the answer accordingly.

[0290] Input Method

[0291] A user enters a question or problem in text format using a device. This device can be a PC, smartphone, tablet, etc. For example, they might enter, "The weather has been bad lately. What will happen this weekend?" This input is sent to the server as an HTTP POST request.

[0292] emotion recognition means

[0293] The server receives the HTTP request sent from the device, analyzes the request data, and extracts the question text. The emotion engine then analyzes the user's emotions. The emotion engine uses natural language processing (NLP) technology to recognize "negative emotions," for example, from a question like, "The weather has been bad lately. What will happen this weekend?"

[0294] Answer generation means

[0295] After recognizing the emotion, the server inputs the question text into the generative AI model and generates an answer. For example, if the prompt is "Tell me what the weather will be like this weekend," the generative AI model will respond with "It's likely to be sunny this weekend."

[0296] Answer adjustment

[0297] The tone and wording of the generated response is adjusted based on emotion recognition: if a negative emotion is detected, the response will be adjusted, such as "Don't worry, it's likely to be sunny over the weekend."

[0298] Certainty assessment method

[0299] The server uses a proprietary rating model to evaluate the reliability of the generated answer and calculates a reliability score. For example, suppose the rating result is 85%.

[0300] Recommendation methods

[0301] Based on the confidence score of the generated answer, the server recommends an appropriate verification method: if the score is below 90%, it suggests checking reliable external databases or expert articles, for example, "check the weather forecasting service" or "visit the official weather bureau website."

[0302] Display means

[0303] The server formats the generated answer, its confidence score, and the recommended verification method and sends it to the user's device as an HTTP response. The device analyzes the received response data and displays it in an easy-to-understand format for the user, allowing the user to check the generated answer, its confidence score, and the recommended verification method.

[0304] For example, the user receives the following result:

[0305] Answer: "Don't worry, it's likely to be sunny over the weekend."

[0306] Probability: 85%

[0307] Recommended verification method:

[0308] Check the weather forecast service

[0309] Visit the official Meteorological Bureau website

[0310] The system of the present invention allows users to efficiently judge the accuracy of answers generated by the AI ​​and, if necessary, to conduct additional verification by referring to reliable external sources. Furthermore, by providing answers that take into consideration the user's feelings, the system can improve the user experience.

[0311] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0312] Step 1: User Input

[0313] User: Enters a question or problem in text format using a terminal. For example, "The weather has been bad lately. What will happen this weekend?"

[0314] Input: The user types text into the terminal.

[0315] Specific actions: The user enters a question into the input field on the device and presses the send button.

[0316] Output: The entered text is sent to the server as an HTTP POST request.

[0317] Step 2: Receiving the request

[0318] Server: Receives the HTTP request sent from the device, analyzes the received data, and extracts the question text.

[0319] Input: HTTP POST request data.

[0320] Specific behavior: The server parses the HTTP request and obtains the question text: "The weather has been bad lately. What will happen this weekend?"

[0321] Output: The extracted question text.

[0322] Step 3: Emotion Recognition

[0323] Server: The extracted question text is input into the emotion engine to analyze the user's emotions.

[0324] Input: Question text.

[0325] How it works: The emotion engine uses natural language processing (NLP) technology to analyze text and recognize emotions. It recognizes "negative emotions" from the question, "The weather has been bad lately. What's going to happen this weekend?"

[0326] Output: Emotion recognition result (e.g., "negative").

[0327] Step 4: Answer Generation

[0328] Server: After recognizing the emotion, the question text is input into the generative AI model to generate an answer.

[0329] Input: Question text, and emotion recognition results.

[0330] Specific operation: The question text is converted into a prompt sentence and input into the generative AI model. Example prompt sentence: "Tell me what the weather will be like this weekend." The generative AI model responds, "It's likely to be sunny this weekend."

[0331] Output: The generated answer.

[0332] Step 5: Adjust your answers

[0333] Server: Adjusts the tone and expression of the generated response based on the emotion recognition results.

[0334] Input: Generated answers, and emotion recognition results.

[0335] Action: If you recognize negative emotions, soften the tone of your response. "Don't worry, it's likely to be sunny over the weekend."

[0336] Output: The adjusted answer.

[0337] Step 6: Assess likelihood

[0338] Server: Uses a dedicated evaluation model to evaluate the reliability of the generated answer and calculates a reliability score.

[0339] Input: The adjusted answer.

[0340] Specific operation: Apply the evaluation model to the adjusted answer and calculate the reliability score. Example evaluation result: 85%.

[0341] Output: A likelihood score (e.g., 85%).

[0342] Step 7: Recommendation Generation

[0343] Server: Recommends an appropriate verification method based on the likelihood score.

[0344] Input: A likelihood score.

[0345] What it does: If the score is below 90%, generate recommendations to check more reliable sources, e.g., "Check the weather forecast service" and "Visit the official weather bureau website."

[0346] Output: Recommended validation method.

[0347] Step 8: Formatting the results

[0348] Server: Formats the generated answers, confidence scores, and verification methods.

[0349] Inputs: Calibrated answer, likelihood score, and recommended validation method.

[0350] Specific operation: Formats data into JSON format.

[0351] Output: The formatted results.

[0352] Step 9: Output to the User

[0353] Server: Sends the formatted results to the user device as an HTTP response.

[0354] Input: The formatted result.

[0355] Specific operation: The result is sent as an HTTP response. The device analyzes the received data and instructs it to display it.

[0356] Output: Display on the user's terminal.

[0357] This allows the user to see the generated answer, its likelihood, and the recommended verification method.

[0358] (Application example 2)

[0359] 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."

[0360] In modern digital communication, users are becoming increasingly concerned about messages suspected of being phishing scams or cyberattacks. There is a need for systems that can quickly and accurately assess the trustworthiness of such messages and recommend appropriate actions to users. However, existing systems lack an integrated approach that takes user emotions into consideration while providing plausible answers using generative AI, assessing their trustworthiness, and recommending appropriate verification methods. This poses a challenge, making it difficult to ensure users' sense of security.

[0361] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[0362] In this invention, the server includes an input means for a user to input a question or problem, an answer generation means for generating an answer using a generation AI, an evaluation means for evaluating the reliability of the generated answer, a recommendation means for recommending an additional verification method based on the reliability, an emotion recognition means for recognizing the emotion of the user when inputting a suspicious message and adjusting the answer by the generation AI, and a display means for displaying to the user the answer by the answer generation means, the reliability evaluation by the evaluation means, and the verification method by the recommendation means. This allows the user to respond quickly and appropriately to security risks such as phishing scams, giving them a sense of security.

[0363] "Input means" refers to a means by which a user inputs a question or problem in text format.

[0364] An "answer generation means" is a means for generating answers using a generation AI based on input questions or problems.

[0365] The "evaluation means" is a means for evaluating the reliability of the generated answer.

[0366] A "recommendation means" is a means for suggesting an appropriate verification method to a user based on the likelihood of success.

[0367] The "display means" is a means for displaying the generated answer, the likelihood assessment, and the recommended verification method to the user.

[0368] An "emotion recognition means" is a means for recognizing emotions from the text entered by the user and adjusting the response of the generating AI according to those emotions.

[0369] This invention provides a system that, when a user inputs a suspicious message or question, uses a generation AI to generate an answer based on the input, evaluates its reliability, and recommends an appropriate verification method. The following describes an embodiment of this system.

[0370] The user first enters a question or problem in text form, for example via a smartphone or computer terminal, and the text is sent to the server, which receives the request and begins analyzing it.

[0371] The server first uses an emotion recognition means to recognize emotions from the text entered by the user. For this emotion recognition, a software library for emotion analysis (e.g., TextBlob) or an emotion analysis model is used. For example, the emotion "anxiety" is recognized from the text "Does this seem like a phishing scam?"

[0372] Next, the server uses an answer generation means to input the input question into a generative AI model and generate an answer. As the generative AI model, for example, a model for text generation (e.g., GPT-3 (registered trademark)) is used. The generative AI model is physically deployed on a server on the cloud or in a high-performance computing environment running locally. The generated answer is obtained in the form of, for example, "This may be a phishing scam. Please be careful."

[0373] The server then uses an evaluation tool to evaluate the reliability of the generated answer. This evaluation uses a dedicated algorithm or model to calculate a reliability score. For example, suppose the reliability of the generated answer is evaluated as 85%.

[0374] Based on the evaluation results, the recommendation tool recommends appropriate verification methods to the user. For example, if the reliability score is less than 90%, verification methods such as "refer to a trusted security site" or "consult an expert" are suggested.

[0375] Finally, these answers, likelihood scores, and verification method recommendations are displayed to the user through a display means, such as an application or web browser that provides a user interface.

[0376] As a concrete example, an example of a prompt sentence input by the user is shown below.

[0377] Example question: "Does this seem like a phishing scam?"

[0378] Sample response: "This may be a phishing scam. Be careful."

[0379] Example of certainty: "85%"

[0380] Recommended verification methods: "Refer to a trusted security site" or "Consult an expert"

[0381] This allows users to respond quickly and appropriately to security risks such as phishing scams, giving them a sense of security.

[0382] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0383] Step 1:

[0384] The user enters a question or problem in text format. The entered text data is sent from the terminal to the server as an HTTP POST request, which transfers the user's input data to the server.

[0385] Step 2:

[0386] The server parses the received HTTP request and extracts the question text. The server passes this text to the emotion engine, which then recognizes the user's emotion from the text (e.g., using TextBlob or an emotion analysis model). The input is the question text, and the output is the emotion recognition result (e.g., "anxious").

[0387] Step 3:

[0388] Based on the emotion recognition results, the server inputs the question text into a generative AI model and generates an answer (e.g., using GPT-3). The input is the question text, and the output is the generated answer text (e.g., "This may be a phishing scam. Please be careful").

[0389] Step 4:

[0390] Based on the generated answer text, the server uses an evaluation method to evaluate the reliability of the answer. A specific algorithm or model is used to calculate a reliability score (e.g., 85%). The input is the generated answer text, and the output is the reliability score.

[0391] Step 5:

[0392] Based on the reliability score, the server uses a recommendation mechanism to suggest additional verification methods to the user (e.g., "refer to a trusted security site" or "consult an expert"). The input is the reliability score, and the output is a list of verification methods.

[0393] Step 6:

[0394] The server formats the generated answer, the reliability score, and the recommended verification method in JSON format and sends it to the user terminal as an HTTP response. The input is the generated answer text, the reliability score, and the list of verification methods, and the output is the formatted JSON data.

[0395] Step 7:

[0396] The terminal parses the received JSON data and displays it in a user-friendly format, allowing the user to see the generated answer, its accuracy, and the recommended verification method. The input is formatted JSON data, and the output is the information displayed on the user interface.

[0397] 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.

[0398] 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.

[0399] 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.

[0400] [Second embodiment]

[0401] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0402] 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.

[0403] 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).

[0404] 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.

[0405] 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.

[0406] 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).

[0407] 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.

[0408] 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.

[0409] 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.

[0410] 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.

[0411] 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.

[0412] 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."

[0413] The system of the present invention executes a series of processes in which a user inputs a question or problem, and a generation AI generates an answer based on that information, evaluates its reliability, and recommends an appropriate verification method.

[0414] User Input

[0415] A user uses a terminal to input a question or problem in text format. For example, the user might input "What is the diameter of the Earth?" This input is sent from the terminal to a server. This process is performed by sending the text data entered by the user to the server as an HTTP request.

[0416] Server-side answer generation

[0417] 1. Request received:

[0418] The server receives an HTTP request sent from a user terminal and extracts the question text. For example, if the request data is "What is the diameter of the Earth?", the server analyzes it and extracts it.

[0419] 2. Enter a question into the generative AI model:

[0420] The extracted question text is input into a generative AI model, which generates an answer based on this question. For example, the generated answer might be "The diameter of the Earth is approximately 12,742 km."

[0421] 3. Rate the confidence of your answer:

[0422] The server uses a model to calculate the likelihood of the generated answer, for example rating it as having a likelihood score of 80%.

[0423] Verification method recommendations

[0424] 1. Recommendation generation based on evaluation results:

[0425] The server determines the verification method to recommend to the user based on the reliability score. For example, if the reliability score is less than 90%, the server will recommend specific verification methods such as "check the NASA database" or "read the Encyclopaedia Britannica article."

[0426] 2. Result format:

[0427] Finally, we format the data to include the answer, a confidence score, and a recommended verification method. For example, the formatted result might look like this:

[0428] Answer: "The diameter of the Earth is approximately 12,742 km."

[0429] Probability: 80%

[0430] Recommended verification method:

[0431] Check the NASA database

[0432] Read the Encyclopaedia Britannica article

[0433] Output to the user

[0434] The server sends the formatted results as an HTTP response to the user's device, which then displays the received data in an easy-to-understand format for the user, who can see the generated answer, its accuracy, and the recommended verification method.

[0435] For example, the user receives the following result:

[0436] Answer: "The diameter of the Earth is approximately 12,742 km."

[0437] Probability: 80%

[0438] Recommended verification method:

[0439] Check the NASA database

[0440] Read the Encyclopaedia Britannica article

[0441] In this way, the system of the present invention helps users efficiently determine the accuracy of answers provided by the generating AI and, if necessary, verify them with more reliable sources of information.

[0442] The processing flow will be explained below.

[0443] Step 1:

[0444] The user enters a question or problem into the terminal. The user enters the question in text format using the touchscreen or keyboard and presses the submit button. At this time, the user's input is sent to the server as an HTTP POST request.

[0445] Step 2:

[0446] The terminal sends the user's input to the server. The terminal sends the text data entered by the user to the server as an HTTP request. For example, if the request data is "What is the diameter of the Earth?", this is sent to the server.

[0447] Step 3:

[0448] The server receives the request. The server receives the HTTP request sent from the device, analyzes the request data, and extracts the question text. Specifically, it extracts it using a method such as request.get("question").

[0449] Step 4:

[0450] The server inputs a question into the generative AI model. The extracted question text is input into the generative AI to generate an answer. If the generative AI generates an answer such as "The diameter of the Earth is approximately 12,742 km," this is accepted.

[0451] Step 5:

[0452] The server evaluates the reliability of the answer. A reliability evaluation model is used to evaluate the reliability of the generated answer. Specifically, the generated answer is input into the evaluation model and a calculated score is obtained. For example, suppose the evaluation result is 80%.

[0453] Step 6:

[0454] The server generates recommendations based on the evaluation results. The server recommends appropriate verification methods to the user based on the reliability score. If the score is less than 90%, the server recommends viewing reliable external databases or articles. For example, the server generates specific recommendations such as "Check the NASA database" or "Read the Encyclopaedia Britannica article."

[0455] Step 7:

[0456] The server formats the results. The server formats the generated answer, confidence score, and recommended verification method together, for example in JSON format like this:

[0457] {

[0458] "answer": "The diameter of the Earth is approximately 12742 km",

[0459] "certainty_score": 80,

[0460] "recommendations": ["Check out the NASA database", "Read the Encyclopaedia Britannica article"]

[0461] }

[0462] Step 8:

[0463] The server sends the formatted results to the terminal. The server sends the formatted data to the user terminal as an HTTP response.

[0464] Step 9:

[0465] The terminal displays the result to the user. The terminal analyzes the received response data and displays it in an easy-to-understand manner for the user. For example, the user receives the result as follows:

[0466] Answer: "The diameter of the Earth is approximately 12,742 km."

[0467] Probability: 80%

[0468] Recommended verification method:

[0469] Check the NASA database

[0470] Read the Encyclopaedia Britannica article

[0471] The user can then perform further verification based on this.

[0472] Example 1

[0473] 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."

[0474] In conventional question-answering systems, users have little means to judge the accuracy and reliability of the generated answers, which creates the risk of receiving incorrect information. Furthermore, users must independently verify the authenticity of the generated answers, which is time-consuming and laborious. Furthermore, due to the lack of functionality to evaluate the reliability of information and recommend appropriate verification methods, users have difficulty finding additional reliable external information sources.

[0475] 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.

[0476] In this invention, the server includes input means for a user to input a question or problem, answer generation means for receiving the question or problem sent from the input means, inputting an inquiry generated based on the question or problem into a generative AI model, and generating an answer from the model, evaluation means for evaluating the reliability of the answer generated by the answer generation means, recommendation means for recommending an additional verification method based on the reliability evaluated by the evaluation means, and display means for formatting and displaying to the user the answer generated by the answer generation means, the reliability evaluation by the evaluation means, and the verification method recommended by the recommendation means. This saves the user the trouble of verifying the reliability of the generated answer, and enables additional verification using a reliable information source.

[0477] "Input means" refers to the means by which a user inputs questions or problems to the system.

[0478] An "answer generation means" is a means for generating a corresponding answer using a generative AI model based on a question or problem from a user.

[0479] The "evaluation means" is a means for calculating and evaluating the likelihood of the generated answer.

[0480] The "recommendation means" is a means for recommending an additional verification method to the user based on the likelihood evaluated by the evaluation means.

[0481] The "display means" is a means for formatting the answer by the answer generation means, the evaluation of the likelihood by the evaluation means, and the verification method by the recommendation means, and displaying them in an easy-to-see manner for the user.

[0482] A "generative AI model" is an artificial intelligence model used to generate answers in natural language based on a user's question or problem.

[0483] "Likelihood" is an evaluation value that indicates the accuracy and reliability of the answer provided by the generative AI model.

[0484] "Validation methods" are additional steps or reference sources used to confirm the accuracy of the answers provided by a generative AI model.

[0485] "Network connection" refers to the communication environment for accessing generative AI models and external information sources.

[0486] This invention is a system in which a user inputs a question or problem, a generative AI model generates an answer based on that information, evaluates its reliability, and recommends an appropriate verification method. The system is composed of a server, terminals, and a communication network.

[0487] System Configuration

[0488] The system mainly consists of the following components:

[0489] 1. Input medium: The interface through which a user enters a question or problem. This can include a web browser or a mobile app.

[0490] 2. Answer generation means: Receives questions or problems from users and generates answers using a generative AI model. The generative AI model can be, for example, a general artificial intelligence model that excels at text generation.

[0491] 3. Evaluation method: The likelihood of the generated answer is calculated and evaluated. Here, a probabilistic evaluation algorithm is used.

[0492] 4. Recommendation: Based on the reliability assessment results, the system recommends additional verification methods to the user. These recommendations are based on databases and reliable external information sources.

[0493] 5. Display: The answers, the confidence scores, and the recommended verification methods are formatted and displayed to the user in an easy-to-read format on the device screen.

[0494] Specific examples

[0495] A user uses a terminal to input a question such as: "What is the diameter of the Earth?" This input is sent to the server as an HTTP request.

[0496] Server Processing

[0497] The server receives an HTTP request sent from the user's device and extracts the question text from it. The extracted question is input into a generative AI model, which then generates an answer to the question. For example, the generated answer might be, "The diameter of the Earth is approximately 12,742 km."

[0498] The generated answer is fed into an algorithm to calculate its likelihood. The algorithm rates the answer as 80% reliable. Based on this rating, it recommends additional verification methods. Because the score is below 90%, it recommends things like "checking the NASA database" or "reading the Encyclopaedia Britannica article."

[0499] Finally, the answer, confidence score, and recommended verification method are formatted into a text like this:

[0500] Answer: "The diameter of the Earth is approximately 12,742 km."

[0501] Probability: 80%

[0502] Recommended verification method:

[0503] Check the NASA database

[0504] Read the Encyclopaedia Britannica article

[0505] What the user sees

[0506] The server sends the formatted results as an HTTP response to the user's device, which receives it and displays it in an easy-to-understand manner for the user.

[0507] In this way, the system of the present invention helps users efficiently determine the accuracy of answers provided by the generating AI and, if necessary, verify them with reliable sources of information.

[0508] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0509] System program processing flow

[0510] Step 1: User Input

[0511] Input: The user types a question or problem into the terminal.

[0512] Output: Text data of the question or problem entered.

[0513] The user inputs a question or problem through the device's interface. For example, they might type, "What is the diameter of the Earth?" The input data is sent to the server as an HTTP request. When the user types a question into the input form and presses the "Submit" button, the device converts the text data into JSON format and sends it to the server.

[0514] Step 2: Receiving the request

[0515] Input: HTTP request sent from the user's device.

[0516] Output: The extracted question text.

[0517] The server receives the HTTP request sent from the user's device and extracts the question text. It then analyzes the JSON format data from the received HTTP request to obtain the specific question. For example, the extracted question might be "What is the diameter of the Earth?"

[0518] Step 3: Input to the generative AI model

[0519] Input: The extracted question text.

[0520] Output: The answer text returned by the generative AI model.

[0521] The server inputs the extracted question text into the generative AI model, which then generates an answer based on this question. Specifically, the prompt "What is the diameter of the Earth?" is input to the generative AI model, and the model responds with an answer such as "The diameter of the Earth is approximately 12,742 km."

[0522] Step 4: Evaluate the likelihood of your answer

[0523] Input: The answer text returned by the generative AI model.

[0524] Output: A likelihood score.

[0525] The server evaluates the reliability of the generated answer. The generated answer is input into the evaluation model and the reliability is calculated. For example, the reliability score is calculated as 80%. Specifically, the evaluation is performed using an algorithm that calculates a reliability score.

[0526] Step 5: Recommend a verification method

[0527] Input: A likelihood score.

[0528] Output: A list of recommended validation methods.

[0529] The server determines the verification method to recommend to the user based on the reliability score. For example, if the score is less than 90%, specific verification methods such as "check the NASA database" or "read the Encyclopaedia Britannica article" are recommended. This is done based on databases and reliable sources.

[0530] Step 6: Formatting the results

[0531] Input: Answer text, confidence score, and a list of recommended verification methods.

[0532] Output: Formatted result data.

[0533] The server formats the answer, confidence score, and recommended verification method, perhaps in a format like JSON or plain text, for example:

[0534] Answer: "The diameter of the Earth is approximately 12,742 km."

[0535] Probability: 80%

[0536] Recommended verification method:

[0537] Check the NASA database

[0538] Read the Encyclopaedia Britannica article

[0539] Step 7: Output to the User

[0540] Input: Formatted result data.

[0541] Output: The result that is displayed to the user.

[0542] The server sends the formatted results as an HTTP response to the user's device. The device then displays the received data in an easy-to-understand format for the user. Specifically, the results are displayed on the screen of a web browser or mobile app. The user can view the generated answer, its accuracy, and the recommended verification method.

[0543] (Application example 1)

[0544] 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."

[0545] In today's world, phishing emails are becoming increasingly sophisticated, making them extremely difficult to identify. Conventional phishing email countermeasures require users to check the emails themselves, which often leads to oversights or incorrect judgments. Furthermore, judgments may be made based on unreliable information. Therefore, there is a need for a system that can quickly and accurately evaluate the reliability of emails received by users and recommend necessary verification methods.

[0546] 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.

[0547] In this invention, the server includes: input means for a user to input a question or problem; answer generation means for generating an answer using a generation AI based on the question or problem sent from the input means; evaluation means for evaluating the reliability of the answer generated by the answer generation means; recommendation means for recommending an additional verification method based on the reliability evaluated by the evaluation means; display means for displaying to the user the answer generated by the answer generation means, the reliability evaluation by the evaluation means, and the verification method recommended by the recommendation means; reliability evaluation means for evaluating the reliability of email text; and warning means for notifying the user that the reliability evaluation means has determined that the email is likely to be a phishing email. This allows the user to quickly determine the reliability of received emails and prevent damage from phishing emails.

[0548] "User" means any person or entity that utilizes the System to enter questions or problems and receive results.

[0549] "Input means" refers to a device or interface that allows a user to input a question or problem in text form.

[0550] An "answer generation means" is a process or system for generating answers using a generation AI based on questions or problems entered by a user.

[0551] An "evaluation method" is a process or system for evaluating the reliability of answers created by generative AI.

[0552] A "recommendation means" is a process or system for recommending an appropriate verification method to a user based on the assessed likelihood.

[0553] A "display means" is a device or interface for visually displaying to a user the generated answer, the likelihood assessment, and the recommended verification method.

[0554] A "trustworthiness assessment means" is a process or system for assessing the trustworthiness of the text of a received email and determining whether it is a phishing email.

[0555] The "warning means" is a process or system for notifying the user when the reliability evaluation means determines that the email is highly likely to be a phishing email.

[0556] A "phishing email" is a fraudulent email sent to steal personal or confidential information from users for malicious purposes.

[0557] This paper describes how to build a system that can quickly and accurately evaluate the reliability of emails received by users and warn them of phishing emails as necessary. This system utilizes a generative AI model to evaluate the reliability of emails and recommend appropriate measures to users.

[0558] The server includes an input means for a user to input a question or problem, an answer generation means for generating an answer using a generation AI based on the question or problem sent from the input means, an evaluation means for evaluating the reliability of the answer generated by the answer generation means, and a recommendation means for recommending an additional verification method based on the reliability evaluated by the evaluation means.The server further includes a display means for displaying to the user the answer generated by the answer generation means, the reliability evaluation by the evaluation means, and the verification method recommended by the recommendation means, a reliability evaluation means for evaluating the reliability of the email text, and a warning means for notifying the user that the email may be a phishing email based on the reliability evaluation means.

[0559] Specifically, the user inputs the text of the email they received and sends it to the server. This input is done via the user's device (smartphone, tablet, PC, etc.). In this process, the email body is input in text format and sent to the server as an HTTP request. For example, the user inputs the following email text:

[0560] "The email said, 'Your account has been locked due to a security issue. Please click the link to verify your account immediately.'"

[0561] The server receives the data sent from the user's device and evaluates the reliability of the email using a generative AI model, such as one from Hugging Face's Transformers library. This model determines the likelihood that the email is a phishing email. For example, it generates a response such as "The email is likely to be phishing."

[0562] The server then recommends appropriate verification methods to the user based on the assessed trustworthiness score. For emails with low trustworthiness, additional verification methods may be recommended. For example, a recommendation such as "Please check official security guidelines" may be given.

[0563] Finally, this information is sent to the user's terminal, where the user can check the results in text format, making it easy for the user to quickly determine the authenticity of the received email and perform further checks if necessary.

[0564] This system allows users to confirm the authenticity of emails in a reliable way, even in an age where it is difficult to distinguish phishing emails.When commercializing the product, it is recommended that devices be standard smartphones, tablets, or PCs, and that software utilize Python or a framework that can handle HTTP requests.

[0565] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0566] Step 1: User enters email text

[0567] The user inputs the text of the received email into the terminal and sends it to the server. The input data could be, for example, the text "Your account has been locked due to a security issue. Please click the link immediately to confirm." This input is sent to the server as an HTTP request.

[0568] Step 2: The server receives the input

[0569] The server receives the HTTP request sent from the user terminal and extracts the email text. For example, it extracts the text "Your account has been locked due to a security issue. Please click the link immediately to confirm." from the request data.

[0570] Step 3: Generative AI reliability assessment

[0571] The server inputs the extracted email text into a generative AI model to evaluate its reliability. The generative AI model uses, for example, Hugging Face's Transformers library to determine whether the email is a phishing email. The model generates an answer such as "This is likely a phishing email." This process uses natural language processing techniques to analyze the input data and calculate a reliability score.

[0572] Step 4: Evaluate the reliability score

[0573] The server analyzes the reliability score provided by the generative AI model and evaluates its reliability. For example, if the reliability score is evaluated as 80%, the next processing step will proceed based on that score. A score analysis algorithm is used for this data processing.

[0574] Step 5: Recommend a verification method

[0575] The server recommends appropriate verification methods to the user based on the reliability score. For example, if the reliability score is low, the server provides a recommendation to "check official security guidelines." This process links the results of the reliability assessment to the user's countermeasure actions.

[0576] Step 6: Formatting the results

[0577] The server then combines the generated answers, confidence ratings, and recommended verification methods into a single format, such as the following:

[0578] "It's likely a phishing email."

[0579] Reliability score: 80%

[0580] Recommended verification: Check official security guidelines

[0581] Step 7: Displaying results to the user

[0582] The device receives the formatted results sent from the server and displays them to the user, who can view the generated answer, its confidence rating, and recommended verification methods, for example, displayed on a smartphone screen.

[0583] Each processing step has been explained in detail above. This system makes it easy for users to quickly determine the authenticity of received emails and to verify them as needed.

[0584] 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.

[0585] The system of the present invention uses generative AI to provide answers to questions or problems entered by users, evaluate their accuracy, and recommend appropriate verification methods. It also incorporates an emotion engine that recognizes emotions from the user's input, adjusts answers based on those emotions, and suggests verification methods.

[0586] User Input

[0587] A user uses a terminal to input a question or problem in text format. For example, a user might type, "The weather has been bad lately. What will happen this weekend?" This input is sent from the terminal to the server. It is sent as an HTTP POST request.

[0588] emotion recognition

[0589] 1. Request received:

[0590] The server receives the HTTP request sent from the terminal, analyzes the request data, and extracts the question text.

[0591] 2. Emotion recognition using emotion engine:

[0592] The extracted text is input into the emotion engine, which analyzes the user's emotions. For example, a question like "The weather has been bad lately. What will happen this weekend?" can be used to identify "negative emotions."

[0593] Answer generation and likelihood assessment

[0594] 1. Enter a question into the generative AI model:

[0595] After recognizing the emotion, the question text is fed into a generative AI model to generate an answer. For example, the generative AI might respond, "It's likely to be sunny over the weekend."

[0596] 2. Adjusting your answers:

[0597] It adjusts the tone and wording of responses based on emotion recognition. If a negative emotion is detected, it will adjust the response, for example, "Don't worry, it's likely to be sunny over the weekend."

[0598] 3. Assessing likelihood:

[0599] The server uses a model to assess the likelihood of the generated answer and calculates a confidence score. For example, suppose the assessment result is 85%.

[0600] Verification method recommendations

[0601] 1. Recommendation generation based on evaluation results:

[0602] Based on the reliability score, the server recommends appropriate verification methods to the user. If the score is below 90%, it suggests checking reliable external databases or expert articles. For example, it recommends "checking weather forecasting services" or "referring to the official website of the Meteorological Agency."

[0603] 2. Result format:

[0604] The server formats the generated answer, the confidence score, and the recommended verification method, for example in JSON format as follows:

[0605] {

[0606] "answer": "Don't worry, it's likely to be sunny over the weekend",

[0607] "certainty_score": 85,

[0608] "recommendations": ["Check the weather forecast service", "See the official Meteorological Agency website"]

[0609] }

[0610] Output to the user

[0611] The server sends the formatted results as an HTTP response to the user's device, which then parses the response data and displays it in a user-friendly format. The user can see the generated answer, its accuracy, and the recommended verification method.

[0612] For example, the user receives the following result:

[0613] Answer: "Don't worry, it's likely to be sunny over the weekend."

[0614] Probability: 85%

[0615] Recommended verification method:

[0616] Check the weather forecast service

[0617] Visit the official Meteorological Bureau website

[0618] In this way, the system of the present invention helps users efficiently judge the accuracy of the answers generated by the AI ​​and, if necessary, allows them to conduct additional verification using more reliable external sources. It also improves the user experience by adjusting the answers to take into account the user's feelings.

[0619] The processing flow will be explained below.

[0620] Step 1:

[0621] The user types a question or problem into the device. The user uses a keyboard or touchscreen to type the question in text format and clicks a submit button. For example, the user might type, "The weather has been bad lately. What's it going to be like this weekend?"

[0622] Step 2:

[0623] The device sends the user's input to the server. The device sends the entered text data to the server as an HTTP POST request. The data sent is in the format {"question": "The weather has been bad lately. What will happen this weekend?"}

[0624] Step 3:

[0625] The server receives the request. The server receives the HTTP request, analyzes its contents, and extracts the question text. Specifically, the question is extracted using request.get("question") .

[0626] Step 4:

[0627] The server inputs the question text into the emotion engine. The server then passes the extracted question text to the emotion engine, which analyzes the user's emotion. For example, the emotion engine recognizes "negative emotion."

[0628] Step 5:

[0629] The server inputs a question into the generative AI model. Based on the results of analysis by the emotion engine, the question text is input into the generative AI model to generate an answer. For example, the generative AI generates the answer "It's likely to be sunny over the weekend."

[0630] Step 6:

[0631] The server adjusts the tone of the generated answer. The tone and expression of the answer are adjusted appropriately depending on the user's emotions. For example, if a negative emotion is detected, the server adjusts the answer to "Don't worry. It's likely to be sunny over the weekend."

[0632] Step 7:

[0633] The server evaluates the reliability of the generated answer. The server inputs the answer into a reliability evaluation model and obtains a reliability score. For example, suppose the evaluation result is 85%.

[0634] Step 8:

[0635] The server generates recommendations based on the evaluation results. If the reliability score is less than 90%, it recommends using a more reliable external database or resource. For example, it suggests "checking the weather forecast service" or "referring to the official website of the Japan Meteorological Agency."

[0636] Step 9:

[0637] The server formats the results: the generated answer, the confidence score, and the recommended verification method in JSON format, for example:

[0638] {

[0639] "answer": "Don't worry, it's likely to be sunny over the weekend",

[0640] "certainty_score": 85,

[0641] "recommendations": ["Check the weather forecast service", "See the official Meteorological Agency website"]

[0642] }

[0643] Step 10:

[0644] The server sends the formatted results to the terminal. The server sends the formatted data to the user terminal as an HTTP response.

[0645] Step 11:

[0646] The terminal displays the result to the user. The terminal analyzes the received response data and displays it in an easy-to-understand manner for the user. For example, the user receives the result as follows:

[0647] Answer: "Don't worry, it's likely to be sunny over the weekend."

[0648] Probability: 85%

[0649] Recommended verification method:

[0650] Check the weather forecast service

[0651] Visit the official Meteorological Bureau website

[0652] Users can then use this information to verify the information with reliable external sources.

[0653] Example 2

[0654] 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."

[0655] Conventional question-answering systems lacked a means for evaluating the reliability of generated answers or a mechanism for adjusting answers based on user emotions. This made it difficult for users to judge the reliability of generated answers, potentially leading to actions based on unreliable information. Furthermore, there were concerns that one-sided answers that did not take the user's emotions into consideration could detract from the user experience. In response to this, the present invention aims to provide more reliable information and improve the user experience by combining reliability evaluation and emotion recognition in an answer generation system using a generative AI model.

[0656] 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.

[0657] In this invention, the server includes: input means for a user to input a question or problem; answer generation means for generating an answer using a generation AI based on the question or problem transmitted from the input means; evaluation means for evaluating the reliability of the answer generated by the answer generation means; recommendation means for recommending an additional verification method based on the reliability evaluated by the evaluation means; emotion recognition means for recognizing emotions in the user's input text and adjusting the answer based on the result; and display means for displaying to the user the answer generated by the answer generation means, the reliability evaluation by the evaluation means, and the verification method recommended by the recommendation means. This makes it possible to evaluate the reliability of the generated answer, provide the user with reliable information, and provide answers that take the user's emotions into consideration, thereby improving the user experience.

[0658] An "input means" is a device or software interface through which a user inputs a question or problem in text form.

[0659] An "answer generation means" is a device or software function that uses generation AI to generate appropriate answers based on questions or problems obtained from an input means.

[0660] The "evaluation means" is a device or software function that evaluates the likelihood of the answer generated by the answer generation means and calculates the reliability as a score.

[0661] The "recommendation means" is a device or software function that recommends additional verification methods to the user based on the likelihood evaluated by the evaluation means.

[0662] An "emotion recognizer" is a device or software function that recognizes the emotion of a user's input text and adjusts the response based on the results.

[0663] The "display means" is a device or software interface that displays to the user the answer from the answer generation means, the evaluation of the likelihood from the evaluation means, and the verification method from the recommendation means.

[0664] MODE FOR CARRYING OUT THE INVENTION

[0665] The system of the present invention uses generative AI to provide answers to questions or problems entered by users, evaluate their accuracy, and recommend appropriate verification methods. This system also incorporates an emotion engine that recognizes the user's emotions and adjusts the answer accordingly.

[0666] Input Method

[0667] A user enters a question or problem in text format using a device. This device can be a PC, smartphone, tablet, etc. For example, they might enter, "The weather has been bad lately. What will happen this weekend?" This input is sent to the server as an HTTP POST request.

[0668] emotion recognition means

[0669] The server receives the HTTP request sent from the device, analyzes the request data, and extracts the question text. The emotion engine then analyzes the user's emotions. The emotion engine uses natural language processing (NLP) technology to recognize "negative emotions," for example, from a question like, "The weather has been bad lately. What will happen this weekend?"

[0670] Answer generation means

[0671] After recognizing the emotion, the server inputs the question text into the generative AI model and generates an answer. For example, if the prompt is "Tell me what the weather will be like this weekend," the generative AI model will respond with "It's likely to be sunny this weekend."

[0672] Answer adjustment

[0673] The tone and wording of the generated response is adjusted based on emotion recognition: if a negative emotion is detected, the response will be adjusted, such as "Don't worry, it's likely to be sunny over the weekend."

[0674] Certainty assessment method

[0675] The server uses a proprietary rating model to evaluate the reliability of the generated answer and calculates a reliability score. For example, suppose the rating result is 85%.

[0676] Recommendation methods

[0677] Based on the confidence score of the generated answer, the server recommends an appropriate verification method: if the score is below 90%, it suggests checking reliable external databases or expert articles, for example, "check the weather forecasting service" or "visit the official weather bureau website."

[0678] Display means

[0679] The server formats the generated answer, its confidence score, and the recommended verification method and sends it to the user's device as an HTTP response. The device analyzes the received response data and displays it in an easy-to-understand format for the user, allowing the user to check the generated answer, its confidence score, and the recommended verification method.

[0680] For example, the user receives the following result:

[0681] Answer: "Don't worry, it's likely to be sunny over the weekend."

[0682] Probability: 85%

[0683] Recommended verification method:

[0684] Check the weather forecast service

[0685] Visit the official Meteorological Bureau website

[0686] The system of the present invention allows users to efficiently judge the accuracy of answers generated by the AI ​​and, if necessary, to conduct additional verification by referring to reliable external sources. Furthermore, by providing answers that take into consideration the user's feelings, the system can improve the user experience.

[0687] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0688] Step 1: User Input

[0689] User: Enters a question or problem in text format using a terminal. For example, "The weather has been bad lately. What will happen this weekend?"

[0690] Input: The user types text into the terminal.

[0691] Specific actions: The user enters a question into the input field on the device and presses the send button.

[0692] Output: The entered text is sent to the server as an HTTP POST request.

[0693] Step 2: Receiving the request

[0694] Server: Receives the HTTP request sent from the device, analyzes the received data, and extracts the question text.

[0695] Input: HTTP POST request data.

[0696] Specific behavior: The server parses the HTTP request and obtains the question text: "The weather has been bad lately. What will happen this weekend?"

[0697] Output: The extracted question text.

[0698] Step 3: Emotion Recognition

[0699] Server: The extracted question text is input into the emotion engine to analyze the user's emotions.

[0700] Input: Question text.

[0701] How it works: The emotion engine uses natural language processing (NLP) technology to analyze text and recognize emotions. It recognizes "negative emotions" from the question, "The weather has been bad lately. What's going to happen this weekend?"

[0702] Output: Emotion recognition result (e.g., "negative").

[0703] Step 4: Answer Generation

[0704] Server: After recognizing the emotion, the question text is input into the generative AI model to generate an answer.

[0705] Input: Question text, and emotion recognition results.

[0706] Specific operation: The question text is converted into a prompt sentence and input into the generative AI model. Example prompt sentence: "Tell me what the weather will be like this weekend." The generative AI model responds, "It's likely to be sunny this weekend."

[0707] Output: The generated answer.

[0708] Step 5: Adjust your answers

[0709] Server: Adjusts the tone and expression of the generated response based on the emotion recognition results.

[0710] Input: Generated answers, and emotion recognition results.

[0711] Action: If you recognize negative emotions, soften the tone of your response. "Don't worry, it's likely to be sunny over the weekend."

[0712] Output: The adjusted answer.

[0713] Step 6: Assess likelihood

[0714] Server: Uses a dedicated evaluation model to evaluate the reliability of the generated answer and calculates a reliability score.

[0715] Input: The adjusted answer.

[0716] Specific operation: Apply the evaluation model to the adjusted answer and calculate the reliability score. Example evaluation result: 85%.

[0717] Output: A likelihood score (e.g., 85%).

[0718] Step 7: Recommendation Generation

[0719] Server: Recommends an appropriate verification method based on the likelihood score.

[0720] Input: A likelihood score.

[0721] What it does: If the score is below 90%, generate recommendations to check more reliable sources, e.g., "Check the weather forecast service" and "Visit the official weather bureau website."

[0722] Output: Recommended validation method.

[0723] Step 8: Formatting the results

[0724] Server: Formats the generated answers, confidence scores, and verification methods.

[0725] Inputs: Calibrated answer, likelihood score, and recommended validation method.

[0726] Specific operation: Formats data into JSON format.

[0727] Output: The formatted results.

[0728] Step 9: Output to the User

[0729] Server: Sends the formatted results to the user device as an HTTP response.

[0730] Input: The formatted result.

[0731] Specific operation: The result is sent as an HTTP response. The device analyzes the received data and instructs it to display it.

[0732] Output: Display on the user's terminal.

[0733] This allows the user to see the generated answer, its likelihood, and the recommended verification method.

[0734] (Application example 2)

[0735] 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."

[0736] In modern digital communication, users are becoming increasingly concerned about messages suspected of being phishing scams or cyberattacks. There is a need for systems that can quickly and accurately assess the trustworthiness of such messages and recommend appropriate actions to users. However, existing systems lack an integrated approach that takes user emotions into consideration while providing plausible answers using generative AI, assessing their trustworthiness, and recommending appropriate verification methods. This poses a challenge, making it difficult to ensure users' sense of security.

[0737] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[0738] In this invention, the server includes an input means for a user to input a question or problem, an answer generation means for generating an answer using a generation AI, an evaluation means for evaluating the reliability of the generated answer, a recommendation means for recommending an additional verification method based on the reliability, an emotion recognition means for recognizing the emotion of the user when inputting a suspicious message and adjusting the answer by the generation AI, and a display means for displaying to the user the answer by the answer generation means, the reliability evaluation by the evaluation means, and the verification method by the recommendation means. This allows the user to respond quickly and appropriately to security risks such as phishing scams, giving them a sense of security.

[0739] "Input means" refers to a means by which a user inputs a question or problem in text format.

[0740] An "answer generation means" is a means for generating answers using a generation AI based on input questions or problems.

[0741] The "evaluation means" is a means for evaluating the reliability of the generated answer.

[0742] A "recommendation means" is a means for suggesting an appropriate verification method to a user based on the likelihood of success.

[0743] The "display means" is a means for displaying the generated answer, the likelihood assessment, and the recommended verification method to the user.

[0744] An "emotion recognition means" is a means for recognizing emotions from the text entered by the user and adjusting the response of the generating AI according to those emotions.

[0745] This invention provides a system that, when a user inputs a suspicious message or question, uses a generation AI to generate an answer based on the input, evaluates its reliability, and recommends an appropriate verification method. The following describes an embodiment of this system.

[0746] The user first enters a question or problem in text form, for example via a smartphone or computer terminal, and the text is sent to the server, which receives the request and begins analyzing it.

[0747] The server first uses an emotion recognition means to recognize emotions from the text entered by the user. For this emotion recognition, a software library for emotion analysis (e.g., TextBlob) or an emotion analysis model is used. For example, the emotion "anxiety" is recognized from the text "Does this seem like a phishing scam?"

[0748] Next, the server uses an answer generation means to input the input question into a generative AI model and generate an answer. As a generative AI model, for example, a text generation model (e.g., GPT-3) is used. The generative AI model is physically located on a server in the cloud or in a high-performance computing environment running locally. The generated answer is provided in the form of, for example, "This may be a phishing scam. Please be careful."

[0749] The server then uses an evaluation tool to evaluate the reliability of the generated answer. This evaluation uses a dedicated algorithm or model to calculate a reliability score. For example, suppose the reliability of the generated answer is evaluated as 85%.

[0750] Based on the evaluation results, the recommendation tool recommends appropriate verification methods to the user. For example, if the reliability score is less than 90%, verification methods such as "refer to a trusted security site" or "consult an expert" are suggested.

[0751] Finally, these answers, likelihood scores, and verification method recommendations are displayed to the user through a display means, such as an application or web browser that provides a user interface.

[0752] As a concrete example, an example of a prompt sentence input by the user is shown below.

[0753] Example question: "Does this seem like a phishing scam?"

[0754] Sample response: "This may be a phishing scam. Be careful."

[0755] Example of certainty: "85%"

[0756] Recommended verification methods: "Refer to a trusted security site" or "Consult an expert"

[0757] This allows users to respond quickly and appropriately to security risks such as phishing scams, giving them a sense of security.

[0758] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0759] Step 1:

[0760] The user enters a question or problem in text format. The entered text data is sent from the terminal to the server as an HTTP POST request, which transfers the user's input data to the server.

[0761] Step 2:

[0762] The server parses the received HTTP request and extracts the question text. The server passes this text to the emotion engine, which then recognizes the user's emotion from the text (e.g., using TextBlob or an emotion analysis model). The input is the question text, and the output is the emotion recognition result (e.g., "anxious").

[0763] Step 3:

[0764] Based on the emotion recognition results, the server inputs the question text into a generative AI model and generates an answer (e.g., using GPT-3). The input is the question text, and the output is the generated answer text (e.g., "This may be a phishing scam. Please be careful").

[0765] Step 4:

[0766] Based on the generated answer text, the server uses an evaluation method to evaluate the reliability of the answer. A specific algorithm or model is used to calculate a reliability score (e.g., 85%). The input is the generated answer text, and the output is the reliability score.

[0767] Step 5:

[0768] Based on the reliability score, the server uses a recommendation mechanism to suggest additional verification methods to the user (e.g., "refer to a trusted security site" or "consult an expert"). The input is the reliability score, and the output is a list of verification methods.

[0769] Step 6:

[0770] The server formats the generated answer, the reliability score, and the recommended verification method in JSON format and sends it to the user terminal as an HTTP response. The input is the generated answer text, the reliability score, and the list of verification methods, and the output is the formatted JSON data.

[0771] Step 7:

[0772] The terminal parses the received JSON data and displays it in a user-friendly format, allowing the user to see the generated answer, its accuracy, and the recommended verification method. The input is formatted JSON data, and the output is the information displayed on the user interface.

[0773] 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.

[0774] 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.

[0775] 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.

[0776] [Third embodiment]

[0777] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[0778] 5, the data processing system 310 includes the data processing device 12 and a headset type terminal 314. An example of the data processing device 12 is a server.

[0779] 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).

[0780] 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.

[0781] 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.

[0782] 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).

[0783] 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.

[0784] 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.

[0785] 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.

[0786] 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.

[0787] 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.

[0788] 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."

[0789] The system of the present invention executes a series of processes in which a user inputs a question or problem, and a generation AI generates an answer based on that information, evaluates its reliability, and recommends an appropriate verification method.

[0790] User Input

[0791] A user uses a terminal to input a question or problem in text format. For example, the user might input "What is the diameter of the Earth?" This input is sent from the terminal to a server. This process is performed by sending the text data entered by the user to the server as an HTTP request.

[0792] Server-side answer generation

[0793] 1. Request received:

[0794] The server receives an HTTP request sent from a user terminal and extracts the question text. For example, if the request data is "What is the diameter of the Earth?", the server analyzes it and extracts it.

[0795] 2. Enter a question into the generative AI model:

[0796] The extracted question text is input into a generative AI model, which generates an answer based on this question. For example, the generated answer might be "The diameter of the Earth is approximately 12,742 km."

[0797] 3. Rate the confidence of your answer:

[0798] The server uses a model to calculate the likelihood of the generated answer, for example rating it as having a likelihood score of 80%.

[0799] Verification method recommendations

[0800] 1. Recommendation generation based on evaluation results:

[0801] The server determines the verification method to recommend to the user based on the reliability score. For example, if the reliability score is less than 90%, the server will recommend specific verification methods such as "check the NASA database" or "read the Encyclopaedia Britannica article."

[0802] 2. Result format:

[0803] Finally, we format the data to include the answer, a confidence score, and a recommended verification method. For example, the formatted result might look like this:

[0804] Answer: "The diameter of the Earth is approximately 12,742 km."

[0805] Probability: 80%

[0806] Recommended verification method:

[0807] Check the NASA database

[0808] Read the Encyclopaedia Britannica article

[0809] Output to the user

[0810] The server sends the formatted results as an HTTP response to the user's device, which then displays the received data in an easy-to-understand format for the user, who can see the generated answer, its accuracy, and the recommended verification method.

[0811] For example, the user receives the following result:

[0812] Answer: "The diameter of the Earth is approximately 12,742 km."

[0813] Probability: 80%

[0814] Recommended verification method:

[0815] Check the NASA database

[0816] Read the Encyclopaedia Britannica article

[0817] In this way, the system of the present invention helps users efficiently determine the accuracy of answers provided by the generating AI and, if necessary, verify them with more reliable sources of information.

[0818] The processing flow will be explained below.

[0819] Step 1:

[0820] The user enters a question or problem into the terminal. The user enters the question in text format using the touchscreen or keyboard and presses the submit button. At this time, the user's input is sent to the server as an HTTP POST request.

[0821] Step 2:

[0822] The terminal sends the user's input to the server. The terminal sends the text data entered by the user to the server as an HTTP request. For example, if the request data is "What is the diameter of the Earth?", this is sent to the server.

[0823] Step 3:

[0824] The server receives the request. The server receives the HTTP request sent from the device, analyzes the request data, and extracts the question text. Specifically, it extracts it using a method such as request.get("question").

[0825] Step 4:

[0826] The server inputs a question into the generative AI model. The extracted question text is input into the generative AI to generate an answer. If the generative AI generates an answer such as "The diameter of the Earth is approximately 12,742 km," this is accepted.

[0827] Step 5:

[0828] The server evaluates the reliability of the answer. A reliability evaluation model is used to evaluate the reliability of the generated answer. Specifically, the generated answer is input into the evaluation model and a calculated score is obtained. For example, suppose the evaluation result is 80%.

[0829] Step 6:

[0830] The server generates recommendations based on the evaluation results. The server recommends appropriate verification methods to the user based on the reliability score. If the score is less than 90%, the server recommends viewing reliable external databases or articles. For example, the server generates specific recommendations such as "Check the NASA database" or "Read the Encyclopaedia Britannica article."

[0831] Step 7:

[0832] The server formats the results. The server formats the generated answer, confidence score, and recommended verification method together, for example in JSON format like this:

[0833] {

[0834] "answer": "The diameter of the Earth is approximately 12742 km",

[0835] "certainty_score": 80,

[0836] "recommendations": ["Check out the NASA database", "Read the Encyclopaedia Britannica article"]

[0837] }

[0838] Step 8:

[0839] The server sends the formatted results to the terminal. The server sends the formatted data to the user terminal as an HTTP response.

[0840] Step 9:

[0841] The terminal displays the result to the user. The terminal analyzes the received response data and displays it in an easy-to-understand manner for the user. For example, the user receives the result as follows:

[0842] Answer: "The diameter of the Earth is approximately 12,742 km."

[0843] Probability: 80%

[0844] Recommended verification method:

[0845] Check the NASA database

[0846] Read the Encyclopaedia Britannica article

[0847] The user can then perform further verification based on this.

[0848] Example 1

[0849] 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."

[0850] In conventional question-answering systems, users have little means to judge the accuracy and reliability of the generated answers, which creates the risk of receiving incorrect information. Furthermore, users must independently verify the authenticity of the generated answers, which is time-consuming and laborious. Furthermore, due to the lack of functionality to evaluate the reliability of information and recommend appropriate verification methods, users have difficulty finding additional reliable external information sources.

[0851] 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.

[0852] In this invention, the server includes input means for a user to input a question or problem, answer generation means for receiving the question or problem sent from the input means, inputting an inquiry generated based on the question or problem into a generative AI model, and generating an answer from the model, evaluation means for evaluating the reliability of the answer generated by the answer generation means, recommendation means for recommending an additional verification method based on the reliability evaluated by the evaluation means, and display means for formatting and displaying to the user the answer generated by the answer generation means, the reliability evaluation by the evaluation means, and the verification method recommended by the recommendation means. This saves the user the trouble of verifying the reliability of the generated answer, and enables additional verification using a reliable information source.

[0853] "Input means" refers to the means by which a user inputs questions or problems to the system.

[0854] An "answer generation means" is a means for generating a corresponding answer using a generative AI model based on a question or problem from a user.

[0855] The "evaluation means" is a means for calculating and evaluating the likelihood of the generated answer.

[0856] The "recommendation means" is a means for recommending an additional verification method to the user based on the likelihood evaluated by the evaluation means.

[0857] The "display means" is a means for formatting the answer by the answer generation means, the evaluation of the likelihood by the evaluation means, and the verification method by the recommendation means, and displaying them in an easy-to-see manner for the user.

[0858] A "generative AI model" is an artificial intelligence model used to generate answers in natural language based on a user's question or problem.

[0859] "Likelihood" is an evaluation value that indicates the accuracy and reliability of the answer provided by the generative AI model.

[0860] "Validation methods" are additional steps or reference sources used to confirm the accuracy of the answers provided by a generative AI model.

[0861] "Network connection" refers to the communication environment for accessing generative AI models and external information sources.

[0862] This invention is a system in which a user inputs a question or problem, a generative AI model generates an answer based on that information, evaluates its reliability, and recommends an appropriate verification method. The system is composed of a server, terminals, and a communication network.

[0863] System Configuration

[0864] The system mainly consists of the following components:

[0865] 1. Input medium: The interface through which a user enters a question or problem. This can include a web browser or a mobile app.

[0866] 2. Answer generation means: Receives questions or problems from users and generates answers using a generative AI model. The generative AI model can be, for example, a general artificial intelligence model that excels at text generation.

[0867] 3. Evaluation method: The likelihood of the generated answer is calculated and evaluated. Here, a probabilistic evaluation algorithm is used.

[0868] 4. Recommendation: Based on the reliability assessment results, the system recommends additional verification methods to the user. These recommendations are based on databases and reliable external information sources.

[0869] 5. Display: The answers, the confidence scores, and the recommended verification methods are formatted and displayed to the user in an easy-to-read format on the device screen.

[0870] Specific examples

[0871] A user uses a terminal to input a question such as: "What is the diameter of the Earth?" This input is sent to the server as an HTTP request.

[0872] Server Processing

[0873] The server receives an HTTP request sent from the user's device and extracts the question text from it. The extracted question is input into a generative AI model, which then generates an answer to the question. For example, the generated answer might be, "The diameter of the Earth is approximately 12,742 km."

[0874] The generated answer is fed into an algorithm to calculate its likelihood. The algorithm rates the answer as 80% reliable. Based on this rating, it recommends additional verification methods. Because the score is below 90%, it recommends things like "checking the NASA database" or "reading the Encyclopaedia Britannica article."

[0875] Finally, the answer, confidence score, and recommended verification method are formatted into a text like this:

[0876] Answer: "The diameter of the Earth is approximately 12,742 km."

[0877] Probability: 80%

[0878] Recommended verification method:

[0879] Check the NASA database

[0880] Read the Encyclopaedia Britannica article

[0881] What the user sees

[0882] The server sends the formatted results as an HTTP response to the user's device, which receives it and displays it in an easy-to-understand manner for the user.

[0883] In this way, the system of the present invention helps users efficiently determine the accuracy of answers provided by the generating AI and, if necessary, verify them with reliable sources of information.

[0884] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0885] System program processing flow

[0886] Step 1: User Input

[0887] Input: The user types a question or problem into the terminal.

[0888] Output: Text data of the question or problem entered.

[0889] The user inputs a question or problem through the device's interface. For example, they might type, "What is the diameter of the Earth?" The input data is sent to the server as an HTTP request. When the user types a question into the input form and presses the "Submit" button, the device converts the text data into JSON format and sends it to the server.

[0890] Step 2: Receiving the request

[0891] Input: HTTP request sent from the user's device.

[0892] Output: The extracted question text.

[0893] The server receives the HTTP request sent from the user's device and extracts the question text. It then analyzes the JSON format data from the received HTTP request to obtain the specific question. For example, the extracted question might be "What is the diameter of the Earth?"

[0894] Step 3: Input to the generative AI model

[0895] Input: The extracted question text.

[0896] Output: The answer text returned by the generative AI model.

[0897] The server inputs the extracted question text into the generative AI model, which then generates an answer based on this question. Specifically, the prompt "What is the diameter of the Earth?" is input to the generative AI model, and the model responds with an answer such as "The diameter of the Earth is approximately 12,742 km."

[0898] Step 4: Evaluate the likelihood of your answer

[0899] Input: The answer text returned by the generative AI model.

[0900] Output: A likelihood score.

[0901] The server evaluates the reliability of the generated answer. The generated answer is input into the evaluation model and the reliability is calculated. For example, the reliability score is calculated as 80%. Specifically, the evaluation is performed using an algorithm that calculates a reliability score.

[0902] Step 5: Recommend a verification method

[0903] Input: A likelihood score.

[0904] Output: A list of recommended validation methods.

[0905] The server determines the verification method to recommend to the user based on the reliability score. For example, if the score is less than 90%, specific verification methods such as "check the NASA database" or "read the Encyclopaedia Britannica article" are recommended. This is done based on databases and reliable sources.

[0906] Step 6: Formatting the results

[0907] Input: Answer text, confidence score, and a list of recommended verification methods.

[0908] Output: Formatted result data.

[0909] The server formats the answer, confidence score, and recommended verification method, perhaps in a format like JSON or plain text, for example:

[0910] Answer: "The diameter of the Earth is approximately 12,742 km."

[0911] Probability: 80%

[0912] Recommended verification method:

[0913] Check the NASA database

[0914] Read the Encyclopaedia Britannica article

[0915] Step 7: Output to the User

[0916] Input: Formatted result data.

[0917] Output: The result that is displayed to the user.

[0918] The server sends the formatted results as an HTTP response to the user's device. The device then displays the received data in an easy-to-understand format for the user. Specifically, the results are displayed on the screen of a web browser or mobile app. The user can view the generated answer, its accuracy, and the recommended verification method.

[0919] (Application example 1)

[0920] 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."

[0921] In today's world, phishing emails are becoming increasingly sophisticated, making them extremely difficult to identify. Conventional phishing email countermeasures require users to check the emails themselves, which often leads to oversights or incorrect judgments. Furthermore, judgments may be made based on unreliable information. Therefore, there is a need for a system that can quickly and accurately evaluate the reliability of emails received by users and recommend necessary verification methods.

[0922] 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.

[0923] In this invention, the server includes: input means for a user to input a question or problem; answer generation means for generating an answer using a generation AI based on the question or problem sent from the input means; evaluation means for evaluating the reliability of the answer generated by the answer generation means; recommendation means for recommending an additional verification method based on the reliability evaluated by the evaluation means; display means for displaying to the user the answer generated by the answer generation means, the reliability evaluation by the evaluation means, and the verification method recommended by the recommendation means; reliability evaluation means for evaluating the reliability of email text; and warning means for notifying the user that the reliability evaluation means has determined that the email is likely to be a phishing email. This allows the user to quickly determine the reliability of received emails and prevent damage from phishing emails.

[0924] "User" means any person or entity that utilizes the System to enter questions or problems and receive results.

[0925] "Input means" refers to a device or interface that allows a user to input a question or problem in text form.

[0926] An "answer generation means" is a process or system for generating answers using a generation AI based on questions or problems entered by a user.

[0927] An "evaluation method" is a process or system for evaluating the reliability of answers created by generative AI.

[0928] A "recommendation means" is a process or system for recommending an appropriate verification method to a user based on the assessed likelihood.

[0929] A "display means" is a device or interface for visually displaying to a user the generated answer, the likelihood assessment, and the recommended verification method.

[0930] A "trustworthiness assessment means" is a process or system for assessing the trustworthiness of the text of a received email and determining whether it is a phishing email.

[0931] The "warning means" is a process or system for notifying the user when the reliability evaluation means determines that the email is highly likely to be a phishing email.

[0932] A "phishing email" is a fraudulent email sent to steal personal or confidential information from users for malicious purposes.

[0933] This paper describes how to build a system that can quickly and accurately evaluate the reliability of emails received by users and warn them of phishing emails as necessary. This system utilizes a generative AI model to evaluate the reliability of emails and recommend appropriate measures to users.

[0934] The server includes an input means for a user to input a question or problem, an answer generation means for generating an answer using a generation AI based on the question or problem sent from the input means, an evaluation means for evaluating the reliability of the answer generated by the answer generation means, and a recommendation means for recommending an additional verification method based on the reliability evaluated by the evaluation means.The server further includes a display means for displaying to the user the answer generated by the answer generation means, the reliability evaluation by the evaluation means, and the verification method recommended by the recommendation means, a reliability evaluation means for evaluating the reliability of the email text, and a warning means for notifying the user that the email may be a phishing email based on the reliability evaluation means.

[0935] Specifically, the user inputs the text of the email they received and sends it to the server. This input is done via the user's device (smartphone, tablet, PC, etc.). In this process, the email body is input in text format and sent to the server as an HTTP request. For example, the user inputs the following email text:

[0936] "The email said, 'Your account has been locked due to a security issue. Please click the link to verify your account immediately.'"

[0937] The server receives the data sent from the user's device and evaluates the reliability of the email using a generative AI model, such as one from Hugging Face's Transformers library. This model determines the likelihood that the email is a phishing email. For example, it generates a response such as "The email is likely to be phishing."

[0938] The server then recommends appropriate verification methods to the user based on the assessed trustworthiness score. For emails with low trustworthiness, additional verification methods may be recommended. For example, a recommendation such as "Please check official security guidelines" may be given.

[0939] Finally, this information is sent to the user's terminal, where the user can check the results in text format, making it easy for the user to quickly determine the authenticity of the received email and perform further checks if necessary.

[0940] This system allows users to confirm the authenticity of emails in a reliable way, even in an age where it is difficult to distinguish phishing emails.When commercializing the product, it is recommended that devices be standard smartphones, tablets, or PCs, and that software utilize Python or a framework that can handle HTTP requests.

[0941] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0942] Step 1: User enters email text

[0943] The user inputs the text of the received email into the terminal and sends it to the server. The input data could be, for example, the text "Your account has been locked due to a security issue. Please click the link immediately to confirm." This input is sent to the server as an HTTP request.

[0944] Step 2: The server receives the input

[0945] The server receives the HTTP request sent from the user terminal and extracts the email text. For example, it extracts the text "Your account has been locked due to a security issue. Please click the link immediately to confirm." from the request data.

[0946] Step 3: Generative AI reliability assessment

[0947] The server inputs the extracted email text into a generative AI model to evaluate its reliability. The generative AI model uses, for example, Hugging Face's Transformers library to determine whether the email is a phishing email. The model generates an answer such as "This is likely a phishing email." This process uses natural language processing techniques to analyze the input data and calculate a reliability score.

[0948] Step 4: Evaluate the reliability score

[0949] The server analyzes the reliability score provided by the generative AI model and evaluates its reliability. For example, if the reliability score is evaluated as 80%, the next processing step will proceed based on that score. A score analysis algorithm is used for this data processing.

[0950] Step 5: Recommend a verification method

[0951] The server recommends appropriate verification methods to the user based on the reliability score. For example, if the reliability score is low, the server provides a recommendation to "check official security guidelines." This process links the results of the reliability assessment to the user's countermeasure actions.

[0952] Step 6: Formatting the results

[0953] The server then combines the generated answers, confidence ratings, and recommended verification methods into a single format, such as the following:

[0954] "It's likely a phishing email."

[0955] Reliability score: 80%

[0956] Recommended verification: Check official security guidelines

[0957] Step 7: Displaying results to the user

[0958] The device receives the formatted results sent from the server and displays them to the user, who can view the generated answer, its confidence rating, and recommended verification methods, for example, displayed on a smartphone screen.

[0959] Each processing step has been explained in detail above. This system makes it easy for users to quickly determine the authenticity of received emails and to verify them as needed.

[0960] 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.

[0961] The system of the present invention uses generative AI to provide answers to questions or problems entered by users, evaluate their accuracy, and recommend appropriate verification methods. It also incorporates an emotion engine that recognizes emotions from the user's input, adjusts answers based on those emotions, and suggests verification methods.

[0962] User Input

[0963] A user uses a terminal to input a question or problem in text format. For example, a user might type, "The weather has been bad lately. What will happen this weekend?" This input is sent from the terminal to the server. It is sent as an HTTP POST request.

[0964] emotion recognition

[0965] 1. Request received:

[0966] The server receives the HTTP request sent from the terminal, analyzes the request data, and extracts the question text.

[0967] 2. Emotion recognition using emotion engine:

[0968] The extracted text is input into the emotion engine, which analyzes the user's emotions. For example, a question like "The weather has been bad lately. What will happen this weekend?" can be used to identify "negative emotions."

[0969] Answer generation and likelihood assessment

[0970] 1. Enter a question into the generative AI model:

[0971] After recognizing the emotion, the question text is fed into a generative AI model to generate an answer. For example, the generative AI might respond, "It's likely to be sunny over the weekend."

[0972] 2. Adjusting your answers:

[0973] It adjusts the tone and wording of responses based on emotion recognition. If a negative emotion is detected, it will adjust the response, for example, "Don't worry, it's likely to be sunny over the weekend."

[0974] 3. Assessing likelihood:

[0975] The server uses a model to assess the likelihood of the generated answer and calculates a confidence score. For example, suppose the assessment result is 85%.

[0976] Verification method recommendations

[0977] 1. Recommendation generation based on evaluation results:

[0978] Based on the reliability score, the server recommends appropriate verification methods to the user. If the score is below 90%, it suggests checking reliable external databases or expert articles. For example, it recommends "checking weather forecasting services" or "referring to the official website of the Meteorological Agency."

[0979] 2. Result format:

[0980] The server formats the generated answer, the confidence score, and the recommended verification method, for example in JSON format as follows:

[0981] {

[0982] "answer": "Don't worry, it's likely to be sunny over the weekend",

[0983] "certainty_score": 85,

[0984] "recommendations": ["Check the weather forecast service", "See the official Meteorological Agency website"]

[0985] }

[0986] Output to the user

[0987] The server sends the formatted results as an HTTP response to the user's device, which then parses the response data and displays it in a user-friendly format. The user can see the generated answer, its accuracy, and the recommended verification method.

[0988] For example, the user receives the following result:

[0989] Answer: "Don't worry, it's likely to be sunny over the weekend."

[0990] Probability: 85%

[0991] Recommended verification method:

[0992] Check the weather forecast service

[0993] Visit the official Meteorological Bureau website

[0994] In this way, the system of the present invention helps users efficiently judge the accuracy of the answers generated by the AI ​​and, if necessary, allows them to conduct additional verification using more reliable external sources. It also improves the user experience by adjusting the answers to take into account the user's feelings.

[0995] The processing flow will be explained below.

[0996] Step 1:

[0997] The user types a question or problem into the device. The user uses a keyboard or touchscreen to type the question in text format and clicks a submit button. For example, the user might type, "The weather has been bad lately. What's it going to be like this weekend?"

[0998] Step 2:

[0999] The device sends the user's input to the server. The device sends the entered text data to the server as an HTTP POST request. The data sent is in the format {"question": "The weather has been bad lately. What will happen this weekend?"}

[1000] Step 3:

[1001] The server receives the request. The server receives the HTTP request, analyzes its contents, and extracts the question text. Specifically, the question is extracted using request.get("question") .

[1002] Step 4:

[1003] The server inputs the question text into the emotion engine. The server then passes the extracted question text to the emotion engine, which analyzes the user's emotion. For example, the emotion engine recognizes "negative emotion."

[1004] Step 5:

[1005] The server inputs a question into the generative AI model. Based on the results of analysis by the emotion engine, the question text is input into the generative AI model to generate an answer. For example, the generative AI generates the answer "It's likely to be sunny over the weekend."

[1006] Step 6:

[1007] The server adjusts the tone of the generated answer. The tone and expression of the answer are adjusted appropriately depending on the user's emotions. For example, if a negative emotion is detected, the server adjusts the answer to "Don't worry. It's likely to be sunny over the weekend."

[1008] Step 7:

[1009] The server evaluates the reliability of the generated answer. The server inputs the answer into a reliability evaluation model and obtains a reliability score. For example, suppose the evaluation result is 85%.

[1010] Step 8:

[1011] The server generates recommendations based on the evaluation results. If the reliability score is less than 90%, it recommends using a more reliable external database or resource. For example, it suggests "checking the weather forecast service" or "referring to the official website of the Japan Meteorological Agency."

[1012] Step 9:

[1013] The server formats the results: the generated answer, the confidence score, and the recommended verification method in JSON format, for example:

[1014] {

[1015] "answer": "Don't worry, it's likely to be sunny over the weekend",

[1016] "certainty_score": 85,

[1017] "recommendations": ["Check the weather forecast service", "See the official Meteorological Agency website"]

[1018] }

[1019] Step 10:

[1020] The server sends the formatted results to the terminal. The server sends the formatted data to the user terminal as an HTTP response.

[1021] Step 11:

[1022] The terminal displays the result to the user. The terminal analyzes the received response data and displays it in an easy-to-understand manner for the user. For example, the user receives the result as follows:

[1023] Answer: "Don't worry, it's likely to be sunny over the weekend."

[1024] Probability: 85%

[1025] Recommended verification method:

[1026] Check the weather forecast service

[1027] Visit the official Meteorological Bureau website

[1028] Users can then use this information to verify the information with reliable external sources.

[1029] Example 2

[1030] 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."

[1031] Conventional question-answering systems lacked a means for evaluating the reliability of generated answers or a mechanism for adjusting answers based on user emotions. This made it difficult for users to judge the reliability of generated answers, potentially leading to actions based on unreliable information. Furthermore, there were concerns that one-sided answers that did not take the user's emotions into consideration could detract from the user experience. In response to this, the present invention aims to provide more reliable information and improve the user experience by combining reliability evaluation and emotion recognition in an answer generation system using a generative AI model.

[1032] 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.

[1033] In this invention, the server includes: input means for a user to input a question or problem; answer generation means for generating an answer using a generation AI based on the question or problem transmitted from the input means; evaluation means for evaluating the reliability of the answer generated by the answer generation means; recommendation means for recommending an additional verification method based on the reliability evaluated by the evaluation means; emotion recognition means for recognizing emotions in the user's input text and adjusting the answer based on the result; and display means for displaying to the user the answer generated by the answer generation means, the reliability evaluation by the evaluation means, and the verification method recommended by the recommendation means. This makes it possible to evaluate the reliability of the generated answer, provide the user with reliable information, and provide answers that take the user's emotions into consideration, thereby improving the user experience.

[1034] An "input means" is a device or software interface through which a user inputs a question or problem in text form.

[1035] An "answer generation means" is a device or software function that uses generation AI to generate appropriate answers based on questions or problems obtained from an input means.

[1036] The "evaluation means" is a device or software function that evaluates the likelihood of the answer generated by the answer generation means and calculates the reliability as a score.

[1037] The "recommendation means" is a device or software function that recommends additional verification methods to the user based on the likelihood evaluated by the evaluation means.

[1038] An "emotion recognizer" is a device or software function that recognizes the emotion of a user's input text and adjusts the response based on the results.

[1039] The "display means" is a device or software interface that displays to the user the answer from the answer generation means, the evaluation of the likelihood from the evaluation means, and the verification method from the recommendation means.

[1040] MODE FOR CARRYING OUT THE INVENTION

[1041] The system of the present invention uses generative AI to provide answers to questions or problems entered by users, evaluate their accuracy, and recommend appropriate verification methods. This system also incorporates an emotion engine that recognizes the user's emotions and adjusts the answer accordingly.

[1042] Input Method

[1043] A user enters a question or problem in text format using a device. This device can be a PC, smartphone, tablet, etc. For example, they might enter, "The weather has been bad lately. What will happen this weekend?" This input is sent to the server as an HTTP POST request.

[1044] emotion recognition means

[1045] The server receives the HTTP request sent from the device, analyzes the request data, and extracts the question text. The emotion engine then analyzes the user's emotions. The emotion engine uses natural language processing (NLP) technology to recognize "negative emotions," for example, from a question like, "The weather has been bad lately. What will happen this weekend?"

[1046] Answer generation means

[1047] After recognizing the emotion, the server inputs the question text into the generative AI model and generates an answer. For example, if the prompt is "Tell me what the weather will be like this weekend," the generative AI model will respond with "It's likely to be sunny this weekend."

[1048] Answer adjustment

[1049] The tone and wording of the generated response is adjusted based on emotion recognition: if a negative emotion is detected, the response will be adjusted, such as "Don't worry, it's likely to be sunny over the weekend."

[1050] Certainty assessment method

[1051] The server uses a proprietary rating model to evaluate the reliability of the generated answer and calculates a reliability score. For example, suppose the rating result is 85%.

[1052] Recommendation methods

[1053] Based on the confidence score of the generated answer, the server recommends an appropriate verification method: if the score is below 90%, it suggests checking reliable external databases or expert articles, for example, "check the weather forecasting service" or "visit the official weather bureau website."

[1054] Display means

[1055] The server formats the generated answer, its confidence score, and the recommended verification method and sends it to the user's device as an HTTP response. The device analyzes the received response data and displays it in an easy-to-understand format for the user, allowing the user to check the generated answer, its confidence score, and the recommended verification method.

[1056] For example, the user receives the following result:

[1057] Answer: "Don't worry, it's likely to be sunny over the weekend."

[1058] Probability: 85%

[1059] Recommended verification method:

[1060] Check the weather forecast service

[1061] Visit the official Meteorological Bureau website

[1062] The system of the present invention allows users to efficiently judge the accuracy of answers generated by the AI ​​and, if necessary, to conduct additional verification by referring to reliable external sources. Furthermore, by providing answers that take into consideration the user's feelings, the system can improve the user experience.

[1063] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1064] Step 1: User Input

[1065] User: Enters a question or problem in text format using a terminal. For example, "The weather has been bad lately. What will happen this weekend?"

[1066] Input: The user types text into the terminal.

[1067] Specific actions: The user enters a question into the input field on the device and presses the send button.

[1068] Output: The entered text is sent to the server as an HTTP POST request.

[1069] Step 2: Receiving the request

[1070] Server: Receives the HTTP request sent from the device, analyzes the received data, and extracts the question text.

[1071] Input: HTTP POST request data.

[1072] Specific behavior: The server parses the HTTP request and obtains the question text: "The weather has been bad lately. What will happen this weekend?"

[1073] Output: The extracted question text.

[1074] Step 3: Emotion Recognition

[1075] Server: The extracted question text is input into the emotion engine to analyze the user's emotions.

[1076] Input: Question text.

[1077] How it works: The emotion engine uses natural language processing (NLP) technology to analyze text and recognize emotions. It recognizes "negative emotions" from the question, "The weather has been bad lately. What's going to happen this weekend?"

[1078] Output: Emotion recognition result (e.g., "negative").

[1079] Step 4: Answer Generation

[1080] Server: After recognizing the emotion, the question text is input into the generative AI model to generate an answer.

[1081] Input: Question text, and emotion recognition results.

[1082] Specific operation: The question text is converted into a prompt sentence and input into the generative AI model. Example prompt sentence: "Tell me what the weather will be like this weekend." The generative AI model responds, "It's likely to be sunny this weekend."

[1083] Output: The generated answer.

[1084] Step 5: Adjust your answers

[1085] Server: Adjusts the tone and expression of the generated response based on the emotion recognition results.

[1086] Input: Generated answers, and emotion recognition results.

[1087] Action: If you recognize negative emotions, soften the tone of your response. "Don't worry, it's likely to be sunny over the weekend."

[1088] Output: The adjusted answer.

[1089] Step 6: Assess likelihood

[1090] Server: Uses a dedicated evaluation model to evaluate the reliability of the generated answer and calculates a reliability score.

[1091] Input: The adjusted answer.

[1092] Specific operation: Apply the evaluation model to the adjusted answer and calculate the reliability score. Example evaluation result: 85%.

[1093] Output: A likelihood score (e.g., 85%).

[1094] Step 7: Recommendation Generation

[1095] Server: Recommends an appropriate verification method based on the likelihood score.

[1096] Input: A likelihood score.

[1097] What it does: If the score is below 90%, generate recommendations to check more reliable sources, e.g., "Check the weather forecast service" and "Visit the official weather bureau website."

[1098] Output: Recommended validation method.

[1099] Step 8: Formatting the results

[1100] Server: Formats the generated answers, confidence scores, and verification methods.

[1101] Inputs: Calibrated answer, likelihood score, and recommended validation method.

[1102] Specific operation: Formats data into JSON format.

[1103] Output: The formatted results.

[1104] Step 9: Output to the User

[1105] Server: Sends the formatted results to the user device as an HTTP response.

[1106] Input: The formatted result.

[1107] Specific operation: The result is sent as an HTTP response. The device analyzes the received data and instructs it to display it.

[1108] Output: Display on the user's terminal.

[1109] This allows the user to see the generated answer, its likelihood, and the recommended verification method.

[1110] (Application example 2)

[1111] 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."

[1112] In modern digital communication, users are becoming increasingly concerned about messages suspected of being phishing scams or cyberattacks. There is a need for systems that can quickly and accurately assess the trustworthiness of such messages and recommend appropriate actions to users. However, existing systems lack an integrated approach that takes user emotions into consideration while providing plausible answers using generative AI, assessing their trustworthiness, and recommending appropriate verification methods. This poses a challenge, making it difficult to ensure users' sense of security.

[1113] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[1114] In this invention, the server includes an input means for a user to input a question or problem, an answer generation means for generating an answer using a generation AI, an evaluation means for evaluating the reliability of the generated answer, a recommendation means for recommending an additional verification method based on the reliability, an emotion recognition means for recognizing the emotion of the user when inputting a suspicious message and adjusting the answer by the generation AI, and a display means for displaying to the user the answer by the answer generation means, the reliability evaluation by the evaluation means, and the verification method by the recommendation means. This allows the user to respond quickly and appropriately to security risks such as phishing scams, giving them a sense of security.

[1115] "Input means" refers to a means by which a user inputs a question or problem in text format.

[1116] An "answer generation means" is a means for generating answers using a generation AI based on input questions or problems.

[1117] The "evaluation means" is a means for evaluating the reliability of the generated answer.

[1118] A "recommendation means" is a means for suggesting an appropriate verification method to a user based on the likelihood of success.

[1119] The "display means" is a means for displaying the generated answer, the likelihood assessment, and the recommended verification method to the user.

[1120] An "emotion recognition means" is a means for recognizing emotions from the text entered by the user and adjusting the response of the generating AI according to those emotions.

[1121] This invention provides a system that, when a user inputs a suspicious message or question, uses a generation AI to generate an answer based on the input, evaluates its reliability, and recommends an appropriate verification method. The following describes an embodiment of this system.

[1122] The user first enters a question or problem in text form, for example via a smartphone or computer terminal, and the text is sent to the server, which receives the request and begins analyzing it.

[1123] The server first uses an emotion recognition means to recognize emotions from the text entered by the user. For this emotion recognition, a software library for emotion analysis (e.g., TextBlob) or an emotion analysis model is used. For example, the emotion "anxiety" is recognized from the text "Does this seem like a phishing scam?"

[1124] Next, the server uses an answer generation means to input the input question into a generative AI model and generate an answer. As a generative AI model, for example, a text generation model (e.g., GPT-3) is used. The generative AI model is physically located on a server in the cloud or in a high-performance computing environment running locally. The generated answer is provided in the form of, for example, "This may be a phishing scam. Please be careful."

[1125] The server then uses an evaluation tool to evaluate the reliability of the generated answer. This evaluation uses a dedicated algorithm or model to calculate a reliability score. For example, suppose the reliability of the generated answer is evaluated as 85%.

[1126] Based on the evaluation results, the recommendation tool recommends appropriate verification methods to the user. For example, if the reliability score is less than 90%, verification methods such as "refer to a trusted security site" or "consult an expert" are suggested.

[1127] Finally, these answers, likelihood scores, and verification method recommendations are displayed to the user through a display means, such as an application or web browser that provides a user interface.

[1128] As a concrete example, an example of a prompt sentence input by the user is shown below.

[1129] Example question: "Does this seem like a phishing scam?"

[1130] Sample response: "This may be a phishing scam. Be careful."

[1131] Example of certainty: "85%"

[1132] Recommended verification methods: "Refer to a trusted security site" or "Consult an expert"

[1133] This allows users to respond quickly and appropriately to security risks such as phishing scams, giving them a sense of security.

[1134] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1135] Step 1:

[1136] The user enters a question or problem in text format. The entered text data is sent from the terminal to the server as an HTTP POST request, which transfers the user's input data to the server.

[1137] Step 2:

[1138] The server parses the received HTTP request and extracts the question text. The server passes this text to the emotion engine, which then recognizes the user's emotion from the text (e.g., using TextBlob or an emotion analysis model). The input is the question text, and the output is the emotion recognition result (e.g., "anxious").

[1139] Step 3:

[1140] Based on the emotion recognition results, the server inputs the question text into a generative AI model and generates an answer (e.g., using GPT-3). The input is the question text, and the output is the generated answer text (e.g., "This may be a phishing scam. Please be careful").

[1141] Step 4:

[1142] Based on the generated answer text, the server uses an evaluation method to evaluate the reliability of the answer. A specific algorithm or model is used to calculate a reliability score (e.g., 85%). The input is the generated answer text, and the output is the reliability score.

[1143] Step 5:

[1144] Based on the reliability score, the server uses a recommendation mechanism to suggest additional verification methods to the user (e.g., "refer to a trusted security site" or "consult an expert"). The input is the reliability score, and the output is a list of verification methods.

[1145] Step 6:

[1146] The server formats the generated answer, the reliability score, and the recommended verification method in JSON format and sends it to the user terminal as an HTTP response. The input is the generated answer text, the reliability score, and the list of verification methods, and the output is the formatted JSON data.

[1147] Step 7:

[1148] The terminal parses the received JSON data and displays it in a user-friendly format, allowing the user to see the generated answer, its accuracy, and the recommended verification method. The input is formatted JSON data, and the output is the information displayed on the user interface.

[1149] 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.

[1150] 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.

[1151] 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.

[1152] [Fourth embodiment]

[1153] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1154] 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.

[1155] 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).

[1156] 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.

[1157] 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.

[1158] 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).

[1159] 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.

[1160] 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.

[1161] 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.

[1162] 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.

[1163] 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.

[1164] 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.

[1165] 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."

[1166] The system of the present invention executes a series of processes in which a user inputs a question or problem, and a generation AI generates an answer based on that information, evaluates its reliability, and recommends an appropriate verification method.

[1167] User Input

[1168] A user uses a terminal to input a question or problem in text format. For example, the user might input "What is the diameter of the Earth?" This input is sent from the terminal to a server. This process is performed by sending the text data entered by the user to the server as an HTTP request.

[1169] Server-side answer generation

[1170] 1. Request received:

[1171] The server receives an HTTP request sent from a user terminal and extracts the question text. For example, if the request data is "What is the diameter of the Earth?", the server analyzes it and extracts it.

[1172] 2. Enter a question into the generative AI model:

[1173] The extracted question text is input into a generative AI model, which generates an answer based on this question. For example, the generated answer might be "The diameter of the Earth is approximately 12,742 km."

[1174] 3. Rate the confidence of your answer:

[1175] The server uses a model to calculate the likelihood of the generated answer, for example rating it as having a likelihood score of 80%.

[1176] Verification method recommendations

[1177] 1. Recommendation generation based on evaluation results:

[1178] The server determines the verification method to recommend to the user based on the reliability score. For example, if the reliability score is less than 90%, the server will recommend specific verification methods such as "check the NASA database" or "read the Encyclopaedia Britannica article."

[1179] 2. Result format:

[1180] Finally, we format the data to include the answer, a confidence score, and a recommended verification method. For example, the formatted result might look like this:

[1181] Answer: "The diameter of the Earth is approximately 12,742 km."

[1182] Probability: 80%

[1183] Recommended verification method:

[1184] Check the NASA database

[1185] Read the Encyclopaedia Britannica article

[1186] Output to the user

[1187] The server sends the formatted results as an HTTP response to the user's device, which then displays the received data in an easy-to-understand format for the user, who can see the generated answer, its accuracy, and the recommended verification method.

[1188] For example, the user receives the following result:

[1189] Answer: "The diameter of the Earth is approximately 12,742 km."

[1190] Probability: 80%

[1191] Recommended verification method:

[1192] Check the NASA database

[1193] Read the Encyclopaedia Britannica article

[1194] In this way, the system of the present invention helps users efficiently determine the accuracy of answers provided by the generating AI and, if necessary, verify them with more reliable sources of information.

[1195] The processing flow will be explained below.

[1196] Step 1:

[1197] The user enters a question or problem into the terminal. The user enters the question in text format using the touchscreen or keyboard and presses the submit button. At this time, the user's input is sent to the server as an HTTP POST request.

[1198] Step 2:

[1199] The terminal sends the user's input to the server. The terminal sends the text data entered by the user to the server as an HTTP request. For example, if the request data is "What is the diameter of the Earth?", this is sent to the server.

[1200] Step 3:

[1201] The server receives the request. The server receives the HTTP request sent from the device, analyzes the request data, and extracts the question text. Specifically, it extracts it using a method such as request.get("question").

[1202] Step 4:

[1203] The server inputs a question into the generative AI model. The extracted question text is input into the generative AI to generate an answer. If the generative AI generates an answer such as "The diameter of the Earth is approximately 12,742 km," this is accepted.

[1204] Step 5:

[1205] The server evaluates the reliability of the answer. A reliability evaluation model is used to evaluate the reliability of the generated answer. Specifically, the generated answer is input into the evaluation model and a calculated score is obtained. For example, suppose the evaluation result is 80%.

[1206] Step 6:

[1207] The server generates recommendations based on the evaluation results. The server recommends appropriate verification methods to the user based on the reliability score. If the score is less than 90%, the server recommends viewing reliable external databases or articles. For example, the server generates specific recommendations such as "Check the NASA database" or "Read the Encyclopaedia Britannica article."

[1208] Step 7:

[1209] The server formats the results. The server formats the generated answer, confidence score, and recommended verification method together, for example in JSON format like this:

[1210] {

[1211] "answer": "The diameter of the Earth is approximately 12742 km",

[1212] "certainty_score": 80,

[1213] "recommendations": ["Check out the NASA database", "Read the Encyclopaedia Britannica article"]

[1214] }

[1215] Step 8:

[1216] The server sends the formatted results to the terminal. The server sends the formatted data to the user terminal as an HTTP response.

[1217] Step 9:

[1218] The terminal displays the result to the user. The terminal analyzes the received response data and displays it in an easy-to-understand manner for the user. For example, the user receives the result as follows:

[1219] Answer: "The diameter of the Earth is approximately 12,742 km."

[1220] Probability: 80%

[1221] Recommended verification method:

[1222] Check the NASA database

[1223] Read the Encyclopaedia Britannica article

[1224] The user can then perform further verification based on this.

[1225] Example 1

[1226] 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."

[1227] In conventional question-answering systems, users have little means to judge the accuracy and reliability of the generated answers, which creates the risk of receiving incorrect information. Furthermore, users must independently verify the authenticity of the generated answers, which is time-consuming and laborious. Furthermore, due to the lack of functionality to evaluate the reliability of information and recommend appropriate verification methods, users have difficulty finding additional reliable external information sources.

[1228] 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.

[1229] In this invention, the server includes input means for a user to input a question or problem, answer generation means for receiving the question or problem sent from the input means, inputting an inquiry generated based on the question or problem into a generative AI model, and generating an answer from the model, evaluation means for evaluating the reliability of the answer generated by the answer generation means, recommendation means for recommending an additional verification method based on the reliability evaluated by the evaluation means, and display means for formatting and displaying to the user the answer generated by the answer generation means, the reliability evaluation by the evaluation means, and the verification method recommended by the recommendation means. This saves the user the trouble of verifying the reliability of the generated answer, and enables additional verification using a reliable information source.

[1230] "Input means" refers to the means by which a user inputs questions or problems to the system.

[1231] An "answer generation means" is a means for generating a corresponding answer using a generative AI model based on a question or problem from a user.

[1232] The "evaluation means" is a means for calculating and evaluating the likelihood of the generated answer.

[1233] The "recommendation means" is a means for recommending an additional verification method to the user based on the likelihood evaluated by the evaluation means.

[1234] The "display means" is a means for formatting the answer by the answer generation means, the evaluation of the likelihood by the evaluation means, and the verification method by the recommendation means, and displaying them in an easy-to-see manner for the user.

[1235] A "generative AI model" is an artificial intelligence model used to generate answers in natural language based on a user's question or problem.

[1236] "Likelihood" is an evaluation value that indicates the accuracy and reliability of the answer provided by the generative AI model.

[1237] "Validation methods" are additional steps or reference sources used to confirm the accuracy of the answers provided by a generative AI model.

[1238] "Network connection" refers to the communication environment for accessing generative AI models and external information sources.

[1239] This invention is a system in which a user inputs a question or problem, a generative AI model generates an answer based on that information, evaluates its reliability, and recommends an appropriate verification method. The system is composed of a server, terminals, and a communication network.

[1240] System Configuration

[1241] The system mainly consists of the following components:

[1242] 1. Input medium: The interface through which a user enters a question or problem. This can include a web browser or a mobile app.

[1243] 2. Answer generation means: Receives questions or problems from users and generates answers using a generative AI model. The generative AI model can be, for example, a general artificial intelligence model that excels at text generation.

[1244] 3. Evaluation method: The likelihood of the generated answer is calculated and evaluated. Here, a probabilistic evaluation algorithm is used.

[1245] 4. Recommendation: Based on the reliability assessment results, the system recommends additional verification methods to the user. These recommendations are based on databases and reliable external information sources.

[1246] 5. Display: The answers, the confidence scores, and the recommended verification methods are formatted and displayed to the user in an easy-to-read format on the device screen.

[1247] Specific examples

[1248] A user uses a terminal to input a question such as: "What is the diameter of the Earth?" This input is sent to the server as an HTTP request.

[1249] Server Processing

[1250] The server receives an HTTP request sent from the user's device and extracts the question text from it. The extracted question is input into a generative AI model, which then generates an answer to the question. For example, the generated answer might be, "The diameter of the Earth is approximately 12,742 km."

[1251] The generated answer is fed into an algorithm to calculate its likelihood. The algorithm rates the answer as 80% reliable. Based on this rating, it recommends additional verification methods. Because the score is below 90%, it recommends things like "checking the NASA database" or "reading the Encyclopaedia Britannica article."

[1252] Finally, the answer, confidence score, and recommended verification method are formatted into a text like this:

[1253] Answer: "The diameter of the Earth is approximately 12,742 km."

[1254] Probability: 80%

[1255] Recommended verification method:

[1256] Check the NASA database

[1257] Read the Encyclopaedia Britannica article

[1258] What the user sees

[1259] The server sends the formatted results as an HTTP response to the user's device, which receives it and displays it in an easy-to-understand manner for the user.

[1260] In this way, the system of the present invention helps users efficiently determine the accuracy of answers provided by the generating AI and, if necessary, verify them with reliable sources of information.

[1261] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1262] System program processing flow

[1263] Step 1: User Input

[1264] Input: The user types a question or problem into the terminal.

[1265] Output: Text data of the question or problem entered.

[1266] The user inputs a question or problem through the device's interface. For example, they might type, "What is the diameter of the Earth?" The input data is sent to the server as an HTTP request. When the user types a question into the input form and presses the "Submit" button, the device converts the text data into JSON format and sends it to the server.

[1267] Step 2: Receiving the request

[1268] Input: HTTP request sent from the user's device.

[1269] Output: The extracted question text.

[1270] The server receives the HTTP request sent from the user's device and extracts the question text. It then analyzes the JSON format data from the received HTTP request to obtain the specific question. For example, the extracted question might be "What is the diameter of the Earth?"

[1271] Step 3: Input to the generative AI model

[1272] Input: The extracted question text.

[1273] Output: The answer text returned by the generative AI model.

[1274] The server inputs the extracted question text into the generative AI model, which then generates an answer based on this question. Specifically, the prompt "What is the diameter of the Earth?" is input to the generative AI model, and the model responds with an answer such as "The diameter of the Earth is approximately 12,742 km."

[1275] Step 4: Evaluate the likelihood of your answer

[1276] Input: The answer text returned by the generative AI model.

[1277] Output: A likelihood score.

[1278] The server evaluates the reliability of the generated answer. The generated answer is input into the evaluation model and the reliability is calculated. For example, the reliability score is calculated as 80%. Specifically, the evaluation is performed using an algorithm that calculates a reliability score.

[1279] Step 5: Recommend a verification method

[1280] Input: A likelihood score.

[1281] Output: A list of recommended validation methods.

[1282] The server determines the verification method to recommend to the user based on the reliability score. For example, if the score is less than 90%, specific verification methods such as "check the NASA database" or "read the Encyclopaedia Britannica article" are recommended. This is done based on databases and reliable sources.

[1283] Step 6: Formatting the results

[1284] Input: Answer text, confidence score, and a list of recommended verification methods.

[1285] Output: Formatted result data.

[1286] The server formats the answer, confidence score, and recommended verification method, perhaps in a format like JSON or plain text, for example:

[1287] Answer: "The diameter of the Earth is approximately 12,742 km."

[1288] Probability: 80%

[1289] Recommended verification method:

[1290] Check the NASA database

[1291] Read the Encyclopaedia Britannica article

[1292] Step 7: Output to the User

[1293] Input: Formatted result data.

[1294] Output: The result that is displayed to the user.

[1295] The server sends the formatted results as an HTTP response to the user's device. The device then displays the received data in an easy-to-understand format for the user. Specifically, the results are displayed on the screen of a web browser or mobile app. The user can view the generated answer, its accuracy, and the recommended verification method.

[1296] (Application example 1)

[1297] 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."

[1298] In today's world, phishing emails are becoming increasingly sophisticated, making them extremely difficult to identify. Conventional phishing email countermeasures require users to check the emails themselves, which often leads to oversights or incorrect judgments. Furthermore, judgments may be made based on unreliable information. Therefore, there is a need for a system that can quickly and accurately evaluate the reliability of emails received by users and recommend necessary verification methods.

[1299] 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.

[1300] In this invention, the server includes: input means for a user to input a question or problem; answer generation means for generating an answer using a generation AI based on the question or problem sent from the input means; evaluation means for evaluating the reliability of the answer generated by the answer generation means; recommendation means for recommending an additional verification method based on the reliability evaluated by the evaluation means; display means for displaying to the user the answer generated by the answer generation means, the reliability evaluation by the evaluation means, and the verification method recommended by the recommendation means; reliability evaluation means for evaluating the reliability of email text; and warning means for notifying the user that the reliability evaluation means has determined that the email is likely to be a phishing email. This allows the user to quickly determine the reliability of received emails and prevent damage from phishing emails.

[1301] "User" means any person or entity that utilizes the System to enter questions or problems and receive results.

[1302] "Input means" refers to a device or interface that allows a user to input a question or problem in text form.

[1303] An "answer generation means" is a process or system for generating answers using a generation AI based on questions or problems entered by a user.

[1304] An "evaluation method" is a process or system for evaluating the reliability of answers created by generative AI.

[1305] A "recommendation means" is a process or system for recommending an appropriate verification method to a user based on the assessed likelihood.

[1306] A "display means" is a device or interface for visually displaying to a user the generated answer, the likelihood assessment, and the recommended verification method.

[1307] A "trustworthiness assessment means" is a process or system for assessing the trustworthiness of the text of a received email and determining whether it is a phishing email.

[1308] The "warning means" is a process or system for notifying the user when the reliability evaluation means determines that the email is highly likely to be a phishing email.

[1309] A "phishing email" is a fraudulent email sent to steal personal or confidential information from users for malicious purposes.

[1310] This paper describes how to build a system that can quickly and accurately evaluate the reliability of emails received by users and warn them of phishing emails as necessary. This system utilizes a generative AI model to evaluate the reliability of emails and recommend appropriate measures to users.

[1311] The server includes an input means for a user to input a question or problem, an answer generation means for generating an answer using a generation AI based on the question or problem sent from the input means, an evaluation means for evaluating the reliability of the answer generated by the answer generation means, and a recommendation means for recommending an additional verification method based on the reliability evaluated by the evaluation means.The server further includes a display means for displaying to the user the answer generated by the answer generation means, the reliability evaluation by the evaluation means, and the verification method recommended by the recommendation means, a reliability evaluation means for evaluating the reliability of the email text, and a warning means for notifying the user that the email may be a phishing email based on the reliability evaluation means.

[1312] Specifically, the user inputs the text of the email they received and sends it to the server. This input is done via the user's device (smartphone, tablet, PC, etc.). In this process, the email body is input in text format and sent to the server as an HTTP request. For example, the user inputs the following email text:

[1313] "The email said, 'Your account has been locked due to a security issue. Please click the link to verify your account immediately.'"

[1314] The server receives the data sent from the user's device and evaluates the reliability of the email using a generative AI model, such as one from Hugging Face's Transformers library. This model determines the likelihood that the email is a phishing email. For example, it generates a response such as "The email is likely to be phishing."

[1315] The server then recommends appropriate verification methods to the user based on the assessed trustworthiness score. For emails with low trustworthiness, additional verification methods may be recommended. For example, a recommendation such as "Please check official security guidelines" may be given.

[1316] Finally, this information is sent to the user's terminal, where the user can check the results in text format, making it easy for the user to quickly determine the authenticity of the received email and perform further checks if necessary.

[1317] This system allows users to confirm the authenticity of emails in a reliable way, even in an age where it is difficult to distinguish phishing emails.When commercializing the product, it is recommended that devices be standard smartphones, tablets, or PCs, and that software utilize Python or a framework that can handle HTTP requests.

[1318] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1319] Step 1: User enters email text

[1320] The user inputs the text of the received email into the terminal and sends it to the server. The input data could be, for example, the text "Your account has been locked due to a security issue. Please click the link immediately to confirm." This input is sent to the server as an HTTP request.

[1321] Step 2: The server receives the input

[1322] The server receives the HTTP request sent from the user terminal and extracts the email text. For example, it extracts the text "Your account has been locked due to a security issue. Please click the link immediately to confirm." from the request data.

[1323] Step 3: Generative AI reliability assessment

[1324] The server inputs the extracted email text into a generative AI model to evaluate its reliability. The generative AI model uses, for example, Hugging Face's Transformers library to determine whether the email is a phishing email. The model generates an answer such as "This is likely a phishing email." This process uses natural language processing techniques to analyze the input data and calculate a reliability score.

[1325] Step 4: Evaluate the reliability score

[1326] The server analyzes the reliability score provided by the generative AI model and evaluates its reliability. For example, if the reliability score is evaluated as 80%, the next processing step will proceed based on that score. A score analysis algorithm is used for this data processing.

[1327] Step 5: Recommend a verification method

[1328] The server recommends appropriate verification methods to the user based on the reliability score. For example, if the reliability score is low, the server provides a recommendation to "check official security guidelines." This process links the results of the reliability assessment to the user's countermeasure actions.

[1329] Step 6: Formatting the results

[1330] The server then combines the generated answers, confidence ratings, and recommended verification methods into a single format, such as the following:

[1331] "It's likely a phishing email."

[1332] Reliability score: 80%

[1333] Recommended verification: Check official security guidelines

[1334] Step 7: Displaying results to the user

[1335] The device receives the formatted results sent from the server and displays them to the user, who can view the generated answer, its confidence rating, and recommended verification methods, for example, displayed on a smartphone screen.

[1336] Each processing step has been explained in detail above. This system makes it easy for users to quickly determine the authenticity of received emails and to verify them as needed.

[1337] 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.

[1338] The system of the present invention uses generative AI to provide answers to questions or problems entered by users, evaluate their accuracy, and recommend appropriate verification methods. It also incorporates an emotion engine that recognizes emotions from the user's input, adjusts answers based on those emotions, and suggests verification methods.

[1339] User Input

[1340] A user uses a terminal to input a question or problem in text format. For example, a user might type, "The weather has been bad lately. What will happen this weekend?" This input is sent from the terminal to the server. It is sent as an HTTP POST request.

[1341] emotion recognition

[1342] 1. Request received:

[1343] The server receives the HTTP request sent from the terminal, analyzes the request data, and extracts the question text.

[1344] 2. Emotion recognition using emotion engine:

[1345] The extracted text is input into the emotion engine, which analyzes the user's emotions. For example, a question like "The weather has been bad lately. What will happen this weekend?" can be used to identify "negative emotions."

[1346] Answer generation and likelihood assessment

[1347] 1. Enter a question into the generative AI model:

[1348] After recognizing the emotion, the question text is fed into a generative AI model to generate an answer. For example, the generative AI might respond, "It's likely to be sunny over the weekend."

[1349] 2. Adjusting your answers:

[1350] It adjusts the tone and wording of responses based on emotion recognition. If a negative emotion is detected, it will adjust the response, for example, "Don't worry, it's likely to be sunny over the weekend."

[1351] 3. Assessing likelihood:

[1352] The server uses a model to assess the likelihood of the generated answer and calculates a confidence score. For example, suppose the assessment result is 85%.

[1353] Verification method recommendations

[1354] 1. Recommendation generation based on evaluation results:

[1355] Based on the reliability score, the server recommends appropriate verification methods to the user. If the score is below 90%, it suggests checking reliable external databases or expert articles. For example, it recommends "checking weather forecasting services" or "referring to the official website of the Meteorological Agency."

[1356] 2. Result format:

[1357] The server formats the generated answer, the confidence score, and the recommended verification method, for example in JSON format as follows:

[1358] {

[1359] "answer": "Don't worry, it's likely to be sunny over the weekend",

[1360] "certainty_score": 85,

[1361] "recommendations": ["Check the weather forecast service", "See the official Meteorological Agency website"]

[1362] }

[1363] Output to the user

[1364] The server sends the formatted results as an HTTP response to the user's device, which then parses the response data and displays it in a user-friendly format. The user can see the generated answer, its accuracy, and the recommended verification method.

[1365] For example, the user receives the following result:

[1366] Answer: "Don't worry, it's likely to be sunny over the weekend."

[1367] Probability: 85%

[1368] Recommended verification method:

[1369] Check the weather forecast service

[1370] Visit the official Meteorological Bureau website

[1371] In this way, the system of the present invention helps users efficiently judge the accuracy of the answers generated by the AI ​​and, if necessary, allows them to conduct additional verification using more reliable external sources. It also improves the user experience by adjusting the answers to take into account the user's feelings.

[1372] The processing flow will be explained below.

[1373] Step 1:

[1374] The user types a question or problem into the device. The user uses a keyboard or touchscreen to type the question in text format and clicks a submit button. For example, the user might type, "The weather has been bad lately. What's it going to be like this weekend?"

[1375] Step 2:

[1376] The device sends the user's input to the server. The device sends the entered text data to the server as an HTTP POST request. The data sent is in the format {"question": "The weather has been bad lately. What will happen this weekend?"}

[1377] Step 3:

[1378] The server receives the request. The server receives the HTTP request, analyzes its contents, and extracts the question text. Specifically, the question is extracted using request.get("question") .

[1379] Step 4:

[1380] The server inputs the question text into the emotion engine. The server then passes the extracted question text to the emotion engine, which analyzes the user's emotion. For example, the emotion engine recognizes "negative emotion."

[1381] Step 5:

[1382] The server inputs a question into the generative AI model. Based on the results of analysis by the emotion engine, the question text is input into the generative AI model to generate an answer. For example, the generative AI generates the answer "It's likely to be sunny over the weekend."

[1383] Step 6:

[1384] The server adjusts the tone of the generated answer. The tone and expression of the answer are adjusted appropriately depending on the user's emotions. For example, if a negative emotion is detected, the server adjusts the answer to "Don't worry. It's likely to be sunny over the weekend."

[1385] Step 7:

[1386] The server evaluates the reliability of the generated answer. The server inputs the answer into a reliability evaluation model and obtains a reliability score. For example, suppose the evaluation result is 85%.

[1387] Step 8:

[1388] The server generates recommendations based on the evaluation results. If the reliability score is less than 90%, it recommends using a more reliable external database or resource. For example, it suggests "checking the weather forecast service" or "referring to the official website of the Japan Meteorological Agency."

[1389] Step 9:

[1390] The server formats the results: the generated answer, the confidence score, and the recommended verification method in JSON format, for example:

[1391] {

[1392] "answer": "Don't worry, it's likely to be sunny over the weekend",

[1393] "certainty_score": 85,

[1394] "recommendations": ["Check the weather forecast service", "See the official Meteorological Agency website"]

[1395] }

[1396] Step 10:

[1397] The server sends the formatted results to the terminal. The server sends the formatted data to the user terminal as an HTTP response.

[1398] Step 11:

[1399] The terminal displays the result to the user. The terminal analyzes the received response data and displays it in an easy-to-understand manner for the user. For example, the user receives the result as follows:

[1400] Answer: "Don't worry, it's likely to be sunny over the weekend."

[1401] Probability: 85%

[1402] Recommended verification method:

[1403] Check the weather forecast service

[1404] Visit the official Meteorological Bureau website

[1405] Users can then use this information to verify the information with reliable external sources.

[1406] Example 2

[1407] 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."

[1408] Conventional question-answering systems lacked a means for evaluating the reliability of generated answers or a mechanism for adjusting answers based on user emotions. This made it difficult for users to judge the reliability of generated answers, potentially leading to actions based on unreliable information. Furthermore, there were concerns that one-sided answers that did not take the user's emotions into consideration could detract from the user experience. In response to this, the present invention aims to provide more reliable information and improve the user experience by combining reliability evaluation and emotion recognition in an answer generation system using a generative AI model.

[1409] 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.

[1410] In this invention, the server includes: input means for a user to input a question or problem; answer generation means for generating an answer using a generation AI based on the question or problem transmitted from the input means; evaluation means for evaluating the reliability of the answer generated by the answer generation means; recommendation means for recommending an additional verification method based on the reliability evaluated by the evaluation means; emotion recognition means for recognizing emotions in the user's input text and adjusting the answer based on the result; and display means for displaying to the user the answer generated by the answer generation means, the reliability evaluation by the evaluation means, and the verification method recommended by the recommendation means. This makes it possible to evaluate the reliability of the generated answer, provide the user with reliable information, and provide answers that take the user's emotions into consideration, thereby improving the user experience.

[1411] An "input means" is a device or software interface through which a user inputs a question or problem in text form.

[1412] An "answer generation means" is a device or software function that uses generation AI to generate appropriate answers based on questions or problems obtained from an input means.

[1413] The "evaluation means" is a device or software function that evaluates the likelihood of the answer generated by the answer generation means and calculates the reliability as a score.

[1414] The "recommendation means" is a device or software function that recommends additional verification methods to the user based on the likelihood evaluated by the evaluation means.

[1415] An "emotion recognizer" is a device or software function that recognizes the emotion of a user's input text and adjusts the response based on the results.

[1416] The "display means" is a device or software interface that displays to the user the answer from the answer generation means, the evaluation of the likelihood from the evaluation means, and the verification method from the recommendation means.

[1417] MODE FOR CARRYING OUT THE INVENTION

[1418] The system of the present invention uses generative AI to provide answers to questions or problems entered by users, evaluate their accuracy, and recommend appropriate verification methods. This system also incorporates an emotion engine that recognizes the user's emotions and adjusts the answer accordingly.

[1419] Input Method

[1420] A user enters a question or problem in text format using a device. This device can be a PC, smartphone, tablet, etc. For example, they might enter, "The weather has been bad lately. What will happen this weekend?" This input is sent to the server as an HTTP POST request.

[1421] emotion recognition means

[1422] The server receives the HTTP request sent from the device, analyzes the request data, and extracts the question text. The emotion engine then analyzes the user's emotions. The emotion engine uses natural language processing (NLP) technology to recognize "negative emotions," for example, from a question like, "The weather has been bad lately. What will happen this weekend?"

[1423] Answer generation means

[1424] After recognizing the emotion, the server inputs the question text into the generative AI model and generates an answer. For example, if the prompt is "Tell me what the weather will be like this weekend," the generative AI model will respond with "It's likely to be sunny this weekend."

[1425] Answer adjustment

[1426] The tone and wording of the generated response is adjusted based on emotion recognition: if a negative emotion is detected, the response will be adjusted, such as "Don't worry, it's likely to be sunny over the weekend."

[1427] Certainty assessment method

[1428] The server uses a proprietary rating model to evaluate the reliability of the generated answer and calculates a reliability score. For example, suppose the rating result is 85%.

[1429] Recommendation methods

[1430] Based on the confidence score of the generated answer, the server recommends an appropriate verification method: if the score is below 90%, it suggests checking reliable external databases or expert articles, for example, "check the weather forecasting service" or "visit the official weather bureau website."

[1431] Display means

[1432] The server formats the generated answer, its confidence score, and the recommended verification method and sends it to the user's device as an HTTP response. The device analyzes the received response data and displays it in an easy-to-understand format for the user, allowing the user to check the generated answer, its confidence score, and the recommended verification method.

[1433] For example, the user receives the following result:

[1434] Answer: "Don't worry, it's likely to be sunny over the weekend."

[1435] Probability: 85%

[1436] Recommended verification method:

[1437] Check the weather forecast service

[1438] Visit the official Meteorological Bureau website

[1439] The system of the present invention allows users to efficiently judge the accuracy of answers generated by the AI ​​and, if necessary, to conduct additional verification by referring to reliable external sources. Furthermore, by providing answers that take into consideration the user's feelings, the system can improve the user experience.

[1440] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1441] Step 1: User Input

[1442] User: Enters a question or problem in text format using a terminal. For example, "The weather has been bad lately. What will happen this weekend?"

[1443] Input: The user types text into the terminal.

[1444] Specific actions: The user enters a question into the input field on the device and presses the send button.

[1445] Output: The entered text is sent to the server as an HTTP POST request.

[1446] Step 2: Receiving the request

[1447] Server: Receives the HTTP request sent from the device, analyzes the received data, and extracts the question text.

[1448] Input: HTTP POST request data.

[1449] Specific behavior: The server parses the HTTP request and obtains the question text: "The weather has been bad lately. What will happen this weekend?"

[1450] Output: The extracted question text.

[1451] Step 3: Emotion Recognition

[1452] Server: The extracted question text is input into the emotion engine to analyze the user's emotions.

[1453] Input: Question text.

[1454] How it works: The emotion engine uses natural language processing (NLP) technology to analyze text and recognize emotions. It recognizes "negative emotions" from the question, "The weather has been bad lately. What's going to happen this weekend?"

[1455] Output: Emotion recognition result (e.g., "negative").

[1456] Step 4: Answer Generation

[1457] Server: After recognizing the emotion, the question text is input into the generative AI model to generate an answer.

[1458] Input: Question text, and emotion recognition results.

[1459] Specific operation: The question text is converted into a prompt sentence and input into the generative AI model. Example prompt sentence: "Tell me what the weather will be like this weekend." The generative AI model responds, "It's likely to be sunny this weekend."

[1460] Output: The generated answer.

[1461] Step 5: Adjust your answers

[1462] Server: Adjusts the tone and expression of the generated response based on the emotion recognition results.

[1463] Input: Generated answers, and emotion recognition results.

[1464] Action: If you recognize negative emotions, soften the tone of your response. "Don't worry, it's likely to be sunny over the weekend."

[1465] Output: The adjusted answer.

[1466] Step 6: Assess likelihood

[1467] Server: Uses a dedicated evaluation model to evaluate the reliability of the generated answer and calculates a reliability score.

[1468] Input: The adjusted answer.

[1469] Specific operation: Apply the evaluation model to the adjusted answer and calculate the reliability score. Example evaluation result: 85%.

[1470] Output: A likelihood score (e.g., 85%).

[1471] Step 7: Recommendation Generation

[1472] Server: Recommends an appropriate verification method based on the likelihood score.

[1473] Input: A likelihood score.

[1474] What it does: If the score is below 90%, generate recommendations to check more reliable sources, e.g., "Check the weather forecast service" and "Visit the official weather bureau website."

[1475] Output: Recommended validation method.

[1476] Step 8: Formatting the results

[1477] Server: Formats the generated answers, confidence scores, and verification methods.

[1478] Inputs: Calibrated answer, likelihood score, and recommended validation method.

[1479] Specific operation: Formats data into JSON format.

[1480] Output: The formatted results.

[1481] Step 9: Output to the User

[1482] Server: Sends the formatted results to the user device as an HTTP response.

[1483] Input: The formatted result.

[1484] Specific operation: The result is sent as an HTTP response. The device analyzes the received data and instructs it to display it.

[1485] Output: Display on the user's terminal.

[1486] This allows the user to see the generated answer, its likelihood, and the recommended verification method.

[1487] (Application example 2)

[1488] 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."

[1489] In modern digital communication, users are becoming increasingly concerned about messages suspected of being phishing scams or cyberattacks. There is a need for systems that can quickly and accurately assess the trustworthiness of such messages and recommend appropriate actions to users. However, existing systems lack an integrated approach that takes user emotions into consideration while providing plausible answers using generative AI, assessing their trustworthiness, and recommending appropriate verification methods. This poses a challenge, making it difficult to ensure users' sense of security.

[1490] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[1491] In this invention, the server includes an input means for a user to input a question or problem, an answer generation means for generating an answer using a generation AI, an evaluation means for evaluating the reliability of the generated answer, a recommendation means for recommending an additional verification method based on the reliability, an emotion recognition means for recognizing the emotion of the user when inputting a suspicious message and adjusting the answer by the generation AI, and a display means for displaying to the user the answer by the answer generation means, the reliability evaluation by the evaluation means, and the verification method by the recommendation means. This allows the user to respond quickly and appropriately to security risks such as phishing scams, giving them a sense of security.

[1492] "Input means" refers to a means by which a user inputs a question or problem in text format.

[1493] An "answer generation means" is a means for generating answers using a generation AI based on input questions or problems.

[1494] The "evaluation means" is a means for evaluating the reliability of the generated answer.

[1495] A "recommendation means" is a means for suggesting an appropriate verification method to a user based on the likelihood of success.

[1496] The "display means" is a means for displaying the generated answer, the likelihood assessment, and the recommended verification method to the user.

[1497] An "emotion recognition means" is a means for recognizing emotions from the text entered by the user and adjusting the response of the generating AI according to those emotions.

[1498] This invention provides a system that, when a user inputs a suspicious message or question, uses a generation AI to generate an answer based on the input, evaluates its reliability, and recommends an appropriate verification method. The following describes an embodiment of this system.

[1499] The user first enters a question or problem in text form, for example via a smartphone or computer terminal, and the text is sent to the server, which receives the request and begins analyzing it.

[1500] The server first uses an emotion recognition means to recognize emotions from the text entered by the user. For this emotion recognition, a software library for emotion analysis (e.g., TextBlob) or an emotion analysis model is used. For example, the emotion "anxiety" is recognized from the text "Does this seem like a phishing scam?"

[1501] Next, the server uses an answer generation means to input the input question into a generative AI model and generate an answer. As a generative AI model, for example, a text generation model (e.g., GPT-3) is used. The generative AI model is physically located on a server in the cloud or in a high-performance computing environment running locally. The generated answer is provided in the form of, for example, "This may be a phishing scam. Please be careful."

[1502] The server then uses an evaluation tool to evaluate the reliability of the generated answer. This evaluation uses a dedicated algorithm or model to calculate a reliability score. For example, suppose the reliability of the generated answer is evaluated as 85%.

[1503] Based on the evaluation results, the recommendation tool recommends appropriate verification methods to the user. For example, if the reliability score is less than 90%, verification methods such as "refer to a trusted security site" or "consult an expert" are suggested.

[1504] Finally, these answers, likelihood scores, and verification method recommendations are displayed to the user through a display means, such as an application or web browser that provides a user interface.

[1505] As a concrete example, an example of a prompt sentence input by the user is shown below.

[1506] Example question: "Does this seem like a phishing scam?"

[1507] Sample response: "This may be a phishing scam. Be careful."

[1508] Example of certainty: "85%"

[1509] Recommended verification methods: "Refer to a trusted security site" or "Consult an expert"

[1510] This allows users to respond quickly and appropriately to security risks such as phishing scams, giving them a sense of security.

[1511] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1512] Step 1:

[1513] The user enters a question or problem in text format. The entered text data is sent from the terminal to the server as an HTTP POST request, which transfers the user's input data to the server.

[1514] Step 2:

[1515] The server parses the received HTTP request and extracts the question text. The server passes this text to the emotion engine, which then recognizes the user's emotion from the text (e.g., using TextBlob or an emotion analysis model). The input is the question text, and the output is the emotion recognition result (e.g., "anxious").

[1516] Step 3:

[1517] Based on the emotion recognition results, the server inputs the question text into a generative AI model and generates an answer (e.g., using GPT-3). The input is the question text, and the output is the generated answer text (e.g., "This may be a phishing scam. Please be careful").

[1518] Step 4:

[1519] Based on the generated answer text, the server uses an evaluation method to evaluate the reliability of the answer. A specific algorithm or model is used to calculate a reliability score (e.g., 85%). The input is the generated answer text, and the output is the reliability score.

[1520] Step 5:

[1521] Based on the reliability score, the server uses a recommendation mechanism to suggest additional verification methods to the user (e.g., "refer to a trusted security site" or "consult an expert"). The input is the reliability score, and the output is a list of verification methods.

[1522] Step 6:

[1523] The server formats the generated answer, the reliability score, and the recommended verification method in JSON format and sends it to the user terminal as an HTTP response. The input is the generated answer text, the reliability score, and the list of verification methods, and the output is the formatted JSON data.

[1524] Step 7:

[1525] The terminal parses the received JSON data and displays it in a user-friendly format, allowing the user to see the generated answer, its accuracy, and the recommended verification method. The input is formatted JSON data, and the output is the information displayed on the user interface.

[1526] 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.

[1527] 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.

[1528] 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.

[1529] 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.

[1530] 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.

[1531] 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.

[1532] 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).

[1533] 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.

[1534] 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."

[1535] 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.

[1536] 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).

[1537] 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.

[1538] 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.

[1539] 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.

[1540] 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.

[1541] 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.

[1542] 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.

[1543] 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.

[1544] 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.

[1545] 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.

[1546] 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.

[1547] The following is further disclosed regarding the above embodiment.

[1548] (Claim 1)

[1549] an input means for a user to input a question or problem;

[1550] an answer generation means for generating an answer using a generation AI based on the question or problem sent from the input means;

[1551] evaluation means for evaluating the reliability of the answer generated by the answer generation means;

[1552] a recommendation means for recommending an additional verification method based on the likelihood evaluated by the evaluation means;

[1553] a display means for displaying to a user the answer by the answer generating means, the evaluation of the likelihood by the evaluating means, and the verification method by the recommending means;

[1554] A system including:

[1555] (Claim 2)

[1556] 2. The system according to claim 1, wherein the evaluation means includes a recommendation means for recommending a highly reliable external database or resource when the likelihood is lower than a predetermined score.

[1557] (Claim 3)

[1558] 2. The system of claim 1, wherein the answer generation means accesses the generative AI model via a network connection to obtain answers to questions or problems.

[1559] "Example 1"

[1560] (Claim 1)

[1561] an input means for a user to input a question or problem;

[1562] an answer generation means for receiving a question or problem transmitted from the input means, inputting an inquiry generated based on the question or problem into a generative AI model, and generating an answer from the model;

[1563] evaluation means for evaluating the reliability of the answer generated by the answer generation means;

[1564] a recommendation means for recommending an additional verification method based on the likelihood evaluated by the evaluation means;

[1565] a display means for formatting and displaying to a user the answer by the answer generating means, the evaluation of the likelihood by the evaluating means, and the verification method by the recommending means;

[1566] A system including:

[1567] (Claim 2)

[1568] 2. The system according to claim 1, wherein the evaluation means includes a recommendation means for recommending a highly reliable external information source when the likelihood is lower than a predetermined score.

[1569] (Claim 3)

[1570] 2. The system of claim 1, wherein the answer generation means accesses the generative AI model via a network connection to obtain answers to questions or problems.

[1571] "Application Example 1"

[1572] (Claim 1)

[1573] an input means for a user to input a question or problem;

[1574] an answer generation means for generating an answer using a generation AI based on the question or problem sent from the input means;

[1575] evaluation means for evaluating the reliability of the answer generated by the answer generation means;

[1576] a recommendation means for recommending an additional verification method based on the likelihood evaluated by the evaluation means;

[1577] a display means for displaying to a user the answer by the answer generating means, the evaluation of the likelihood by the evaluating means, and the verification method by the recommending means;

[1578] a reliability evaluation means for evaluating the reliability of the email text;

[1579] a warning means for notifying a user that the email is likely to be a phishing email based on the reliability evaluation means;

[1580] A system including:

[1581] (Claim 2)

[1582] 2. The system according to claim 1, wherein the evaluation means includes a recommendation means for recommending a highly reliable external database or resource when the likelihood is lower than a predetermined score.

[1583] (Claim 3)

[1584] 2. The system of claim 1, wherein the answer generation means accesses the generative AI model via a network connection to obtain answers to questions or problems.

[1585] "Example 2: Combining Emotion Engines"

[1586] (Claim 1)

[1587] an input means for a user to input a question or problem;

[1588] an answer generation means for generating an answer using a generation AI based on the question or problem sent from the input means;

[1589] evaluation means for evaluating the reliability of the answer generated by the answer generation means;

[1590] a recommendation means for recommending an additional verification method based on the likelihood evaluated by the evaluation means;

[1591] an emotion recognition means for recognizing emotions in a user's input text and adjusting a response based on the emotion recognition result;

[1592] a display means for displaying to a user the answer by the answer generating means, the evaluation of the likelihood by the evaluating means, and the verification method by the recommending means;

[1593] A system including:

[1594] (Claim 2)

[1595] 2. The system according to claim 1, wherein the evaluation means includes a recommendation means for recommending a highly reliable external database or resource when the likelihood is lower than a predetermined score.

[1596] (Claim 3)

[1597] 2. The system of claim 1, wherein the answer generation means accesses the generative AI model via a network connection to obtain answers to questions or problems.

[1598] "Application example 2 when combining emotion engines"

[1599] (Claim 1)

[1600] an input means for a user to input a question or problem;

[1601] an answer generation means for generating an answer using a generation AI based on the question or problem sent from the input means;

[1602] evaluation means for evaluating the reliability of the answer generated by the answer generation means;

[1603] a recommendation means for recommending an additional verification method based on the likelihood evaluated by the evaluation means;

[1604] a display means for displaying to a user the answer by the answer generating means, the evaluation of the likelihood by the evaluating means, and the verification method by the recommending means;

[1605] An emotion recognition means for recognizing the emotion of a user when they input a suspicious message and adjusting the response by the generative AI;

[1606] A system including:

[1607] (Claim 2)

[1608] 2. The system according to claim 1, wherein the evaluation means includes a recommendation means for recommending a highly reliable external database or resource when the likelihood is lower than a predetermined score.

[1609] (Claim 3)

[1610] 2. The system of claim 1, wherein the answer generation means accesses the generative AI model via a network connection to obtain answers to questions or problems. [Explanation of symbols]

[1611] 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. an input means for a user to input a question or problem; an answer generation means for generating an answer using a generation AI based on the question or problem sent from the input means; evaluation means for evaluating the reliability of the answer generated by the answer generation means; a recommendation means for recommending an additional verification method based on the likelihood evaluated by the evaluation means; a display means for displaying to a user the answer by the answer generating means, the evaluation of the likelihood by the evaluating means, and the verification method by the recommending means; A system including:

2. 2. The system according to claim 1, wherein the evaluation means includes a recommendation means for recommending a highly reliable external database or resource when the likelihood is lower than a predetermined score.

3. 2. The system of claim 1, wherein the answer generation means accesses the generative AI model via a network connection to obtain answers to questions or problems.

Images