system

Abstract

We provide the system. [Solution] A means for receiving basic information about the disability, A means of searching past incident reports and evaluating similarities, A method for automatically generating new incident reports based on previously searched reports, A means of generating predicted Q&A by referring to the customer's industry and past inquiry history, A means of preparing answers to the generated Q&A, A system that includes means for receiving actual Q&A and evaluating its accuracy by comparing it with predicted Q&A.

Description

Technical Field

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

Background Art

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

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] In a conventional failure reporting system, it is necessary to create each report from scratch every time a failure occurs, which is time-consuming and laborious. Also, in communication with customers, there is a risk that the quality of customer support will decline because appropriate preparation cannot be made for predicted questions. Furthermore, since feedback after failure reporting is not sufficiently carried out, there is a problem that the response in case of recurrence of the same failure is not made more efficient.

Means for Solving the Problems

[0005] This invention provides a means for receiving basic information about a malfunction, searching and referencing past malfunction reports, and automatically generating a new malfunction report. Furthermore, by providing a system that includes means for referencing the customer's industry and past inquiry history to generate predicted Q&A and prepare answers, and means for receiving actual Q&A and comparing and evaluating them with the predicted Q&A, the invention aims to improve the efficiency of reporting and the quality of customer service.

[0006] "Basic information regarding the incident" refers to information necessary to provide an overview of the incident, such as the name of the incident, the date and time of occurrence, the scope of impact, and a detailed description.

[0007] A "past incident report" is a document that describes incidents that have occurred in the past, and includes information such as the circumstances of the incident, its cause, its impact, and the countermeasures taken.

[0008] "Means for evaluating similarity" refers to a function that compares newly received failure information with the contents of past failure reports to determine the degree of similarity.

[0009] "Method for automatically generating incident reports" refers to a function that automatically creates reports about new incidents based on past incident reports.

[0010] "Customer's industry" refers to information about the industry sector or type of business to which the customer receiving the report belongs.

[0011] "Inquiry history" refers to a record of questions previously submitted by customers and the answers provided to those questions.

[0012] The "means of generating Q&A" refer to a function that analyzes the contents of a report, lists anticipated questions from customers, and prepares appropriate answers to those questions in advance.

[0013] "Predicted Q&A" refers to a set of questions and answers that the system has anticipated in advance.

[0014] "Actual Q&A" refers to a set of questions actually raised by customers when they receive a report and their corresponding answers.

[0015] "Means for evaluating accuracy" is a function for determining how well the predicted Q&A matches the actual Q&A and evaluating its accuracy.

Brief Description of Drawings

[0016] [Figure 1] It is a conceptual diagram showing an example of the configuration of a data processing system according to the first embodiment. [Figure 2] It is a conceptual diagram showing an example of the main functions of a data processing device and a smart device according to the first embodiment. [Figure 3] It is a conceptual diagram showing an example of the configuration of a data processing system according to the second embodiment. [Figure 4] It is a conceptual diagram showing an example of the main functions of a data processing device and smart glasses according to the second embodiment. [Figure 5] It is a conceptual diagram showing an example of the configuration of a data processing system according to the third embodiment. [Figure 6] It is a conceptual diagram showing an example of the main functions of a data processing device and a headset-type terminal according to the third embodiment. [Figure 7] It is a conceptual diagram showing an example of the configuration of a data processing system according to the fourth embodiment. [Figure 8] It is a conceptual diagram showing an example of the main functions of a data processing device and a robot according to the fourth embodiment. [Figure 9] It shows an emotion map to which a plurality of emotions are mapped. [Figure 10] It shows an emotion map to which a plurality of emotions are mapped. [Figure 11] It is a sequence diagram showing the processing flow of the data processing system in Example 1. [Figure 12] It is a sequence diagram showing the processing flow of the data processing system in Application Example 1. [Figure 13]It is a sequence diagram showing the processing flow of the data processing system in Embodiment 2 when combined with an emotion engine. [Figure 14] It is a sequence diagram showing the processing flow of the data processing system in Application Example 2 when combined with an emotion engine.

Mode for Carrying Out the Invention

[0017] Hereinafter, an example of an embodiment of a system according to the technology of the present disclosure will be described with reference to the accompanying drawings.

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

[0019] In the following embodiments, a numbered processor (hereinafter simply referred to as "processor") may be a single arithmetic unit or a combination of multiple arithmetic units. Also, the processor may be a single type of arithmetic unit or a combination of multiple types of arithmetic units. Examples of arithmetic units include a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), a GPGPU (General-Purpose computing on Graphics Processing Units), an APU (Accelerated Processing Unit), and the like.

[0020] In the following embodiments, a numbered RAM (Random Access Memory) is a memory in which information is temporarily stored and is used as a work memory by the processor.

[0021] In the following embodiments, a numbered storage is one or more non-volatile storage devices that store various programs and various parameters, etc. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), magnetic disks (e.g., hard disks), or magnetic tapes, etc.

[0022] In the following embodiments, the signed communication interface (I / F) is an interface that includes a communication processor and an antenna, etc. The communication interface manages communication between multiple computers. Examples of communication standards applicable to the communication interface include wireless communication standards such as 5G (5th Generation Mobile Communication System), Wi-Fi (registered trademark), or Bluetooth (registered trademark).

[0023] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B." That is, "A and / or B" means that it may be A alone, or B alone, or a combination of A and B. Furthermore, in this specification, the same concept as "A and / or B" applies when expressing three or more things linked by "and / or."

[0024] [First Embodiment]

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

[0026] As shown in Figure 1, the data processing system 10 includes a data processing device 12 and a smart device 14. An example of the data processing device 12 is a server.

[0027] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0028] The smart device 14 comprises a computer 36, a reception device 38, an output device 40, a camera 42, and a communication interface 44. The computer 36 comprises a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The reception device 38, output device 40, and camera 42 are also connected to the bus 52.

[0029] The reception device 38 is equipped with a touch panel 38A and a microphone 38B, etc., and receives user input. The touch panel 38A receives user input by detecting contact with an object (e.g., a pen or finger). The microphone 38B receives user input by detecting the user's voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the data indicating the user input.

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

[0031] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various types of information between processor 46 and processor 28 via network 54.

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

[0033] As shown in Figure 2, in the data processing device 12, a specific processing is performed by the processor 28. A specific processing program 56 is stored in the storage 32. The specific processing program 56 is an example of a "program" related to the technology of this disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 according to the specific processing program 56 executed on the RAM 30.

[0034] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0035] In the smart device 14, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. The reception output program 60 is used in conjunction with a specific processing program 56 by the data processing system 10. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[0036] Next, the specific processing performed by the specific processing unit 290 of the data processing device 12 will be described. In the following description, the data processing device 12 will be referred to as the "server" and the smart device 14 as the "terminal".

[0037] The present invention aims to improve the quality of customer support after reporting by enabling companies to quickly and effectively create and submit incident reports to customers, predict appropriate Q&A based on the customer's industry and past inquiry history, and prepare answers accordingly.

[0038] Program processing

[0039] Entering and sending error information

[0040] 1. The user logs into the terminal and enters basic information about the problem. Specifically, they enter the name of the problem, the date and time of occurrence, the scope of impact, and a detailed description.

[0041] 2. The terminal sends the entered error information to the server.

[0042] Automatic generation of incident reports

[0043] 3. Based on the received failure information, the server searches past failure reports and evaluates their similarity.

[0044] 4. The server automatically generates a new incident report based on the most similar past incident reports. The new report includes an overview of the incident, its cause, its scope of impact, and countermeasures.

[0045] 5. The server sends the generated report to the terminal.

[0046] 6. The terminal displays the generated report to the user, who can review and modify its contents.

[0047] 7. After the user has finished making corrections, they press the confirm button to send the final report to the server.

[0048] Q&A prediction and preparation

[0049] 8. The server analyzes the contents of the confirmed report and refers to the customer's industry and past inquiry history.

[0050] 9. The server generates anticipated Q&A and prepares appropriate answers for each.

[0051] 10. The server sends the generated Q&A to the terminal.

[0052] 11. The terminal displays the generated Q&A list and its answers to the user.

[0053] 12. Users review the Q&A and answers to prepare for their visit.

[0054] Reports and Feedback

[0055] 13. The user actually visits the customer and submits a report. They respond to customer questions based on the anticipated Q&A during the visit.

[0056] 14. After the visit, the user enters the questions and answers they received into the terminal.

[0057] 15. The terminal sends the actual Q&A entered to the server.

[0058] Accuracy evaluation and algorithm improvement

[0059] 16. The server compares the actual received Q&A with the predicted Q&A and evaluates the accuracy of the prediction.

[0060] 17. The server improves its prediction algorithm based on the evaluation results to improve the accuracy of the next prediction.

[0061] 18. The server generates a new prompt based on the improved algorithm and provides feedback to the user.

[0062] Specific example

[0063] 1. Input and transmission of failure information

[0064] The user enters a message into their terminal indicating that the server is down, and then submits a detailed description of the date and time of the outage and the extent of the impact.

[0065] 2. Automatic generation of incident reports

[0066] The server searches past reports using the keyword "server down" and selects the most similar report.

[0067] Based on the selected reports, the server automatically generates a "Server Downtime Incident Report," which includes detailed explanations, causes, and countermeasures.

[0068] 3. Predicting and preparing for the Q&A session

[0069] The server anticipates questions related to "server downtime," such as "server recovery time" and "whether data has been lost," and prepares appropriate answers for each.

[0070] 4. Reporting and Feedback

[0071] The system responds to questions based on anticipated Q&A from the user during their customer visit. For example, in response to the question, "When will the server be restored?", it might answer, "The estimated restoration time is 2 hours from now."

[0072] After the visit, the terminal inputs the questions the user actually received (for example, "Will there be any data loss?") and the answer ("There will be no data loss"), and sends them to the server.

[0073] 5. Accuracy evaluation and algorithm improvement

[0074] The server compares the predicted Q&A with the actual Q&A to evaluate prediction accuracy and use the results to improve the algorithm.

[0075] The system of this invention streamlines the creation of fault reports and customer support, thereby improving customer satisfaction.

[0076] The following describes the processing flow.

[0077] Step 1:

[0078] The user logs into the device and accesses the incident report form. The device displays the incident report form.

[0079] Step 2:

[0080] The user enters basic information about the incident (incident name, date and time of occurrence, scope of impact, and detailed description) and presses the submit button. The terminal sends the entered incident information to the server.

[0081] Step 3:

[0082] Based on the failure information received by the server, it searches the database for past failure reports. The server uses a similarity evaluation algorithm to select the most similar past report.

[0083] Step 4:

[0084] Based on past reports where servers were selected, a new incident report is automatically generated. The generated report includes an overview of the incident, its cause, the scope of its impact, and the countermeasures taken.

[0085] Step 5:

[0086] The server sends the generated failure report to the terminal. The terminal displays the generated report to the user.

[0087] Step 6:

[0088] The user reviews the report content and makes corrections as needed. Once the corrections are complete, they press the confirm button to send the report to the server.

[0089] Step 7:

[0090] The server analyzes the contents of the confirmed report. The server refers to the customer's industry and past inquiry history to generate predicted Q&A.

[0091] Step 8:

[0092] The server prepares appropriate answers to the predicted Q&A. Once the Q&A is ready, it sends it to the terminal.

[0093] Step 9:

[0094] The terminal displays a generated list of Q&A and its answers to the user. The user reviews the Q&A and answers to prepare for their visit.

[0095] Step 10:

[0096] Users visit customers in person and submit reports. They respond to customer questions based on anticipated Q&A from the visit.

[0097] Step 11:

[0098] After a user visits, the terminal inputs the actual questions and answers they received. The terminal then sends the entered Q&A to the server.

[0099] Step 12:

[0100] The server compares the actual Q&A received with the predicted Q&A to evaluate the accuracy of the prediction.

[0101] Step 13:

[0102] The server improves the prediction algorithm based on the evaluation results. It generates a new prompt based on the improved algorithm and provides feedback to the user.

[0103] By following the steps outlined above, this system can streamline the creation of reports and customer support in the event of a failure, thereby improving accuracy.

[0104] (Example 1)

[0105] Next, we will describe Example 1. In the following description, the data processing device 12 will be referred to as the "server," and the smart device 14 will be referred to as the "terminal."

[0106] Traditional incident reporting and customer support processes are often manual, time-consuming and labor-intensive, and can lead to insufficient preparation for anticipated Q&A. This can result in decreased customer satisfaction and damage to the company's reputation. Furthermore, it's difficult to extract relevant information from past reports and inquiry histories to provide efficient and accurate responses.

[0107] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0108] In this invention, the server includes means for receiving basic information about a failure, means for searching past failure reports and evaluating similarity, means for automatically generating a new failure report based on the retrieved past reports, means for displaying the generated failure report and accepting corrections, means for referencing the customer's industry and past inquiry history to generate predicted Q&A, means for preparing answers to the generated Q&A, means for receiving actual Q&A and evaluating its accuracy by comparing it with the predicted Q&A, means for improving the prediction algorithm based on the evaluation results, and means for generating new prompt statements based on the improved algorithm and providing feedback. This streamlines the process of creating failure reports and responding to customers, enabling highly accurate responses.

[0109] "Means for receiving basic information about a malfunction" refers to a device or program that has an interface for a user to input details of a malfunction and a function to transmit that information to a server.

[0110] "Means for searching past incident reports and evaluating similarity" refers to a device or program for searching past incident reports in a database using string search or natural language processing techniques and calculating their similarity to current incident information.

[0111] "Means for automatically generating new incident reports based on retrieved past reports" refers to a device or program that automatically creates a new incident report by using the most similar report from the search results as a template and filling in the current incident information.

[0112] "Means for displaying generated failure reports and accepting corrections" refers to a device or program that displays generated failure reports to the user, has an interface that allows the user to confirm and correct the contents, and has the function of saving and transmitting the corrected contents.

[0113] "Means for generating predicted Q&A by referring to the customer's industry and past inquiry history" refers to a device or program that searches a database for the customer's industry and past inquiry history, and generates predicted questions and answers based on this information.

[0114] "Means for preparing answers to generated Q&A" refers to a device or program that uses natural language generation technology or past answer data to automatically generate appropriate answers to predicted questions.

[0115] "A means of receiving actual Q&A and evaluating its accuracy by comparing it with predicted Q&A" refers to a device or program that takes actual questions and answers received by a user at a visited location as input and evaluates the accuracy of the prediction by comparing it with the predicted Q&A.

[0116] "Means for improving the prediction algorithm based on evaluation results" refers to a device or program that analyzes the results of accuracy evaluation and adjusts or improves the parameters and model of the prediction algorithm.

[0117] "Means for generating new prompt statements based on an improved algorithm and providing feedback" refers to a device or program that generates new prompt statements using an improved algorithm and has a notification function to inform the user of their contents.

[0118] The present invention aims to enable companies to quickly and effectively create and submit incident reports to their customers, predict appropriate Q&A based on the customer's industry and past inquiry history, and prepare answers accordingly. This system primarily consists of server, terminal, and user-centric processing.

[0119] First, the user logs into the company's internal system and enters basic information about the incident into the terminal. Specifically, this includes the incident name, date and time of occurrence, scope of impact, and a detailed description. The terminal verifies this information and sends it to the server. This process utilizes form input and HTTP requests.

[0120] Next, the server analyzes the received failure information and searches for past failure reports. Natural language processing technology is used for the search, and a similarity score is calculated. Then, a new failure report is automatically generated based on the most similar report. The generated report includes an overview of the failure, its cause, scope of impact, and countermeasures. This report is returned to the terminal, where the user can review and correct its contents.

[0121] Once the corrections are complete, the user confirms the report and sends it back to the server. Based on the confirmed report, the server references the customer's industry and past inquiry history to generate predicted Q&A. This uses database searches and a generation AI model. Appropriate answers are automatically added to the generated Q&A and sent back to the terminal.

[0122] Users check a Q&A list on their terminal to prepare for customer visits. After the visit, they input the actual questions and answers into their terminal and send them to the server. The server compares these actual Q&A with the predicted Q&A and evaluates the accuracy. The evaluation results are used to improve the algorithm, aiming for better prediction accuracy next time. Furthermore, new prompts are generated based on the improved algorithm and provided to the user as feedback.

[0123] Hardware and software details

[0124] Terminals: Desktop PCs, laptops, tablets, etc., within the company are used, and information is entered and displayed via web browsers or dedicated applications.

[0125] Servers: Cloud servers will be used to host large-scale databases and AI models, performing tasks such as data analysis, automated report generation, and Q&A generation. Specifically, Amazon Web Services (AWS®) and Microsoft Azure® are envisioned.

[0126] Software: For natural language processing, Python libraries such as NLTK and SpaCy are used, and the generative AI model utilizes OpenAI's GPT model.

[0127] Specific example

[0128] Example prompt: "Please prepare a failure report required when the server goes down. Include the date and time of the incident, the scope of the impact, and a detailed description."

[0129] Example of customer interaction: When a user visits a customer's site, they might ask, "When will the server be restored?" The response would be, "The estimated restoration time is in 2 hours." The actual question is entered into the server, and for the question, "Will there be any data loss?", the response "There will be no data loss" is recorded.

[0130] Through the processes described above, this system streamlines the creation of incident reports and customer support, enabling highly accurate responses.

[0131] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0132] Step 1:

[0133] The user logs into the company's internal system and enters basic information about the incident into the terminal. Specifically, they enter the incident name, date and time of occurrence, scope of impact, and a detailed description (input). The terminal validates the input and sends it to the server (output). During this process, form input occurs and an HTTP request is generated (specific action).

[0134] Step 2:

[0135] The server analyzes the received failure information and searches past failure reports in the database to evaluate similarity (input). It calculates a similarity score using natural language processing techniques (data calculation). It generates a new failure report using the most similar report as a template (output). This involves keyword search and similarity calculation (such as Cosine Similarity) (specific operation).

[0136] Step 3:

[0137] The server sends the generated failure report to the terminal (output). The terminal displays the report to the user and provides an editing form that allows the user to review and modify the contents (specific action). The user reviews the report and makes corrections as needed (input).

[0138] Step 4:

[0139] Once the user has completed the corrections, they press the confirm button to send the final error report to the server (output). The terminal then sends the corrected report to the server as data in JSON format (specific action).

[0140] Step 5:

[0141] The server references the customer's industry and past inquiry history based on the confirmed report (input). It uses database searches and generative AI models to generate predicted Q&A (data computation). The generated Q&A is automatically assigned appropriate answers (output). This uses natural language generation technology to produce answers customized for specific industries (specific behavior).

[0142] Step 6:

[0143] The server sends the generated Q&A list to the terminal (output). The terminal displays the Q&A list and its answers to the user (specific action). The user reviews the Q&A and prepares for their visit (input).

[0144] Step 7:

[0145] During customer visits, users respond to questions based on predicted Q&A (input). After the visit, users input the actual questions received and their answers into the terminal (input).

[0146] Step 8:

[0147] The terminal sends the actual Q&A input to the server (output). The server compares the actual Q&A with the predicted Q&A and evaluates the accuracy of the prediction (data calculation). Based on the evaluation results, the prediction algorithm is improved (output). This uses statistical evaluation metrics (e.g., accuracy, recall) (specific operation).

[0148] Step 9:

[0149] The server generates a new prompt message based on the improved algorithm (output) and provides feedback on its contents to the user (specific action). The user receives the improved prompt message and uses it to help with troubleshooting in the future (input).

[0150] (Application Example 1)

[0151] Next, we will explain Application Example 1. In the following explanation, the data processing device 12 will be referred to as the "server," and the smart device 14 will be referred to as the "terminal."

[0152] For companies to quickly and effectively create and submit incident reports to customers, and to provide prompt and high-quality customer support afterward, accurate reporting and appropriate Q&A preparation are required when incidents occur. However, incident response is time-consuming and labor-intensive, and preparing appropriate Q&A based on past inquiry history and the customer's industry is a particularly difficult challenge. Furthermore, human error can occur in report generation and Q&A preparation, so it is necessary to improve the efficiency and accuracy of the entire system.

[0153] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0154] In this invention, the server includes means for receiving basic information about a failure; means for searching past failure reports and evaluating similarity; means for automatically generating a new failure report based on the searched past reports; means for referencing the customer's industry and past inquiry history to generate predicted Q&A; means for preparing answers to the generated Q&A; means for receiving actual Q&A and evaluating its accuracy by comparing it with the predicted Q&A; means for sending the generated failure report and predicted Q&A to a terminal so that the user can review and correct them; means for regenerating the final report and Q&A based on the corrected content; means for inputting the questions and answers actually received into the terminal and sending them to the server; and means for comparing the actual Q&A received by the server with the predicted Q&A and improving the prediction algorithm. This enables increased efficiency and improved quality in the generation of failure reports and customer support.

[0155] "Basic information about the disruption" refers to fundamental information related to the disruption, such as the name of the disruption, the date and time of occurrence, the scope of its impact, and a detailed explanation.

[0156] "Past incident reports" refer to incident reports created in the past, and are used as reference information by analyzing similar incidents.

[0157] "Means for evaluating similarity" refers to a method or system for identifying the most similar case from past incident reports based on the incident information received.

[0158] "Means for automatically generating new incident reports" refers to a method or system for automatically creating new incident reports based on current incident information, using past reports as templates.

[0159] "Customer's industry" refers to the specific industry or sector to which the customer belongs, and is information that reflects the unique requirements and problems associated with that particular industry.

[0160] "Past inquiry history" refers to the content of inquiries received from customers in the past and the history of how those inquiries were handled. This serves as reference material for providing appropriate customer service.

[0161] "Means for generating predicted Q&A" refers to a method or system for analyzing fault information and past inquiry history to generate questions that are predicted to be asked by customers in the future, along with their answers.

[0162] "Means for preparing answers to generated Q&A" refers to a method or system for preparing appropriate answers to anticipated questions.

[0163] "Means for receiving actual Q&A" refers to a method or system for collecting questions and answers actually submitted by customers.

[0164] "Means for evaluating accuracy by comparing with predicted Q&A" refers to a method or system for comparing predicted questions with actual questions to evaluate the accuracy and usefulness of the predictions.

[0165] A "terminal" refers to a device used by system administrators or users, and includes desktop computers, notebooks, smartphones, and other similar devices.

[0166] "Means for users to review and correct" refers to interfaces and tools that allow users to review generated reports and Q&A and make corrections as needed.

[0167] "Means for generating the final report and Q&A based on the revisions" refers to a method or system for generating the final version of the report and Q&A, reflecting the user's revisions.

[0168] "Means for improving prediction algorithms" refers to methods or systems for adjusting prediction models based on actual data to improve their accuracy.

[0169] This invention is a system that enables companies to quickly and effectively create and submit fault reports to customers, and to provide high-quality customer support thereafter. This system consists of a smartphone application (hereinafter referred to as "the app") and a server system.

[0170] First, the user logs into the app and enters basic information about the outage. Specifically, this includes the outage name, date and time of occurrence, scope of impact, and a detailed description. Next, the app sends the entered outage information to the server. The server then compares the received outage information with past outage reports and evaluates the similarity. At this time, the server uses a database (e.g., MongoDB) to search for past reports.

[0171] The server automatically generates a new incident report based on the most similar past incident reports. This generated report includes an overview of the incident, its cause, scope of impact, and countermeasures. The server then sends the generated report to the app, where the user can review it and make corrections as needed. Once corrections are complete, the user presses the confirm button to send the final report to the server.

[0172] Furthermore, the server analyzes the contents of the confirmed incident report and refers to the customer's industry and past inquiry history. This generates predicted Q&A. To generate the Q&A, the server uses an artificial intelligence (AI) model (e.g., TENSORFLOW®). The generated Q&A list and its answers are sent to the app, where the user can review them and prepare for customer visits.

[0173] After the visit, the user enters the actual questions and answers into the app and sends them to the server. The server compares the received actual Q&A with the predicted Q&A and evaluates the accuracy of the prediction. Based on the evaluation results, the server improves the prediction algorithm to improve the accuracy of the next prediction.

[0174] As a concrete example, if a live stream is interrupted due to a malfunction, the user enters the malfunction information, such as "live stream interruption," and sends it to the server, detailing the date and time of the incident and the scope of the impact. The server searches past reports using the keyword "live stream interruption" and generates a new malfunction report based on the most similar report. The server sends the generated report to the app for the user to review and correct. After that, the server generates predictive Q&A related to "live stream interruption" (for example, "When is it expected to be restored?", "Can I rewatch past streamed videos?") and provides appropriate answers.

[0175] Examples of prompt statements are as follows:

[0176] "Input example:

[0177] Incident Name: Live Stream Interruption, Date and Time: 2023-10-10 14:00, Affected Users: All Users, Details: The live stream was interrupted midway.

[0178] Example output:

[0179] Outage Summary: Live stream stopped for all users. Cause: Server load. Countermeasure: Server upgrade. Predicted Q&A: When is it expected to be restored? Answer: It is expected to be restored in 1 hour. Can I watch past streamed videos again? Answer: Yes, you can watch them again.

[0180] This system allows companies to efficiently create reports and handle customer inquiries in the event of a system failure, which is expected to improve customer satisfaction.

[0181] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0182] Step 1:

[0183] The user logs into the smartphone app and enters basic information about the outage (outage name, date and time of occurrence, scope of impact, and detailed description).

[0184] Input: Name of the problem, date and time of occurrence, scope of impact, detailed description

[0185] Output: Failure information data

[0186] Specific action: The user fills in the required information in the app's input form and presses the submit button.

[0187] Step 2:

[0188] The terminal sends the entered error information to the server.

[0189] Input: Failure information data

[0190] Output: Failure information sent to the server

[0191] Specific operation: The app sends an HTTP request to the server via the API.

[0192] Step 3:

[0193] Based on the failure information received by the server, past failure reports are searched in the database (MongoDB) and their similarity is evaluated.

[0194] Input: Received fault information

[0195] Output: Similar past incident reports

[0196] Specific operation: The server extracts keywords related to the failure information and queries the database to find the most similar report.

[0197] Step 4:

[0198] The server automatically generates a new incident report based on the most similar past incident report.

[0199] Input: Similar past incident reports, received incident information

[0200] Output: New Incident Report

[0201] Specific operation: The server uses past reports as templates and combines basic information and detailed descriptions to generate new reports.

[0202] Step 5:

[0203] The server sends the generated report to the terminal, allowing the user to review and correct it.

[0204] Input: New Incident Report

[0205] Output: Report displayed on the user's device

[0206] Specific operation: The server sends the generated report to the app in JSON format, and the app displays it.

[0207] Step 6:

[0208] The user reviews the report and makes corrections as needed. After completing the corrections, they press the confirm button to send the final report to the server.

[0209] Input: User modifications, final report

[0210] Output: Final report sent to the server

[0211] Specific operation: When the user modifies the report content and presses the confirm button, the app sends the final report, including the modifications, to the server.

[0212] Step 7:

[0213] The system analyzes the contents of confirmed server failure reports, referencing the customer's industry and past inquiry history. It then generates anticipated Q&A and prepares appropriate answers.

[0214] Input: Finalized report, customer industry, past inquiry history

[0215] Output: Predicted Q&A list and its answers

[0216] Specific operation: The server uses a prediction algorithm (TensorFlow) to analyze reports and historical data and generate predicted Q&A.

[0217] Step 8:

[0218] The server sends the generated Q&A list and its answers to the user's device for review.

[0219] Input: Predicted Q&A list and its answers

[0220] Output: Q&A displayed on the user's device

[0221] Specific operation: The server sends the generated Q&A to the app, and the app displays it.

[0222] Step 9:

[0223] Users actually visit customers and answer questions based on predicted Q&A.

[0224] Input: Customer questions, predicted Q&A

[0225] Output: Actual Q&A from customer support

[0226] Specific operation: The user refers to a Q&A list and responds to customer questions.

[0227] Step 10:

[0228] After a user visits, the device inputs the actual questions and answers they submitted and sends them to the server.

[0229] Input: Actual Q&A

[0230] Output: Actual Q&A sent to the server

[0231] Specific operation: The user enters the question and answer into the terminal and sends them to the server.

[0232] Step 11:

[0233] The server compares the actual Q&A received with the predicted Q&A to evaluate the accuracy of the prediction.

[0234] Input: Actual Q&A, Predicted Q&A

[0235] Output: Evaluation results and feedback for algorithm improvement

[0236] Specific operation: The server performs comparison calculations and improves the prediction algorithm based on the evaluation results.

[0237] Step 12:

[0238] The server generates new prompts based on an improved algorithm and provides feedback to the user.

[0239] Input: Evaluation results, improved algorithm

[0240] Output: New prompts and feedback content

[0241] Specific action: The server generates a new prompt and notifies the user.

[0242] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[0243] The system of this invention enables companies to quickly and effectively create and submit incident reports to customers, predict appropriate Q&A based on the customer's industry and past inquiry history, prepare answers, and improve the quality of customer service by recognizing user emotions and adjusting responses accordingly. The specific program processing is described below.

[0244] Program processing

[0245] Entering and sending error information

[0246] 1. The user logs into the device and accesses the incident report form. The device displays the incident report form.

[0247] 2. The user enters basic information about the incident (incident name, date and time of occurrence, scope of impact, and detailed description) and presses the submit button. The terminal sends the entered incident information to the server.

[0248] Automatic generation of incident reports

[0249] 3. Based on the failure information received by the server, it searches the database for past failure reports. The server uses a similarity evaluation algorithm to select the most similar past report.

[0250] 4. Based on past reports where the server was selected, a new incident report is automatically generated. The generated report includes an overview of the incident, its cause, the scope of its impact, and the countermeasures taken.

[0251] 5. The server sends the generated report to the terminal. The terminal displays the generated report to the user.

[0252] 6. The user reviews the report content and makes corrections as needed. After completing the corrections, they press the confirm button to send the report to the server.

[0253] Adjusting user emotion recognition and responses

[0254] 7. The device's built-in emotion engine analyzes user input and interactions to recognize the user's emotions. For example, it can detect when the user is anxious or angry.

[0255] 8. The server adjusts the content of the incident report based on the perceived emotions. For example, if the user is anxious, the report is revised to include more detailed steps or additional explanations.

[0256] Q&A prediction and preparation

[0257] 9. The server analyzes the contents of the confirmed report and refers to the customer's industry and past inquiry history.

[0258] 10. The server generates anticipated Q&A and prepares appropriate answers for each. The Q&A and answers are also adjusted based on the perceived user sentiment.

[0259] 11. The server sends the generated Q&A to the terminal. The terminal displays the generated Q&A list and its answers to the user.

[0260] 12. Users review the Q&A and answers to prepare for their visit.

[0261] Reports and Feedback

[0262] 13. The user actually visits the customer and submits a report. They respond to customer questions based on the anticipated Q&A during the visit.

[0263] 14. After the visit, the user enters the actual question and answer into the terminal. The terminal then sends the actual Q&A entered to the server.

[0264] Accuracy evaluation and algorithm improvement

[0265] 15. Compare the actual Q&A received by the server with the predicted Q&A and evaluate the accuracy of the prediction.

[0266] 16. The server improves the prediction algorithm based on the evaluation results. It generates a new prompt based on the improved algorithm and provides feedback to the user.

[0267] 17. The server analyzes user sentiment during actual Q&A sessions and uses the feedback to further improve the accuracy of the generated Q&A and its answers.

[0268] Specific example

[0269] 1. Input and transmission of failure information

[0270] The user enters a message into their terminal indicating that the server is down, and then submits a detailed description of the date and time of the outage and the extent of the impact. The emotion engine detects the user's anxiety.

[0271] 2. Automatic generation of incident reports

[0272] The server searches past reports using the keyword "server down" and selects the most similar report.

[0273] Based on the server selection report, an "Incident Report Regarding Server Downtime" is automatically generated, including detailed explanations, causes, and countermeasures.

[0274] 3. Recognizing and responding to user emotions

[0275] The emotion engine recognizes the user's impatience and adjusts the report to include more detailed steps and additional explanations.

[0276] 4. Predicting and preparing for the Q&A session

[0277] The server anticipates questions related to "server downtime," such as "server recovery time" and "whether data has been lost," and prepares appropriate answers for each.

[0278] Based on the user's emotions recognized by the emotion engine, adjust the response content to be more detailed and understandable.

[0279] 5. Report and Feedback

[0280] The user responds to questions based on the predicted Q&A during customer visits. For example, in response to the question "When will the server be restored?", the answer is "The scheduled restoration time is in 2 hours".

[0281] After the visit, the user inputs the actual questions received (e.g., "Is there any data loss?") and their answers ("There is no data loss") into the terminal and sends them to the server.

[0282] 6. Accuracy Evaluation and Algorithm Improvement

[0283] The server compares the predicted Q&A with the actual Q&A, evaluates the prediction accuracy, and uses it to improve the algorithm.

[0284] Based on the feedback, the server further improves the accuracy of the Q&A and its answers generated using the emotion engine.

[0285] The system of the present invention enables the creation of trouble reports and customer response to be more efficient, enables responses considering the user's emotions, and can improve customer satisfaction.

[0286] The following describes the processing flow.

[0287] Step 1:

[0288] The user logs in to the terminal and accesses the trouble report form. The terminal displays the trouble report form.

[0289] Step 2:

[0290] The user enters basic information about the incident (incident name, date and time of occurrence, scope of impact, and detailed description) and presses the submit button. The terminal sends the entered incident information to the server.

[0291] Step 3:

[0292] Based on the failure information received by the server, it searches the database for past failure reports. The server uses a similarity evaluation algorithm to select the most similar past report.

[0293] Step 4:

[0294] Based on past reports where servers were selected, a new incident report is automatically generated. The generated report includes an overview of the incident, its cause, the scope of its impact, and the countermeasures taken.

[0295] Step 5:

[0296] The server sends the generated report to the terminal. The terminal displays the generated report to the user.

[0297] Step 6:

[0298] The user reviews the report content and makes corrections as needed. Once the corrections are complete, they press the confirm button to send the report to the server.

[0299] Step 7:

[0300] The device's built-in emotion engine analyzes user input and interactions to recognize the user's emotions. The device can detect situations such as when the user is anxious or angry.

[0301] Step 8:

[0302] The server adjusts the content of the incident report based on the perceived emotions. For example, if the user is anxious, the report is revised to include more detailed steps and additional explanations.

[0303] Step 9:

[0304] The server analyzes the content of the finalized report and refers to the customer's industry type and past inquiry history.

[0305] Step 10:

[0306] The server generates predicted Q&A and prepares appropriate answers for each. Based on the recognized user sentiment, the Q&A and answers are adjusted.

[0307] Step 11:

[0308] The server transmits the generated Q&A and their answers to the terminal. The terminal displays the generated Q&A list and their answers to the user.

[0309] [[ID=?]]Step 12:

[0310] The user checks the Q&A and answers in preparation for the visit.

[0311] Step 13:

[0312] The user actually visits the customer and submits the report. Respond to the customer's questions based on the predicted Q&A during the visit.

[0313] Step 14:

[0314] After the user's visit, the user inputs the actual questions received and their answers into the terminal. The terminal transmits the input actual Q&A to the server.

[0315] Step 15:

[0316] The server compares the actual Q&A received with the predicted Q&A and evaluates the accuracy of the prediction.

[0317] Step 16:

[0318] )]] The server improves the prediction algorithm based on the evaluation results. It generates a new prompt based on the improved algorithm and provides feedback to the user.

[0319] Step 17:

[0320] The server analyzes user sentiment during actual Q&A sessions and uses the feedback to further improve the accuracy of the generated Q&A and its answers.

[0321] Through the steps outlined above, this system streamlines the creation of reports and customer support in the event of a failure, enabling responses that take user emotions into consideration, and ultimately improving customer satisfaction.

[0322] (Example 2)

[0323] Next, we will describe Example 2. In the following description, the data processing device 12 will be referred to as the "server" and the smart device 14 as the "terminal".

[0324] It is crucial for companies to quickly and effectively create and submit incident reports to their customers, but this usually requires considerable time and effort. Furthermore, preparing appropriate Q&A based on the customer's industry and past inquiry history is not easy. Additionally, considering user emotions in customer service is difficult, leading to inconsistent service quality. A system is needed to address these challenges and improve the quality of customer service.

[0325] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[0326] In this invention, the server includes means for receiving basic information about a failure, means for searching past failure reports and evaluating similarity, means for automatically generating a new failure report based on the retrieved past reports, means for recognizing emotions from user input and operations, means for adjusting the content of the failure report based on the recognized emotion information, means for generating predicted Q&A by referring to the customer's industry and past inquiry history, means for preparing answers to the generated Q&A, means for receiving actual Q&A and evaluating its accuracy by comparing it with the predicted Q&A, and means for improving the prediction algorithm. This enables the rapid and effective creation of failure reports, appropriate responses to customer inquiries, and flexible responses that take user emotions into consideration.

[0327] "Means for receiving basic information about a failure" refers to a function that allows users to input basic information such as the name of the failure, the date and time of occurrence, the scope of impact, and a detailed description, and for the system to retrieve this information.

[0328] "Means for searching past failure reports and evaluating similarity" refers to a function that searches a database of previously recorded failure reports and uses a similarity evaluation algorithm (e.g., cosine similarity) to find reports similar to current failure information.

[0329] "A means of automatically generating a new incident report based on searched past reports" refers to a function that uses a searched and selected past incident report as a template and automatically creates a new incident report based on it.

[0330] "Means of recognizing emotions from user input and actions" refers to a function that analyzes the patterns of text and actions entered by the user to identify the emotions the user is currently feeling.

[0331] "Means for adjusting the content of incident reports based on recognized emotional information" refers to a function that adjusts the content of incident reports to make them more detailed or supplementary in explanations according to the recognized emotions of the user, thereby ensuring the user feels more at ease.

[0332] "A means of generating predicted Q&A by referring to the customer's industry and past inquiry history" refers to a function that generates questions that are expected to be asked by the customer in the future, based on the customer's industry information and past inquiry history.

[0333] "Means for preparing answers to generated Q&A" refers to a function that pre-prepares appropriate answers for each generated question.

[0334] "A means of receiving actual Q&A and evaluating its accuracy by comparing it with predicted Q&A" refers to a function that receives questions and answers actually submitted by users and evaluates how well they match the predicted Q&A.

[0335] "Means for improving the prediction algorithm" refers to a function that improves the question and answer prediction algorithm based on the results obtained from the evaluation of prediction accuracy, thereby improving the accuracy of predictions in subsequent instances.

[0336] This invention provides a system that enables companies to quickly and effectively create and submit incident reports to customers, predict appropriate Q&A based on the customer's industry and past inquiry history, and prepare responses accordingly. Furthermore, it improves the quality of customer service by recognizing user emotions and adjusting responses accordingly.

[0337] Entering and sending error information

[0338] The user logs into the device and accesses the incident report form. The device displays the incident report form using a web page created with HTML and CSS. The user enters the incident name, date and time of occurrence, scope of impact, and detailed description, and clicks the submit button. The device converts the entered incident information into JSON format and sends it to the server using a REST API.

[0339] Automatic generation of incident reports

[0340] When the server receives failure information, it queries past failure reports in the database using SQL statements. It then runs a similarity evaluation algorithm (e.g., cosine similarity) using a Python library (e.g., scikit-learn) to select the most similar past report. Based on this selected report, the server automatically generates a new failure report using a template engine (e.g., Jinja2).

[0341] Adjusting user emotion recognition and responses

[0342] The device has an emotion analysis engine (e.g., Google Cloud Natural Language API) installed, which analyzes the user's emotions based on their input and the timing of their actions. If the user is feeling anxious or angry, emotion information is sent to the server.

[0343] The server adjusts the content of the generated incident report based on the emotional information it receives. For example, if the user is anxious, the report will be revised to include more detailed steps and specific examples.

[0344] Q&A prediction and generation

[0345] The server analyzes the content of the finalized report, referencing the customer's industry and past inquiry history. Based on this information, a natural language processing model (e.g., GPT-3®) is used to generate anticipated Q&A for the customer and prepare appropriate answers for each. Information from the sentiment engine is also utilized to refine the answers.

[0346] Actual visits and feedback

[0347] Based on the generated incident reports and Q&A, users visit customers, submit reports, and answer customer questions. After the visit, users input the questions and answers they received from the customer into a terminal and send them to the server. The server receives this data and evaluates the accuracy of the prediction algorithm. At the same time, sentiment data is also analyzed to help improve the generation algorithm for future generations.

[0348] Example of a prompt

[0349] "Please write a program to automatically generate incident reports. Analyze the incident information entered by the user, compare it to past reports, and generate a new report based on the most similar report. Also, recognize the user's sentiment and adjust the report content accordingly."

[0350] In summary, the system of the present invention enables the rapid and effective creation of fault reports and customer support, allows for flexible responses that take into account user emotions, and can improve customer satisfaction.

[0351] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0352] Step 1:

[0353] The user logs into their device and accesses the bug report form. The device displays the bug report form screen using HTML and CSS.

[0354] Input: User login information and request for the incident report form.

[0355] Data processing: None

[0356] Output: Incident Report Form Screen

[0357] Step 2:

[0358] The user enters basic information about the incident (incident name, date and time of occurrence, scope of impact, and detailed description) into a form and presses the submit button. The device sends the incident information to the server in JSON format.

[0359] Input: Name of the problem, date and time of occurrence, scope of impact, detailed description

[0360] Data processing: Convert form data to JSON format

[0361] Output: Send failure information in JSON format to the server.

[0362] Step 3:

[0363] When the server receives failure information, it searches the database for past failure reports. It uses a similarity evaluation algorithm (e.g., cosine similarity) to select the most similar past report.

[0364] Input: Failure information in JSON format

[0365] Data processing: Database search using SQL queries, similarity evaluation.

[0366] Output: Most similar past incident reports

[0367] Step 4:

[0368] A new incident report is generated based on past reports where the server was selected. A template engine (e.g., Jinja2) is used to fill in each field of the report and generate the completed report.

[0369] Input: Most similar past incident reports

[0370] Data processing: Field embedding using a template engine

[0371] Output: New Incident Report

[0372] Step 5:

[0373] The server sends the generated failure report to the terminal, and the terminal displays the report to the user.

[0374] Input: New Incident Report

[0375] Data processing: None

[0376] Output: Display of the failure report screen

[0377] Step 6:

[0378] The user reviews the report content and makes any necessary corrections. After completing the corrections, they press the confirm button to send the report to the server.

[0379] Input: Revised report content

[0380] Data processing: None

[0381] Output: Send the revised report to the server.

[0382] Step 7:

[0383] The device's built-in emotion analysis engine analyzes user input and interactions to recognize the user's emotions. It then generates specific emotion labels (e.g., impatience, anger).

[0384] Input: User input and operation data

[0385] Data processing: Emotion recognition using emotion analysis algorithms

[0386] Output: Sentiment Labels

[0387] Step 8:

[0388] Based on the emotional information recognized by the server, the content of the incident report will be adjusted. If the user is anxious, detailed steps and additional explanations will be added.

[0389] Input: Revised report, sentiment label

[0390] Data processing: Report adjustment using text generation algorithms

[0391] Output: Adjusted Incident Report

[0392] Step 9:

[0393] The server analyzes the contents of the confirmed report, references the customer's industry and past inquiry history, and generates predicted Q&A. A natural language processing model (e.g., GPT-3) is used.

[0394] Input: Confirmed incident report, customer industry information, past inquiry history

[0395] Data processing: Text analysis, Q&A generation

[0396] Output: Predicted Q&A

[0397] Step 10:

[0398] The server prepares answers to the generated Q&A and adjusts the content of the answers based on sentiment information.

[0399] Input: Predicted Q&A, sentiment information

[0400] Data processing: Response text generation, content adjustment based on sentiment.

[0401] Output: Adjusted Q&A and their answers

[0402] Step 11:

[0403] The server sends the generated Q&A and its answer to the terminal, which then displays it to the user.

[0404] Input: Adjusted Q&A and their answers

[0405] Data processing: None

[0406] Output: Display a Q&A list and its answers on the screen.

[0407] Step 12:

[0408] Users review the Q&A and answers to prepare for their visit. After the visit, they input the actual questions and answers into their terminal and send them to the server.

[0409] Input: Actual questions and answers received

[0410] Data processing: None

[0411] Output: Send actual Q&A data to the server

[0412] Step 13:

[0413] The server compares the actual Q&A with the predicted Q&A and evaluates the accuracy of the prediction. Based on the evaluation results, the prediction algorithm is improved.

[0414] Input: Actual Q&A, Predicted Q&A

[0415] Data processing: Accuracy evaluation, algorithm improvement

[0416] Output: Improved prediction algorithm

[0417] As described above, the system's processing steps involve searching past reports based on the input failure information, referencing similar reports, and generating a new failure report. Furthermore, it adjusts the report considering the user's sentiment, generates anticipated Q&A, and prepares appropriate answers. This enables rapid and effective customer support.

[0418] (Application Example 2)

[0419] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as the "server," and the smart device 14 will be referred to as the "terminal."

[0420] Creating security-related incident reports and handling customer inquiries requires speed and accuracy, but manual processes are time-consuming and labor-intensive, and it's difficult to respond while considering user emotions. Furthermore, predicting the information needed to solve actual problems and preparing appropriate answers in advance is challenging.

[0421] In Application Example 2, the identification processing by the identification processing unit 290 of the data processing device 12 is realized by the following means. In this invention, the server includes means for receiving basic information about a failure, means for searching past failure reports and evaluating similarity, means for automatically generating a new failure report based on the searched past reports, means for referencing the customer's industry and past inquiry history and generating predicted Q&A, means for preparing answers to the generated Q&A, means for receiving actual Q&A and evaluating its accuracy by comparing it with the predicted Q&A, means for recognizing the user's emotions and adjusting the content of the report and Q&A, and means for operating as an application installed on a smartphone or head-mounted display. This streamlines the creation of failure reports and customer support, and enables high-quality support that takes user emotions into consideration.

[0422] "Means for receiving basic information about a failure" refers to a function in the system that allows users to input information such as the name of the failure, the date and time of occurrence, the scope of impact, and a detailed explanation.

[0423] "Means for searching past failure reports and evaluating similarity" refers to a function in which the system searches past failure reports in the database and uses an algorithm to evaluate and identify similarities with those reports.

[0424] The "means for automatically generating new incident reports" refer to a function that automatically creates reports addressing newly occurring incidents by using past reports, which are searched based on similarity assessments, as templates.

[0425] "A means of generating predicted Q&A by referring to the customer's industry and past inquiry history" refers to a function that prepares anticipated questions and their answers in advance based on the customer's industry information and past inquiry history.

[0426] "Means for preparing answers to generated Q&A" refers to a function that provides appropriate information for anticipated questions and prepares the answers in advance.

[0427] "A means of receiving actual Q&A and evaluating its accuracy by comparing it with predicted Q&A" refers to a function that compares the actual questions asked with the predicted results to evaluate and improve prediction accuracy.

[0428] "Means for recognizing user emotions and adjusting the content of reports and Q&A" refers to a function that analyzes the user's emotional state (e.g., impatience or anger) using an emotion recognition engine and appropriately modifies and adjusts the content of reports and Q&A according to those emotions.

[0429] "Means of operation using applications installed on smartphones and head-mounted displays" refers to a method of using this system by installing a dedicated application on devices such as smartphones and head-mounted displays, enabling the system to function on these devices.

[0430] The system for carrying out this invention mainly uses the following hardware and software. Specific examples will be described below.

[0431] Hardware and software to be used

[0432] Device: Smartphone (iOS / ANDROID®) or head-mounted display (HoloLens®, etc.)

[0433] Servers: Web server, database (such as MySQL®)

[0434] Emotion recognition engine: IBM Watson® and Azure Cognitive Services

[0435] AI inference engine: Generative AI models such as GPT-4(registered trademark)

[0436] System Processing Overview

[0437] The server performs the following actions based on the fault report entered by the user using a terminal.

[0438] 1. Enter error information

[0439] The user logs in to their smartphone or head-mounted display (hereinafter referred to as "device") and accesses a dedicated incident reporting form. They enter basic information about the incident (incident name, date and time of occurrence, scope of impact, and detailed description) and press the submit button. The device sends the entered incident information to the server.

[0440] 2. Automatic generation of incident reports

[0441] The server searches its database for past incident reports based on the received incident information. A similarity evaluation algorithm selects the most similar past report, and then automatically generates a new incident report based on it. This report includes an overview of the incident, its cause, scope of impact, and countermeasures.

[0442] 3. User emotion recognition

[0443] The device's built-in emotion recognition engine (e.g., IBM Watson) analyzes user input and interaction to recognize emotions. For example, it can detect if the user is anxious or angry. The server then adjusts the report generated based on the emotion. For instance, it might include detailed instructions or additional explanations for an anxious user.

[0444] 4. Predicting and preparing for the Q&A session

[0445] The server analyzes the generated incident report, referencing the customer's industry and past inquiry history. It generates anticipated questions and prepares appropriate answers for each. The server also adjusts the Q&A responses based on sentiment recognition results.

[0446] 5. Feedback and accuracy evaluation

[0447] The user enters the Q&A used when actually visiting a customer and submitting a report into a terminal and sends it to the server. The server compares the predicted Q&A with the actual Q&A, evaluates the accuracy of the prediction, and improves the algorithm as needed.

[0448] Specific examples and prompt statements

[0449] Example: Scenario for when a security camera malfunction is reported.

[0450] 1. User input fields:

[0451] Problem name: Security camera footage is distorted.

[0452] Date and time of occurrence: October 12, 2023, 14:00

[0453] Scope of impact: All store monitoring systems

[0454] Detailed description: The video frequently cuts out and is noisy.

[0455] 2. Emotion recognition:

[0456] Recognize that the user is anxious and include more detailed explanations in the report.

[0457] 3. Generated report:

[0458] The server automatically generates a report based on the "distorted security camera footage" issue, including the cause and solution.

[0459] 4. Prediction Q&A:

[0460] Q: When will the video become stable?

[0461] A: Expected to be restored within 2 hours.

[0462] Examples of prompts for a generative AI model:

[0463] Please generate the most appropriate report based on past troubleshooting data regarding the issue of distorted security camera footage. Also, please provide anticipated Q&A and their answers.

[0464] Problem name: Security camera footage is distorted.

[0465] Date and time of occurrence: October 12, 2023, 14:00

[0466] Scope of impact: All store monitoring systems

[0467] Detailed description: The video frequently cuts out and is noisy.

[0468] User's emotion: impatience

[0469] This will enable more efficient creation of incident reports and customer support, as well as responses that take user emotions into consideration.

[0470] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0471] Step 1:

[0472] The user logs in using their smartphone or head-mounted display and accesses a dedicated incident reporting form. They enter basic information about the incident, such as the incident name, date and time of occurrence, scope of impact, and a detailed description, and then press the submit button. The entered information is sent from the device to the server.

[0473] Input: Name of the problem, date and time of occurrence, scope of impact, detailed description

[0474] Output: Failure information transferred to the server

[0475] Step 2:

[0476] Based on the failure information received by the server, past failure reports are searched in the database. A similarity evaluation algorithm is used to select the past report with the highest similarity. Database queries and similarity calculations are performed during this search and selection process.

[0477] Input: Error information received by the server

[0478] Output: Past incident reports with high similarity

[0479] Step 3:

[0480] The server automatically generates a new incident report using a previously selected report as a template. The new report includes an overview of the incident, its cause, scope of impact, and countermeasures. Template and text generation algorithms are used in this process.

[0481] Input: Similar past incident reports

[0482] Output: Newly generated incident report

[0483] Step 4:

[0484] The device's built-in emotion recognition engine analyzes user input and interaction to recognize the user's emotions. The recognized emotion information is sent to a server. Text and speech analysis algorithms are used for emotion recognition.

[0485] Input: User input and interaction data

[0486] Output: Recognized user sentiment information

[0487] Step 5:

[0488] The server adjusts the content of the generated incident report based on the perceived emotions. For example, if the user is anxious, the report will include more detailed steps and additional explanations. A text generation algorithm is used in this process.

[0489] Input: Recognized emotion information and automatically generated new incident report

[0490] Output: Adjusted Incident Report

[0491] Step 6:

[0492] The server analyzes the generated incident report and references the customer's industry and past inquiry history. Based on this, it generates predicted Q&A and prepares appropriate answers for each. An AI inference engine is used for prediction and answer generation.

[0493] Input: Adjusted incident report, customer industry information, past inquiry history

[0494] Output: Predicted Q&A and their answers

[0495] Step 7:

[0496] Users visit customers in person and respond to Q&A based on submitted incident reports. After the visit, they input the questions and answers into a terminal and send them to the server.

[0497] Input: Actual questions and answers

[0498] Output: Actual Q&A as feedback to the server

[0499] Step 8:

[0500] The server compares predicted Q&A with actual Q&A to evaluate the accuracy of the prediction. Based on the evaluation results, the prediction algorithm is improved to enhance the accuracy of future answers. An accuracy evaluation algorithm and a feedback loop are used in this process.

[0501] Input: Actual Q&A, Predicted Q&A

[0502] Output: Prediction accuracy evaluation results and improved algorithm

[0503] The specific processing unit 290 transmits the result of the specific processing to the smart device 14. In the smart device 14, the control unit 46A causes the output device 40 to output the result of the specific processing. The microphone 38B acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0504] Data generation model 58 is a type of so-called generative AI (Artificial Intelligence). An example of data generation model 58 is ChatGPT (registered trademark) (Internet search).<URL: https: / / openai.com / blog / chatgpt> ), Gemini (registered trademark) (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0505] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and the specific processing may also be performed by the smart device 14.

[0506] [Second Embodiment]

[0507] Figure 3 shows an example of the configuration of the data processing system 210 according to the second embodiment.

[0508] As shown in Figure 3, the data processing system 210 includes a data processing device 12 and smart glasses 214. An example of the data processing device 12 is a server.

[0509] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0510] The smart glasses 214 include a computer 36, a microphone 238, a speaker 240, a camera 42, and a communication interface 44. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, and camera 42 are also connected to the bus 52.

[0511] The microphone 238 receives voice signals from the user 20 and receives instructions from the user 20. The microphone 238 captures the voice signals from the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[0512] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the area around the user 20 (for example, an imaging range defined by a field of view equivalent to the width of a typical healthy person's field of vision).

[0513] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[0514] Figure 4 shows an example of the main functions of the data processing device 12 and the smart glasses 214. As shown in Figure 4, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[0515] The specific processing program 56 is an example of a "program" relating to the technology of this disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0516] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0517] In the smart glasses 214, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[0518] Next, the identification processing performed by the identification processing unit 290 of the data processing device 12 will be described. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal".

[0519] The present invention aims to improve the quality of customer support after reporting by enabling companies to quickly and effectively create and submit incident reports to customers, predict appropriate Q&A based on the customer's industry and past inquiry history, and prepare answers accordingly.

[0520] Program processing

[0521] Entering and sending error information

[0522] 1. The user logs into the terminal and enters basic information about the problem. Specifically, they enter the name of the problem, the date and time of occurrence, the scope of impact, and a detailed description.

[0523] 2. The terminal sends the entered error information to the server.

[0524] Automatic generation of incident reports

[0525] 3. Based on the received failure information, the server searches past failure reports and evaluates their similarity.

[0526] 4. The server automatically generates a new incident report based on the most similar past incident reports. The new report includes an overview of the incident, its cause, its scope of impact, and countermeasures.

[0527] 5. The server sends the generated report to the terminal.

[0528] 6. The terminal displays the generated report to the user, who can review and modify its contents.

[0529] 7. After the user has finished making corrections, they press the confirm button to send the final report to the server.

[0530] Q&A prediction and preparation

[0531] 8. The server analyzes the contents of the confirmed report and refers to the customer's industry and past inquiry history.

[0532] 9. The server generates anticipated Q&A and prepares appropriate answers for each.

[0533] 10. The server sends the generated Q&A to the terminal.

[0534] 11. The terminal displays the generated Q&A list and its answers to the user.

[0535] 12. Users review the Q&A and answers to prepare for their visit.

[0536] Reports and Feedback

[0537] 13. The user actually visits the customer and submits a report. They respond to customer questions based on the anticipated Q&A during the visit.

[0538] 14. After the visit, the user enters the questions and answers they received into the terminal.

[0539] 15. The terminal sends the actual Q&A entered to the server.

[0540] Accuracy evaluation and algorithm improvement

[0541] 16. The server compares the actual received Q&A with the predicted Q&A and evaluates the accuracy of the prediction.

[0542] 17. The server improves its prediction algorithm based on the evaluation results to improve the accuracy of the next prediction.

[0543] 18. The server generates a new prompt based on an improved algorithm and provides feedback to the user.

[0544] Specific example

[0545] 1. Input and transmission of failure information

[0546] The user enters a message into their terminal indicating that the server is down, and then submits a detailed description of the date and time of the outage and the extent of the impact.

[0547] 2. Automatic generation of incident reports

[0548] The server searches past reports using the keyword "server down" and selects the report that is most similar.

[0549] Based on the selected reports, the server automatically generates a "Server Downtime Incident Report," which includes detailed explanations, causes, and countermeasures.

[0550] 3. Predicting and preparing for the Q&A session

[0551] The server anticipates questions related to "server downtime," such as "server recovery time" and "whether or not data was lost," and prepares appropriate answers for each.

[0552] 4. Reporting and Feedback

[0553] The system responds to questions based on anticipated Q&A from the user during their customer visit. For example, in response to the question, "When will the server be restored?", it might answer, "The estimated restoration time is 2 hours from now."

[0554] After the visit, the terminal inputs the questions the user actually received (for example, "Will there be any data loss?") and the answer ("There will be no data loss"), and sends them to the server.

[0555] 5. Accuracy evaluation and algorithm improvement

[0556] The server compares the predicted Q&A with the actual Q&A to evaluate prediction accuracy and use the results to improve the algorithm.

[0557] The system of this invention streamlines the creation of fault reports and customer support, thereby improving customer satisfaction.

[0558] The following describes the processing flow.

[0559] Step 1:

[0560] The user logs into the device and accesses the incident report form. The device displays the incident report form.

[0561] Step 2:

[0562] The user enters basic information about the incident (incident name, date and time of occurrence, scope of impact, and detailed description) and presses the submit button. The terminal sends the entered incident information to the server.

[0563] Step 3:

[0564] Based on the failure information received by the server, it searches the database for past failure reports. The server uses a similarity evaluation algorithm to select the most similar past report.

[0565] Step 4:

[0566] Based on past reports where servers were selected, a new incident report is automatically generated. The generated report includes an overview of the incident, its cause, the scope of its impact, and the countermeasures taken.

[0567] Step 5:

[0568] The server sends the generated failure report to the terminal. The terminal displays the generated report to the user.

[0569] Step 6:

[0570] The user reviews the report content and makes corrections as needed. Once the corrections are complete, they press the confirm button to send the report to the server.

[0571] Step 7:

[0572] The server analyzes the contents of the confirmed report. The server refers to the customer's industry and past inquiry history to generate predicted Q&A.

[0573] Step 8:

[0574] The server prepares appropriate answers to the predicted Q&A. Once the Q&A is ready, it sends it to the terminal.

[0575] Step 9:

[0576] The terminal displays a generated list of Q&A and its answers to the user. The user reviews the Q&A and answers to prepare for their visit.

[0577] Step 10:

[0578] Users visit customers in person and submit reports. They respond to customer questions based on anticipated Q&A from the visit.

[0579] Step 11:

[0580] After a user visits, the terminal inputs the actual questions and answers they received. The terminal then sends the entered Q&A to the server.

[0581] Step 12:

[0582] The server compares the actual Q&A received with the predicted Q&A to evaluate the accuracy of the prediction.

[0583] Step 13:

[0584] The server improves the prediction algorithm based on the evaluation results. It generates a new prompt based on the improved algorithm and provides feedback to the user.

[0585] By following the steps outlined above, this system can streamline the creation of reports and customer support in the event of a failure, thereby improving accuracy.

[0586] (Example 1)

[0587] Next, we will describe Example 1. In the following description, the data processing device 12 will be referred to as the "server," and the smart glasses 214 will be referred to as the "terminal."

[0588] Traditional incident reporting and customer support processes are often manual, time-consuming and labor-intensive, and can lead to insufficient preparation for anticipated Q&A. This can result in decreased customer satisfaction and damage to the company's reputation. Furthermore, it's difficult to extract relevant information from past reports and inquiry histories to provide efficient and accurate responses.

[0589] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0590] In this invention, the server includes means for receiving basic information about a failure, means for searching past failure reports and evaluating similarity, means for automatically generating a new failure report based on the retrieved past reports, means for displaying the generated failure report and accepting corrections, means for referencing the customer's industry and past inquiry history to generate predicted Q&A, means for preparing answers to the generated Q&A, means for receiving actual Q&A and evaluating its accuracy by comparing it with the predicted Q&A, means for improving the prediction algorithm based on the evaluation results, and means for generating new prompt statements based on the improved algorithm and providing feedback. This streamlines the process of creating failure reports and responding to customers, enabling highly accurate responses.

[0591] "Means for receiving basic information about a malfunction" refers to a device or program that has an interface for a user to input details of a malfunction and a function to transmit that information to a server.

[0592] "Means for searching past incident reports and evaluating similarity" refers to a device or program for searching past incident reports in a database using string search or natural language processing techniques and calculating their similarity to current incident information.

[0593] "Means for automatically generating new incident reports based on retrieved past reports" refers to a device or program that automatically creates a new incident report by using the most similar report from the search results as a template and filling in the current incident information.

[0594] "Means for displaying generated failure reports and accepting corrections" refers to a device or program that displays generated failure reports to the user, has an interface that allows the user to confirm and correct the contents, and has the function of saving and transmitting the corrected contents.

[0595] "Means for generating predicted Q&A by referring to the customer's industry and past inquiry history" refers to a device or program that searches a database for the customer's industry and past inquiry history, and generates predicted questions and answers based on this information.

[0596] "Means for preparing answers to generated Q&A" refers to a device or program that uses natural language generation technology or past answer data to automatically generate appropriate answers to predicted questions.

[0597] "A means of receiving actual Q&A and evaluating its accuracy by comparing it with predicted Q&A" refers to a device or program that takes actual questions and answers received by a user at a visited location as input and evaluates the accuracy of the prediction by comparing it with the predicted Q&A.

[0598] "Means for improving the prediction algorithm based on evaluation results" refers to a device or program that analyzes the results of the accuracy evaluation and adjusts or improves the parameters and model of the prediction algorithm.

[0599] "Means for generating new prompt statements based on an improved algorithm and providing feedback" refers to a device or program that generates new prompt statements using an improved algorithm and has a notification function to inform the user of their contents.

[0600] The present invention aims to enable companies to quickly and effectively create and submit incident reports to their customers, predict appropriate Q&A based on the customer's industry and past inquiry history, and prepare answers accordingly. This system primarily consists of server, terminal, and user-centric processing.

[0601] First, the user logs into the company's internal system and enters basic information about the incident into the terminal. Specifically, this includes the incident name, date and time of occurrence, scope of impact, and a detailed description. The terminal verifies this information and sends it to the server. This process utilizes form input and HTTP requests.

[0602] Next, the server analyzes the received failure information and searches for past failure reports. Natural language processing technology is used for the search, and a similarity score is calculated. Then, a new failure report is automatically generated based on the most similar report. The generated report includes an overview of the failure, its cause, scope of impact, and countermeasures. This report is returned to the terminal, where the user can review and correct its contents.

[0603] Once the corrections are complete, the user confirms the report and sends it back to the server. Based on the confirmed report, the server references the customer's industry and past inquiry history to generate predicted Q&A. This uses database searches and a generation AI model. Appropriate answers are automatically added to the generated Q&A and sent back to the terminal.

[0604] Users check a Q&A list on their terminal to prepare for customer visits. After the visit, they input the actual questions and answers into their terminal and send them to the server. The server compares these actual Q&A with the predicted Q&A and evaluates the accuracy. The evaluation results are used to improve the algorithm, aiming for better prediction accuracy next time. Furthermore, new prompts are generated based on the improved algorithm and provided to the user as feedback.

[0605] Hardware and software details

[0606] Terminals: Desktop PCs, laptops, tablets, etc., within the company are used, and information is entered and displayed via web browsers or dedicated applications.

[0607] Servers: Cloud servers will be used to host large-scale databases and AI models, performing tasks such as data analysis, automated report generation, and Q&A generation. Specifically, Amazon Web Services (AWS) and Microsoft Azure are envisioned.

[0608] Software: For natural language processing, Python libraries such as NLTK and SpaCy are used, and OpenAI's GPT model is used for generative AI models.

[0609] Specific example

[0610] Example prompt: "Please prepare a failure report required when the server goes down. Include the date and time of the incident, the scope of the impact, and a detailed description."

[0611] Example of customer interaction: When a user visits a customer's site, they might ask, "When will the server be restored?" The response would be, "The estimated restoration time is in 2 hours." The actual question is entered into the server, and for the question, "Will there be any data loss?", the response "There will be no data loss" is recorded.

[0612] Through the processes described above, this system streamlines the creation of incident reports and customer support, enabling highly accurate responses.

[0613] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0614] Step 1:

[0615] The user logs into the company's internal system and enters basic information about the incident into the terminal. Specifically, they enter the incident name, date and time of occurrence, scope of impact, and a detailed description (input). The terminal validates the input and sends it to the server (output). During this process, form input occurs and an HTTP request is generated (specific action).

[0616] Step 2:

[0617] The server analyzes the received failure information and searches past failure reports in the database to evaluate similarity (input). It calculates a similarity score using natural language processing techniques (data calculation). It generates a new failure report using the most similar report as a template (output). This involves keyword search and similarity calculation (such as Cosine Similarity) (specific operation).

[0618] Step 3:

[0619] The server sends the generated failure report to the terminal (output). The terminal displays the report to the user and provides an editing form that allows the user to review and modify the contents (specific action). The user reviews the report and makes corrections as needed (input).

[0620] Step 4:

[0621] Once the user has completed the corrections, they press the confirm button to send the final error report to the server (output). The terminal then sends the corrected report to the server as data in JSON format (specific action).

[0622] Step 5:

[0623] The server references the customer's industry and past inquiry history based on the confirmed report (input). It uses database searches and generative AI models to generate predicted Q&A (data computation). The generated Q&A is automatically assigned appropriate answers (output). This uses natural language generation technology to produce answers customized for specific industries (specific behavior).

[0624] Step 6:

[0625] The server sends the generated Q&A list to the terminal (output). The terminal displays the Q&A list and its answers to the user (specific action). The user reviews the Q&A and prepares for their visit (input).

[0626] Step 7:

[0627] During customer visits, users respond to questions based on predicted Q&A (input). After the visit, users input the actual questions received and their answers into the terminal (input).

[0628] Step 8:

[0629] The terminal sends the actual Q&A input to the server (output). The server compares the actual Q&A with the predicted Q&A and evaluates the accuracy of the prediction (data calculation). Based on the evaluation results, the prediction algorithm is improved (output). This uses statistical evaluation metrics (e.g., accuracy, recall) (specific operation).

[0630] Step 9:

[0631] The server generates a new prompt message based on the improved algorithm (output) and provides feedback on its contents to the user (specific action). The user receives the improved prompt message and uses it to help with troubleshooting in the future (input).

[0632] (Application Example 1)

[0633] Next, we will explain Application Example 1. In the following explanation, the data processing device 12 will be referred to as the "server," and the smart glasses 214 will be referred to as the "terminal."

[0634] For companies to quickly and effectively create and submit incident reports to customers, and to provide prompt and high-quality customer support afterward, accurate reporting and appropriate Q&A preparation are required when incidents occur. However, incident response is time-consuming and labor-intensive, and preparing appropriate Q&A based on past inquiry history and the customer's industry is a particularly difficult challenge. Furthermore, human error can occur in report generation and Q&A preparation, so it is necessary to improve the efficiency and accuracy of the entire system.

[0635] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0636] In this invention, the server includes means for receiving basic information about a failure; means for searching past failure reports and evaluating similarity; means for automatically generating a new failure report based on the searched past reports; means for referencing the customer's industry and past inquiry history to generate predicted Q&A; means for preparing answers to the generated Q&A; means for receiving actual Q&A and evaluating its accuracy by comparing it with the predicted Q&A; means for sending the generated failure report and predicted Q&A to a terminal so that the user can review and correct them; means for regenerating the final report and Q&A based on the corrected content; means for inputting the questions and answers actually received into the terminal and sending them to the server; and means for comparing the actual Q&A received by the server with the predicted Q&A and improving the prediction algorithm. This enables increased efficiency and improved quality in the generation of failure reports and customer support.

[0637] "Basic information about the disruption" refers to fundamental information related to the disruption, such as the name of the disruption, the date and time of occurrence, the scope of its impact, and a detailed explanation.

[0638] "Past incident reports" refer to incident reports created in the past, and are used as reference information by analyzing similar incidents.

[0639] "Means for evaluating similarity" refers to a method or system for identifying the most similar case from past incident reports based on the incident information received.

[0640] "Means for automatically generating new incident reports" refers to a method or system for automatically creating new incident reports based on current incident information, using past reports as templates.

[0641] "Customer's industry" refers to the specific industry or sector to which the customer belongs, and is information that reflects the unique requirements and problems associated with that particular industry.

[0642] "Past inquiry history" refers to the content of inquiries received from customers in the past and the history of how those inquiries were handled. This serves as reference material for providing appropriate customer service.

[0643] "Means for generating predicted Q&A" refers to a method or system for analyzing fault information and past inquiry history to generate questions that are predicted to be asked by customers in the future, along with their answers.

[0644] "Means for preparing answers to generated Q&A" refers to a method or system for preparing appropriate answers to anticipated questions.

[0645] "Means for receiving actual Q&A" refers to a method or system for collecting questions and answers actually submitted by customers.

[0646] "Means for evaluating accuracy by comparing with predicted Q&A" refers to a method or system for comparing predicted questions with actual questions to evaluate the accuracy and usefulness of the predictions.

[0647] A "terminal" refers to a device used by system administrators or users, and includes desktop computers, notebooks, smartphones, and other similar devices.

[0648] "Means for users to review and correct" refers to interfaces and tools that allow users to review generated reports and Q&A and make corrections as needed.

[0649] "Means for generating the final report and Q&A based on the revisions" refers to a method or system for generating the final version of the report and Q&A, reflecting the user's revisions.

[0650] "Means of improving prediction algorithms" refers to methods or systems for adjusting prediction models based on actual data to improve their accuracy.

[0651] This invention is a system that enables companies to quickly and effectively create and submit fault reports to customers, and to provide high-quality customer support thereafter. This system consists of a smartphone application (hereinafter referred to as "the app") and a server system.

[0652] First, the user logs into the app and enters basic information about the outage. Specifically, this includes the outage name, date and time of occurrence, scope of impact, and a detailed description. Next, the app sends the entered outage information to the server. The server then compares the received outage information with past outage reports and evaluates the similarity. At this time, the server uses a database (e.g., MongoDB) to search for past reports.

[0653] The server automatically generates a new incident report based on the most similar past incident reports. This generated report includes an overview of the incident, its cause, scope of impact, and countermeasures. The server then sends the generated report to the app, where the user can review it and make corrections as needed. Once corrections are complete, the user presses the confirm button to send the final report to the server.

[0654] Furthermore, the server analyzes the contents of the confirmed incident report and refers to the customer's industry and past inquiry history. This generates predicted Q&A. To generate the Q&A, the server uses an artificial intelligence (AI) model (e.g., TensorFlow). The generated Q&A list and its answers are sent to the app, where the user can review them and prepare for customer visits.

[0655] After the visit, the user enters the actual questions and answers into the app and sends them to the server. The server compares the received actual Q&A with the predicted Q&A and evaluates the accuracy of the prediction. Based on the evaluation results, the server improves the prediction algorithm to improve the accuracy of the next prediction.

[0656] As a concrete example, if a live stream is interrupted due to a malfunction, the user enters the malfunction information, such as "live stream interruption," and sends it to the server, detailing the date and time of the incident and the scope of the impact. The server searches past reports using the keyword "live stream interruption" and generates a new malfunction report based on the most similar report. The server sends the generated report to the app for the user to review and correct. After that, the server generates predictive Q&A related to "live stream interruption" (for example, "When is it expected to be restored?", "Can I rewatch past streamed videos?") and provides appropriate answers.

[0657] Examples of prompt statements are as follows:

[0658] "Input example:

[0659] Incident Name: Live Stream Interruption, Date and Time: 2023-10-10 14:00, Affected Users: All Users, Details: The live stream was interrupted midway.

[0660] Example output:

[0661] Outage Summary: Live stream stopped for all users. Cause: Server load. Countermeasure: Server upgrade. Predicted Q&A: When is it expected to be restored? Answer: It is expected to be restored in 1 hour. Can I watch past streamed videos again? Answer: Yes, you can watch them again.

[0662] This system allows companies to efficiently create reports and handle customer inquiries in the event of a system failure, which is expected to improve customer satisfaction.

[0663] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0664] Step 1:

[0665] The user logs into the smartphone app and enters basic information about the outage (outage name, date and time of occurrence, scope of impact, and detailed description).

[0666] Input: Name of the problem, date and time of occurrence, scope of impact, detailed description

[0667] Output: Failure information data

[0668] Specific action: The user fills in the required information in the app's input form and presses the submit button.

[0669] Step 2:

[0670] The terminal sends the entered error information to the server.

[0671] Input: Failure information data

[0672] Output: Failure information sent to the server

[0673] Specific operation: The app sends an HTTP request to the server via the API.

[0674] Step 3:

[0675] Based on the failure information received by the server, past failure reports are searched in the database (MongoDB) and their similarity is evaluated.

[0676] Input: Received fault information

[0677] Output: Similar past incident reports

[0678] Specific operation: The server extracts keywords related to the failure information and queries the database to find the most similar report.

[0679] Step 4:

[0680] The server automatically generates a new incident report based on the most similar past incident report.

[0681] Input: Similar past incident reports, received incident information

[0682] Output: New Incident Report

[0683] Specific operation: The server uses past reports as templates and combines basic information and detailed descriptions to generate new reports.

[0684] Step 5:

[0685] The server sends the generated report to the terminal, allowing the user to review and correct it.

[0686] Input: New Incident Report

[0687] Output: Report displayed on the user's device

[0688] Specific operation: The server sends the generated report to the app in JSON format, and the app displays it.

[0689] Step 6:

[0690] The user reviews the report and makes corrections as needed. After completing the corrections, they press the confirm button to send the final report to the server.

[0691] Input: User modifications, final report

[0692] Output: Final report sent to the server

[0693] Specific operation: When the user modifies the report content and presses the confirm button, the app sends the final report, including the modifications, to the server.

[0694] Step 7:

[0695] The system analyzes the contents of confirmed server failure reports, referencing the customer's industry and past inquiry history. It then generates anticipated Q&A and prepares appropriate answers.

[0696] Input: Finalized report, customer industry, past inquiry history

[0697] Output: Predicted Q&A list and its answers

[0698] Specific operation: The server uses a prediction algorithm (TensorFlow) to analyze reports and historical data and generate predicted Q&A.

[0699] Step 8:

[0700] The server sends the generated Q&A list and its answers to the user's device for review.

[0701] Input: Predicted Q&A list and its answers

[0702] Output: Q&A displayed on the user's device

[0703] Specific operation: The server sends the generated Q&A to the app, and the app displays it.

[0704] Step 9:

[0705] Users actually visit customers and answer questions based on predicted Q&A.

[0706] Input: Customer questions, predicted Q&A

[0707] Output: Actual Q&A from customer support

[0708] Specific operation: The user refers to a Q&A list and responds to customer questions.

[0709] Step 10:

[0710] After a user visits, the device inputs the actual questions and answers they submitted and sends them to the server.

[0711] Input: Actual Q&A

[0712] Output: Actual Q&A sent to the server

[0713] Specific operation: The user enters the question and answer into the terminal and sends them to the server.

[0714] Step 11:

[0715] The server compares the actual Q&A received with the predicted Q&A to evaluate the accuracy of the prediction.

[0716] Input: Actual Q&A, Predicted Q&A

[0717] Output: Evaluation results and feedback for algorithm improvement

[0718] Specific operation: The server performs comparison calculations and improves the prediction algorithm based on the evaluation results.

[0719] Step 12:

[0720] The server generates new prompts based on an improved algorithm and provides feedback to the user.

[0721] Input: Evaluation results, improved algorithm

[0722] Output: New prompts and feedback content

[0723] Specific action: The server generates a new prompt and notifies the user.

[0724] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[0725] The system of this invention enables companies to quickly and effectively create and submit incident reports to customers, predict appropriate Q&A based on the customer's industry and past inquiry history, prepare answers, and improve the quality of customer service by recognizing user emotions and adjusting responses accordingly. The specific program processing is described below.

[0726] Program processing

[0727] Entering and sending error information

[0728] 1. The user logs into the device and accesses the incident report form. The device displays the incident report form.

[0729] 2. The user enters basic information about the incident (incident name, date and time of occurrence, scope of impact, and detailed description) and presses the submit button. The terminal sends the entered incident information to the server.

[0730] Automatic generation of incident reports

[0731] 3. Based on the failure information received by the server, it searches the database for past failure reports. The server uses a similarity evaluation algorithm to select the most similar past report.

[0732] 4. Based on past reports where the server was selected, a new incident report is automatically generated. The generated report includes an overview of the incident, its cause, the scope of its impact, and the countermeasures taken.

[0733] 5. The server sends the generated report to the terminal. The terminal displays the generated report to the user.

[0734] 6. The user reviews the report content and makes corrections as needed. After completing the corrections, they press the confirm button to send the report to the server.

[0735] Adjusting user emotion recognition and responses

[0736] 7. The device's built-in emotion engine analyzes user input and interactions to recognize the user's emotions. For example, it can detect when the user is anxious or angry.

[0737] 8. The server adjusts the content of the incident report based on the perceived emotions. For example, if the user is anxious, the report is revised to include more detailed steps or additional explanations.

[0738] Q&A prediction and preparation

[0739] 9. The server analyzes the contents of the confirmed report and refers to the customer's industry and past inquiry history.

[0740] 10. The server generates anticipated Q&A and prepares appropriate answers for each. The Q&A and answers are also adjusted based on the perceived user sentiment.

[0741] 11. The server sends the generated Q&A to the terminal. The terminal displays the generated Q&A list and its answers to the user.

[0742] 12. Users review the Q&A and answers to prepare for their visit.

[0743] Reports and Feedback

[0744] 13. The user actually visits the customer and submits a report. They respond to customer questions based on the anticipated Q&A during the visit.

[0745] 14. After the visit, the user enters the actual question and answer into the terminal. The terminal then sends the actual Q&A entered to the server.

[0746] Accuracy evaluation and algorithm improvement

[0747] 15. Compare the actual Q&A received by the server with the predicted Q&A and evaluate the accuracy of the prediction.

[0748] 16. The server improves the prediction algorithm based on the evaluation results. It generates a new prompt based on the improved algorithm and provides feedback to the user.

[0749] 17. The server analyzes user sentiment during actual Q&A sessions and uses the feedback to further improve the accuracy of the generated Q&A and its answers.

[0750] Specific example

[0751] 1. Input and transmission of failure information

[0752] The user enters a message into their terminal indicating that the server is down, and then submits a detailed description of the date and time of the outage and the extent of the impact. The emotion engine detects the user's anxiety.

[0753] 2. Automatic generation of incident reports

[0754] The server searches past reports using the keyword "server down" and selects the most similar report.

[0755] Based on the server selection report, an "Incident Report Regarding Server Downtime" is automatically generated, including detailed explanations, causes, and countermeasures.

[0756] 3. Recognizing and responding to user emotions

[0757] The emotion engine recognizes the user's impatience and adjusts the report to include more detailed steps and additional explanations.

[0758] 4. Predicting and preparing for the Q&A session

[0759] The server anticipates questions related to "server downtime," such as "server recovery time" and "whether data has been lost," and prepares appropriate answers for each.

[0760] Based on the emotions recognized by the emotion engine, the response content is adjusted to be more detailed and easier to understand.

[0761] 5. Reporting and Feedback

[0762] The system responds to questions based on anticipated Q&A from the user during their customer visit. For example, in response to the question, "When will the server be restored?", it might answer, "The estimated restoration time is 2 hours from now."

[0763] After the visit, the terminal inputs the questions the user actually received (for example, "Will there be any data loss?") and the answer ("There will be no data loss"), and sends them to the server.

[0764] 6. Accuracy evaluation and algorithm improvement

[0765] The server compares the predicted Q&A with the actual Q&A to evaluate prediction accuracy and use the results to improve the algorithm.

[0766] The server uses feedback to further improve the accuracy of the Q&A and answers generated using the emotion engine.

[0767] The system of the present invention streamlines the creation of fault reports and customer support, enables responses that take user emotions into consideration, and improves customer satisfaction.

[0768] The following describes the processing flow.

[0769] Step 1:

[0770] The user logs into the device and accesses the incident report form. The device displays the incident report form.

[0771] Step 2:

[0772] The user enters basic information about the incident (incident name, date and time of occurrence, scope of impact, and detailed description) and presses the submit button. The terminal sends the entered incident information to the server.

[0773] Step 3:

[0774] Based on the failure information received by the server, it searches the database for past failure reports. The server uses a similarity evaluation algorithm to select the most similar past report.

[0775] Step 4:

[0776] Based on past reports where servers were selected, a new incident report is automatically generated. The generated report includes an overview of the incident, its cause, the scope of its impact, and the countermeasures taken.

[0777] Step 5:

[0778] The server sends the generated report to the terminal. The terminal displays the generated report to the user.

[0779] Step 6:

[0780] The user reviews the report content and makes corrections as needed. Once the corrections are complete, they press the confirm button to send the report to the server.

[0781] Step 7:

[0782] The device's built-in emotion engine analyzes user input and interactions to recognize the user's emotions. The device can detect situations such as when the user is anxious or angry.

[0783] Step 8:

[0784] The server adjusts the content of the incident report based on the perceived emotions. For example, if the user is anxious, the report will be revised to include more detailed steps and additional explanations.

[0785] Step 9:

[0786] The server analyzes the contents of the confirmed report and refers to the customer's industry and past inquiry history.

[0787] Step 10:

[0788] The server generates anticipated Q&A and prepares appropriate answers for each. It also adjusts the Q&A and answers based on the perceived user sentiment.

[0789] Step 11:

[0790] The server sends the generated Q&A and its answers to the terminal. The terminal displays the generated Q&A list and its answers to the user.

[0791] Step 12:

[0792] Users can review the Q&A and answers to prepare for their visit.

[0793] Step 13:

[0794] Users visit customers in person and submit reports. They respond to customer questions based on anticipated Q&A from the visit.

[0795] Step 14:

[0796] After a user visits, the terminal inputs the actual questions and answers they received. The terminal then sends the entered Q&A to the server.

[0797] Step 15:

[0798] The server compares the actual Q&A received with the predicted Q&A to evaluate the accuracy of the prediction.

[0799] Step 16:

[0800] The server improves the prediction algorithm based on the evaluation results. It generates a new prompt based on the improved algorithm and provides feedback to the user.

[0801] Step 17:

[0802] The server analyzes user sentiment during actual Q&A sessions and uses the feedback to further improve the accuracy of the generated Q&A and its answers.

[0803] Through the steps outlined above, this system streamlines the creation of reports and customer support in the event of a failure, enabling responses that take user emotions into consideration, and ultimately improving customer satisfaction.

[0804] (Example 2)

[0805] Next, we will describe Example 2. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal".

[0806] It is crucial for companies to quickly and effectively create and submit incident reports to their customers, but this usually requires considerable time and effort. Furthermore, preparing appropriate Q&A based on the customer's industry and past inquiry history is not easy. Additionally, considering user emotions in customer service is difficult, leading to inconsistent service quality. A system is needed to address these challenges and improve the quality of customer service.

[0807] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[0808] In this invention, the server includes means for receiving basic information about a failure, means for searching past failure reports and evaluating similarity, means for automatically generating a new failure report based on the retrieved past reports, means for recognizing emotions from user input and operations, means for adjusting the content of the failure report based on the recognized emotion information, means for generating predicted Q&A by referring to the customer's industry and past inquiry history, means for preparing answers to the generated Q&A, means for receiving actual Q&A and evaluating its accuracy by comparing it with the predicted Q&A, and means for improving the prediction algorithm. This enables the rapid and effective creation of failure reports, appropriate responses to customer inquiries, and flexible responses that take user emotions into consideration.

[0809] "Means for receiving basic information about a failure" refers to a function that allows users to input basic information such as the name of the failure, the date and time of occurrence, the scope of impact, and a detailed description, and for the system to retrieve this information.

[0810] "Means for searching past failure reports and evaluating similarity" refers to a function that searches a database of previously recorded failure reports and uses a similarity evaluation algorithm (e.g., cosine similarity) to find reports similar to current failure information.

[0811] "A means of automatically generating new incident reports based on searched past reports" refers to a function that uses searched and selected past incident reports as templates and automatically creates new incident reports based on them.

[0812] "Means of recognizing emotions from user input and actions" refers to a function that analyzes the patterns of text and actions entered by the user to identify the emotions the user is currently feeling.

[0813] "Means for adjusting the content of incident reports based on recognized emotional information" refers to a function that adjusts the content of incident reports to make them more detailed or supplementary in explanations according to the recognized emotions of the user, thereby reassuring the user.

[0814] "A means of generating predicted Q&A by referring to the customer's industry and past inquiry history" refers to a function that generates questions that are expected to be asked by the customer in the future, based on the customer's industry information and past inquiry history.

[0815] "Means for preparing answers to generated Q&A" refers to a function that pre-prepares appropriate answers for each generated question.

[0816] "A means of receiving actual Q&A and evaluating its accuracy by comparing it with predicted Q&A" refers to a function that receives questions and answers actually submitted by users and evaluates how well they match the predicted Q&A.

[0817] "Means for improving the prediction algorithm" refers to a function that improves the question and answer prediction algorithm based on the results obtained from the evaluation of prediction accuracy, thereby improving the accuracy of predictions in subsequent instances.

[0818] This invention provides a system that enables companies to quickly and effectively create and submit incident reports to customers, predict appropriate Q&A based on the customer's industry and past inquiry history, and prepare responses accordingly. Furthermore, it improves the quality of customer service by recognizing user emotions and adjusting responses accordingly.

[0819] Entering and sending error information

[0820] The user logs into the device and accesses the incident report form. The device displays the incident report form using a web page created with HTML and CSS. The user enters the incident name, date and time of occurrence, scope of impact, and detailed description, and clicks the submit button. The device converts the entered incident information into JSON format and sends it to the server using a REST API.

[0821] Automatic generation of incident reports

[0822] When the server receives failure information, it queries past failure reports in the database using SQL statements. It then runs a similarity evaluation algorithm (e.g., cosine similarity) using a Python library (e.g., scikit-learn) to select the most similar past report. Based on this selected report, the server automatically generates a new failure report using a template engine (e.g., Jinja2).

[0823] Adjusting user emotion recognition and responses

[0824] The device has an emotion analysis engine (e.g., Google Cloud Natural Language API) installed, which analyzes the user's emotions based on their input and the timing of their actions. If the user is feeling anxious or angry, emotion information is sent to the server.

[0825] The server adjusts the content of the generated incident report based on the emotional information it receives. For example, if the user is anxious, the report will be revised to include more detailed steps and specific examples.

[0826] Q&A prediction and generation

[0827] The server analyzes the content of the finalized report, referencing the customer's industry and past inquiry history. Based on this information, a natural language processing model (e.g., GPT-3) is used to generate anticipated Q&A for the customer and prepare appropriate answers for each. Information from the sentiment engine is also utilized to refine the answers.

[0828] Actual visits and feedback

[0829] Based on the generated incident reports and Q&A, users visit customers, submit reports, and answer customer questions. After the visit, users input the questions and answers they received from the customer into a terminal and send them to the server. The server receives this data and evaluates the accuracy of the prediction algorithm. At the same time, sentiment data is also analyzed to help improve the generation algorithm for future generations.

[0830] Example of a prompt

[0831] "Please write a program to automatically generate incident reports. Analyze the incident information entered by the user, compare it to past reports, and generate a new report based on the most similar report. Also, recognize the user's sentiment and adjust the report content accordingly."

[0832] In summary, the system of the present invention enables the rapid and effective creation of fault reports and customer support, allows for flexible responses that take into account user emotions, and can improve customer satisfaction.

[0833] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0834] Step 1:

[0835] The user logs into their device and accesses the bug report form. The device displays the bug report form screen using HTML and CSS.

[0836] Input: User login information and request for the incident report form.

[0837] Data processing: None

[0838] Output: Incident Report Form Screen

[0839] Step 2:

[0840] The user enters basic information about the incident (incident name, date and time of occurrence, scope of impact, and detailed description) into a form and presses the submit button. The device sends the incident information to the server in JSON format.

[0841] Input: Name of the problem, date and time of occurrence, scope of impact, detailed description

[0842] Data processing: Convert form data to JSON format

[0843] Output: Send failure information in JSON format to the server.

[0844] Step 3:

[0845] When the server receives failure information, it searches the database for past failure reports. It uses a similarity evaluation algorithm (e.g., cosine similarity) to select the most similar past report.

[0846] Input: Failure information in JSON format

[0847] Data processing: Database search using SQL queries, similarity evaluation.

[0848] Output: Most similar past incident reports

[0849] Step 4:

[0850] A new incident report is generated based on past reports where the server was selected. A template engine (e.g., Jinja2) is used to fill in each field of the report and generate the completed report.

[0851] Input: Most similar past incident reports

[0852] Data processing: Field embedding using a template engine

[0853] Output: New Incident Report

[0854] Step 5:

[0855] The server sends the generated failure report to the terminal, and the terminal displays the report to the user.

[0856] Input: New Incident Report

[0857] Data processing: None

[0858] Output: Display of the failure report screen

[0859] Step 6:

[0860] The user reviews the report content and makes any necessary corrections. After completing the corrections, they press the confirm button to send the report to the server.

[0861] Input: Revised report content

[0862] Data processing: None

[0863] Output: Send the revised report to the server.

[0864] Step 7:

[0865] The device's built-in emotion analysis engine analyzes user input and interactions to recognize the user's emotions. It then generates specific emotion labels (e.g., impatience, anger).

[0866] Input: User input and operation data

[0867] Data processing: Emotion recognition using emotion analysis algorithms

[0868] Output: Sentiment Labels

[0869] Step 8:

[0870] Based on the emotional information recognized by the server, the content of the incident report will be adjusted. If the user is anxious, detailed steps and additional explanations will be added.

[0871] Input: Revised report, sentiment label

[0872] Data processing: Report adjustment using text generation algorithms

[0873] Output: Adjusted Incident Report

[0874] Step 9:

[0875] The server analyzes the contents of the confirmed report, references the customer's industry and past inquiry history, and generates predicted Q&A. A natural language processing model (e.g., GPT-3) is used.

[0876] Input: Confirmed incident report, customer industry information, past inquiry history

[0877] Data processing: Text analysis, Q&A generation

[0878] Output: Predicted Q&A

[0879] Step 10:

[0880] The server prepares answers to the generated Q&A and adjusts the content of the answers based on sentiment information.

[0881] Input: Predicted Q&A, sentiment information

[0882] Data processing: Response text generation, content adjustment based on sentiment.

[0883] Output: Adjusted Q&A and their answers

[0884] Step 11:

[0885] The server sends the generated Q&A and its answer to the terminal, which then displays it to the user.

[0886] Input: Adjusted Q&A and their answers

[0887] Data processing: None

[0888] Output: Display a Q&A list and its answers on the screen.

[0889] Step 12:

[0890] Users review the Q&A and answers to prepare for their visit. After the visit, they input the actual questions and answers into their terminal and send them to the server.

[0891] Input: Actual questions and answers received

[0892] Data processing: None

[0893] Output: Send actual Q&A data to the server

[0894] Step 13:

[0895] The server compares the actual Q&A with the predicted Q&A and evaluates the accuracy of the prediction. Based on the evaluation results, the prediction algorithm is improved.

[0896] Input: Actual Q&A, Predicted Q&A

[0897] Data processing: Accuracy evaluation, algorithm improvement

[0898] Output: Improved prediction algorithm

[0899] As described above, the system's processing steps involve searching past reports based on the input failure information, referencing similar reports, and generating a new failure report. Furthermore, it adjusts the report considering the user's sentiment, generates anticipated Q&A, and prepares appropriate answers. This enables rapid and effective customer support.

[0900] (Application Example 2)

[0901] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as the "server," and the smart glasses 214 will be referred to as the "terminal."

[0902] Creating security-related incident reports and handling customer inquiries requires speed and accuracy, but manual processes are time-consuming and labor-intensive, and it's difficult to respond while considering user emotions. Furthermore, predicting the information needed to solve actual problems and preparing appropriate answers in advance is challenging.

[0903] In Application Example 2, the identification processing by the identification processing unit 290 of the data processing device 12 is realized by the following means. In this invention, the server includes means for receiving basic information about a failure, means for searching past failure reports and evaluating similarity, means for automatically generating a new failure report based on the searched past reports, means for referencing the customer's industry and past inquiry history and generating predicted Q&A, means for preparing answers to the generated Q&A, means for receiving actual Q&A and evaluating its accuracy by comparing it with the predicted Q&A, means for recognizing the user's emotions and adjusting the content of the report and Q&A, and means for operating as an application installed on a smartphone or head-mounted display. This streamlines the creation of failure reports and customer support, and enables high-quality support that takes user emotions into consideration.

[0904] "Means for receiving basic information about a failure" refers to a function in the system that allows users to input information such as the name of the failure, the date and time of occurrence, the scope of impact, and a detailed explanation.

[0905] "Means for searching past failure reports and evaluating similarity" refers to a function in which the system searches past failure reports in the database and uses an algorithm to evaluate and identify similarities with those reports.

[0906] The "means for automatically generating new incident reports" refer to a function that automatically creates reports addressing newly occurring incidents by using past reports, which are searched based on similarity assessments, as templates.

[0907] "A means of generating predicted Q&A by referring to the customer's industry and past inquiry history" refers to a function that prepares anticipated questions and their answers in advance based on the customer's industry information and past inquiry history.

[0908] "Means for preparing answers to generated Q&A" refers to a function that provides appropriate information for anticipated questions and prepares the answers in advance.

[0909] "A means of receiving actual Q&A and evaluating its accuracy by comparing it with predicted Q&A" refers to a function that compares the actual questions asked with the predicted results to evaluate and improve prediction accuracy.

[0910] "Means for recognizing user emotions and adjusting the content of reports and Q&A" refers to a function that analyzes the user's emotional state (e.g., impatience or anger) using an emotion recognition engine and appropriately modifies and adjusts the content of reports and Q&A according to those emotions.

[0911] "Means of operation using applications installed on smartphones and head-mounted displays" refers to a method of using this system by installing a dedicated application on devices such as smartphones and head-mounted displays, enabling the system to function on these devices.

[0912] The system for carrying out this invention mainly uses the following hardware and software. Specific examples will be described below.

[0913] Hardware and software to be used

[0914] Device: Smartphone (iOS / Android) or head-mounted display (HoloLens, etc.)

[0915] Servers: Web server, database (MySQL, etc.)

[0916] Emotion recognition engine: IBM Watson and Azure Cognitive Services

[0917] AI inference engine: Generative AI models such as GPT-4

[0918] System Processing Overview

[0919] The server performs the following actions based on the fault report entered by the user using a terminal.

[0920] 1. Enter error information

[0921] The user logs in to their smartphone or head-mounted display (hereinafter referred to as "device") and accesses a dedicated incident reporting form. They enter basic information about the incident (incident name, date and time of occurrence, scope of impact, and detailed description) and press the submit button. The device sends the entered incident information to the server.

[0922] 2. Automatic generation of incident reports

[0923] The server searches its database for past incident reports based on the received incident information. A similarity evaluation algorithm selects the most similar past report, and then automatically generates a new incident report based on it. This report includes an overview of the incident, its cause, scope of impact, and countermeasures.

[0924] 3. User emotion recognition

[0925] The device's built-in emotion recognition engine (e.g., IBM Watson) analyzes user input and interaction to recognize emotions. For example, it can detect if the user is anxious or angry. The server then adjusts the report generated based on the emotion. For instance, it might include detailed instructions or additional explanations for an anxious user.

[0926] 4. Predicting and preparing for the Q&A session

[0927] The server analyzes the generated incident report, referencing the customer's industry and past inquiry history. It generates anticipated questions and prepares appropriate answers for each. The server also adjusts the Q&A responses based on sentiment recognition results.

[0928] 5. Feedback and accuracy evaluation

[0929] The user enters the Q&A used when actually visiting a customer and submitting a report into a terminal and sends it to the server. The server compares the predicted Q&A with the actual Q&A, evaluates the accuracy of the prediction, and improves the algorithm as needed.

[0930] Specific examples and prompt statements

[0931] Example: Scenario for when a security camera malfunction is reported.

[0932] 1. User input fields:

[0933] Problem name: Security camera footage is distorted.

[0934] Date and time of occurrence: October 12, 2023, 14:00

[0935] Scope of impact: All store monitoring systems

[0936] Detailed description: The video frequently cuts out and is noisy.

[0937] 2. Emotion recognition:

[0938] Recognize that the user is anxious and include more detailed explanations in the report.

[0939] 3. Generated report:

[0940] The server automatically generates a report based on the "distorted security camera footage" issue, including the cause and solution.

[0941] 4. Prediction Q&A:

[0942] Q: When will the video become stable?

[0943] A: Expected to be restored within 2 hours.

[0944] Examples of prompts for a generative AI model:

[0945] Please generate the most appropriate report based on past troubleshooting data regarding the issue of distorted security camera footage. Also, please provide anticipated Q&A and their answers.

[0946] Problem name: Security camera footage is distorted.

[0947] Date and time of occurrence: October 12, 2023, 14:00

[0948] Scope of impact: All store monitoring systems

[0949] Detailed description: The video frequently cuts out and is noisy.

[0950] User's emotion: impatience

[0951] This will enable more efficient creation of incident reports and customer support, as well as responses that take user emotions into consideration.

[0952] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0953] Step 1:

[0954] The user logs in using their smartphone or head-mounted display and accesses a dedicated incident reporting form. They enter basic information about the incident, such as the incident name, date and time of occurrence, scope of impact, and a detailed description, and then press the submit button. The entered information is sent from the device to the server.

[0955] Input: Name of the problem, date and time of occurrence, scope of impact, detailed description

[0956] Output: Failure information transferred to the server

[0957] Step 2:

[0958] Based on the failure information received by the server, past failure reports are searched in the database. A similarity evaluation algorithm is used to select the past report with the highest similarity. Database queries and similarity calculations are performed during this search and selection process.

[0959] Input: Error information received by the server

[0960] Output: Past incident reports with high similarity

[0961] Step 3:

[0962] The server automatically generates a new incident report using a previously selected report as a template. The new report includes an overview of the incident, its cause, scope of impact, and countermeasures. Template and text generation algorithms are used in this process.

[0963] Input: Similar past incident reports

[0964] Output: Newly generated incident report

[0965] Step 4:

[0966] The device's built-in emotion recognition engine analyzes user input and interaction to recognize the user's emotions. The recognized emotion information is sent to a server. Text and speech analysis algorithms are used for emotion recognition.

[0967] Input: User input and interaction data

[0968] Output: Recognized user sentiment information

[0969] Step 5:

[0970] The server adjusts the content of the generated incident report based on the perceived emotions. For example, if the user is anxious, the report will include more detailed steps and additional explanations. A text generation algorithm is used in this process.

[0971] Input: Recognized emotion information and automatically generated new incident report

[0972] Output: Adjusted Incident Report

[0973] Step 6:

[0974] The server analyzes the generated incident report and references the customer's industry and past inquiry history. Based on this, it generates predicted Q&A and prepares appropriate answers for each. An AI inference engine is used for prediction and answer generation.

[0975] Input: Adjusted incident report, customer industry information, past inquiry history

[0976] Output: Predicted Q&A and their answers

[0977] Step 7:

[0978] Users visit customers in person and respond to Q&A based on submitted incident reports. After the visit, they input the questions and answers into a terminal and send them to the server.

[0979] Input: Actual questions and answers

[0980] Output: Actual Q&A as feedback to the server

[0981] Step 8:

[0982] The server compares predicted Q&A with actual Q&A to evaluate the accuracy of the prediction. Based on the evaluation results, the prediction algorithm is improved to enhance the accuracy of future answers. An accuracy evaluation algorithm and a feedback loop are used in this process.

[0983] Input: Actual Q&A, Predicted Q&A

[0984] Output: Prediction accuracy evaluation results and improved algorithm

[0985] The specific processing unit 290 transmits the result of the specific processing to the smart glasses 214. In the smart glasses 214, the control unit 46A causes the speaker 240 to output the result of the specific processing. The microphone 238 acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[0986] Data generation model 58 is a type of so-called generative AI (Artificial Intelligence). An example of data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0987] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and the specific processing may also be performed by the smart glasses 214.

[0988] [Third Embodiment]

[0989] Figure 5 shows an example of the configuration of the data processing system 310 according to the third embodiment.

[0990] As shown in Figure 5, the data processing system 310 includes a data processing device 12 and a headset terminal 314. An example of the data processing device 12 is a server.

[0991] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0992] The headset terminal 314 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication interface 44, and a display 343. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, camera 42, and display 343 are also connected to the bus 52.

[0993] The microphone 238 receives voice signals from the user 20 and receives instructions from the user 20. The microphone 238 captures the voice signals from the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[0994] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the area around the user 20 (for example, an imaging range defined by a field of view equivalent to the width of a typical healthy person's field of vision).

[0995] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[0996] Figure 6 shows an example of the main functions of the data processing device 12 and the headset terminal 314. As shown in Figure 6, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[0997] The specific processing program 56 is an example of a "program" relating to the technology of this disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0998] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0999] In the headset terminal 314, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[1000] Next, the specific processing performed by the specific processing unit 290 of the data processing device 12 will be described. In the following description, the data processing device 12 will be referred to as the "server" and the headset terminal 314 will be referred to as the "terminal".

[1001] The present invention aims to improve the quality of customer support after reporting by enabling companies to quickly and effectively create and submit incident reports to customers, predict appropriate Q&A based on the customer's industry and past inquiry history, and prepare answers accordingly.

[1002] Program processing

[1003] Entering and sending error information

[1004] 1. The user logs into the terminal and enters basic information about the problem. Specifically, they enter the name of the problem, the date and time of occurrence, the scope of impact, and a detailed description.

[1005] 2. The terminal sends the entered error information to the server.

[1006] Automatic generation of incident reports

[1007] 3. Based on the received failure information, the server searches past failure reports and evaluates their similarity.

[1008] 4. The server automatically generates a new incident report based on the most similar past incident reports. The new report includes an overview of the incident, its cause, its scope of impact, and countermeasures.

[1009] 5. The server sends the generated report to the terminal.

[1010] 6. The terminal displays the generated report to the user, who can review and modify its contents.

[1011] 7. After the user has finished making corrections, they press the confirm button to send the final report to the server.

[1012] Q&A prediction and preparation

[1013] 8. The server analyzes the contents of the confirmed report and refers to the customer's industry and past inquiry history.

[1014] 9. The server generates anticipated Q&A and prepares appropriate answers for each.

[1015] 10. The server sends the generated Q&A to the terminal.

[1016] 11. The terminal displays the generated Q&A list and its answers to the user.

[1017] 12. Users review the Q&A and answers to prepare for their visit.

[1018] Reports and Feedback

[1019] 13. The user actually visits the customer and submits a report. They respond to customer questions based on the anticipated Q&A during the visit.

[1020] 14. After the visit, the user enters the questions and answers they received into the terminal.

[1021] 15. The terminal sends the actual Q&A entered to the server.

[1022] Accuracy evaluation and algorithm improvement

[1023] 16. The server compares the actual received Q&A with the predicted Q&A and evaluates the accuracy of the prediction.

[1024] 17. The server improves its prediction algorithm based on the evaluation results to improve the accuracy of the next prediction.

[1025] 18. The server generates a new prompt based on an improved algorithm and provides feedback to the user.

[1026] Specific example

[1027] 1. Input and transmission of failure information

[1028] The user enters a message into their terminal indicating that the server is down, and then submits a detailed description of the date and time of the outage and the extent of the impact.

[1029] 2. Automatic generation of incident reports

[1030] The server searches past reports using the keyword "server down" and selects the report that is most similar.

[1031] Based on the selected reports, the server automatically generates a "Server Downtime Incident Report," which includes detailed explanations, causes, and countermeasures.

[1032] 3. Predicting and preparing for the Q&A session

[1033] The server anticipates questions related to "server downtime," such as "server recovery time" and "whether or not data was lost," and prepares appropriate answers for each.

[1034] 4. Reporting and Feedback

[1035] The system responds to questions based on anticipated Q&A from the user during their customer visit. For example, in response to the question, "When will the server be restored?", it might answer, "The estimated restoration time is 2 hours from now."

[1036] After the visit, the terminal inputs the questions the user actually received (for example, "Will there be any data loss?") and the answer ("There will be no data loss"), and sends them to the server.

[1037] 5. Accuracy evaluation and algorithm improvement

[1038] The server compares the predicted Q&A with the actual Q&A to evaluate prediction accuracy and use the results to improve the algorithm.

[1039] The system of this invention streamlines the creation of fault reports and customer support, thereby improving customer satisfaction.

[1040] The following describes the processing flow.

[1041] Step 1:

[1042] The user logs into the device and accesses the incident report form. The device displays the incident report form.

[1043] Step 2:

[1044] The user enters basic information about the incident (incident name, date and time of occurrence, scope of impact, and detailed description) and presses the submit button. The terminal sends the entered incident information to the server.

[1045] Step 3:

[1046] Based on the failure information received by the server, it searches the database for past failure reports. The server uses a similarity evaluation algorithm to select the most similar past report.

[1047] Step 4:

[1048] Based on past reports where servers were selected, a new incident report is automatically generated. The generated report includes an overview of the incident, its cause, the scope of its impact, and the countermeasures taken.

[1049] Step 5:

[1050] The server sends the generated failure report to the terminal. The terminal displays the generated report to the user.

[1051] Step 6:

[1052] The user reviews the report content and makes corrections as needed. Once the corrections are complete, they press the confirm button to send the report to the server.

[1053] Step 7:

[1054] The server analyzes the contents of the confirmed report. The server refers to the customer's industry and past inquiry history to generate predicted Q&A.

[1055] Step 8:

[1056] The server prepares appropriate answers to the predicted Q&A. Once the Q&A is ready, it sends it to the terminal.

[1057] Step 9:

[1058] The terminal displays a generated list of Q&A and its answers to the user. The user reviews the Q&A and answers to prepare for their visit.

[1059] Step 10:

[1060] Users visit customers in person and submit reports. They respond to customer questions based on anticipated Q&A from the visit.

[1061] Step 11:

[1062] After a user visits, the terminal inputs the actual questions and answers they received. The terminal then sends the entered Q&A to the server.

[1063] Step 12:

[1064] The server compares the actual Q&A received with the predicted Q&A to evaluate the accuracy of the prediction.

[1065] Step 13:

[1066] The server improves the prediction algorithm based on the evaluation results. It generates a new prompt based on the improved algorithm and provides feedback to the user.

[1067] By following the steps outlined above, this system can streamline the creation of reports and customer support in the event of a failure, thereby improving accuracy.

[1068] (Example 1)

[1069] Next, we will describe Example 1. In the following description, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[1070] Traditional incident reporting and customer support processes are often manual, time-consuming and labor-intensive, and can lead to insufficient preparation for anticipated Q&A. This can result in decreased customer satisfaction and damage to the company's reputation. Furthermore, it's difficult to extract relevant information from past reports and inquiry histories to provide efficient and accurate responses.

[1071] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[1072] In this invention, the server includes means for receiving basic information about a failure, means for searching past failure reports and evaluating similarity, means for automatically generating a new failure report based on the retrieved past reports, means for displaying the generated failure report and accepting corrections, means for referencing the customer's industry and past inquiry history to generate predicted Q&A, means for preparing answers to the generated Q&A, means for receiving actual Q&A and evaluating its accuracy by comparing it with the predicted Q&A, means for improving the prediction algorithm based on the evaluation results, and means for generating new prompt statements based on the improved algorithm and providing feedback. This streamlines the process of creating failure reports and responding to customers, enabling highly accurate responses.

[1073] "Means for receiving basic information about a malfunction" refers to a device or program that has an interface for a user to input details of a malfunction and a function to transmit that information to a server.

[1074] "Means for searching past incident reports and evaluating similarity" refers to a device or program for searching past incident reports in a database using string search or natural language processing techniques and calculating their similarity to current incident information.

[1075] "Means for automatically generating new incident reports based on retrieved past reports" refers to a device or program that automatically creates a new incident report by using the most similar report from the search results as a template and filling in the current incident information.

[1076] "Means for displaying generated failure reports and accepting corrections" refers to a device or program that displays generated failure reports to the user, has an interface that allows the user to confirm and correct the contents, and has the function of saving and transmitting the corrected contents.

[1077] "Means for generating predicted Q&A by referring to the customer's industry and past inquiry history" refers to a device or program that searches a database for the customer's industry and past inquiry history, and generates predicted questions and answers based on this information.

[1078] "Means for preparing answers to generated Q&A" refers to a device or program that uses natural language generation technology or past answer data to automatically generate appropriate answers to predicted questions.

[1079] "A means of receiving actual Q&A and evaluating its accuracy by comparing it with predicted Q&A" refers to a device or program that takes actual questions and answers received by a user at a visited location as input and evaluates the accuracy of the prediction by comparing it with the predicted Q&A.

[1080] "Means for improving the prediction algorithm based on evaluation results" refers to a device or program that analyzes the results of the accuracy evaluation and adjusts or improves the parameters and model of the prediction algorithm.

[1081] "Means for generating new prompt statements based on an improved algorithm and providing feedback" refers to a device or program that generates new prompt statements using an improved algorithm and has a notification function to inform the user of their contents.

[1082] The present invention aims to enable companies to quickly and effectively create and submit incident reports to their customers, predict appropriate Q&A based on the customer's industry and past inquiry history, and prepare answers accordingly. This system primarily consists of server, terminal, and user-centric processing.

[1083] First, the user logs into the company's internal system and enters basic information about the incident into the terminal. Specifically, this includes the incident name, date and time of occurrence, scope of impact, and a detailed description. The terminal verifies this information and sends it to the server. This process utilizes form input and HTTP requests.

[1084] Next, the server analyzes the received failure information and searches for past failure reports. Natural language processing technology is used for the search, and a similarity score is calculated. Then, a new failure report is automatically generated based on the most similar report. The generated report includes an overview of the failure, its cause, scope of impact, and countermeasures. This report is returned to the terminal, where the user can review and correct its contents.

[1085] Once the corrections are complete, the user confirms the report and sends it back to the server. Based on the confirmed report, the server references the customer's industry and past inquiry history to generate predicted Q&A. This uses database searches and a generation AI model. Appropriate answers are automatically added to the generated Q&A and sent back to the terminal.

[1086] Users check a Q&A list on their terminal to prepare for customer visits. After the visit, they input the actual questions and answers into their terminal and send them to the server. The server compares these actual Q&A with the predicted Q&A and evaluates the accuracy. The evaluation results are used to improve the algorithm, aiming for better prediction accuracy next time. Furthermore, new prompts are generated based on the improved algorithm and provided to the user as feedback.

[1087] Hardware and software details

[1088] Terminals: Desktop PCs, laptops, tablets, etc., within the company are used, and information is entered and displayed via web browsers or dedicated applications.

[1089] Servers: Cloud servers will be used to host large-scale databases and AI models, performing tasks such as data analysis, automated report generation, and Q&A generation. Specifically, Amazon Web Services (AWS) and Microsoft Azure are envisioned.

[1090] Software: For natural language processing, Python libraries such as NLTK and SpaCy are used, and OpenAI's GPT model is used for generative AI models.

[1091] Specific example

[1092] Example prompt: "Please prepare a failure report required when the server goes down. Include the date and time of the incident, the scope of the impact, and a detailed description."

[1093] Example of customer interaction: When a user visits a customer's site, they might ask, "When will the server be restored?" The response would be, "The estimated restoration time is in 2 hours." The actual question is entered into the server, and for the question, "Will there be any data loss?", the response "There will be no data loss" is recorded.

[1094] Through the processes described above, this system streamlines the creation of incident reports and customer support, enabling highly accurate responses.

[1095] The flow of the specific processing in Example 1 will be explained using Figure 11.

[1096] Step 1:

[1097] The user logs into the company's internal system and enters basic information about the incident into the terminal. Specifically, they enter the incident name, date and time of occurrence, scope of impact, and a detailed description (input). The terminal validates the input and sends it to the server (output). During this process, form input occurs and an HTTP request is generated (specific action).

[1098] Step 2:

[1099] The server analyzes the received failure information and searches past failure reports in the database to evaluate similarity (input). It calculates a similarity score using natural language processing techniques (data calculation). It generates a new failure report using the most similar report as a template (output). This involves keyword search and similarity calculation (such as Cosine Similarity) (specific operation).

[1100] Step 3:

[1101] The server sends the generated failure report to the terminal (output). The terminal displays the report to the user and provides an editing form that allows the user to review and modify the contents (specific action). The user reviews the report and makes corrections as needed (input).

[1102] Step 4:

[1103] Once the user has completed the corrections, they press the confirm button to send the final error report to the server (output). The terminal then sends the corrected report to the server as data in JSON format (specific action).

[1104] Step 5:

[1105] The server references the customer's industry and past inquiry history based on the confirmed report (input). It uses database searches and generative AI models to generate predicted Q&A (data computation). The generated Q&A is automatically assigned appropriate answers (output). This uses natural language generation technology to produce answers customized for specific industries (specific behavior).

[1106] Step 6:

[1107] The server sends the generated Q&A list to the terminal (output). The terminal displays the Q&A list and its answers to the user (specific action). The user reviews the Q&A and prepares for their visit (input).

[1108] Step 7:

[1109] During customer visits, users respond to questions based on predicted Q&A (input). After the visit, users input the actual questions received and their answers into the terminal (input).

[1110] Step 8:

[1111] The terminal sends the actual Q&A input to the server (output). The server compares the actual Q&A with the predicted Q&A and evaluates the accuracy of the prediction (data calculation). Based on the evaluation results, the prediction algorithm is improved (output). This uses statistical evaluation metrics (e.g., accuracy, recall) (specific operation).

[1112] Step 9:

[1113] The server generates a new prompt message based on the improved algorithm (output) and provides feedback on its contents to the user (specific action). The user receives the improved prompt message and uses it to help with troubleshooting in the future (input).

[1114] (Application Example 1)

[1115] Next, we will explain Application Example 1. In the following explanation, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[1116] For companies to quickly and effectively create and submit incident reports to customers, and to provide prompt and high-quality customer support afterward, accurate reporting and appropriate Q&A preparation are required when incidents occur. However, incident response is time-consuming and labor-intensive, and preparing appropriate Q&A based on past inquiry history and the customer's industry is a particularly difficult challenge. Furthermore, human error can occur in report generation and Q&A preparation, so it is necessary to improve the efficiency and accuracy of the entire system.

[1117] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[1118] In this invention, the server includes means for receiving basic information about a failure; means for searching past failure reports and evaluating similarity; means for automatically generating a new failure report based on the searched past reports; means for referencing the customer's industry and past inquiry history to generate predicted Q&A; means for preparing answers to the generated Q&A; means for receiving actual Q&A and evaluating its accuracy by comparing it with the predicted Q&A; means for sending the generated failure report and predicted Q&A to a terminal so that the user can review and correct them; means for regenerating the final report and Q&A based on the corrected content; means for inputting the questions and answers actually received into the terminal and sending them to the server; and means for comparing the actual Q&A received by the server with the predicted Q&A and improving the prediction algorithm. This enables increased efficiency and improved quality in the generation of failure reports and customer support.

[1119] "Basic information about the disruption" refers to fundamental information related to the disruption, such as the name of the disruption, the date and time of occurrence, the scope of its impact, and a detailed explanation.

[1120] "Past incident reports" refer to incident reports created in the past, and are used as reference information by analyzing similar incidents.

[1121] "Means for evaluating similarity" refers to a method or system for identifying the most similar case from past incident reports based on the incident information received.

[1122] "Means for automatically generating new incident reports" refers to a method or system for automatically creating new incident reports based on current incident information, using past reports as templates.

[1123] "Customer's industry" refers to the specific industry or sector to which the customer belongs, and is information that reflects the unique requirements and problems associated with that particular industry.

[1124] "Past inquiry history" refers to the content of inquiries received from customers in the past and the history of how those inquiries were handled. This serves as reference material for providing appropriate customer service.

[1125] "Means for generating predicted Q&A" refers to a method or system for analyzing fault information and past inquiry history to generate questions that are predicted to be asked by customers in the future, along with their answers.

[1126] "Means for preparing answers to generated Q&A" refers to a method or system for preparing appropriate answers to anticipated questions.

[1127] "Means for receiving actual Q&A" refers to a method or system for collecting questions and answers actually submitted by customers.

[1128] "Means for evaluating accuracy by comparing with predicted Q&A" refers to a method or system for comparing predicted questions with actual questions to evaluate the accuracy and usefulness of the predictions.

[1129] A "terminal" refers to a device used by system administrators or users, and includes desktop computers, notebooks, smartphones, and other similar devices.

[1130] "Means for users to review and correct" refers to interfaces and tools that allow users to review generated reports and Q&A and make corrections as needed.

[1131] "Means for generating the final report and Q&A based on the revisions" refers to a method or system for generating the final version of the report and Q&A, reflecting the user's revisions.

[1132] "Means of improving prediction algorithms" refers to methods or systems for adjusting prediction models based on actual data to improve their accuracy.

[1133] This invention is a system that enables companies to quickly and effectively create and submit fault reports to customers, and to provide high-quality customer support thereafter. This system consists of a smartphone application (hereinafter referred to as "the app") and a server system.

[1134] First, the user logs into the app and enters basic information about the outage. Specifically, this includes the outage name, date and time of occurrence, scope of impact, and a detailed description. Next, the app sends the entered outage information to the server. The server then compares the received outage information with past outage reports and evaluates the similarity. At this time, the server uses a database (e.g., MongoDB) to search for past reports.

[1135] The server automatically generates a new incident report based on the most similar past incident reports. This generated report includes an overview of the incident, its cause, scope of impact, and countermeasures. The server then sends the generated report to the app, where the user can review it and make corrections as needed. Once corrections are complete, the user presses the confirm button to send the final report to the server.

[1136] Furthermore, the server analyzes the contents of the confirmed incident report and refers to the customer's industry and past inquiry history. This generates predicted Q&A. To generate the Q&A, the server uses an artificial intelligence (AI) model (e.g., TensorFlow). The generated Q&A list and its answers are sent to the app, where the user can review them and prepare for customer visits.

[1137] After the visit, the user enters the actual questions and answers into the app and sends them to the server. The server compares the received actual Q&A with the predicted Q&A and evaluates the accuracy of the prediction. Based on the evaluation results, the server improves the prediction algorithm to improve the accuracy of the next prediction.

[1138] As a concrete example, if a live stream is interrupted due to a malfunction, the user enters the malfunction information, such as "live stream interruption," and sends it to the server, detailing the date and time of the incident and the scope of the impact. The server searches past reports using the keyword "live stream interruption" and generates a new malfunction report based on the most similar report. The server sends the generated report to the app for the user to review and correct. After that, the server generates predictive Q&A related to "live stream interruption" (for example, "When is it expected to be restored?", "Can I rewatch past streamed videos?") and provides appropriate answers.

[1139] Examples of prompt statements are as follows:

[1140] "Input example:

[1141] Incident Name: Live Stream Interruption, Date and Time: 2023-10-10 14:00, Affected Users: All Users, Details: The live stream was interrupted midway.

[1142] Example output:

[1143] Outage Summary: Live stream stopped for all users. Cause: Server load. Countermeasure: Server upgrade. Predicted Q&A: When is it expected to be restored? Answer: It is expected to be restored in 1 hour. Can I watch past streamed videos again? Answer: Yes, you can watch them again.

[1144] This system allows companies to efficiently create reports and handle customer inquiries in the event of a system failure, which is expected to improve customer satisfaction.

[1145] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[1146] Step 1:

[1147] The user logs into the smartphone app and enters basic information about the outage (outage name, date and time of occurrence, scope of impact, and detailed description).

[1148] Input: Name of the problem, date and time of occurrence, scope of impact, detailed description

[1149] Output: Failure information data

[1150] Specific action: The user fills in the required information in the app's input form and presses the submit button.

[1151] Step 2:

[1152] The terminal sends the entered error information to the server.

[1153] Input: Failure information data

[1154] Output: Failure information sent to the server

[1155] Specific operation: The app sends an HTTP request to the server via the API.

[1156] Step 3:

[1157] Based on the failure information received by the server, past failure reports are searched in the database (MongoDB) and their similarity is evaluated.

[1158] Input: Received fault information

[1159] Output: Similar past incident reports

[1160] Specific operation: The server extracts keywords related to the failure information and queries the database to find the most similar report.

[1161] Step 4:

[1162] The server automatically generates a new incident report based on the most similar past incident report.

[1163] Input: Similar past incident reports, received incident information

[1164] Output: New Incident Report

[1165] Specific operation: The server uses past reports as templates and combines basic information and detailed descriptions to generate new reports.

[1166] Step 5:

[1167] The server sends the generated report to the terminal, allowing the user to review and correct it.

[1168] Input: New Incident Report

[1169] Output: Report displayed on the user's device

[1170] Specific operation: The server sends the generated report to the app in JSON format, and the app displays it.

[1171] Step 6:

[1172] The user reviews the report and makes corrections as needed. After completing the corrections, they press the confirm button to send the final report to the server.

[1173] Input: User modifications, final report

[1174] Output: Final report sent to the server

[1175] Specific operation: When the user modifies the report content and presses the confirm button, the app sends the final report, including the modifications, to the server.

[1176] Step 7:

[1177] The system analyzes the contents of confirmed server failure reports, referencing the customer's industry and past inquiry history. It then generates anticipated Q&A and prepares appropriate answers.

[1178] Input: Finalized report, customer industry, past inquiry history

[1179] Output: Predicted Q&A list and its answers

[1180] Specific operation: The server uses a prediction algorithm (TensorFlow) to analyze reports and historical data and generate predicted Q&A.

[1181] Step 8:

[1182] The server sends the generated Q&A list and its answers to the user's device for review.

[1183] Input: Predicted Q&A list and its answers

[1184] Output: Q&A displayed on the user's device

[1185] Specific operation: The server sends the generated Q&A to the app, and the app displays it.

[1186] Step 9:

[1187] Users actually visit customers and answer questions based on predicted Q&A.

[1188] Input: Customer questions, predicted Q&A

[1189] Output: Actual Q&A from customer support

[1190] Specific operation: The user refers to a Q&A list and responds to customer questions.

[1191] Step 10:

[1192] After a user visits, the device inputs the actual questions and answers they submitted and sends them to the server.

[1193] Input: Actual Q&A

[1194] Output: Actual Q&A sent to the server

[1195] Specific operation: The user enters the question and answer into the terminal and sends them to the server.

[1196] Step 11:

[1197] The server compares the actual Q&A received with the predicted Q&A to evaluate the accuracy of the prediction.

[1198] Input: Actual Q&A, Predicted Q&A

[1199] Output: Evaluation results and feedback for algorithm improvement

[1200] Specific operation: The server performs comparison calculations and improves the prediction algorithm based on the evaluation results.

[1201] Step 12:

[1202] The server generates new prompts based on an improved algorithm and provides feedback to the user.

[1203] Input: Evaluation results, improved algorithm

[1204] Output: New prompts and feedback content

[1205] Specific action: The server generates a new prompt and notifies the user.

[1206] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[1207] The system of this invention enables companies to quickly and effectively create and submit incident reports to customers, predict appropriate Q&A based on the customer's industry and past inquiry history, prepare answers, and improve the quality of customer service by recognizing user emotions and adjusting responses accordingly. The specific program processing is described below.

[1208] Program processing

[1209] Entering and sending error information

[1210] 1. The user logs into the device and accesses the incident report form. The device displays the incident report form.

[1211] 2. The user enters basic information about the incident (incident name, date and time of occurrence, scope of impact, and detailed description) and presses the submit button. The terminal sends the entered incident information to the server.

[1212] Automatic generation of incident reports

[1213] 3. Based on the failure information received by the server, it searches the database for past failure reports. The server uses a similarity evaluation algorithm to select the most similar past report.

[1214] 4. Based on past reports where the server was selected, a new incident report is automatically generated. The generated report includes an overview of the incident, its cause, the scope of its impact, and the countermeasures taken.

[1215] 5. The server sends the generated report to the terminal. The terminal displays the generated report to the user.

[1216] 6. The user reviews the report content and makes corrections as needed. After completing the corrections, they press the confirm button to send the report to the server.

[1217] Adjusting user emotion recognition and responses

[1218] 7. The device's built-in emotion engine analyzes user input and interactions to recognize the user's emotions. For example, it can detect when the user is anxious or angry.

[1219] 8. The server adjusts the content of the incident report based on the perceived emotions. For example, if the user is anxious, the report is revised to include more detailed steps or additional explanations.

[1220] Q&A prediction and preparation

[1221] 9. The server analyzes the contents of the confirmed report and refers to the customer's industry and past inquiry history.

[1222] 10. The server generates anticipated Q&A and prepares appropriate answers for each. The Q&A and answers are also adjusted based on the perceived user sentiment.

[1223] 11. The server sends the generated Q&A to the terminal. The terminal displays the generated Q&A list and its answers to the user.

[1224] 12. Users review the Q&A and answers to prepare for their visit.

[1225] Reports and Feedback

[1226] 13. The user actually visits the customer and submits a report. They respond to customer questions based on the anticipated Q&A during the visit.

[1227] 14. After the visit, the user enters the actual question and answer into the terminal. The terminal then sends the actual Q&A entered to the server.

[1228] Accuracy evaluation and algorithm improvement

[1229] 15. Compare the actual Q&A received by the server with the predicted Q&A and evaluate the accuracy of the prediction.

[1230] 16. The server improves the prediction algorithm based on the evaluation results. It generates a new prompt based on the improved algorithm and provides feedback to the user.

[1231] 17. The server analyzes user sentiment during actual Q&A sessions and uses the feedback to further improve the accuracy of the generated Q&A and its answers.

[1232] Specific example

[1233] 1. Input and transmission of failure information

[1234] The user enters a message into their terminal indicating that the server is down, and then submits a detailed description of the date and time of the outage and the extent of the impact. The emotion engine detects the user's anxiety.

[1235] 2. Automatic generation of incident reports

[1236] The server searches past reports using the keyword "server down" and selects the most similar report.

[1237] Based on the server selection report, an "Incident Report Regarding Server Downtime" is automatically generated, including detailed explanations, causes, and countermeasures.

[1238] 3. Recognizing and responding to user emotions

[1239] The emotion engine recognizes the user's impatience and adjusts the report to include more detailed steps and additional explanations.

[1240] 4. Predicting and preparing for the Q&A session

[1241] The server anticipates questions related to "server downtime," such as "server recovery time" and "whether data has been lost," and prepares appropriate answers for each.

[1242] Based on the emotions recognized by the emotion engine, the response content is adjusted to be more detailed and easier to understand.

[1243] 5. Reporting and Feedback

[1244] The system responds to questions based on anticipated Q&A from the user during their customer visit. For example, in response to the question, "When will the server be restored?", it might answer, "The estimated restoration time is 2 hours from now."

[1245] After the visit, the terminal inputs the questions the user actually received (for example, "Will there be any data loss?") and the answer ("There will be no data loss"), and sends them to the server.

[1246] 6. Accuracy evaluation and algorithm improvement

[1247] The server compares the predicted Q&A with the actual Q&A to evaluate prediction accuracy and use the results to improve the algorithm.

[1248] The server uses feedback to further improve the accuracy of the Q&A and answers generated using the emotion engine.

[1249] The system of the present invention streamlines the creation of fault reports and customer support, enables responses that take user emotions into consideration, and improves customer satisfaction.

[1250] The following describes the processing flow.

[1251] Step 1:

[1252] The user logs into the device and accesses the incident report form. The device displays the incident report form.

[1253] Step 2:

[1254] The user enters basic information about the incident (incident name, date and time of occurrence, scope of impact, and detailed description) and presses the submit button. The terminal sends the entered incident information to the server.

[1255] Step 3:

[1256] Based on the failure information received by the server, it searches the database for past failure reports. The server uses a similarity evaluation algorithm to select the most similar past report.

[1257] Step 4:

[1258] Based on past reports where servers were selected, a new incident report is automatically generated. The generated report includes an overview of the incident, its cause, the scope of its impact, and the countermeasures taken.

[1259] Step 5:

[1260] The server sends the generated report to the terminal. The terminal displays the generated report to the user.

[1261] Step 6:

[1262] The user reviews the report content and makes corrections as needed. Once the corrections are complete, they press the confirm button to send the report to the server.

[1263] Step 7:

[1264] The device's built-in emotion engine analyzes user input and interactions to recognize the user's emotions. The device can detect situations such as when the user is anxious or angry.

[1265] Step 8:

[1266] The server adjusts the content of the incident report based on the perceived emotions. For example, if the user is anxious, the report will be revised to include more detailed steps and additional explanations.

[1267] Step 9:

[1268] The server analyzes the contents of the confirmed report and refers to the customer's industry and past inquiry history.

[1269] Step 10:

[1270] The server generates anticipated Q&A and prepares appropriate answers for each. It also adjusts the Q&A and answers based on the perceived user sentiment.

[1271] Step 11:

[1272] The server sends the generated Q&A and its answers to the terminal. The terminal displays the generated Q&A list and its answers to the user.

[1273] Step 12:

[1274] Users can review the Q&A and answers to prepare for their visit.

[1275] Step 13:

[1276] Users visit customers in person and submit reports. They respond to customer questions based on anticipated Q&A from the visit.

[1277] Step 14:

[1278] After a user visits, the terminal inputs the actual questions and answers they received. The terminal then sends the entered Q&A to the server.

[1279] Step 15:

[1280] The server compares the actual Q&A received with the predicted Q&A to evaluate the accuracy of the prediction.

[1281] Step 16:

[1282] The server improves the prediction algorithm based on the evaluation results. It generates a new prompt based on the improved algorithm and provides feedback to the user.

[1283] Step 17:

[1284] The server analyzes user sentiment during actual Q&A sessions and uses the feedback to further improve the accuracy of the generated Q&A and its answers.

[1285] Through the steps outlined above, this system streamlines the creation of reports and customer support in the event of a failure, enabling responses that take user emotions into consideration, and ultimately improving customer satisfaction.

[1286] (Example 2)

[1287] Next, we will describe Example 2. In the following description, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[1288] It is crucial for companies to quickly and effectively create and submit incident reports to their customers, but this usually requires considerable time and effort. Furthermore, preparing appropriate Q&A based on the customer's industry and past inquiry history is not easy. Additionally, considering user emotions in customer service is difficult, leading to inconsistent service quality. A system is needed to address these challenges and improve the quality of customer service.

[1289] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[1290] In this invention, the server includes means for receiving basic information about a failure, means for searching past failure reports and evaluating similarity, means for automatically generating a new failure report based on the retrieved past reports, means for recognizing emotions from user input and operations, means for adjusting the content of the failure report based on the recognized emotion information, means for generating predicted Q&A by referring to the customer's industry and past inquiry history, means for preparing answers to the generated Q&A, means for receiving actual Q&A and evaluating its accuracy by comparing it with the predicted Q&A, and means for improving the prediction algorithm. This enables the rapid and effective creation of failure reports, appropriate responses to customer inquiries, and flexible responses that take user emotions into consideration.

[1291] "Means for receiving basic information about a failure" refers to a function that allows users to input basic information such as the name of the failure, the date and time of occurrence, the scope of impact, and a detailed description, and for the system to retrieve this information.

[1292] "Means for searching past failure reports and evaluating similarity" refers to a function that searches a database of previously recorded failure reports and uses a similarity evaluation algorithm (e.g., cosine similarity) to find reports similar to current failure information.

[1293] "A means of automatically generating new incident reports based on searched past reports" refers to a function that uses searched and selected past incident reports as templates and automatically creates new incident reports based on them.

[1294] "Means of recognizing emotions from user input and actions" refers to a function that analyzes the patterns of text and actions entered by the user to identify the emotions the user is currently feeling.

[1295] "Means for adjusting the content of incident reports based on recognized emotional information" refers to a function that adjusts the content of incident reports to make them more detailed or supplementary in explanations according to the recognized emotions of the user, thereby reassuring the user.

[1296] "A means of generating predicted Q&A by referring to the customer's industry and past inquiry history" refers to a function that generates questions that are expected to be asked by the customer in the future, based on the customer's industry information and past inquiry history.

[1297] "Means for preparing answers to generated Q&A" refers to a function that pre-prepares appropriate answers for each generated question.

[1298] "A means of receiving actual Q&A and evaluating its accuracy by comparing it with predicted Q&A" refers to a function that receives questions and answers actually submitted by users and evaluates how well they match the predicted Q&A.

[1299] "Means for improving the prediction algorithm" refers to a function that improves the question and answer prediction algorithm based on the results obtained from the evaluation of prediction accuracy, thereby improving the accuracy of predictions in subsequent instances.

[1300] This invention provides a system that enables companies to quickly and effectively create and submit incident reports to customers, predict appropriate Q&A based on the customer's industry and past inquiry history, and prepare responses accordingly. Furthermore, it improves the quality of customer service by recognizing user emotions and adjusting responses accordingly.

[1301] Entering and sending error information

[1302] The user logs into the device and accesses the incident report form. The device displays the incident report form using a web page created with HTML and CSS. The user enters the incident name, date and time of occurrence, scope of impact, and detailed description, and clicks the submit button. The device converts the entered incident information into JSON format and sends it to the server using a REST API.

[1303] Automatic generation of incident reports

[1304] When the server receives failure information, it queries past failure reports in the database using SQL statements. It then runs a similarity evaluation algorithm (e.g., cosine similarity) using a Python library (e.g., scikit-learn) to select the most similar past report. Based on this selected report, the server automatically generates a new failure report using a template engine (e.g., Jinja2).

[1305] Adjusting user emotion recognition and responses

[1306] The device has an emotion analysis engine (e.g., Google Cloud Natural Language API) installed, which analyzes the user's emotions based on their input and the timing of their actions. If the user is feeling anxious or angry, emotion information is sent to the server.

[1307] The server adjusts the content of the generated incident report based on the emotional information it receives. For example, if the user is anxious, the report will be revised to include more detailed steps and specific examples.

[1308] Q&A prediction and generation

[1309] The server analyzes the content of the finalized report, referencing the customer's industry and past inquiry history. Based on this information, a natural language processing model (e.g., GPT-3) is used to generate anticipated Q&A for the customer and prepare appropriate answers for each. Information from the sentiment engine is also utilized to refine the answers.

[1310] Actual visits and feedback

[1311] Based on the generated incident reports and Q&A, users visit customers, submit reports, and answer customer questions. After the visit, users input the questions and answers they received from the customer into a terminal and send them to the server. The server receives this data and evaluates the accuracy of the prediction algorithm. At the same time, sentiment data is also analyzed to help improve the generation algorithm for future generations.

[1312] Example of a prompt

[1313] "Please write a program to automatically generate incident reports. Analyze the incident information entered by the user, compare it to past reports, and generate a new report based on the most similar report. Also, recognize the user's sentiment and adjust the report content accordingly."

[1314] In summary, the system of the present invention enables the rapid and effective creation of fault reports and customer support, allows for flexible responses that take into account user emotions, and can improve customer satisfaction.

[1315] The flow of the specific processing in Example 2 will be explained using Figure 13.

[1316] Step 1:

[1317] The user logs into their device and accesses the bug report form. The device displays the bug report form screen using HTML and CSS.

[1318] Input: User login information and request for the incident report form.

[1319] Data processing: None

[1320] Output: Incident Report Form Screen

[1321] Step 2:

[1322] The user enters basic information about the incident (incident name, date and time of occurrence, scope of impact, and detailed description) into a form and presses the submit button. The device sends the incident information to the server in JSON format.

[1323] Input: Name of the problem, date and time of occurrence, scope of impact, detailed description

[1324] Data processing: Convert form data to JSON format

[1325] Output: Send failure information in JSON format to the server.

[1326] Step 3:

[1327] When the server receives failure information, it searches the database for past failure reports. It uses a similarity evaluation algorithm (e.g., cosine similarity) to select the most similar past report.

[1328] Input: Failure information in JSON format

[1329] Data processing: Database search using SQL queries, similarity evaluation.

[1330] Output: Most similar past incident reports

[1331] Step 4:

[1332] A new incident report is generated based on past reports where the server was selected. A template engine (e.g., Jinja2) is used to fill in each field of the report and generate the completed report.

[1333] Input: Most similar past incident reports

[1334] Data processing: Field embedding using a template engine

[1335] Output: New Incident Report

[1336] Step 5:

[1337] The server sends the generated failure report to the terminal, and the terminal displays the report to the user.

[1338] Input: New Incident Report

[1339] Data processing: None

[1340] Output: Display of the failure report screen

[1341] Step 6:

[1342] The user reviews the report content and makes any necessary corrections. After completing the corrections, they press the confirm button to send the report to the server.

[1343] Input: Revised report content

[1344] Data processing: None

[1345] Output: Send the revised report to the server.

[1346] Step 7:

[1347] The device's built-in emotion analysis engine analyzes user input and interactions to recognize the user's emotions. It then generates specific emotion labels (e.g., impatience, anger).

[1348] Input: User input and operation data

[1349] Data processing: Emotion recognition using emotion analysis algorithms

[1350] Output: Sentiment labels

[1351] Step 8:

[1352] Based on the emotional information recognized by the server, the content of the incident report will be adjusted. If the user is anxious, detailed steps and additional explanations will be added.

[1353] Input: Revised report, sentiment label

[1354] Data processing: Report adjustment using text generation algorithms

[1355] Output: Adjusted Incident Report

[1356] Step 9:

[1357] The server analyzes the content of the confirmed report, references the customer's industry and past inquiry history, and generates predicted Q&A. A natural language processing model (e.g., GPT-3) is used.

[1358] Input: Confirmed incident report, customer industry information, past inquiry history

[1359] Data processing: Text analysis, Q&A generation

[1360] Output: Predicted Q&A

[1361] Step 10:

[1362] The server prepares answers to the generated Q&A and adjusts the content of the answers based on sentiment information.

[1363] Input: Predicted Q&A, sentiment information

[1364] Data processing: Response text generation, content adjustment based on sentiment.

[1365] Output: Adjusted Q&A and their answers

[1366] Step 11:

[1367] The server sends the generated Q&A and its answer to the terminal, which then displays it to the user.

[1368] Input: Adjusted Q&A and their answers

[1369] Data processing: None

[1370] Output: Display a Q&A list and its answers on the screen.

[1371] Step 12:

[1372] Users review the Q&A and answers to prepare for their visit. After the visit, they input the actual questions and answers into their terminal and send them to the server.

[1373] Input: Actual questions and answers received

[1374] Data processing: None

[1375] Output: Send actual Q&A data to the server

[1376] Step 13:

[1377] The server compares the actual Q&A with the predicted Q&A and evaluates the accuracy of the prediction. Based on the evaluation results, the prediction algorithm is improved.

[1378] Input: Actual Q&A, Predicted Q&A

[1379] Data processing: Accuracy evaluation, algorithm improvement

[1380] Output: Improved prediction algorithm

[1381] As described above, the system's processing steps involve searching past reports based on the input failure information, referencing similar reports, and generating a new failure report. Furthermore, it adjusts the report considering the user's sentiment, generates anticipated Q&A, and prepares appropriate answers. This enables rapid and effective customer support.

[1382] (Application Example 2)

[1383] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[1384] Creating security-related incident reports and handling customer inquiries requires speed and accuracy, but manual processes are time-consuming and labor-intensive, and it's difficult to respond while considering user emotions. Furthermore, predicting the information needed to solve actual problems and preparing appropriate answers in advance is challenging.

[1385] In Application Example 2, the identification processing by the identification processing unit 290 of the data processing device 12 is realized by the following means. In this invention, the server includes means for receiving basic information about a failure, means for searching past failure reports and evaluating similarity, means for automatically generating a new failure report based on the searched past reports, means for referencing the customer's industry and past inquiry history and generating predicted Q&A, means for preparing answers to the generated Q&A, means for receiving actual Q&A and evaluating its accuracy by comparing it with the predicted Q&A, means for recognizing the user's emotions and adjusting the content of the report and Q&A, and means for operating as an application installed on a smartphone or head-mounted display. This streamlines the creation of failure reports and customer support, and enables high-quality support that takes user emotions into consideration.

[1386] "Means for receiving basic information about a failure" refers to a function in the system that allows users to input information such as the name of the failure, the date and time of occurrence, the scope of impact, and a detailed description.

[1387] "Means for searching past failure reports and evaluating similarity" refers to a function in which the system searches past failure reports in the database and uses an algorithm to evaluate and identify similarities with those reports.

[1388] The "means for automatically generating new incident reports" refer to a function that automatically creates reports addressing newly occurring incidents by using past reports, which are searched based on similarity assessments, as templates.

[1389] "A means of generating predicted Q&A by referring to the customer's industry and past inquiry history" refers to a function that prepares anticipated questions and their answers in advance based on the customer's industry information and past inquiry history.

[1390] "Means for preparing answers to generated Q&A" refers to a function that provides appropriate information for anticipated questions and prepares the answers in advance.

[1391] "A means of receiving actual Q&A and evaluating its accuracy by comparing it with predicted Q&A" refers to a function that compares the actual questions asked with the predicted results to evaluate and improve prediction accuracy.

[1392] "Means for recognizing user emotions and adjusting the content of reports and Q&A" refers to a function that analyzes the user's emotional state (e.g., impatience or anger) using an emotion recognition engine and appropriately modifies and adjusts the content of reports and Q&A according to those emotions.

[1393] "Means of operation using applications installed on smartphones and head-mounted displays" refers to a method of using this system by installing a dedicated application on devices such as smartphones and head-mounted displays, enabling the system to function on these devices.

[1394] The system for carrying out this invention mainly uses the following hardware and software. Specific examples will be described below.

[1395] Hardware and software to be used

[1396] Device: Smartphone (iOS / Android) or head-mounted display (HoloLens, etc.)

[1397] Servers: Web server, database (MySQL, etc.)

[1398] Emotion recognition engine: IBM Watson and Azure Cognitive Services

[1399] AI inference engine: Generative AI models such as GPT-4

[1400] System Processing Overview

[1401] The server performs the following actions based on the fault report entered by the user using a terminal.

[1402] 1. Enter error information

[1403] The user logs in to their smartphone or head-mounted display (hereinafter referred to as "device") and accesses a dedicated incident reporting form. They enter basic information about the incident (incident name, date and time of occurrence, scope of impact, and detailed description) and press the submit button. The device sends the entered incident information to the server.

[1404] 2. Automatic generation of incident reports

[1405] The server searches its database for past incident reports based on the received incident information. A similarity evaluation algorithm selects the most similar past report, and then automatically generates a new incident report based on it. This report includes an overview of the incident, its cause, scope of impact, and countermeasures.

[1406] 3. User emotion recognition

[1407] The device's built-in emotion recognition engine (e.g., IBM Watson) analyzes user input and interaction to recognize emotions. For example, it can detect if the user is anxious or angry. The server then adjusts the report generated based on the emotion. For instance, it might include detailed instructions or additional explanations for an anxious user.

[1408] 4. Predicting and preparing for the Q&A session

[1409] The server analyzes the generated incident report, referencing the customer's industry and past inquiry history. It generates anticipated questions and prepares appropriate answers for each. The server also adjusts the Q&A responses based on sentiment recognition results.

[1410] 5. Feedback and accuracy evaluation

[1411] The user enters the Q&A used during actual customer visits and report submissions into a terminal and sends it to the server. The server compares the predicted Q&A with the actual Q&A, evaluates the accuracy of the prediction, and improves the algorithm as needed.

[1412] Specific examples and prompt statements

[1413] Example: Scenario for when a security camera malfunction is reported.

[1414] 1. User input fields:

[1415] Problem name: Security camera footage is distorted.

[1416] Date and time of occurrence: October 12, 2023, 14:00

[1417] Scope of impact: All store monitoring systems

[1418] Detailed description: The video frequently cuts out and is noisy.

[1419] 2. Emotion recognition:

[1420] Recognize that the user is anxious and include more detailed explanations in the report.

[1421] 3. Generated report:

[1422] The server automatically generates a report based on the "distorted security camera footage" issue, including the cause and solution.

[1423] 4. Prediction Q&A:

[1424] Q: When will the video become stable?

[1425] A: Expected to be restored within 2 hours.

[1426] Examples of prompts for a generative AI model:

[1427] Please generate the most appropriate report based on past troubleshooting data regarding the issue of distorted security camera footage. Also, please provide anticipated Q&A and their answers.

[1428] Problem name: Security camera footage is distorted.

[1429] Date and time of occurrence: October 12, 2023, 14:00

[1430] Scope of impact: All store monitoring systems

[1431] Detailed description: The video frequently cuts out and is noisy.

[1432] User's emotion: impatience

[1433] This will enable more efficient creation of incident reports and customer support, as well as responses that take user emotions into consideration.

[1434] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[1435] Step 1:

[1436] The user logs in using their smartphone or head-mounted display and accesses a dedicated incident reporting form. They enter basic information about the incident, such as the incident name, date and time of occurrence, scope of impact, and detailed description, and then press the submit button. The entered information is sent from the device to the server.

[1437] Input: Name of the problem, date and time of occurrence, scope of impact, detailed description

[1438] Output: Failure information transferred to the server

[1439] Step 2:

[1440] Based on the failure information received by the server, past failure reports are searched in the database. A similarity evaluation algorithm is used to select the past report with the highest similarity. Database queries and similarity calculations are performed during this search and selection process.

[1441] Input: Error information received by the server

[1442] Output: Past incident reports with high similarity

[1443] Step 3:

[1444] The server automatically generates a new incident report using a previously selected report as a template. The new report includes an overview of the incident, its cause, scope of impact, and countermeasures. Template and text generation algorithms are used in this process.

[1445] Input: Similar past incident reports

[1446] Output: Newly generated incident report

[1447] Step 4:

[1448] The device's built-in emotion recognition engine analyzes user input and interaction to recognize the user's emotions. The recognized emotion information is sent to a server. Text and speech analysis algorithms are used for emotion recognition.

[1449] Input: User input and interaction data

[1450] Output: Recognized user sentiment information

[1451] Step 5:

[1452] The server adjusts the content of the generated incident report based on the perceived emotions. For example, if the user is anxious, the report will include more detailed steps and additional explanations. A text generation algorithm is used in this process.

[1453] Input: Recognized emotion information and automatically generated new incident report

[1454] Output: Adjusted Incident Report

[1455] Step 6:

[1456] The server analyzes the generated incident report and references the customer's industry and past inquiry history. Based on this, it generates predicted Q&A and prepares appropriate answers for each. An AI inference engine is used for prediction and answer generation.

[1457] Input: Adjusted incident report, customer industry information, past inquiry history

[1458] Output: Predicted Q&A and their answers

[1459] Step 7:

[1460] Users visit customers in person and respond to Q&A based on submitted incident reports. After the visit, they input the questions and answers into a terminal and send them to the server.

[1461] Input: Actual questions and answers

[1462] Output: Actual Q&A as feedback to the server

[1463] Step 8:

[1464] The server compares predicted Q&A with actual Q&A to evaluate the accuracy of the prediction. Based on the evaluation results, the prediction algorithm is improved to enhance the accuracy of future answers. An accuracy evaluation algorithm and a feedback loop are used in this process.

[1465] Input: Actual Q&A, Predicted Q&A

[1466] Output: Prediction accuracy evaluation results and improved algorithm

[1467] The specific processing unit 290 transmits the result of the specific processing to the headset terminal 314. In the headset terminal 314, the control unit 46A causes the speaker 240 and display 343 to output the result of the specific processing. The microphone 238 acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[1468] Data generation model 58 is a type of so-called generative AI (Artificial Intelligence). One example of data generation model 58 is ChatGPT (Internet search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1469] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and specific processing may also be performed by the headset terminal 314.

[1470] [Fourth Embodiment]

[1471] Figure 7 shows an example of the configuration of the data processing system 410 according to the fourth embodiment.

[1472] As shown in Figure 7, the data processing system 410 includes a data processing device 12 and a robot 414. An example of the data processing device 12 is a server.

[1473] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[1474] The robot 414 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication interface 44, and a controlled object 443. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, camera 42, and controlled object 443 are also connected to the bus 52.

[1475] The microphone 238 receives voice signals from the user 20 and receives instructions from the user 20. The microphone 238 captures the voice signals from the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[1476] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the area around the user 20 (for example, an imaging range defined by a field of view equivalent to the width of a typical healthy person's field of vision).

[1477] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[1478] The controlled object 443 includes a display device, LEDs in the eyes, and motors that drive the arms, hands, and feet. The posture and gestures of the robot 414 are controlled by controlling the motors of the arms, hands, and feet. Some of the robot 414's emotions can be expressed by controlling these motors. Furthermore, the robot 414's facial expressions can also be expressed by controlling the illumination state of the LEDs in its eyes.

[1479] Figure 8 shows an example of the main functions of the data processing device 12 and the robot 414. As shown in Figure 8, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[1480] The specific processing program 56 is an example of a "program" relating to the technology of this disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[1481] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[1482] In robot 414, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[1483] Next, the specific processing performed by the specific processing unit 290 of the data processing device 12 will be described. In the following description, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[1484] The present invention aims to improve the quality of customer support after reporting by enabling companies to quickly and effectively create and submit incident reports to customers, predict appropriate Q&A based on the customer's industry and past inquiry history, and prepare answers accordingly.

[1485] Program processing

[1486] Entering and sending error information

[1487] 1. The user logs into the terminal and enters basic information about the problem. Specifically, they enter the name of the problem, the date and time of occurrence, the scope of impact, and a detailed description.

[1488] 2. The terminal sends the entered error information to the server.

[1489] Automatic generation of incident reports

[1490] 3. Based on the received failure information, the server searches past failure reports and evaluates their similarity.

[1491] 4. The server automatically generates a new incident report based on the most similar past incident reports. The new report includes an overview of the incident, its cause, its scope of impact, and countermeasures.

[1492] 5. The server sends the generated report to the terminal.

[1493] 6. The terminal displays the generated report to the user, who can review and modify its contents.

[1494] 7. After the user has finished making corrections, they press the confirm button to send the final report to the server.

[1495] Q&A prediction and preparation

[1496] 8. The server analyzes the contents of the confirmed report and refers to the customer's industry and past inquiry history.

[1497] 9. The server generates anticipated Q&A and prepares appropriate answers for each.

[1498] 10. The server sends the generated Q&A to the terminal.

[1499] 11. The terminal displays the generated Q&A list and its answers to the user.

[1500] 12. Users review the Q&A and answers to prepare for their visit.

[1501] Reports and Feedback

[1502] 13. The user actually visits the customer and submits a report. They respond to customer questions based on the anticipated Q&A during the visit.

[1503] 14. After the visit, the user enters the questions and answers they received into the terminal.

[1504] 15. The terminal sends the actual Q&A entered to the server.

[1505] Accuracy evaluation and algorithm improvement

[1506] 16. The server compares the actual received Q&A with the predicted Q&A and evaluates the accuracy of the prediction.

[1507] 17. The server improves its prediction algorithm based on the evaluation results to improve the accuracy of the next prediction.

[1508] 18. The server generates a new prompt based on an improved algorithm and provides feedback to the user.

[1509] Specific example

[1510] 1. Input and transmission of failure information

[1511] The user enters a message into their terminal indicating that the server is down, and then submits a detailed description of the date and time of the outage and the extent of the impact.

[1512] 2. Automatic generation of incident reports

[1513] The server searches past reports using the keyword "server down" and selects the report that is most similar.

[1514] Based on the selected reports, the server automatically generates a "Server Downtime Incident Report," which includes detailed explanations, causes, and countermeasures.

[1515] 3. Predicting and preparing for the Q&A session

[1516] The server anticipates questions related to "server downtime," such as "server recovery time" and "whether or not data was lost," and prepares appropriate answers for each.

[1517] 4. Reporting and Feedback

[1518] The system responds to questions based on anticipated Q&A from the user during their customer visit. For example, in response to the question, "When will the server be restored?", it might answer, "The estimated restoration time is 2 hours from now."

[1519] After the visit, the terminal inputs the questions the user actually received (for example, "Will there be any data loss?") and the answer ("There will be no data loss"), and sends them to the server.

[1520] 5. Accuracy evaluation and algorithm improvement

[1521] The server compares the predicted Q&A with the actual Q&A to evaluate prediction accuracy and use the results to improve the algorithm.

[1522] The system of this invention streamlines the creation of fault reports and customer support, thereby improving customer satisfaction.

[1523] The following describes the processing flow.

[1524] Step 1:

[1525] The user logs into the device and accesses the incident report form. The device displays the incident report form.

[1526] Step 2:

[1527] The user enters basic information about the incident (incident name, date and time of occurrence, scope of impact, and detailed description) and presses the submit button. The terminal sends the entered incident information to the server.

[1528] Step 3:

[1529] Based on the failure information received by the server, it searches the database for past failure reports. The server uses a similarity evaluation algorithm to select the most similar past report.

[1530] Step 4:

[1531] Based on past reports where servers were selected, a new incident report is automatically generated. The generated report includes an overview of the incident, its cause, the scope of its impact, and the countermeasures taken.

[1532] Step 5:

[1533] The server sends the generated failure report to the terminal. The terminal displays the generated report to the user.

[1534] Step 6:

[1535] The user reviews the report content and makes corrections as needed. Once the corrections are complete, they press the confirm button to send the report to the server.

[1536] Step 7:

[1537] The server analyzes the contents of the confirmed report. The server refers to the customer's industry and past inquiry history to generate predicted Q&A.

[1538] Step 8:

[1539] The server prepares appropriate answers to the predicted Q&A. Once the Q&A is ready, it sends it to the terminal.

[1540] Step 9:

[1541] The terminal displays a generated list of Q&A and its answers to the user. The user reviews the Q&A and answers to prepare for their visit.

[1542] Step 10:

[1543] Users visit customers in person and submit reports. They respond to customer questions based on anticipated Q&A from the visit.

[1544] Step 11:

[1545] After a user visits, the terminal inputs the actual questions and answers they received. The terminal then sends the entered Q&A to the server.

[1546] Step 12:

[1547] The server compares the actual Q&A received with the predicted Q&A to evaluate the accuracy of the prediction.

[1548] Step 13:

[1549] The server improves the prediction algorithm based on the evaluation results. It generates a new prompt based on the improved algorithm and provides feedback to the user.

[1550] By following the steps outlined above, this system can streamline the creation of reports and customer support in the event of a failure, thereby improving accuracy.

[1551] (Example 1)

[1552] Next, we will describe Example 1. In the following description, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[1553] Traditional incident reporting and customer support processes are often manual, time-consuming and labor-intensive, and can lead to insufficient preparation for anticipated Q&A. This can result in decreased customer satisfaction and damage to the company's reputation. Furthermore, it's difficult to extract relevant information from past reports and inquiry histories to provide efficient and accurate responses.

[1554] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[1555] In this invention, the server includes means for receiving basic information about a failure, means for searching past failure reports and evaluating similarity, means for automatically generating a new failure report based on the retrieved past reports, means for displaying the generated failure report and accepting corrections, means for referencing the customer's industry and past inquiry history to generate predicted Q&A, means for preparing answers to the generated Q&A, means for receiving actual Q&A and evaluating its accuracy by comparing it with the predicted Q&A, means for improving the prediction algorithm based on the evaluation results, and means for generating new prompt statements based on the improved algorithm and providing feedback. This streamlines the process of creating failure reports and responding to customers, enabling highly accurate responses.

[1556] "Means for receiving basic information about a malfunction" refers to a device or program that has an interface for a user to input details of a malfunction and a function to transmit that information to a server.

[1557] "Means for searching past incident reports and evaluating similarity" refers to a device or program for searching past incident reports in a database using string search or natural language processing techniques and calculating their similarity to current incident information.

[1558] "Means for automatically generating new incident reports based on retrieved past reports" refers to a device or program that automatically creates a new incident report by using the most similar report from the search results as a template and filling in the current incident information.

[1559] "Means for displaying generated failure reports and accepting corrections" refers to a device or program that displays generated failure reports to the user, has an interface that allows the user to confirm and correct the contents, and has the function of saving and transmitting the corrected contents.

[1560] "Means for generating predicted Q&A by referring to the customer's industry and past inquiry history" refers to a device or program that searches a database for the customer's industry and past inquiry history, and generates predicted questions and answers based on this information.

[1561] "Means for preparing answers to generated Q&A" refers to a device or program that uses natural language generation technology or past answer data to automatically generate appropriate answers to predicted questions.

[1562] "A means of receiving actual Q&A and evaluating its accuracy by comparing it with predicted Q&A" refers to a device or program that takes actual questions and answers received by a user at a visited location as input and evaluates the accuracy of the prediction by comparing it with the predicted Q&A.

[1563] "Means for improving the prediction algorithm based on evaluation results" refers to a device or program that analyzes the results of the accuracy evaluation and adjusts or improves the parameters and model of the prediction algorithm.

[1564] "Means for generating new prompt statements based on an improved algorithm and providing feedback" refers to a device or program that generates new prompt statements using an improved algorithm and has a notification function to inform the user of their contents.

[1565] The present invention aims to enable companies to quickly and effectively create and submit incident reports to their customers, predict appropriate Q&A based on the customer's industry and past inquiry history, and prepare answers accordingly. This system primarily consists of server, terminal, and user-centric processing.

[1566] First, the user logs into the company's internal system and enters basic information about the incident into the terminal. Specifically, this includes the incident name, date and time of occurrence, scope of impact, and a detailed description. The terminal verifies this information and sends it to the server. This process utilizes form input and HTTP requests.

[1567] Next, the server analyzes the received failure information and searches for past failure reports. Natural language processing technology is used for the search, and a similarity score is calculated. Then, a new failure report is automatically generated based on the most similar report. The generated report includes an overview of the failure, its cause, scope of impact, and countermeasures. This report is returned to the terminal, where the user can review and correct its contents.

[1568] Once the corrections are complete, the user confirms the report and sends it back to the server. Based on the confirmed report, the server references the customer's industry and past inquiry history to generate predicted Q&A. This uses database searches and a generation AI model. Appropriate answers are automatically added to the generated Q&A and sent back to the terminal.

[1569] Users check a Q&A list on their terminal to prepare for customer visits. After the visit, they input the actual questions and answers into their terminal and send them to the server. The server compares these actual Q&A with the predicted Q&A and evaluates the accuracy. The evaluation results are used to improve the algorithm, aiming for better prediction accuracy next time. Furthermore, new prompts are generated based on the improved algorithm and provided to the user as feedback.

[1570] Hardware and software details

[1571] Terminals: Desktop PCs, laptops, tablets, etc., within the company are used, and information is entered and displayed via web browsers or dedicated applications.

[1572] Servers: Cloud servers will be used to host large-scale databases and AI models, performing tasks such as data analysis, automated report generation, and Q&A generation. Specifically, Amazon Web Services (AWS) and Microsoft Azure are envisioned.

[1573] Software: For natural language processing, Python libraries such as NLTK and SpaCy are used, and OpenAI's GPT model is used for generative AI models.

[1574] Specific example

[1575] Example prompt: "Please prepare a failure report required when the server goes down. Include the date and time of the incident, the scope of the impact, and a detailed description."

[1576] Example of customer interaction: When a user visits a customer's site, they might ask, "When will the server be restored?" The response would be, "The estimated restoration time is in 2 hours." The actual question is entered into the server, and for the question, "Will there be any data loss?", the response "There will be no data loss" is recorded.

[1577] Through the processes described above, this system streamlines the creation of incident reports and customer support, enabling highly accurate responses.

[1578] The flow of the specific processing in Example 1 will be explained using Figure 11.

[1579] Step 1:

[1580] The user logs into the company's internal system and enters basic information about the incident into the terminal. Specifically, they enter the incident name, date and time of occurrence, scope of impact, and a detailed description (input). The terminal validates the input and sends it to the server (output). During this process, form input occurs and an HTTP request is generated (specific action).

[1581] Step 2:

[1582] The server analyzes the received failure information and searches past failure reports in the database to evaluate similarity (input). It calculates a similarity score using natural language processing techniques (data calculation). It generates a new failure report using the most similar report as a template (output). This involves keyword search and similarity calculation (such as Cosine Similarity) (specific operation).

[1583] Step 3:

[1584] The server sends the generated failure report to the terminal (output). The terminal displays the report to the user and provides an editing form that allows the user to review and modify the contents (specific action). The user reviews the report and makes corrections as needed (input).

[1585] Step 4:

[1586] Once the user has completed the corrections, they press the confirm button to send the final error report to the server (output). The terminal then sends the corrected report to the server as data in JSON format (specific action).

[1587] Step 5:

[1588] The server references the customer's industry and past inquiry history based on the confirmed report (input). It uses database searches and generative AI models to generate predicted Q&A (data computation). The generated Q&A is automatically assigned appropriate answers (output). This uses natural language generation technology to produce answers customized for specific industries (specific behavior).

[1589] Step 6:

[1590] The server sends the generated Q&A list to the terminal (output). The terminal displays the Q&A list and its answers to the user (specific action). The user reviews the Q&A and prepares for their visit (input).

[1591] Step 7:

[1592] During customer visits, users respond to questions based on predicted Q&A (input). After the visit, users input the actual questions received and their answers into the terminal (input).

[1593] Step 8:

[1594] The terminal sends the actual Q&A input to the server (output). The server compares the actual Q&A with the predicted Q&A and evaluates the accuracy of the prediction (data calculation). Based on the evaluation results, the prediction algorithm is improved (output). This uses statistical evaluation metrics (e.g., accuracy, recall) (specific operation).

[1595] Step 9:

[1596] The server generates a new prompt message based on the improved algorithm (output) and provides feedback on its contents to the user (specific action). The user receives the improved prompt message and uses it to help with troubleshooting in the future (input).

[1597] (Application Example 1)

[1598] Next, we will explain Application Example 1. In the following explanation, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[1599] For companies to quickly and effectively create and submit incident reports to customers, and to provide prompt and high-quality customer support afterward, accurate reporting and appropriate Q&A preparation are required when incidents occur. However, incident response is time-consuming and labor-intensive, and preparing appropriate Q&A based on past inquiry history and the customer's industry is a particularly difficult challenge. Furthermore, human error can occur in report generation and Q&A preparation, so it is necessary to improve the efficiency and accuracy of the entire system.

[1600] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[1601] In this invention, the server includes means for receiving basic information about a failure; means for searching past failure reports and evaluating similarity; means for automatically generating a new failure report based on the searched past reports; means for referencing the customer's industry and past inquiry history to generate predicted Q&A; means for preparing answers to the generated Q&A; means for receiving actual Q&A and evaluating its accuracy by comparing it with the predicted Q&A; means for sending the generated failure report and predicted Q&A to a terminal so that the user can review and correct them; means for regenerating the final report and Q&A based on the corrected content; means for inputting the questions and answers actually received into the terminal and sending them to the server; and means for comparing the actual Q&A received by the server with the predicted Q&A and improving the prediction algorithm. This enables increased efficiency and improved quality in the generation of failure reports and customer support.

[1602] "Basic information about the disruption" refers to fundamental information related to the disruption, such as the name of the disruption, the date and time of occurrence, the scope of its impact, and a detailed explanation.

[1603] "Past incident reports" refer to incident reports created in the past, and are used as reference information by analyzing similar incidents.

[1604] "Means for evaluating similarity" refers to a method or system for identifying the most similar case from past incident reports based on the incident information received.

[1605] "Means for automatically generating new incident reports" refers to a method or system for automatically creating new incident reports based on current incident information, using past reports as templates.

[1606] "Customer's industry" refers to the specific industry or sector to which the customer belongs, and is information that reflects the unique requirements and problems associated with that particular industry.

[1607] "Past inquiry history" refers to the content of inquiries received from customers in the past and the history of how those inquiries were handled. This serves as reference material for providing appropriate customer service.

[1608] "Means for generating predicted Q&A" refers to a method or system for analyzing fault information and past inquiry history to generate questions that are predicted to be asked by customers in the future, along with their answers.

[1609] "Means for preparing answers to generated Q&A" refers to a method or system for preparing appropriate answers to anticipated questions.

[1610] "Means for receiving actual Q&A" refers to a method or system for collecting questions and answers actually submitted by customers.

[1611] "Means for evaluating accuracy by comparing with predicted Q&A" refers to a method or system for comparing predicted questions with actual questions to evaluate the accuracy and usefulness of the predictions.

[1612] A "terminal" refers to a device used by system administrators or users, and includes desktop computers, notebooks, smartphones, and other similar devices.

[1613] "Means for users to review and correct" refers to interfaces and tools that allow users to review generated reports and Q&A and make corrections as needed.

[1614] "Means for generating the final report and Q&A based on the revisions" refers to a method or system for generating the final version of the report and Q&A, reflecting the user's revisions.

[1615] "Means of improving prediction algorithms" refers to methods or systems for adjusting prediction models based on actual data to improve their accuracy.

[1616] This invention is a system that enables companies to quickly and effectively create and submit fault reports to customers, and to provide high-quality customer support thereafter. This system consists of a smartphone application (hereinafter referred to as "the app") and a server system.

[1617] First, the user logs into the app and enters basic information about the outage. Specifically, this includes the outage name, date and time of occurrence, scope of impact, and a detailed description. Next, the app sends the entered outage information to the server. The server then compares the received outage information with past outage reports and evaluates the similarity. At this time, the server uses a database (e.g., MongoDB) to search for past reports.

[1618] The server automatically generates a new incident report based on the most similar past incident reports. This generated report includes an overview of the incident, its cause, scope of impact, and countermeasures. The server then sends the generated report to the app, where the user can review it and make corrections as needed. Once corrections are complete, the user presses the confirm button to send the final report to the server.

[1619] Furthermore, the server analyzes the contents of the confirmed incident report and refers to the customer's industry and past inquiry history. This generates predicted Q&A. To generate the Q&A, the server uses an artificial intelligence (AI) model (e.g., TensorFlow). The generated Q&A list and its answers are sent to the app, where the user can review them and prepare for customer visits.

[1620] After the visit, the user enters the actual questions and answers into the app and sends them to the server. The server compares the received actual Q&A with the predicted Q&A and evaluates the accuracy of the prediction. Based on the evaluation results, the server improves the prediction algorithm to improve the accuracy of the next prediction.

[1621] As a concrete example, if a live stream is interrupted due to a malfunction, the user enters the malfunction information, such as "live stream interruption," and sends it to the server, detailing the date and time of the incident and the scope of the impact. The server searches past reports using the keyword "live stream interruption" and generates a new malfunction report based on the most similar report. The server sends the generated report to the app for the user to review and correct. After that, the server generates predictive Q&A related to "live stream interruption" (for example, "When is it expected to be restored?", "Can I rewatch past streamed videos?") and provides appropriate answers.

[1622] Examples of prompt statements are as follows:

[1623] "Input example:

[1624] Incident Name: Live Stream Interruption, Date and Time: 2023-10-10 14:00, Affected Users: All Users, Details: The live stream was interrupted midway.

[1625] Example output:

[1626] Outage Summary: Live stream stopped for all users. Cause: Server load. Countermeasure: Server upgrade. Predicted Q&A: When is it expected to be restored? Answer: It is expected to be restored in 1 hour. Can I watch past streamed videos again? Answer: Yes, you can watch them again.

[1627] This system allows companies to efficiently create reports and handle customer inquiries in the event of a system failure, which is expected to improve customer satisfaction.

[1628] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[1629] Step 1:

[1630] The user logs into the smartphone app and enters basic information about the outage (outage name, date and time of occurrence, scope of impact, and detailed description).

[1631] Input: Name of the problem, date and time of occurrence, scope of impact, detailed description

[1632] Output: Failure information data

[1633] Specific action: The user fills in the required information in the app's input form and presses the submit button.

[1634] Step 2:

[1635] The terminal sends the entered error information to the server.

[1636] Input: Failure information data

[1637] Output: Failure information sent to the server

[1638] Specific operation: The app sends an HTTP request to the server via the API.

[1639] Step 3:

[1640] Based on the failure information received by the server, past failure reports are searched in the database (MongoDB) and their similarity is evaluated.

[1641] Input: Received fault information

[1642] Output: Similar past incident reports

[1643] Specific operation: The server extracts keywords related to the failure information and queries the database to find the most similar report.

[1644] Step 4:

[1645] The server automatically generates a new incident report based on the most similar past incident report.

[1646] Input: Similar past incident reports, received incident information

[1647] Output: New Incident Report

[1648] Specific operation: The server uses past reports as templates and combines basic information and detailed descriptions to generate new reports.

[1649] Step 5:

[1650] The server sends the generated report to the terminal, allowing the user to review and correct it.

[1651] Input: New Incident Report

[1652] Output: Report displayed on the user's device

[1653] Specific operation: The server sends the generated report to the app in JSON format, and the app displays it.

[1654] Step 6:

[1655] The user reviews the report and makes corrections as needed. After completing the corrections, they press the confirm button to send the final report to the server.

[1656] Input: User modifications, final report

[1657] Output: Final report sent to the server

[1658] Specific operation: When the user modifies the report content and presses the confirm button, the app sends the final report, including the modifications, to the server.

[1659] Step 7:

[1660] The system analyzes the contents of confirmed server failure reports, referencing the customer's industry and past inquiry history. It then generates anticipated Q&A and prepares appropriate answers.

[1661] Input: Finalized report, customer industry, past inquiry history

[1662] Output: Predicted Q&A list and its answers

[1663] Specific operation: The server uses a prediction algorithm (TensorFlow) to analyze reports and historical data and generate predicted Q&A.

[1664] Step 8:

[1665] The server sends the generated Q&A list and its answers to the user's device for review.

[1666] Input: Predicted Q&A list and its answers

[1667] Output: Q&A displayed on the user's device

[1668] Specific operation: The server sends the generated Q&A to the app, and the app displays it.

[1669] Step 9:

[1670] Users actually visit customers and answer questions based on predicted Q&A.

[1671] Input: Customer questions, predicted Q&A

[1672] Output: Actual Q&A from customer support

[1673] Specific operation: The user refers to a Q&A list and responds to customer questions.

[1674] Step 10:

[1675] After a user visits, the device inputs the actual questions and answers they submitted and sends them to the server.

[1676] Input: Actual Q&A

[1677] Output: Actual Q&A sent to the server

[1678] Specific operation: The user enters the question and answer into the terminal and sends them to the server.

[1679] Step 11:

[1680] The server compares the actual Q&A received with the predicted Q&A to evaluate the accuracy of the prediction.

[1681] Input: Actual Q&A, Predicted Q&A

[1682] Output: Evaluation results and feedback for algorithm improvement

[1683] Specific operation: The server performs comparison calculations and improves the prediction algorithm based on the evaluation results.

[1684] Step 12:

[1685] The server generates new prompts based on an improved algorithm and provides feedback to the user.

[1686] Input: Evaluation results, improved algorithm

[1687] Output: New prompts and feedback content

[1688] Specific action: The server generates a new prompt and notifies the user.

[1689] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[1690] The system of this invention enables companies to quickly and effectively create and submit incident reports to customers, predict appropriate Q&A based on the customer's industry and past inquiry history, prepare answers, and improve the quality of customer service by recognizing user emotions and adjusting responses accordingly. The specific program processing is described below.

[1691] Program processing

[1692] Entering and sending error information

[1693] 1. The user logs into the device and accesses the incident report form. The device displays the incident report form.

[1694] 2. The user enters basic information about the incident (incident name, date and time of occurrence, scope of impact, and detailed description) and presses the submit button. The terminal sends the entered incident information to the server.

[1695] Automatic generation of incident reports

[1696] 3. Based on the failure information received by the server, it searches the database for past failure reports. The server uses a similarity evaluation algorithm to select the most similar past report.

[1697] 4. Based on past reports where the server was selected, a new incident report is automatically generated. The generated report includes an overview of the incident, its cause, the scope of its impact, and the countermeasures taken.

[1698] 5. The server sends the generated report to the terminal. The terminal displays the generated report to the user.

[1699] 6. The user reviews the report content and makes corrections as needed. After completing the corrections, they press the confirm button to send the report to the server.

[1700] Adjusting user emotion recognition and responses

[1701] 7. The device's built-in emotion engine analyzes user input and interactions to recognize the user's emotions. For example, it can detect when the user is anxious or angry.

[1702] 8. The server adjusts the content of the incident report based on the perceived emotions. For example, if the user is anxious, the report is revised to include more detailed steps or additional explanations.

[1703] Q&A prediction and preparation

[1704] 9. The server analyzes the contents of the confirmed report and refers to the customer's industry and past inquiry history.

[1705] 10. The server generates anticipated Q&A and prepares appropriate answers for each. The Q&A and answers are also adjusted based on the perceived user sentiment.

[1706] 11. The server sends the generated Q&A to the terminal. The terminal displays the generated Q&A list and its answers to the user.

[1707] 12. Users review the Q&A and answers to prepare for their visit.

[1708] Reports and Feedback

[1709] 13. The user actually visits the customer and submits a report. They respond to customer questions based on the anticipated Q&A during the visit.

[1710] 14. After the visit, the user enters the actual question and answer into the terminal. The terminal then sends the actual Q&A entered to the server.

[1711] Accuracy evaluation and algorithm improvement

[1712] 15. Compare the actual Q&A received by the server with the predicted Q&A and evaluate the accuracy of the prediction.

[1713] 16. The server improves the prediction algorithm based on the evaluation results. It generates a new prompt based on the improved algorithm and provides feedback to the user.

[1714] 17. The server analyzes user sentiment during actual Q&A sessions and uses the feedback to further improve the accuracy of the generated Q&A and its answers.

[1715] Specific example

[1716] 1. Input and transmission of failure information

[1717] The user enters a message into their terminal indicating that the server is down, and then submits a detailed description of the date and time of the outage and the extent of the impact. The emotion engine detects the user's anxiety.

[1718] 2. Automatic generation of incident reports

[1719] The server searches past reports using the keyword "server down" and selects the most similar report.

[1720] Based on the server selection report, an "Incident Report Regarding Server Downtime" is automatically generated, including detailed explanations, causes, and countermeasures.

[1721] 3. Recognizing and responding to user emotions

[1722] The emotion engine recognizes the user's impatience and adjusts the report to include more detailed steps and additional explanations.

[1723] 4. Predicting and preparing for the Q&A session

[1724] The server anticipates questions related to "server downtime," such as "server recovery time" and "whether data has been lost," and prepares appropriate answers for each.

[1725] Based on the emotions recognized by the emotion engine, the response content is adjusted to be more detailed and easier to understand.

[1726] 5. Reporting and Feedback

[1727] The system responds to questions based on anticipated Q&A from the user during their customer visit. For example, in response to the question, "When will the server be restored?", it might answer, "The estimated restoration time is 2 hours from now."

[1728] After the visit, the terminal inputs the questions the user actually received (for example, "Will there be any data loss?") and the answer ("There will be no data loss"), and sends them to the server.

[1729] 6. Accuracy evaluation and algorithm improvement

[1730] The server compares the predicted Q&A with the actual Q&A to evaluate prediction accuracy and use the results to improve the algorithm.

[1731] The server uses feedback to further improve the accuracy of the Q&A and answers generated using the emotion engine.

[1732] The system of the present invention streamlines the creation of fault reports and customer support, enables responses that take user emotions into consideration, and improves customer satisfaction.

[1733] The following describes the processing flow.

[1734] Step 1:

[1735] The user logs into the device and accesses the incident report form. The device displays the incident report form.

[1736] Step 2:

[1737] The user enters basic information about the incident (incident name, date and time of occurrence, scope of impact, and detailed description) and presses the submit button. The terminal sends the entered incident information to the server.

[1738] Step 3:

[1739] Based on the failure information received by the server, it searches the database for past failure reports. The server uses a similarity evaluation algorithm to select the most similar past report.

[1740] Step 4:

[1741] Based on past reports where servers were selected, a new incident report is automatically generated. The generated report includes an overview of the incident, its cause, the scope of its impact, and the countermeasures taken.

[1742] Step 5:

[1743] The server sends the generated report to the terminal. The terminal displays the generated report to the user.

[1744] Step 6:

[1745] The user reviews the report content and makes corrections as needed. Once the corrections are complete, they press the confirm button to send the report to the server.

[1746] Step 7:

[1747] The device's built-in emotion engine analyzes user input and interactions to recognize the user's emotions. The device can detect situations such as when the user is anxious or angry.

[1748] Step 8:

[1749] The server adjusts the content of the incident report based on the perceived emotions. For example, if the user is anxious, the report will be revised to include more detailed steps and additional explanations.

[1750] Step 9:

[1751] The server analyzes the contents of the confirmed report and refers to the customer's industry and past inquiry history.

[1752] Step 10:

[1753] The server generates anticipated Q&A and prepares appropriate answers for each. It also adjusts the Q&A and answers based on the perceived user sentiment.

[1754] Step 11:

[1755] The server sends the generated Q&A and its answers to the terminal. The terminal displays the generated Q&A list and its answers to the user.

[1756] Step 12:

[1757] Users can review the Q&A and answers to prepare for their visit.

[1758] Step 13:

[1759] Users visit customers in person and submit reports. They respond to customer questions based on anticipated Q&A from the visit.

[1760] Step 14:

[1761] After a user visits, the terminal inputs the actual questions and answers they received. The terminal then sends the entered Q&A to the server.

[1762] Step 15:

[1763] The server compares the actual Q&A received with the predicted Q&A to evaluate the accuracy of the prediction.

[1764] Step 16:

[1765] The server improves the prediction algorithm based on the evaluation results. It generates a new prompt based on the improved algorithm and provides feedback to the user.

[1766] Step 17:

[1767] The server analyzes user sentiment during actual Q&A sessions and uses the feedback to further improve the accuracy of the generated Q&A and its answers.

[1768] Through the steps outlined above, this system streamlines the creation of reports and customer support in the event of a failure, enabling responses that take user emotions into consideration, and ultimately improving customer satisfaction.

[1769] (Example 2)

[1770] Next, we will describe Example 2. In the following description, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[1771] It is crucial for companies to quickly and effectively create and submit incident reports to their customers, but this usually requires considerable time and effort. Furthermore, preparing appropriate Q&A based on the customer's industry and past inquiry history is not easy. Additionally, considering user emotions in customer service is difficult, leading to inconsistent service quality. A system is needed to address these challenges and improve the quality of customer service.

[1772] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[1773] In this invention, the server includes means for receiving basic information about a failure, means for searching past failure reports and evaluating similarity, means for automatically generating a new failure report based on the retrieved past reports, means for recognizing emotions from user input and operations, means for adjusting the content of the failure report based on the recognized emotion information, means for generating predicted Q&A by referring to the customer's industry and past inquiry history, means for preparing answers to the generated Q&A, means for receiving actual Q&A and evaluating its accuracy by comparing it with the predicted Q&A, and means for improving the prediction algorithm. This enables the rapid and effective creation of failure reports, appropriate responses to customer inquiries, and flexible responses that take user emotions into consideration.

[1774] "Means for receiving basic information about a failure" refers to a function that allows users to input basic information such as the name of the failure, the date and time of occurrence, the scope of impact, and a detailed description, and for the system to retrieve this information.

[1775] "Means for searching past failure reports and evaluating similarity" refers to a function that searches a database of previously recorded failure reports and uses a similarity evaluation algorithm (e.g., cosine similarity) to find reports similar to current failure information.

[1776] "A means of automatically generating new incident reports based on searched past reports" refers to a function that uses searched and selected past incident reports as templates and automatically creates new incident reports based on them.

[1777] "Means of recognizing emotions from user input and actions" refers to a function that analyzes the patterns of text and actions entered by the user to identify the emotions the user is currently feeling.

[1778] "Means for adjusting the content of incident reports based on recognized emotional information" refers to a function that adjusts the content of incident reports to make them more detailed or supplementary in explanations according to the recognized emotions of the user, thereby reassuring the user.

[1779] "A means of generating predicted Q&A by referring to the customer's industry and past inquiry history" refers to a function that generates questions that are expected to be asked by the customer in the future, based on the customer's industry information and past inquiry history.

[1780] "Means for preparing answers to generated Q&A" refers to a function that pre-prepares appropriate answers for each generated question.

[1781] "A means of receiving actual Q&A and evaluating its accuracy by comparing it with predicted Q&A" refers to a function that receives questions and answers actually submitted by users and evaluates how well they match the predicted Q&A.

[1782] "Means for improving the prediction algorithm" refers to a function that improves the question and answer prediction algorithm based on the results obtained from the evaluation of prediction accuracy, thereby improving the accuracy of predictions in subsequent instances.

[1783] This invention provides a system that enables companies to quickly and effectively create and submit incident reports to customers, predict appropriate Q&A based on the customer's industry and past inquiry history, and prepare responses accordingly. Furthermore, it improves the quality of customer service by recognizing user emotions and adjusting responses accordingly.

[1784] Entering and sending error information

[1785] The user logs into the device and accesses the incident report form. The device displays the incident report form using a web page created with HTML and CSS. The user enters the incident name, date and time of occurrence, scope of impact, and detailed description, and clicks the submit button. The device converts the entered incident information into JSON format and sends it to the server using a REST API.

[1786] Automatic generation of incident reports

[1787] When the server receives failure information, it queries past failure reports in the database using SQL statements. It then runs a similarity evaluation algorithm (e.g., cosine similarity) using a Python library (e.g., scikit-learn) to select the most similar past report. Based on this selected report, the server automatically generates a new failure report using a template engine (e.g., Jinja2).

[1788] Adjusting user emotion recognition and responses

[1789] The device has an emotion analysis engine (e.g., Google Cloud Natural Language API) installed, which analyzes the user's emotions based on their input and the timing of their actions. If the user is feeling anxious or angry, emotion information is sent to the server.

[1790] The server adjusts the content of the generated incident report based on the emotional information it receives. For example, if the user is anxious, the report will be revised to include more detailed steps and specific examples.

[1791] Q&A prediction and generation

[1792] The server analyzes the content of the finalized report, referencing the customer's industry and past inquiry history. Based on this information, a natural language processing model (e.g., GPT-3) is used to generate anticipated Q&A for the customer and prepare appropriate answers for each. Information from the sentiment engine is also utilized to refine the answers.

[1793] Actual visits and feedback

[1794] Based on the generated incident reports and Q&A, users visit customers, submit reports, and answer customer questions. After the visit, users input the questions and answers they received from the customer into a terminal and send them to the server. The server receives this data and evaluates the accuracy of the prediction algorithm. At the same time, sentiment data is also analyzed to help improve the generation algorithm for future generations.

[1795] Example of a prompt

[1796] "Please write a program to automatically generate incident reports. Analyze the incident information entered by the user, compare it to past reports, and generate a new report based on the most similar report. Also, recognize the user's sentiment and adjust the report content accordingly."

[1797] In summary, the system of the present invention enables the rapid and effective creation of fault reports and customer support, allows for flexible responses that take into account user emotions, and can improve customer satisfaction.

[1798] The flow of the specific processing in Example 2 will be explained using Figure 13.

[1799] Step 1:

[1800] The user logs into their device and accesses the bug report form. The device displays the bug report form screen using HTML and CSS.

[1801] Input: User login information and request for the incident report form.

[1802] Data processing: None

[1803] Output: Incident Report Form Screen

[1804] Step 2:

[1805] The user enters basic information about the incident (incident name, date and time of occurrence, scope of impact, and detailed description) into a form and presses the submit button. The device sends the incident information to the server in JSON format.

[1806] Input: Name of the problem, date and time of occurrence, scope of impact, detailed description

[1807] Data processing: Convert form data to JSON format

[1808] Output: Send failure information in JSON format to the server.

[1809] Step 3:

[1810] When the server receives failure information, it searches the database for past failure reports. It uses a similarity evaluation algorithm (e.g., cosine similarity) to select the most similar past report.

[1811] Input: Failure information in JSON format

[1812] Data processing: Database search using SQL queries, similarity evaluation.

[1813] Output: Most similar past incident reports

[1814] Step 4:

[1815] A new incident report is generated based on past reports where the server was selected. A template engine (e.g., Jinja2) is used to fill in each field of the report and generate the completed report.

[1816] Input: Most similar past incident reports

[1817] Data processing: Field embedding using a template engine

[1818] Output: New Incident Report

[1819] Step 5:

[1820] The server sends the generated failure report to the terminal, and the terminal displays the report to the user.

[1821] Input: New Incident Report

[1822] Data processing: None

[1823] Output: Display of the failure report screen

[1824] Step 6:

[1825] The user reviews the report content and makes any necessary corrections. After completing the corrections, they press the confirm button to send the report to the server.

[1826] Input: Revised report content

[1827] Data processing: None

[1828] Output: Send the revised report to the server.

[1829] Step 7:

[1830] The device's built-in emotion analysis engine analyzes user input and interactions to recognize the user's emotions. It then generates specific emotion labels (e.g., impatience, anger).

[1831] Input: User input and operation data

[1832] Data processing: Emotion recognition using emotion analysis algorithms

[1833] Output: Sentiment labels

[1834] Step 8:

[1835] Based on the emotional information recognized by the server, the content of the incident report will be adjusted. If the user is anxious, detailed steps and additional explanations will be added.

[1836] Input: Revised report, sentiment label

[1837] Data processing: Report adjustment using text generation algorithms

[1838] Output: Adjusted Incident Report

[1839] Step 9:

[1840] The server analyzes the content of the confirmed report, references the customer's industry and past inquiry history, and generates predicted Q&A. A natural language processing model (e.g., GPT-3) is used.

[1841] Input: Confirmed incident report, customer industry information, past inquiry history

[1842] Data processing: Text analysis, Q&A generation

[1843] Output: Predicted Q&A

[1844] Step 10:

[1845] The server prepares answers to the generated Q&A and adjusts the content of the answers based on sentiment information.

[1846] Input: Predicted Q&A, sentiment information

[1847] Data processing: Response text generation, content adjustment based on sentiment.

[1848] Output: Adjusted Q&A and their answers

[1849] Step 11:

[1850] The server sends the generated Q&A and its answer to the terminal, which then displays it to the user.

[1851] Input: Adjusted Q&A and their answers

[1852] Data processing: None

[1853] Output: Display a Q&A list and its answers on the screen.

[1854] Step 12:

[1855] Users review the Q&A and answers to prepare for their visit. After the visit, they input the actual questions and answers into their terminal and send them to the server.

[1856] Input: Actual questions and answers received

[1857] Data processing: None

[1858] Output: Send actual Q&A data to the server

[1859] Step 13:

[1860] The server compares the actual Q&A with the predicted Q&A and evaluates the accuracy of the prediction. Based on the evaluation results, the prediction algorithm is improved.

[1861] Input: Actual Q&A, Predicted Q&A

[1862] Data processing: Accuracy evaluation, algorithm improvement

[1863] Output: Improved prediction algorithm

[1864] As described above, the system's processing steps involve searching past reports based on the input failure information, referencing similar reports, and generating a new failure report. Furthermore, it adjusts the report considering the user's sentiment, generates anticipated Q&A, and prepares appropriate answers. This enables rapid and effective customer support.

[1865] (Application Example 2)

[1866] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[1867] Creating security-related incident reports and handling customer inquiries requires speed and accuracy, but manual processes are time-consuming and labor-intensive, and it's difficult to respond while considering user emotions. Furthermore, predicting the information needed to solve actual problems and preparing appropriate answers in advance is challenging.

[1868] In Application Example 2, the identification processing by the identification processing unit 290 of the data processing device 12 is realized by the following means. In this invention, the server includes means for receiving basic information about a failure, means for searching past failure reports and evaluating similarity, means for automatically generating a new failure report based on the searched past reports, means for referencing the customer's industry and past inquiry history and generating predicted Q&A, means for preparing answers to the generated Q&A, means for receiving actual Q&A and evaluating its accuracy by comparing it with the predicted Q&A, means for recognizing the user's emotions and adjusting the content of the report and Q&A, and means for operating as an application installed on a smartphone or head-mounted display. This streamlines the creation of failure reports and customer support, and enables high-quality support that takes user emotions into consideration.

[1869] "Means for receiving basic information about a failure" refers to a function in the system that allows users to input information such as the name of the failure, the date and time of occurrence, the scope of impact, and a detailed description.

[1870] "Means for searching past failure reports and evaluating similarity" refers to a function in which the system searches past failure reports in the database and uses an algorithm to evaluate and identify similarities with those reports.

[1871] The "means for automatically generating new incident reports" refer to a function that automatically creates reports addressing newly occurring incidents by using past reports, which are searched based on similarity assessments, as templates.

[1872] "A means of generating predicted Q&A by referring to the customer's industry and past inquiry history" refers to a function that prepares anticipated questions and their answers in advance based on the customer's industry information and past inquiry history.

[1873] "Means for preparing answers to generated Q&A" refers to a function that provides appropriate information for anticipated questions and prepares the answers in advance.

[1874] "A means of receiving actual Q&A and evaluating its accuracy by comparing it with predicted Q&A" refers to a function that compares the actual questions asked with the predicted results to evaluate and improve prediction accuracy.

[1875] "Means for recognizing user emotions and adjusting the content of reports and Q&A" refers to a function that analyzes the user's emotional state (e.g., impatience or anger) using an emotion recognition engine and appropriately modifies and adjusts the content of reports and Q&A according to those emotions.

[1876] "Means of operation using applications installed on smartphones and head-mounted displays" refers to a method of using this system by installing a dedicated application on devices such as smartphones and head-mounted displays, enabling the system to function on these devices.

[1877] The system for carrying out this invention mainly uses the following hardware and software. Specific examples will be described below.

[1878] Hardware and software to be used

[1879] Device: Smartphone (iOS / Android) or head-mounted display (HoloLens, etc.)

[1880] Servers: Web server, database (MySQL, etc.)

[1881] Emotion recognition engine: IBM Watson and Azure Cognitive Services

[1882] AI inference engine: Generative AI models such as GPT-4

[1883] System Processing Overview

[1884] The server performs the following actions based on the fault report entered by the user using a terminal.

[1885] 1. Enter error information

[1886] The user logs in to their smartphone or head-mounted display (hereinafter referred to as "device") and accesses a dedicated incident reporting form. They enter basic information about the incident (incident name, date and time of occurrence, scope of impact, and detailed description) and press the submit button. The device sends the entered incident information to the server.

[1887] 2. Automatic generation of incident reports

[1888] The server searches its database for past incident reports based on the received incident information. A similarity evaluation algorithm selects the most similar past report, and then automatically generates a new incident report based on it. This report includes an overview of the incident, its cause, scope of impact, and countermeasures.

[1889] 3. User emotion recognition

[1890] The device's built-in emotion recognition engine (e.g., IBM Watson) analyzes user input and interaction to recognize emotions. For example, it can detect if the user is anxious or angry. The server then adjusts the report generated based on the emotion. For instance, it might include detailed instructions or additional explanations for an anxious user.

[1891] 4. Predicting and preparing for the Q&A session

[1892] The server analyzes the generated incident report, referencing the customer's industry and past inquiry history. It generates anticipated questions and prepares appropriate answers for each. The server also adjusts the Q&A responses based on sentiment recognition results.

[1893] 5. Feedback and accuracy evaluation

[1894] The user enters the Q&A used during actual customer visits and report submissions into a terminal and sends it to the server. The server compares the predicted Q&A with the actual Q&A, evaluates the accuracy of the prediction, and improves the algorithm as needed.

[1895] Specific examples and prompt statements

[1896] Example: Scenario for when a security camera malfunction is reported.

[1897] 1. User input fields:

[1898] Problem name: Security camera footage is distorted.

[1899] Date and time of occurrence: October 12, 2023, 14:00

[1900] Scope of impact: All store monitoring systems

[1901] Detailed description: The video frequently cuts out and is noisy.

[1902] 2. Emotion recognition:

[1903] Recognize that the user is anxious and include more detailed explanations in the report.

[1904] 3. Generated report:

[1905] The server automatically generates a report based on the "distorted security camera footage" issue, including the cause and solution.

[1906] 4. Prediction Q&A:

[1907] Q: When will the video become stable?

[1908] A: Expected to be restored within 2 hours.

[1909] Examples of prompts for a generative AI model:

[1910] Please generate the most appropriate report based on past troubleshooting data regarding the issue of distorted security camera footage. Also, please provide anticipated Q&A and their answers.

[1911] Problem name: Security camera footage is distorted.

[1912] Date and time of occurrence: October 12, 2023, 14:00

[1913] Scope of impact: All store monitoring systems

[1914] Detailed description: The video frequently cuts out and is noisy.

[1915] User's emotion: impatience

[1916] This will enable more efficient creation of incident reports and customer support, as well as responses that take user emotions into consideration.

[1917] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[1918] Step 1:

[1919] The user logs in using their smartphone or head-mounted display and accesses a dedicated incident reporting form. They enter basic information about the incident, such as the incident name, date and time of occurrence, scope of impact, and detailed description, and then press the submit button. The entered information is sent from the device to the server.

[1920] Input: Name of the problem, date and time of occurrence, scope of impact, detailed description

[1921] Output: Failure information transferred to the server

[1922] Step 2:

[1923] Based on the failure information received by the server, past failure reports are searched in the database. A similarity evaluation algorithm is used to select the past report with the highest similarity. Database queries and similarity calculations are performed during this search and selection process.

[1924] Input: Error information received by the server

[1925] Output: Past incident reports with high similarity

[1926] Step 3:

[1927] The server automatically generates a new incident report using a previously selected report as a template. The new report includes an overview of the incident, its cause, scope of impact, and countermeasures. Template and text generation algorithms are used in this process.

[1928] Input: Similar past incident reports

[1929] Output: Newly generated incident report

[1930] Step 4:

[1931] The device's built-in emotion recognition engine analyzes user input and interaction to recognize the user's emotions. The recognized emotion information is sent to a server. Text and speech analysis algorithms are used for emotion recognition.

[1932] Input: User input and interaction data

[1933] Output: Recognized user sentiment information

[1934] Step 5:

[1935] The server adjusts the content of the generated incident report based on the perceived emotions. For example, if the user is anxious, the report will include more detailed steps and additional explanations. A text generation algorithm is used in this process.

[1936] Input: Recognized emotion information and automatically generated new incident report

[1937] Output: Adjusted Incident Report

[1938] Step 6:

[1939] The server analyzes the generated incident report and references the customer's industry and past inquiry history. Based on this, it generates predicted Q&A and prepares appropriate answers for each. An AI inference engine is used for prediction and answer generation.

[1940] Input: Adjusted incident report, customer industry information, past inquiry history

[1941] Output: Predicted Q&A and their answers

[1942] Step 7:

[1943] Users visit customers in person and respond to Q&A based on submitted incident reports. After the visit, they input the questions and answers into a terminal and send them to the server.

[1944] Input: Actual questions and answers

[1945] Output: Actual Q&A as feedback to the server

[1946] Step 8:

[1947] The server compares predicted Q&A with actual Q&A to evaluate the accuracy of the prediction. Based on the evaluation results, the prediction algorithm is improved to enhance the accuracy of future answers. An accuracy evaluation algorithm and a feedback loop are used in this process.

[1948] Input: Actual Q&A, Predicted Q&A

[1949] Output: Prediction accuracy evaluation results and improved algorithm

[1950] The specific processing unit 290 transmits the result of the specific processing to the robot 414. In the robot 414, the control unit 46A causes the speaker 240 and the controlled object 443 to output the result of the specific processing. The microphone 238 acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[1951] Data generation model 58 is a type of so-called generative AI (Artificial Intelligence). One example of data generation model 58 is ChatGPT (Internet search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1952] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and the specific processing may also be performed by the robot 414.

[1953] Furthermore, the emotion identification model 59, acting as an emotion engine, may determine the user's emotion according to a specific mapping. Specifically, the emotion identification model 59 may determine the user's emotion according to a specific mapping, which is an emotion map (see Figure 9). Similarly, the emotion identification model 59 may also determine the robot's emotion, and the identification processing unit 290 may perform identification processing using the robot's emotion.

[1954] Figure 9 shows an emotion map 400 in which multiple emotions are mapped. In the emotion map 400, emotions are arranged in concentric circles radiating from the center. The closer to the center of the concentric circles, the more primitive the emotions are located. Further out of the concentric circles, emotions representing states and actions arising from mental states are located. Emotion is a concept that includes feelings and mental states. On the left side of the concentric circles, emotions that are generally generated from reactions occurring in the brain are located. On the right side of the concentric circles, emotions that are generally induced by situational judgment are located. Above and below the concentric circles, emotions that are generally generated from reactions occurring in the brain and induced by situational judgment are located. In addition, the emotion of "pleasure" is located on the upper side of the concentric circles, and the emotion of "displeasure" is located on the lower side. Thus, in the emotion map 400, multiple emotions are mapped based on the structure in which emotions arise, and emotions that are likely to occur simultaneously are mapped close together.

[1955] These emotions are distributed at the 3 o'clock position on the Emotion Map 400, and usually fluctuate between feelings of security and anxiety. In the right half of the Emotion Map 400, situational awareness takes precedence over internal feelings, resulting in a calm impression.

[1956] The inside of the Emotion Map 400 represents inner thoughts, while the outside represents actions. Therefore, the further you go from the outside of the Emotion Map 400, the more visible (expressed in actions) your emotions become.

[1957] Here, human emotions are based on various balances, such as posture ...

Claims

1. A means of receiving basic information about disabilities, A means of searching past incident reports and evaluating similarities, A method for automatically generating new incident reports based on previously searched reports, A means of generating predicted Q&A by referring to the customer's industry and past inquiry history, A means of preparing answers to the generated Q&A, A system that includes means for receiving actual Q&A and evaluating its accuracy by comparing it with predicted Q&A.

2. The system according to claim 1, wherein the means for automatically generating failure reports uses past failure reports as templates to generate new failure reports.

3. The system according to claim 1, wherein the means for generating predicted Q&A refers to the customer's industry and past inquiry history to prepare the necessary answers.

Images