Processing device, processing method, and processing program

Abstract

It is possible to extract hidden problems of the user from the quantitative information and subjective information of the user, and to provide countermeasures for the extracted problems. [Solution] The support server 10 causes a generation AI (Artificial Intelligence) to extract the user's issues and / or worries for each user based on the user's quantitative information and the user's subjective information, which is the user's dialogue history regarding the user's past quantitative information. The support server 10 causes the generation AI to classify the issues and / or worries of female employees from integrated data that integrates the issues and / or worries of multiple users into one data set, by fitting them into a matrix diagram combining two items, and outputs countermeasures for the user's issues and / or worries for at least one combination of the combinations.

Description

[Technical Field]

[0001] The present invention relates to a processing device, a processing method, and a processing program. [Background technology]

[0002] In various fields, such as solving women's health issues and promoting women's advancement, creating nursing care plans, and providing instruction at cram schools, questionnaire surveys and dialogues through interviews and chats are conducted to identify issues users face. For example, amid a shrinking labor force due to a declining birthrate, women's health issues are affecting work efficiency and employment continuity. As a result, companies are being asked by government agencies to improve their work environments, including by providing solutions to women's health issues. Companies often use questionnaire surveys and interviews to investigate what services should be introduced as employee benefits. [Prior art documents] [Non-patent literature]

[0003] [Non-Patent Document 1] Cabinet Office, "Creating a new normal for a society where women can work comfortably," [Retrieved August 13, 2024], Internet<URL:https: / / www.femtech-projects.jp / > Summary of the Invention [Problem to be solved by the invention]

[0004] Here, there is a demand from companies and the like to extract hidden issues of users other than the answers to the questionnaire from the questionnaire survey and the dialogue history of interviews, chats, etc. related to the questionnaire content, and to select appropriate measures.

[0005] Therefore, a method has been proposed that uses predetermined logic and generation AI to extract user issues from questionnaire surveys and conversation histories with users, such as interviews and chats, regarding the survey content.However, with this method, issues that are almost identical to the survey results are often extracted, making it difficult to meet customer needs.

[0006] The present invention has been made in consideration of the above, and aims to provide a processing device, processing method, and processing program that can accurately extract a user's hidden issues from the user's quantitative information and the user's subjective information. [Means for solving the problem]

[0007] In order to solve the above-mentioned problems and achieve the object, the processing device of the present invention includes a first acquisition unit that acquires quantitative information of a user, a second acquisition unit that acquires a dialogue history with the user regarding the user's past quantitative information as subjective information of the user, and a generation AI (artificial intelligence) that generates, for each user, the user's problems and / or worries based on the user's quantitative information and the user's subjective information. an integration unit that generates integrated data that integrates the issues and / or worries of the multiple users into one data; a second extraction unit that causes the generative AI model to classify the issues and / or worries of the multiple users from the integrated data for each combination of the degree of a first item related to the issues and / or worries of the users and the degree of a second item related to the issues and / or worries of the users, and output countermeasures for the issues and / or worries of the users for at least one of the combinations; and a presentation unit that presents the issues and / or worries of the users and the countermeasures for the issues to a source that requests the countermeasures.

[0008] The processing device of the present invention is characterized by having a first extraction unit that causes a generative AI (Artificial Intelligence) model to extract the user's issues and / or concerns for each user based on the user's quantitative information and the user's subjective information, which is a dialogue history with the user regarding the user's past quantitative information; an integration unit that generates integrated data that integrates the issues and / or concerns of multiple users into one data; and a second extraction unit that causes the generative AI model to classify the issues and / or concerns of the multiple users from the integrated data for each combination of the degree of each first item related to the user's issues and / or concerns and the degree of each second item related to the user's issues and / or concerns.

[0009] The processing method of the present invention is a processing method executed by a processing device, and includes a first acquisition step of acquiring quantitative information of a user, a second acquisition step of acquiring a dialogue history with the user regarding the user's past quantitative information as subjective information of the user, and a generation step of generating, for each user, the user's issues and / or worries based on the user's quantitative information and the user's subjective information, using artificial intelligence (AI). a first extraction step of causing a generative AI model to extract the user's issues and / or worries; an integration step of generating integrated data that integrates the user's issues and / or worries into one data; a second extraction step of causing the generative AI model to classify the user's issues and / or worries from the integrated data for each combination of the degree of a first item related to the user's issues and / or worries and the degree of a second item related to the user's issues and / or worries, and outputting countermeasures for the user's issues and / or worries for at least one of the combinations; and a presentation step of presenting the user's issues and / or worries and the countermeasures for the issues to a source that requests the countermeasures.

[0010] Furthermore, the processing method of the present invention is a processing method executed by a processing device, and is characterized by including: a first extraction step of having a generative AI (Artificial Intelligence) model extract the user's issues and / or concerns for each user based on the user's quantitative information and the user's subjective information, which is a dialogue history with the user regarding the user's past quantitative information; an integration step of generating integrated data that integrates the issues and / or concerns of multiple users into one data; and a second extraction step of having the generative AI model classify the issues and / or concerns of the multiple users from the integrated data for each combination of the degree of each first item related to the user's issues and / or concerns and the degree of each second item related to the user's issues and / or concerns.

[0011] In addition, the processing program of the present invention causes a computer to execute the following steps: a first acquisition step of acquiring quantitative information of a user; a second acquisition step of acquiring a dialogue history with the user regarding the user's past quantitative information as the user's subjective information; a first extraction step of having a generative AI (Artificial Intelligence) model extract the user's issues and / or concerns for each user based on the user's quantitative information and the user's subjective information; an integration step of generating integrated data that integrates the issues and / or concerns of multiple users into one data; a second extraction step of having the generative AI model classify the issues and / or concerns of the multiple users from the integrated data for each combination of the degree of each first item related to the user's issues and / or concerns and the degree of each second item related to the user's issues and / or concerns, and output countermeasures for the user's issues and / or concerns for at least one of the combinations; and a presentation step of presenting the user's issues and / or concerns and the countermeasures for the issues to a source that requests the countermeasures.

[0012] In addition, the processing program of the present invention causes a computer to execute the following steps: a first extraction step of having a generative AI (Artificial Intelligence) model extract the user's issues and / or concerns for each user based on the user's quantitative information and the user's subjective information, which is a dialogue history with the user regarding the user's past quantitative information; an integration step of generating integrated data that integrates the issues and / or concerns of multiple users into one data; and a second extraction step of having the generative AI model classify the issues and / or concerns of the multiple users from the integrated data for each combination of the degree of each first item related to the user's issues and / or concerns and the degree of each second item related to the user's issues and / or concerns. [Effects of the Invention]

[0013] According to the present invention, a user's hidden problems can be accurately extracted from the user's quantitative information and subjective information. [Brief explanation of the drawings]

[0014] [Figure 1] FIG. 1 is a diagram showing an outline of processing in a processing system according to an embodiment. [Figure 2] FIG. 2 is a diagram illustrating an example of a configuration of a processing system according to an embodiment. [Figure 3] FIG. 3 is a diagram illustrating an example of the configuration of the support server illustrated in FIG. [Figure 4] FIG. 4 is a diagram for explaining an outline of the questions in the questionnaire. [Figure 5] FIG. 5 is a diagram illustrating input data to the generation AI. [Figure 6] FIG. 6 is a diagram showing an example of a prompt set in the generation AI. [Figure 7] FIG. 7 is a diagram illustrating the processing of the support server shown in FIG. [Figure 8] FIG. 8 is a diagram showing an example of a prompt for problem extraction. [Figure 9]FIG. 9 is a diagram illustrating an example of a prompt for classification. [Figure 10] FIG. 10 is a diagram showing the determination contents of severity, feasibility, and possibility. [Figure 11] FIG. 11 is a diagram showing an example of a report screen created by the report creation unit shown in FIG. [Figure 12] FIG. 12 is a diagram showing an example of a conversation script plan. [Figure 13] FIG. 13 is a diagram showing an example of a conversation script plan. [Figure 14] FIG. 14 is a diagram showing an example of a conversation script plan. [Figure 15] FIG. 15 is a diagram showing an example of a conversation script plan. [Figure 16] FIG. 16 is a diagram showing an example of a conversation script plan. [Figure 17] FIG. 17 is a diagram showing an example of a prompt for problem extraction. [Figure 18] FIG. 18 shows an example of a hidden assignment file extracted by the generation AI. [Figure 19] FIG. 19 shows an example of a hidden assignment file extracted by the generation AI. [Figure 20] FIG. 20 shows an example of a hidden assignment file extracted by the generation AI. [Figure 21] FIG. 21 is a diagram showing an example of a prompt for classification. [Figure 22] FIG. 22 is a diagram showing an example of a matrix diagram generated by the generation AI. [Figure 23] FIG. 23 shows countermeasures for each problem generated by the generation AI. [Figure 24] FIG. 24 is a diagram for explaining input and output information of the generation AI. [Figure 25] FIG. 25 is a diagram showing an example of a screen of a user terminal. [Figure 26] FIG. 26 is a diagram showing an example of a screen of a user terminal. [Figure 27] FIG. 27 is a diagram showing an example of a prompt set in the generation AI. [Figure 28] FIG. 28 is a diagram showing an example of a chat on a user terminal. [Figure 29] FIG. 29 is a diagram showing an example of a screen of a user terminal. [Figure 30] FIG. 30 is a sequence diagram illustrating a processing procedure of the support processing according to the embodiment. [Figure 31] FIG. 31 is a sequence diagram illustrating a processing procedure of the support processing according to the embodiment. [Figure 32] FIG. 32 is a sequence diagram illustrating a processing procedure of the support processing according to the embodiment. [Figure 33] FIG. 33 is a diagram illustrating an example of a computer that implements the support server by executing a program. DETAILED DESCRIPTION OF THE INVENTION

[0015] Hereinafter, an embodiment of the present invention will be described in detail with reference to the drawings. Note that the present invention is not limited to this embodiment. In addition, in the description of the drawings, the same parts are designated by the same reference numerals.

[0016] [Embodiment Mode] First, an embodiment will be described. In the embodiment, the user is a female employee, and hidden issues of the female employee other than the answers to the questionnaire are extracted, and countermeasures for the extracted issues are provided.

[0017] In this case, the support server (processing device) according to the embodiment picks up the inner thoughts of female employees through chat using a generative AI (artificial intelligence) (generative AI model) that can hold conversations that are empathetic to female employees, based on health management information specific to women and a questionnaire survey for female employees (users) regarding women's specific health issues and women's working styles (for example, policies to promote women's participation in the workforce, promotion of women's participation in the workforce).

[0018] The support server causes the generation AI to extract, for each female employee, the issues and / or worries of the female employee that could not be collected through the questionnaire. The support server generates integrated issues that integrate the issues and / or worries of multiple female employees, and causes the generation AI to classify the issues of the multiple female employees for each combination of the degree of a first item related to the issue and the degree of a second item related to the issue. The support server also causes the generation AI to output countermeasures for the issues and / or worries for at least one or more combinations of the combinations.

[0019] In this way, the support server performs extraction and classification of female employee's issues and / or worries in two stages, thereby accurately extracting hidden issues and worries of female employees other than the answers to the questionnaire.The support server 10 then provides appropriate solutions to the extracted issues and / or worries to, for example, the welfare officer of company A where the female employee works.

[0020] The support server provides companies with information on the hidden issues and / or concerns of female employees and proposed solutions to address them, thereby supporting companies in introducing services and planning measures to create workplaces that are easy for female employees to work in.

[0021] [Outline of the embodiment] Fig. 1 is a diagram showing an overview of the processing of a processing system according to an embodiment. As shown in Fig. 1, in the embodiment, a welfare service is provided to the user terminal of each female employee as a curation site or application (app) that extracts the issues and concerns of each female employee and presents solutions to them.

[0022] The support server (processing device) 10 of the processing system collects information about female employee H to confirm her needs (step S1). The support server acquires female employee H's answers 12a-1 (user's subjective information) to a questionnaire survey (questionnaire questions (user's quantitative information)) about health issues specific to women and women's working styles. A questionnaire survey about women's empowerment measures may be conducted on the user terminal used by female employee H via a generation AI chat 13c-1.

[0023] Next, the support server 10 uses the generated AI chat 13c-1 with female employee H to dig deeper into the survey results and collect female employee H's current issues and / or concerns that were not picked up by the survey. At this time, the support server 10 collects information while engaging in a supportive conversation with female employee H via the generated AI chat 13c-1. The generated AI chat 13c-1 collects female employee H's current chat history (dialogue history) (user subjective information) regarding, for example, career aspirations, concerns about health issues, concerns about life issues, dissatisfaction with systems and policies, requests, and literacy issues. The support server 10 also performs the above process on female employees other than female employee H, acquiring chat histories from multiple female employees.

[0024] Next, the support server 10 executes the following process to analyze the employee satisfaction and hidden needs ((1) in FIG. 1).

[0025] The support server 10 uses a generation AI to analyze the results of a survey of multiple female employees and the chat history (dialogue history) related to the survey results, extracts the issues and / or concerns of the female employees, and outputs countermeasures for the issues and / or concerns ((2) in FIG. 1). The support server 10 may also perform anonymization.

[0026] Here, the support server 10 applies a two-stage process to extract the issues and / or concerns of female employees. As the first stage of the process, the support server 10 executes a process in which the generation AI extracts the issues and / or concerns of each female employee. Next, as the second stage of the process, the support server 10 causes the generation AI to classify the issues and / or concerns of the female employees from integrated data that integrates the issues and / or concerns of multiple female employees so that they fit into a matrix diagram combining two items, and outputs a countermeasure for the issues and / or concerns of the female employees for any combination.

[0027] The support server 10 then provides the company where the female employees work with a dashboard that allows the company to view, in report format, the statistical results of the survey, as well as various issues, including hidden issues and / or worries of female employees, regarding their issues (e.g., health issues) and / or worries (work style, etc.), and necessary countermeasures ((3) in Figure 1).

[0028] Using the classification results and advice on countermeasures output on the dashboard, company welfare staff can efficiently devise more effective measures for hidden issues and / or concerns of female employees that are not reflected in the questionnaire responses (Figure 1 (4) and (5)). Measures include providing health management apps, making work flexible, establishing support systems for mental health, replacing fixtures in rooms, partnering with medical institutions, and conducting in-house training.

[0029] For example, in order to solve the problems of female employee H, the company provides female employee H with services such as a menstrual management app F1, a pregnancy / infertility support app F2, a health management app F3, and other personal health records (PHR) for women, as well as femtech-related services, as employee benefit services via a curation site or app (steps S2, S3). Specifically, the support server 10, in a state where it can link with the service provider, presents proposed solutions to the user terminal of female employee H. The presented services correspond to proposed solutions to the problems and worries of female employee H.

[0030] The support server 10 then works in conjunction with each PHR application and Femtech-related services (Femcare recommendation service W1, lifestyle improvement recommendation service W2, online dispensing / prescription W3, online medical consultation / diagnosis W4) to analyze the female employee's condition (analysis result 12i-1), and presents solutions to the concerns narrowed down from female employee H's chat history (dialogue history) etc. to female employee H's terminal (step S2).

[0031] Furthermore, after implementing the proposed countermeasures, the support server 10 uses the generation AI to hold a generation AI chat 13c-2 with female employee H. The generation AI chat 13c-2 is conducted to draw out the results of implementing the proposed countermeasures and the issues and / or worries of female employee H thereafter. The support server 10 extracts the effects of implementing the proposed countermeasures based on the chat history here and / or the health management information of female employee H obtained from the PHR service.

[0032] The support server 10 feeds back the effects of the implementation of the extracted countermeasures to the company where female employee H works. In addition, the support server 10 can anonymize the effects of the implementation of the countermeasures and then feed them back to the partner companies of the PHR service as the effects of the service and user feedback ((6) in Figure 1), allowing them to use the data to improve the service, etc. ((7) in Figure 1, step S4).

[0033] [Processing System] Next, the configuration of the processing system will be described with reference to Fig. 2. Fig. 2 shows an example of the configuration of the processing system according to the embodiment.

[0034] As shown in Fig. 2, the processing system according to the embodiment has a support server 10 provided on a platform that provides support services. The support server 10 communicates with a company terminal 20 of company A, which is the service target, and user terminals 30h, 30i, and 30j of female employees working at company A. The support server 10 communicates with a generation AI server 200.

[0035] The generation AI server 200 is equipped with a generation AI 210 (a generation AI model, a generation AI model for dialogue), which is a large-scale natural language processing model. The generation AI 210 performs natural language processing on input text data in accordance with set prompts, creates text, and outputs it. The generation AI 210, for example, generates text in accordance with information input by a female employee and outputs it, thereby engaging in dialogue with the female employee. The generation AI 210 also performs natural language processing on input text data in accordance with set prompts, extracts data, and creates and outputs a report (summary).

[0036] The support server 10 sets a prompt in the generation AI 210 and inputs text data (or transcription data of voice data) input to the user terminals 30h, 30i, and 30j into the generation AI 210. The support server 10 transmits the text data output by the generation AI 210 to the user terminals 30h, 30i, and 30j, and outputs it in chat. In this way, the support server 10 uses the generation AI 210 to have a dialogue with the female employees to uncover hidden issues and / or concerns of the female employees that cannot be collected through a questionnaire.

[0037] The support server 10 extracts hidden issues and worries (such as causes of illness) of female employees through chats using conversations that are considerate to the female employees, and outputs measures to address the issues and / or worries of the female employees. The support server 10 then transmits the extracted issues and / or worries of the female employees and the measures to the company terminal 20.

[0038] The company terminal 20 is, for example, a PC (Personal Computer), a notebook PC, or a tablet terminal. The user terminals 30h to 30j are, for example, a PC, a notebook PC, a tablet terminal, a smartphone, etc. The user terminals 30h to 30i are collectively referred to as user terminals 30.

[0039] [Support Server] Fig. 3 is a diagram showing an example of the configuration of the support server 10 shown in Fig. 2. The support server 10 includes, for example, a communication unit 11, a storage unit 12, and a control unit 13.

[0040] The communication unit 11 controls communication related to various information. For example, the communication unit 11 controls communication between the generation AI server 200, the company terminal 20, and the user terminal 30.

[0041] The memory unit 12 stores data and programs necessary for various processes by the control unit 13. For example, the memory unit 12 may be a semiconductor memory element such as a random access memory (RAM) or a flash memory, or a storage device such as a hard disk or an optical disk. The memory unit 12 has questionnaire survey results 12a, user information 12b, health management information 12c, chat history 12d, prompt data 12e, task data group 12f, task integrated data 12g, classification and countermeasure data 12h, analysis results 12i, countermeasure information 12j, and report data 12k.

[0042] The questionnaire survey results 12a are the questions and answers to a questionnaire administered to each female employee regarding women's specific health issues and women's working styles (for example, women's empowerment policies and promotion of women's empowerment). The questionnaire survey results 12a may be received from the company terminal 20 or collected via the user terminal 30 used by each female employee. The questions in the questionnaire correspond to the user's quantitative information. The user's responses to the questionnaire correspond to the user's subjective information.

[0043] The user information 12b is information about each female employee, and includes, for example, the age, address, and family structure of the female employee.

[0044] Health management information 12c is information about the health of each female employee, and is information acquired, for example, via a menstrual management app F1, a pregnancy / infertility support app F2, a health management app F3, etc., installed on user terminal 30. For example, health management information 12c includes, as menstrual management information, the status of the menstrual cycle (including the start date), basal body temperature, and mood (for example, mental and physical symptoms of instability due to premenstrual syndrome (PMS)). Health management information 12c also includes, as pregnancy / infertility information, the timing of hospital visits, the purpose of hospital visits, treatment and medication details, etc. Health management information also includes vital information such as the number of steps, exercise, distance traveled, and heart rate, as well as sleep information, dietary history, and headache status.

[0045] The chat history 12d is a chat history between each female employee and the chatbot provided by the support server 10. The chat history 12d is a combination of prompts, questions, solutions, etc. output by the generation AI 210 and the text input by the female employee in response to them. Since the text input by the female employee is based on the female employee's subjective opinion, the chat history 12d is one of the female employee's past subjective information. The subjective information is, for example, the female employee's responses to a questionnaire regarding women's specific health issues and women's working styles, and / or the female employee's chat history regarding women's specific health issues and women's working styles, and the female employee's issues and / or worries.

[0046] The prompt data 12e is each prompt set in the generation AI 210. For example, it includes a prompt that instructs to conduct an in-depth questionnaire survey of female employees during chat, a prompt that instructs to extract issues and / or concerns of female employees that could not be collected through the questionnaire, a prompt that instructs to classify issues and / or concerns of female employees based on integrated data that integrates the issues and / or concerns of multiple female employees and output measures for the issues and / or concerns of female employees, a prompt that instructs to create a report for the company, and the like.

[0047] The problem data group 12f is a data group indicating the problems and / or worries of a plurality of female employees, each extracted by a problem extraction unit 13d (described later).

[0048] The integrated problem data 12g is data that integrates the problems and / or worries of multiple female employees.

[0049] The classification and countermeasure data 12h is data showing the issues and / or worries of female employees that could not be collected in the questionnaire classified by the classification presentation unit 13f (described later) and the countermeasures therefor.

[0050] The analysis result 12i includes the analysis result by the analysis unit 13i (described later) and includes the state of illness of the female employee, and prediction results of possible causes and timing of the illness.

[0051] The countermeasure information 12j includes various countermeasures that can be adopted when a female employee is unwell. For example, the countermeasure information 12j is information that associates, for each type of ailment, methods for improving lifestyle habits such as drinking a warm caffeine-free beverage before going to bed or going to bed earlier, examples of recommended medicines, recommendations for visiting a medical institution and the appropriate medical department, etc. The countermeasure information 12j is, for example, based on content supervised by medical professionals.

[0052] The report data 12k is a report created by the generation AI 210 in response to an instruction from the report creation unit 13g (described later). The report data 12k includes a report summarizing the issues and / or worries of female employees, such as health issues, lifestyle issues, dissatisfaction with systems and policies, attitudes toward careers, and / or dissatisfaction with the workplace (literacy, work environment), as well as proposed solutions.

[0053] The control unit 13 has an internal memory for storing programs that define various processing procedures and necessary data, and executes various processes using these. Here, the control unit 13 may be, for example, an electronic circuit such as a CPU (Central Processing Unit) or an MPU (Micro Processing Unit), or an integrated circuit such as an ASIC (Application Specific Integrated Circuit) or an FPGA (Field Programmable Gate Array).

[0054] The control unit 13 has a questionnaire data collection unit 13a (first acquisition unit), a health management information acquisition unit 13b (first acquisition unit), a chat control unit 13c (second acquisition unit), a problem extraction unit 13d (first extraction unit), a problem integration unit 13e (integration unit), a classification presentation unit 13f (second extraction unit), a report creation unit 13g (presentation unit), an effect extraction unit 13h (third extraction unit), an analysis unit 13i, and a solution presentation unit 13j.

[0055] The questionnaire data collection unit 13a collects questionnaire survey results (questionnaire responses) from female employees. The questionnaires are about health issues specific to women and women's working styles. For example, the questionnaires are conducted via the web or a chatbot using the generation AI 210, and the questionnaire response results are collected in the company terminal 20 with the permission of each female employee.

[0056] The health management information acquisition unit 13b acquires health management information indicating the past health condition of each female employee. The health management information is used as past quantitative information of the female employee.

[0057] Health management information acquisition unit 13b acquires health management information for each female employee via a menstrual management app F1, a pregnancy / infertility support app F2, a health management app F3, etc. installed on user terminal 30. For example, health management information 12c includes menstrual management information such as menstrual cycle (including start date), basal body temperature, and mood (PMS symptoms, etc.). Health management information 12c also includes pregnancy / infertility information such as timing of hospital visits, purpose of hospital visits, treatment / medication details, etc. Health management information also includes vital information such as number of steps, exercise, distance traveled, and heart rate, sleep information, diet history, and headache status. Health management information acquisition unit 13b acquires weather information, such as time-series data of barometric pressure.

[0058] The chat control unit 13c controls chats with female employees via a chatbot using the generation AI 210. The chat control unit 13c digs into at least the survey results via the chatbot and acquires the chat history (dialogue history) (user subjective information) of the female employees regarding their issues and / or worries. The chat control unit 13c instructs the generation AI 210 to have a dialogue that narrows down the issues and / or worries of the female employees while listening to their worries.

[0059] At the same time, the chat control unit 13c sets a prompt instructing the user to have a friendly conversation with the woman in the generation AI 210.

[0060] The problem extraction unit 13d causes the extraction generation AI 210 to extract, for each female employee, the problems and / or concerns of the female employee that could not be collected through the questionnaire, based on the content of the questionnaire questions, the questionnaire answers, and the chat history (dialogue history) with the female employee regarding the questionnaire results.

[0061] The problem integration unit 13e generates integrated data in which the problems and / or worries of a plurality of female employees are integrated into one piece of data.

[0062] The classification and presentation unit 13f causes the generation AI 210 to classify the issues and / or worries of multiple female employees from the integrated data for each combination of the degree of each first item related to the issues and / or worries of female employees and the degree of each second item related to the issues and / or worries of female employees. At the same time, the classification and presentation unit 13f causes the generation AI 210 to output countermeasures for the issues for at least one or more combinations of the combinations.

[0063] The report creation unit 13g creates a report showing the issues and / or concerns of female employees classified by the classification presentation unit 13f and proposed solutions, and presents the report so that it can be viewed on a dashboard to the requester of the proposed solutions (company A's corporate terminal 20).

[0064] After the proposed countermeasures are implemented, the chat control unit 13c acquires the results of the proposed countermeasures after being implemented and the chat history (dialogue history) of the female employee regarding her subsequent issues and / or concerns through chat with the female employee using the generation AI 210.

[0065] The effect extraction unit 13h causes the generation AI model to extract the effects after the implementation of the proposed countermeasures based on the chat history of the female employee and / or the health management information of the female employee after the implementation of the proposed countermeasures, and outputs the extracted effects to the source of the request for the proposed countermeasures (company A's corporate terminal 20).

[0066] The analysis unit 13i estimates the causal relationship between the information based on the past quantitative information and past subjective information of the female employee to be supported, and predicts the female employee's ill health, as well as possible causes and timing of the ill health, based on the causal relationship. The past quantitative information is health management information of the female employee to be analyzed. The past subjective information is the past chat history between the female employee to be analyzed and the generation AI 210. The analysis unit 13i transmits the predicted ill health of the female employee, as well as possible causes and timing of the ill health, to the user terminal 30.

[0067] The solution suggestion unit 13j narrows down the possible causes of the female employee's illness from the possible causes. To narrow down the possible causes, the solution suggestion unit 13j uses the current subjective information (current chat history) of the female employee acquired through chat using the generation AI 210.

[0068] The solution presentation unit 13j presents one or more solutions to the narrowed-down causes to the female employee's user terminal 30. The solution presentation unit 13j presents one or more solutions to the female employee's user terminal 30 in a state in which it is possible to link with a service provider according to the solution corresponding to the narrowed-down cause. The solution presentation unit 13j suggests a solution that is easy for the female employee to accept based on subjective information about how she is feeling through a conversation with a chatbot using the generation AI 210.

[0069] The solution presentation unit 13j may, for example, narrow down the cause of female employee H's illness from candidate causes based on current subjective information of female employee H acquired through chat, and set a prompt in the generation AI 210 to instruct the presentation of one or more solutions to the narrowed down cause. The solution presentation unit 13j may, for example, extract one or more solutions corresponding to the narrowed down cause from among the solutions in the solution information 12j, and present them to the female employee's user terminal 30. The solution presentation unit 13j may also estimate one or more solutions to the cause using a trained machine learning model.

[0070] [questionnaire] We will explain the results of a survey on women's specific health issues and women's working styles.

[0071] Figure 4 is a diagram outlining the questions in the survey. For example, the survey was about promoting women's participation in the workforce and included questions about attributes, attitudes toward careers, health issues specific to women, the workplace atmosphere, things that were given up due to health issues specific to women, internal policies and systems, and what was felt to be necessary to promote women's participation in the workforce.

[0072] [Chat control section] The processing of the chat control unit 13c will be described.

[0073] FIG. 5 is a diagram illustrating input data to the generation AI 210. The chat control unit 13c inputs questionnaire survey responses to the generation AI 210. If there is a chat history from past chats, the chat control unit 13c inputs the past chat history to the generation AI 210. The generation AI 210 is then instructed by the support server 10 to ask various probing questions, exemplified in FIG. 5, while providing insight into the questionnaire responses. As exemplified in FIG. 5, the questions inquire into details regarding health issues, lifestyle issues, systems and policies, attitudes toward careers, and dissatisfaction with the workplace.

[0074] Fig. 6 is a diagram showing an example of a prompt set in the generation AI 210. Prompt Pa shown in Fig. 6 instructs asking probing questions while being considerate to the user about the answers to the questionnaire.

[0075] The chat control unit 13c causes the generation AI 210 to generate text that is sympathetic to the woman and includes words that show empathy (for example, "I understand how you feel," "That's right," etc.).

[0076] The chat control unit 13c also instructs the generation AI 210 to change the tone of the conversation depending on the physical condition of the female employee. For example, the chat control unit 13c instructs the generation AI 210 to end the conversation more gently for a female employee who has menstrual pain, a headache, or PMS.

[0077] Depending on the question, the chat control unit 13c changes the timing of the response from the generation AI 210. For example, when a specific type of question is input by a female employee, the chat control unit 13c does not return an immediate answer, but rather makes the generation AI 210 respond slowly, for example, after two seconds, so as not to rush the female employee's chat input.

[0078] The chat control unit 13c also instructs the generation AI 210 to have a conversation with the female employee about health issues specific to women and concerns about women's work styles. For example, the chat control unit 13c causes the generation AI 210 to ask questions about health issues, lifestyle issues, dissatisfaction with systems and policies, attitudes toward careers, and / or dissatisfaction with the workplace (literacy, work environment). The chat control unit 13c then acquires the female employee's dissatisfaction and concerns in response to these questions as subjective information about the female employee. The chat control unit 13c acquires chat histories from multiple female employees about health issues specific to women and women's work styles.

[0079] Prompt Pa instructs the user to create a character that exudes familiarity, to use polite language to create a sense of trust, and to first conduct a survey via chat. Prompt Pa instructs the user to ask probing questions about the survey answers, to repeat the answers and include words of sympathy (that must be tough, that must be difficult, etc.), while also indicating that further questions will be asked. Prompt Pa also instructs the user to conclude by expressing gratitude for the answers, such as "Thank you for teaching me so much." Prompt Pa instructs the user to shorten long sentences by dividing them into speech bubbles to make them easier to read, and to slightly slow down the tempo at which the speech bubbles appear so as not to rush the user into answering.

[0080] [Two-stage extraction] Next, a description will be given of a two-stage process for extracting the issues and / or worries of female employees in the support server 10. FIG. 7 is a diagram for explaining the process of the support server 10 shown in FIG.

[0081] 7, for example, a questionnaire was administered to each of female employees H, I, and J regarding health issues specific to women and women's working styles, and a dialogue was held between the female employees and the employees regarding their issues and / or concerns based on the results of the questionnaire. Note that the dialogue may be a chat with the generation AI 210 under the control of the chat control unit 13c, or a dialogue with a public health nurse or an industrial physician.

[0082] Then, the support server 10 acquires the survey results (questions and answers) 12a-H, 12a-I, 12a-J of female employees H, I, J, and the dialogue history 12d-I, 12d-H, 12d-J ((1) in FIG. 7).

[0083] The problem extraction unit 13d causes the generation AI 210 to extract the problems and / or worries of the female employees H, I, and J that could not be collected through the questionnaire survey ((2) in FIG. 7).

[0084] For example, for female employee H, the problem extracting unit 13d sets a prompt P1 for problem extraction and inputs data Dh including the questionnaire results 12a-H and the dialogue history 12d-H to the generation AI 210.

[0085] Fig. 8 is a diagram showing an example of a prompt P1 for extracting issues. As shown in Fig. 8, prompt P1 instructs that issues and concerns that could not be collected from the employee's responses to the questionnaire or free comments should be extracted from the employee's chat conversation portion, and that responses related to issues and concerns that were answered in the questionnaire should be excluded.

[0086] As a result, the problem extraction unit 13d causes the extraction generation AI 210 to generate the problems and / or worries of the female employee H that could not be collected through the questionnaire, that is, the hidden problem file Th.

[0087] Similarly, the problem extraction unit 13d sets a prompt P1 for female employee I, and inputs data Di including the survey results 12a-I and the dialogue history 12d-I into the generation AI 210 to obtain a hidden problem file Ti for female employee I. The problem extraction unit 13d also sets a prompt P1 for female employee J, and inputs data Di including the survey results 12a-J and the dialogue history 12d-J into the generation AI 210 to obtain a hidden problem file Tj for female employee J.

[0088] Next, the task integration unit 13e generates integrated data DA by integrating the hidden task files Th to Tj of all the female employees H, I, and J who are respondents into one file data ((3) in FIG. 7).

[0089] Then, the classification presentation unit 13f sets a classification prompt P2 in the generation AI 210, inputs the integrated data DA, and causes the generation AI 210 to classify the problems and / or worries for each combination in the matrix diagram M1 ((4) in FIG. 7).

[0090] The first axis (first item) of the matrix diagram M1 is urgency (importance), and the second axis (second item) of the matrix diagram M1 is ease of implementation (difficulty of achievement). The urgency is set as "high" or "low." The ease of implementation is set as "easy to implement" or "difficult to implement." The classification presentation unit 13f causes the generation AI 210 to classify the issues and / or worries for each combination (A) to (D) of the degrees of the first and second axes of the matrix diagram M1 ((4) in FIG. 2).

[0091] FIG. 9 is a diagram showing an example of a prompt P2 for classification. As shown in FIG. 9, the prompt P2 instructs the respondent to classify their issues and concerns into (A) to (D) and display them in a matrix diagram M1. Here, as shown in the matrix diagram M1, (A) is an issue with a high level of urgency and easy to address. (B) is an issue with a high level of urgency and difficult to address. (C) is an issue with a low level of urgency and easy to address. (D) is an issue with a low level of urgency and difficult to address. FIG. 10 is a diagram showing the determination contents of the severity, feasibility, and visibility. For example, the data shown in FIG. 10 is also input to the generation AI 210 for issue classification and countermeasure generation.

[0092] Furthermore, the classification presentation unit 13f causes the generation AI 210 to generate countermeasures for the issues and / or concerns for (A) to (C) among the combinations (A) to (D) of the degrees of the first and second axes of the matrix diagram M1 ((5) in FIG. 7). The report creation unit 13g creates a report showing each of the issues and / or concerns (A) to (D) classified by the classification presentation unit 13f and the countermeasures Ra to Rc for the issues (A) to (C), and presents it on a dashboard on the company terminal 20 of company A.

[0093] FIG. 11 is a diagram showing an example of a report screen created by the report creation unit 13g shown in FIG. 3. For example, the report screen W23 is about women's working styles. The report screen W23 is about a questionnaire item about things that have been given up due to health issues specific to women. The report screen W23 lists the classification results (R23-1) of hidden issues and / or worries collected from the inner thoughts of female employees, classified by the generation AI 210, and necessary countermeasures (R23-2) (area R23).

[0094] In addition, reports include, for example, information on awareness and usage experience of Company A's policies and services, women's working styles, career aspirations, and concerns about health issues specific to women.The dashboard screen allows you to select the attributes of female employees you want to visualize by selecting the survey time, affiliation, position, age, and pregnancy / childbirth experience, and check statistical data, issues, and countermeasures from the survey response results.

[0095] In addition, the report creation unit 13g may embed text indicating each classified issue in the (A) to (D) sections of the matrix diagram M1 and create a report that describes countermeasures for the issues (A) to (C).

[0096] [Demonstration experiment] [Draft dialogue script] Next, an example of a conversation script proposal for a chat between female employees H to J and the generation AI 210 will be described. In a demonstration experiment, for example, an interview was conducted between female employees H to J and a public health nurse via chat to dig deeper into questionnaire responses. The conversation script proposal for this case is shown in FIGS. 12 to 16. In addition to the interview with the public health nurse, the conversation script proposal may be reflected and, under the control of the chat control unit 13c, a chat may also be executed between female employees H to J and the generation AI 210, or only a chat with the generation AI 210 may be executed.

[0097] 12 to 16 are diagrams showing examples of conversation script proposals. As shown in Fig. 12, the conversation script proposal describes that the generation AI 210 is to conduct a dialogue regarding greetings, personal information protection, and purpose clarification. As shown in Fig. 13, the conversation script proposal describes that the generation AI 210 is to conduct a dialogue regarding the symptoms and specific troublesome experiences of a female employee (e.g., female employee H) based on the responses to a questionnaire from the female employee.

[0098] As shown in Fig. 14, the conversation script proposal describes that the generation AI 210 is to have a conversation regarding the support that female employee H feels is necessary. As shown in Fig. 15, the conversation script proposal describes that the generation AI 210 is to have a conversation regarding how to deal with female employee H. As shown in Fig. 16, the conversation script proposal describes that the generation AI 210 is to offer thanks and closing remarks at the end.

[0099] Next, the problem extraction unit 13d extracted the problems and / or concerns of female employees H, I, J that could not be collected through the questionnaire from the questionnaire results 12a-H, 12a-I, 12a-J and the dialogue histories 12d-H, 12d-I, 12d-J of female employees H to J obtained through chat based on the proposed conversation script, for each of female employees H, I, J.

[0100] Figure 17 is a diagram showing an example of a prompt for issue extraction. As shown in Figure 17, prompt P11 instructs the user that "you are a specialist consultant in the field of human resources and welfare," that the document to be input "is a compilation of employee questionnaire responses and chat conversations in which a public health nurse asks questions about the questionnaire responses," that "issues and concerns that could not be collected from the questionnaire responses or free comments should be extracted from the employee conversation portions of the chat conversations," and that "questions related to issues and concerns that were answered in the questionnaire should be excluded."

[0101] The problem extraction unit 13d set the prompt P11 in the generation AI 210 and input the survey results 12a-H, 12a-I, 12a-J and the dialogue history 12d-H, 12d-I, 12d-J separately for each female employee H, I, J.

[0102] 18 to 20 are diagrams showing examples of hidden task files extracted by the generation AI 210. Fig. 18 shows the hidden task file Th of female employee H. Fig. 19 shows the hidden task file Ti of female employee I. Fig. 20 shows the hidden task file Tj of female employee J.

[0103] As shown in the hidden problem file Th (Figure 18), the problem extraction unit 13d extracted issues and concerns about female employee H that could not be collected from the questionnaire responses or free comments, such as surgery for uterine fibroids and its effects, the effects of PMS, and the effects of menopausal symptoms.

[0104] As shown in the hidden issue file Ti (Figure 19), the issue extraction unit 13d extracted issues and concerns about female employee I that could not be collected from the questionnaire responses or free comments, such as a lack of mental health support, the difficulty of balancing infertility treatment with work, and improving the work environment.

[0105] As shown in the hidden problem file Tj (Figure 20), the problem extraction unit 13d extracted issues and concerns about female employee J that could not be collected from the questionnaire responses or free comments, such as difficulty in adjusting schedules, difficulty in communicating with superiors, and anxiety about staff shortages in the department.

[0106] In this way, the issue extraction unit 13d was able to use the generation AI 210 to obtain the hidden issues and concerns of female employees H, I, and J for each of the female employees H, I, and J, which could not be collected through questionnaire responses or free-form writing.

[0107] Next, the task integration unit 13e generates integrated data DA by integrating the hidden task files Th to Tj of all the female employees H, I, and J who are respondents into one file data.

[0108] Then, the classification presentation unit 13f sets a classification prompt in the generation AI 210, inputs the integrated data DA, and causes the generation AI 210 to classify the tasks for each combination in the matrix diagram M1.

[0109] Fig. 21 is a diagram showing an example of a classification prompt. As shown in Fig. 21, classification prompt P21 instructs the user that "You are in charge of promoting women's activities in the human resources and welfare field," that the document to be input "is a collection of three employees' chat conversations, from which their respective issues and concerns are extracted (integrated data DA)," that "The concerns of these three employees are to be classified as follows (the aforementioned (A) to (D)) and shown in a matrix diagram (the aforementioned matrix diagram M1)," and that "Specific measures are to be shown for the issues (A), (B), and (C) in the matrix diagram."

[0110] The classification presentation unit 13f sets the prompt P21 in the generation AI210 and inputs the integrated data DA.

[0111] Fig. 22 is a diagram showing an example of a matrix diagram generated by the generation AI 210. Fig. 23 is a diagram showing countermeasures for each problem generated by the generation AI 210.

[0112] As shown in Fig. 22, the classification presentation unit 13f was able to obtain a matrix diagram M11 in which the problems and worries corresponding to each combination of (A) to (D) were classified. Also, as shown in Fig. 23, the classification presentation unit 13f was able to obtain a countermeasure plan R1 for each of the problems (A) to (C).

[0113] For example, as shown in Figure 22, issues and concerns that are highly urgent and easy to address (A) are classified as "Deterioration in work quality due to the impact of PMS (Mr. H)" (A-1) and "Lack of mental health support (Mr. I)" (A-2). Furthermore, countermeasure proposal R1 in Figure 23 lists countermeasure proposals for the issues and concerns that are highly urgent and easy to address (A-1) and (A-2), such as "Providing regular health consultations and mental health support, utilizing remote work and flextime systems," and "Strengthening the internal support system for mental health, and establishing specialized counselors and consultation desks."

[0114] The classification presentation unit 13f can clearly present the problems and concerns classified as (A-1) and (A-2) as well as countermeasures for each problem (A-1) and (A-2) to the welfare officer of company A. This allows the welfare officer of company A to recognize the problems and concerns of (A-1) and (A-2) and, by looking at the presented countermeasures R1, to quickly put the countermeasures into action without having to think of a countermeasure himself.

[0115] Furthermore, as shown in matrix M11, issues and concerns that are low in urgency but easy to address (C) are classified as "Advantages of remote work (Mr. H)" (C-1) and "Improving the color and material of chairs in the annex (Mr. I)" (C-2). Furthermore, countermeasure proposal R1 in Figure 23 lists countermeasure proposals for issues and / or concerns that are low in urgency but easy to address (C-1) and (C-2), such as "Further promoting the remote work system" and "Improving the color and material of chairs in the annex."

[0116] The issues and / or concerns classified as (C) here have been difficult to extract using previous systems.

[0117] In response to this, the classification presentation unit 13f can categorize issues and concerns into (C) which have a low level of urgency but are easy to address, and clearly display them together with proposed countermeasures. This allows the welfare officer of company A to accurately recognize issues and concerns classified as (C) which have been difficult to recognize in the past. Furthermore, by acquiring the proposed countermeasures presented by the support server 10, the welfare officer of company A can quickly put the proposed countermeasures into action. In particular, since issues and concerns in (C) are low in urgency but are easy to address, the welfare officer of company A can immediately take measures to resolve the issues and concerns of female employees.

[0118] As shown in matrix diagram M11, issues that are highly urgent but difficult to address (B) and issues that are less urgent but difficult to address (D) are specifically classified, so the welfare officer at Company A can consider how to address both issues.

[0119] In this way, the support server 10 can extract issues for each combination of the two degrees on the first axis and the two degrees on the second axis, as shown in the matrix diagram M11. By using the urgency as the first axis and the ease of addressing as the second axis as in the matrix diagram M11, the support server 10 can classify issues and / or worries into combinations (A) to (D) of the two degrees of urgency and ease of addressing, and clearly display them together with proposed countermeasures.

[0120] Therefore, the welfare officer of company A can recognize the issues and / or concerns by dividing them into (A) to (D), and can clearly recognize the urgency and ease of addressing each issue and / or concern. At the same time, the welfare officer of company A can quickly decide which issue to start with based on the matrix diagram M11. In this way, the support server 10 supports the company's consideration of measures to promote women's participation in the workforce.

[0121] [Providing individual solutions to female employees] Next, the presentation of measures to individual female employees (steps S3 and S2 in FIG. 1) will be described.

[0122] For example, Company A proposes a countermeasure to provide each female employee H with welfare services such as menstrual management app F1, fertility support app F2, health management app F3, and other HR services for women, as well as Femtech-related services (Femcare recommendation service W1, lifestyle improvement recommendation service W2, online dispensing / prescription W3, online medical consultation / diagnosis W4) via curation sites and apps. The support server 10 shares information with each application and service system and presents solutions appropriate for each individual female employee.

[0123] Fig. 24 is a diagram illustrating input and output information of the generation AI 210. Fig. 24 illustrates a case where a solution is output to the user terminal 30h of the female employee H, among the outputs of the generation AI 210.

[0124] In the support server 10, the analysis unit 13i receives input of a female employee (e.g., female employee H)'s questionnaire responses 12a-1, chat history 12d-1, attributes, health management information (menstrual management information, fertility and infertility information, vital signs information, sleep information, dietary history, headache status, etc.), and weather information (time-series data on atmospheric pressure) ((A) in FIG. 24). Based on this input information, the analysis unit 13i estimates causal relationships between the information, and predicts the state of illness of female employee H, as well as potential causes and timing of the illness, based on the causal relationships ((1) in FIG. 24). The analysis results by the analysis unit 13i are linked to the identification number of female employee H and stored in the memory unit 12.

[0125] The support server 10 displays visualized data of the analysis results on the user terminal 30h of the female employee H ((2) in FIG. 24). The visualized data includes the time of onset of the illness, the state of the illness, and possible causes.

[0126] 25 and 26 are diagrams showing examples of the screen of the user terminal 30h. As shown in Fig. 25, the screen of the user terminal 30h displays time-series data T11 of a physical condition forecast, weather, atmospheric pressure, a schedule, and hormone balance, as well as a comment C11 saying, "Headache and dizziness are forecast for the 16th. Start preparing now. We'll give you advice on what to do."

[0127] By checking this screen, female employee H can see that on September 16th, she is predicted to experience headaches and dizziness due to fluctuations in atmospheric pressure and hormone balance, which will cause her to feel unwell. Then, when female employee H selects the chat start button B11 on the generated AI 210, the screen of the user terminal 30h transitions to screen M12 (Figure 26), and she can start chatting with the generated AI 210.

[0128] Returning to Fig. 24, the explanation will be continued. Under the control of the chat control unit 13c, in response to the female employee H's instruction to start a chat with the generation AI 210, the support server 10 instructs the generation AI 210 to have a dialogue to narrow down the predicted cause candidates while having a friendly conversation ((3) in Fig. 24).

[0129] Fig. 27 is a diagram showing an example of a prompt set in the generation AI 210. In order to have the generation AI 210 execute such a dialogue, the support server 10 sets the prompt Pb shown in Fig. 27 in the generation AI 210.

[0130] Prompt Pb, for example, instructs the support server 10 to dig deeper into the current mood (stress), plans for the next few days, what would be troublesome if an illness occurs, and the countermeasures that have been taken so far, their effects, and impressions (box 210-1 in FIG. 24) while having a conversation based on the condition, cause, and time predicted by the support server 10. Prompt Pb instructs the support server 10 to select an appropriate countermeasure based on the predicted condition, cause, and time and the contents that have been dug up, and to propose x countermeasures in order of likelihood.

[0131] Then, the generation AI 210 narrows down the cause of female employee H's illness from the candidate causes through a conversation with female employee H. The generation AI 210 creates a list of candidate countermeasures for the narrowed down causes, for example, by referring to the countermeasure information 12j ((4) in FIG. 24). Then, in the support server 10, the countermeasure presentation unit 13j adds in-depth content to the prediction result by the analysis unit 13i and proposes the optimal countermeasure to female employee H ((5) in FIG. 24).

[0132] For example, the solution presentation unit 13j displays, on the user terminal 30h via chat, an example of self-care by the female employee H herself, or a guidance screen for visiting a medical institution, such as online medical treatment, as a solution ((5) in Figure 24).

[0133] Fig. 28 is a diagram showing an example of a chat in the user terminal 30h. As shown in Fig. 28, the generation AI 210 (AI) sends a message of support to female employee H and suggests self-care (messages K11 to K13). The generation AI 210 suggests going to bed earlier in accordance with female employee H's schedule for September 16th (message K13).

[0134] Next, female employee H inputs a question Q14 about what to do if she cannot sleep well. The generation AI 210 sends a message K15 suggesting what to eat and drink before going to bed to the user terminal 30h in a gentle tone of voice.

[0135] At the same time, the generation AI 210 suggests a recommended product for a drink that is good to drink before going to bed (message K16). Clicking on the product description section of this message K16 will take you to the e-commerce site of the partner. This allows female employee H to smoothly purchase the recommended product.

[0136] The generation AI 210 also suggests measures to regulate the autonomic nervous system (message K17) and introduces a column that explains the details. Clicking on "column" in this message K17 will take you to this column. This allows female employee H to learn more about measures to regulate the autonomic nervous system.

[0137] The generation AI 210 may refer to the chat history of other female employees and present one or more methods that other female employees with symptoms similar to those of female employee H have actually performed. Furthermore, if female employee H's symptoms are severe, the generation AI 210 may display a message on the user terminal 30h recommending that she visit a medical institution rather than self-care, and may direct her to an online medical consultation site.

[0138] In addition to presenting specific solutions, the solution presentation unit 13j may also display a message about health issues faced by female employees with careers similar to that of female employee H. The solution presentation unit 13j may also display a message about specific career advances for female employees based on menstrual symptoms. The solution presentation unit 13j may also display a message about life events (marriage, pregnancy, childbirth) of other female employees and how to advance their careers. By providing such messages to female employee H, the solution presentation unit 13j allows female employee H to know the situation of other female employees and to resolve her own doubts and anxieties.

[0139] In addition, the solution presentation unit 13j may display a curation site exclusively for female employees on the screen of the PC (user terminal 30i) used by female employee H, thereby providing female employee H with flexible options for choosing solutions to her health issues.

[0140] FIG. 29 is a diagram showing an example of a screen of the user terminal 30i. As shown in FIG. 29, screen M13 leads to answering a questionnaire about women's working styles. Then, when the female employee selects the chat start button A13 of the generated AI210, a chat screen C13 is displayed, and the female employee can start chatting with the generated AI210 ((1) in FIG. 29). By having a sympathetic conversation on the chat screen C13, the generated AI210 narrows down the possible causes of the predicted female employee's ill health.

[0141] The curation site allows employees to view websites and videos related to solutions to the causes of their ailments, and to transition to online medical consultation sites. By moving and clicking cursor K1, female employee H can view the desired website or video, or transition to the online medical consultation site, and consider specific solutions to her ailments.

[0142] [Processing Procedure] Next, a processing procedure of the support processing according to the embodiment will be described. Figures 30 to 32 are sequence diagrams showing the processing procedure of the support processing according to the embodiment.

[0143] As shown in FIG. 30, the support server 10 conducts a survey from the company terminal 20 (step S11). The survey information includes the survey questions and the survey responses from the female employees H to J. The support server 10 receives health management information (past quantitative information) indicating the past health condition of female employee H from the user terminal 30h used by female employee H (step S12).

[0144] For example, when the support server 10 receives a request to start a chat from the user terminal 30h (step S13), the support server 10 creates a prompt for the chat (for example, prompt Pa in FIG. 6) (step S14). Note that if the female employee H has had an interview with a public health nurse or the like, for example, via a web conference, the support server 10 acquires the dialogue history.

[0145] The support server 10 sets a prompt in the generation AI 210 of the generation AI server 200, inputs data such as female employee H's attributes, questionnaire results (questions and answers), past health management information, weather information, and past chat history (including dialogue history) (step S15), and starts a chat with the user terminal 30h (steps S16-1 and S16-2). At this time, the support server 10 uses the chatbot to dig deeper into at least the questionnaire results and collects female employee H's chat history (dialogue history) related to her issues and / or concerns (step S17). The support server 10 narrows down female employee H's issues and / or concerns via the chatbot. In the prompt, the support server 10 instructs the female employee to have a conversation that asks about health issues specific to women and concerns related to women's working styles. The support server 10 records this chat history as female employee H's current subjective information.

[0146] Similarly, for female employee I, the support server 10 receives health management information (past quantitative information) indicating the past health condition of female employee I from the user terminal 30i used by female employee I (step S18). When a chat start request is received from the user terminal 30i (step S19), the support server 10 performs the processes of steps S20 to S23, similar to steps S14 to S17, and collects the chat history (dialogue history) of female employee I regarding her issues and / or worries. For female employee J, the support server 10 performs the same processes as steps S12 to S17 and collects the chat history (dialogue history) of female employee J regarding her issues and / or worries.

[0147] As shown in FIG. 31, the support server 10 creates prompts for extracting issues (e.g., prompts P1 and P11) (step S31), and inputs data including the prompts, the survey results (questions and answers), and chat histories with female employees regarding the survey results to the generation AI 210 of the generation AI server 200 (step S32). The support server 10 acquires issue data including the issues and / or concerns of female employees that could not be collected through the survey from the generation AI 210 (step S33). The support server 10 performs steps S31 to S33 for each of female employees H, I, and J, and causes the extraction generation AI 210 to extract the issues and / or concerns of the female employees that could not be collected through the survey.

[0148] The support server 10 generates integrated data by integrating the assignment data of the female employees H, I, and J into one data (step S34).

[0149] The support server 10 creates prompts for classification (e.g., prompts P2, P21) (step S35). The support server 10 inputs the created prompts and integrated data into the generation AI 210 (step S36), causes the AI ​​210 to classify the issues and / or worries for each combination in the matrix diagram M1, and generates countermeasures for each issue and / or worry (step S37).

[0150] The support server 10 receives the classified issues and / or worries and their countermeasures from the generation AI 210 (step S38), creates a report (step S39), and presents it to the company terminal 20 of company A (step S40).

[0151] At company A, for example, a welfare officer provides services to female employees H, I, and J in accordance with the proposed measures based on the report (steps S41, S42).

[0152] For example, the support server 10 recommends services (e.g., PHR services for women (such as a menstrual management app F1, a pregnancy / infertility support app F2, and a health management app F3 installed on the user terminal 30)) and Femtech-related services (such as a Femcare recommendation service W1, a lifestyle improvement recommendation service W2, an online dispensing / prescription W3, and an online medical consultation / diagnosis W4) to the user terminals 30h, 30i, and 30j (steps S43, S45, and S47). The female employees H, I, and J install the applications for the recommended services on the user terminals 30h, 30i, and 30j and start using the services (steps S44, S46, and S48).

[0153] The support server 10 then provides support to each of the female employees H to J individually. First, as shown in FIG. 32, the support server 10 receives survey information from the female employees H to J from the company terminal 20 (step S51). The survey information includes the content of the survey questions and the responses to the survey by the female employees H to J. In FIG. 32, support for female employee H will be described as an example.

[0154] The support server 10 receives health management information (past quantitative information) indicating the past health condition of the female employee H from the user terminal 30h used by the female employee H (step S52).

[0155] The support server 10 refers to the past chat history (past subjective information) of the female employee H from the storage unit 12 (step S53).

[0156] The support server 10 estimates the causal relationship between each piece of information based on female employee H's past health management information and past chat history, and performs an analysis process to predict female employee H's state of illness, as well as possible causes and timing of the illness, based on the causal relationship (step S54).

[0157] The support server 10 transmits the predicted state of the female employee's ill health, and the possible causes and timing of the ill health, as analysis results, to the user terminal 30h (step S55), and causes the user terminal 30h to display them.

[0158] When the support server 10 receives a chat start request from the user terminal 30h (step S56), it creates a prompt (for example, prompt Pa) (step S57).

[0159] The support server 10 sets a prompt in the generation AI 210 of the generation AI server 200, inputs female employee H's attributes, questionnaire results (questions and answers), past health management information, weather information, and past chat history (step S58), and starts a chat with the user terminal 30h (steps S59-1 and S59-2). At this time, the support server 10 causes the generation AI 210 to engage in a dialogue regarding the results of implementing the proposed countermeasures and female employee H's subsequent challenges and / or concerns. For example, the support server 10 sets a prompt in the generation AI 210 that instructs the AI ​​210 to narrow down potential causes of her illness while listening to female employee H's concerns. In the prompt, the support server 10 instructs the AI ​​210 to engage in a conversation in which the AI ​​210 listens to female employee H's concerns about health challenges specific to women and concerns about women's working styles. The support server 10 records this chat history as female employee H's current subjective information.

[0160] The support server 10 narrows down the possible causes of female employee H's illness based on the current subjective information of female employee H acquired through chat, and sets a prompt (e.g., prompt Pb) in the generation AI 210 to instruct the generation AI 210 to present one or more solutions to the narrowed down causes. The support server 10 receives the one or more solutions presented by the generation AI 210 in response to this prompt from the generation AI server 200 (step S60-1) and transmits them to the user terminal 30h (step S60-2). The support server 10 also executes the processes of steps S52 to S60-2 on the user terminals 30i and 30j of other female employees I and J.

[0161] In addition, the support server 10 may have the generation AI 210 conduct a chat with the user terminal 30h to collect current subjective information of multiple female employees regarding the introduction of the measures or services in order to confirm the effectiveness of the introduction of the measures or services.

[0162] The support server 10 creates a prompt for extracting effects (step S61), and inputs the created prompt and data into the generation AI 210 to extract effects obtained by using the service provided by company A and the countermeasures presented by the support server 10, as well as dissatisfaction with the service, etc. (steps S62, S63). The input data includes female employee H's attributes, questionnaire questions and answers, past health management information, weather information, and past chat history, and the chat history of each of female employees H, I, and J obtained in steps S59-1 and S59-2.

[0163] Then, the support server 10 instructs the generation AI 210 to create a report summarizing the effects obtained by multiple female employees H, I, and J from using the services offered by company A and utilizing the countermeasures presented by the support server 10 regarding the introduction of measures and services, as well as their dissatisfaction with the services, etc. (step S64).

[0164] In response to this, the generation AI 210 creates a report (step S65), and the generation AI server 200 sends the report to the support server 10 (step S66-1). The support server 10 sends the report created by the generation AI 210 to the company terminal 20 (step S66-2), and displays it (step S67).

[0165] [Effects of the embodiment] In this way, the support server 10 according to the embodiment can accurately extract hidden issues and concerns of female employees other than those found in the questionnaire by extracting and classifying the issues and / or concerns of female employees in two stages.The support server 10 then provides appropriate solutions to the extracted issues and / or concerns to, for example, the welfare officer of company A where the female employee works.

[0166] In particular, it is now possible to accurately identify issues that were difficult to identify with previous systems and clearly present them along with proposed solutions.

[0167] By providing companies with information on the hidden issues and / or concerns of female employees and proposed solutions, the support server 10 can smoothly support companies in introducing services and planning measures to create workplaces that are easy for female employees to work in.

[0168] [Application example 1] A description will be given of other application examples of the support server 10. For example, the support server 10 can also be applied to online prescriptions (including online consultations and preliminary medical interviews).

[0169] In this case, the input data for the generation AI 210 is the user's PHR (Personal Health Record) data and / or medication notebook data (medication history) (quantitative information of the user) and the dialogue history between the generation AI 210 and the user (subjective information of the user). The first axis of the matrix diagram represents the severity of the illness, and the second axis represents the purpose of taking the medication. The support server 10 outputs appropriate supplements and over-the-counter medicines as countermeasures. For example, this can be used for AI consultations at online pharmacies.

[0170] [Application example 2] It is also believed that this system can be applied to support rehabilitation and health guidance in the medical field. In this case, the input data for the generation AI 210 is the user's electronic medical record (quantitative information about the user) and a medical interview or online consultation history by a doctor or other medical professional (subjective information about the user). The first axis of the matrix diagram represents severity, and the second axis represents health awareness (literacy). The support server 10 outputs the optimal rehabilitation and health initiatives for each individual as countermeasures. For example, this system can be used for rehabilitation and guidance on preventing lifestyle-related diseases.

[0171] [Application example 3] It is also believed that this system can be applied to support the creation of care plans for nursing care. In this case, the input data for the generation AI 210 is the user's doctor's opinion and level of care required (quantitative information of the user), and the conversation history between the user and the care manager or family (subjective information of the user). The first axis of the matrix diagram represents the urgency and / or severity, and the second axis represents the nursing care fee points. The support server 10 outputs the contents of the optimal nursing care service as a countermeasure proposal. For example, it can be used as an auxiliary tool for creating a care plan.

[0172] [Application example 4] It is also considered applicable to learning support for students at cram schools and other institutions. In this case, the input data for the generation AI 210 is the user's student's report card, curriculum guidelines, and past question trends (e.g., past exam questions) at the desired school (quantitative information about the user), as well as the contents of interviews between the instructor (including chats with the generation AI 210) and the student (regarding the learning situation by subject / field) (subjective information about the user). The first axis of the matrix diagram represents the degree of achievement toward the goal and the difference between the current situation and the goal (e.g., deviation score). The first axis is plotted by subject / field. The second axis of the matrix diagram represents study time (amount), plotted by subject / field. The support server 10 outputs the necessary learning content for each subject / field as a countermeasure. For example, this can be used to develop appropriate learning methods for individuals.

[0173] [Application example 5] It is also believed that this system can be applied to support insurance consultations. In this case, the input data for the generation AI 210 is a questionnaire (quantitative information about the user) and a conversation history (about insurance and life goals, etc.) (subjective information about the user) with an insurance concierge (including chats with the generation AI 210). The first axis of the matrix diagram is income and affordable costs, and the second axis is the goal (whether to prioritize insurance or asset formation). The support server 10 outputs appropriate insurance products as countermeasures. For example, this system can be used for AI online insurance consultations.

[0174] [Application Example 6] It is also believed to be applicable to support job changes and employment. In this case, the input data for the generation AI 210 is job-change website registration information (career history, annual income, etc.) (quantitative information of the user), a resume, and the contents of an interview with an AI agent (work orientation, experience, skills) (subjective information of the user). The first axis of the matrix diagram represents work orientation (whether to prioritize work or life), and the second axis represents experience and skill compatibility. The support server 10 proposes the most suitable job destination (industry, industry, occupation, etc.) as a countermeasure. For example, it can be used in AI pre-interviews before an agent interview.

[0175] [Application Example 7] It is also believed that this system can be applied to childcare support in nurseries. In this case, the input data for the generation AI 210 is the child's height and weight, amount of food eaten, and amount of exercise (quantitative information of the user), as well as the child's resume and the contents of interviews between the childcare worker (including chats with the generation AI 210) and the child (subjective information of the user). The first axis of the matrix diagram is motor function, and the second axis is language function. The support server 10 proposes the child's characteristics as a countermeasure. For example, this can be used to formulate educational guidance policies and collaborate with welfare organizations and local governments.

[0176] In this way, the support server 10 can extract hidden problems of the user and provide countermeasures for the extracted problems, and therefore can be utilized for a wide range of user support.

[0177] [System configuration of the embodiment] The support server 10 and various terminals are conceptual functional entities and do not necessarily need to be physically configured as shown in the figure. In other words, the specific form of distribution and integration of the functions of the support server 10 and various terminals is not limited to that shown in the figure, and all or part of them can be functionally or physically distributed or integrated in any unit depending on various loads, usage conditions, etc.

[0178] Furthermore, all or any part of the processes performed in the support server 10 and various terminals may be realized by a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), and a program analyzed and executed by the CPU and GPU. Furthermore, each process performed in the support server 10 and various terminals may be realized as hardware using wired logic.

[0179] Furthermore, among the processes described in the embodiments, all or part of the processes described as being performed automatically can be performed manually. Alternatively, all or part of the processes described as being performed manually can be performed automatically using a known method. In addition, the processing procedures, control procedures, specific names, and information including various data and parameters described above and illustrated can be changed as appropriate unless otherwise specified.

[0180] [program] 33 is a diagram showing an example of a computer in which the support server 10 is realized by executing a program. The computer 1000 has, for example, a memory 1010 and a CPU 1020. The computer 1000 also has a hard disk drive interface 1030, a disk drive interface 1040, a serial port interface 1050, a video adapter 1060, and a network interface 1070. These components are connected by a bus 1080.

[0181] The memory 1010 includes a ROM 1011 and a RAM 1012. The ROM 1011 stores a boot program such as a BIOS (Basic Input Output System). The hard disk drive interface 1030 is connected to a hard disk drive 1090. The disk drive interface 1040 is connected to a disk drive 1100. A removable storage medium such as a magnetic disk or optical disk is inserted into the disk drive 1100. The serial port interface 1050 is connected to a mouse 1110 and a keyboard 1120, for example. The video adapter 1060 is connected to a display 1130, for example.

[0182] The hard disk drive 1090 stores, for example, an OS (Operating System) 1091, an application program 1092, a program module 1093, and program data 1094. That is, a program that defines each process of the support server 10 is implemented as a program module 1093 in which code executable by the computer 1000 is written. The program module 1093 is stored, for example, in the hard disk drive 1090. For example, a program module 1093 for executing the same process as the functional configuration of the support server 10 is stored in the hard disk drive 1090. The hard disk drive 1090 may be replaced by an SSD (Solid State Drive).

[0183] Furthermore, setting data used in the processing of the above-described embodiment is stored as program data 1094, for example, in memory 1010 or hard disk drive 1090. Then, CPU 1020 reads program module 1093 and program data 1094 stored in memory 1010 or hard disk drive 1090 into RAM 1012 as necessary and executes them.

[0184] The program module 1093 and program data 1094 are not limited to being stored in the hard disk drive 1090, but may also be stored in, for example, a removable storage medium and read by the CPU 1020 via the disk drive 1100 or the like. Alternatively, the program module 1093 and program data 1094 may be stored in another computer connected via a network (such as a local area network (LAN) or a wide area network (WAN)). The program module 1093 and program data 1094 may then be read by the CPU 1020 from the other computer via the network interface 1070.

[0185] Although the present invention has been described above as an embodiment, the present invention is not limited to the descriptions and drawings that form part of the disclosure of the present invention. In other words, other embodiments, examples, and operational techniques that can be made by those skilled in the art based on the present invention are all included in the scope of the present invention. [Explanation of symbols]

[0186] 10 Support Server 11 Communications Department 12 Storage section 13 Control Unit 20 Corporate terminals 30, 30h~30i User terminal 200 Generation AI Server 210 Generation AI

Claims

1. a first acquisition unit that acquires quantitative information of a user; a second acquisition unit that acquires a dialogue history with the user regarding the user's past quantitative information as subjective information of the user; a first extraction unit that causes a generative AI (Artificial Intelligence) model to extract, for each user, a problem and / or a worry of the user based on the user's quantitative information and the user's subjective information; an integration unit that generates integrated data by integrating the issues and / or worries of a plurality of users into one data; a second extraction unit that causes the generative AI model to classify the issues and / or concerns of the plurality of users for each combination of the degree of each first item related to the issues and / or concerns of the users and the degree of each second item related to the issues and / or concerns of the users from the integrated data, and to output countermeasures for the issues and / or concerns of the users for at least one of the combinations; a presentation unit that presents the user's problems and / or worries and countermeasures for the problems to a source that requests the countermeasures; A processing device comprising:

2. The processing device according to claim 1, characterized in that the second acquisition unit further includes a first acquisition unit that acquires the user's dialogue history, in which a generative AI model for dialogue that generates text based on information input by the user and engages in dialogue by digging into at least the user's subjective information and having dialogue with the user regarding the user's issues and / or concerns.

3. The processing device according to claim 2, characterized in that the second acquisition unit instructs the generative AI model for dialogue to conduct a dialogue that narrows down the user's issues while listening to the user's concerns, to generate text that is sympathetic to the user and includes words that show empathy, and to change the timing of responses depending on the question.

4. The processing device according to claim 1 , wherein the presentation unit presents the proposed measures to the user's terminal in a state capable of linking with a provider of a service corresponding to the proposed measures for the user's issues and / or concerns.

5. The second acquisition unit, after the countermeasure plan is implemented, uses the generative AI model to acquire a dialogue history of the user regarding a result after the implementation of the countermeasure plan and the user's subsequent issues and / or worries; a third extraction unit that causes the generation AI model to extract an effect after the implementation of the countermeasure plan based on the user's dialogue history and / or health management information after the implementation of the countermeasure plan, and outputs the extracted effect to a source that requests the countermeasure plan.

2. The processing apparatus according to claim 1, further comprising:

6. The quantitative information of the user is a questionnaire about health issues specific to women and working styles of women to the user, The subjective information of the user is a response by the female user to a questionnaire regarding the health issues specific to women and the working styles of women, and / or a dialogue history of the user regarding the health issues specific to women and the working styles of women, the first extraction unit causes the generative AI model to extract, for each user, issues and / or concerns of the user that could not be collected through the questionnaire, based on the questionnaire, the user's answers to the questionnaire, and / or the user's dialogue history; The processing device according to claim 1, characterized in that the second extraction unit causes the generative AI model to classify the issues of the multiple users for each combination of each degree of urgency and each degree of ease of addressing, and to output countermeasures for the issues for at least one of the combinations for each of the combinations.

7. a first extraction unit that causes a generative AI (Artificial Intelligence) model to extract, for each user, issues and / or concerns of the user based on the user's quantitative information and the user's subjective information, which is a dialogue history with the user regarding the user's past quantitative information; an integration unit that generates integrated data by integrating the issues and / or worries of a plurality of users into one data; a second extraction unit that causes the generative AI model to classify the issues and / or concerns of the plurality of users from the integrated data for each combination of a degree of each first item related to the issues and / or concerns of the users and a degree of each second item related to the issues and / or concerns of the users; A processing device comprising:

8. A processing method executed by a processing device, a first acquisition step of acquiring quantitative information of a user; a second acquisition step of acquiring a dialogue history with the user regarding the user's past quantitative information as subjective information of the user; a first extraction step of extracting, for each user, the user's issues and / or concerns based on the user's quantitative information and the user's subjective information using a generative AI (Artificial Intelligence) model; an integration step of generating integrated data in which the issues and / or worries of the plurality of users are integrated into one data; a second extraction step of causing the generative AI model to classify the issues and / or concerns of the plurality of users for each combination of the degree of each first item related to the issues and / or concerns of the users and the degree of each second item related to the issues and / or concerns of the users from the integrated data, and outputting countermeasures for the issues and / or concerns of the users for at least one of the combinations; a presentation step of presenting the user's problems and / or worries and countermeasures for the problems to a source of a request for the countermeasures; A processing method comprising:

9. A processing method executed by a processing device, a first extraction step of extracting, for each user, the user's issues and / or concerns using a generative AI (Artificial Intelligence) model based on the user's quantitative information and the user's subjective information, which is a dialogue history with the user regarding the user's past quantitative information; an integration step of generating integrated data in which the issues and / or worries of the plurality of users are integrated into one data; a second extraction step of causing the generative AI model to classify the issues and / or concerns of the plurality of users from the integrated data for each combination of the degree of each first item related to the issues and / or concerns of the users and the degree of each second item related to the issues and / or concerns of the users; A processing method comprising:

10. a first acquisition step of acquiring quantitative information of a user; a second acquisition step of acquiring a dialogue history with the user regarding the user's past quantitative information as subjective information of the user; a first extraction step of extracting, for each user, the user's issues and / or worries using a generative AI (Artificial Intelligence) model based on the user's quantitative information and the user's subjective information; an integration step of generating integrated data in which the issues and / or worries of the plurality of users are integrated into one data; a second extraction step of causing the generative AI model to classify the issues and / or concerns of the plurality of users for each combination of the degree of each first item related to the issues and / or concerns of the users and the degree of each second item related to the issues and / or concerns of the users from the integrated data, and outputting countermeasures for the issues and / or concerns of the users for at least one or more of the combinations; a presentation step of presenting the user's problems and / or worries and countermeasures for the problems to a source of a request for the countermeasures; A processing program that causes a computer to execute the above.

11. a first extraction step of extracting, for each user, the user's issues and / or concerns using a generative AI (Artificial Intelligence) model based on the user's quantitative information and the user's subjective information, which is a dialogue history with the user regarding the user's past quantitative information; an integration step of generating integrated data in which the issues and / or worries of the plurality of users are integrated into one data; a second extraction step of causing the generative AI model to classify the issues and / or concerns of the plurality of users from the integrated data for each combination of the degree of each first item related to the issues and / or concerns of the users and the degree of each second item related to the issues and / or concerns of the users; A processing program that causes a computer to execute the above.

Images