JP7708449B2 - Answer generation system, answer generation device, and program - Google Patents

Description

Application of Article 30, Paragraph 2 of the Patent Act (1) Website listing ● June 7, 2023, company website https://estoma.world/ https://estoma.world/index.html https://estoma.world/esg-starter.html https://estoma.world/esg-pro.html https://estoma.world/demo.html https://estoma.world/basic-demo.html https://estoma.world/standard-demo.html html https://estoma. world/enterprise-demo. html https://estoma. world/human_capital. html https://estoma. world/tcfd. html https://estoma. world/co2. html https://estoma. world/rating. html https://estoma. world/supplier. html ● June 7, 2020, PR TIMES https://prtimes. jp/main/html/rd/p/000000009.000103789. html

The present invention relates to an answer generation system, an answer generation device, and a program.

In recent years, attention has been focused worldwide on the environment, society, and governance, and companies are being asked to take ESG into consideration, an acronym made up of the first letters of these terms. In addition, third-party organizations (hereafter referred to as "assessment organizations") evaluate companies' ESG-conscious efforts, and an indicator for this is the ESG score.

For example, the ESG score evaluates a company's environmental efforts, such as reducing carbon dioxide (CO2) emissions, improving water pollution caused by wastewater, and addressing environmental issues such as microplastics in the ocean, as well as the use of renewable energy and preserving biodiversity. The ESG score also evaluates a company's efforts to improve the work environment for employees and address social human rights issues, such as fair working conditions and human rights measures in the workplace such as gender equality, diversity, work-life balance, child labor issues, and contributions to the local community. Furthermore, the ESG score evaluates a company's management system, such as the creation of mechanisms to avoid scandals that directly lead to a decline in performance, information disclosure for risk management, compliance with laws and regulations, and a high level of awareness of capital efficiency.

In addition, some investors often compare and consider investment targets based on the ESG scores given by rating agencies. For this reason, companies considering fundraising make an effort to respond to questions from rating agencies as much as possible in order to improve their ESG scores. Various technologies have been proposed as systems for evaluating and improving corporate activities related to ESG (see, for example, Patent Document 1 and Patent Document 2).

JP 2023-025783 A JP 2023-095083 A

Incidentally, there are multiple rating agencies around the world that calculate ESG scores, such as DJSI (The Dow Jones Sustainability Indices) and newspapers, and responding to each of these rating agencies' requests for information disclosure through questionnaires is extremely difficult, taking time, effort, and cost. Moreover, to the applicant's knowledge, no system has yet been proposed to reduce the effort required for responding to these requests.

The present invention aims to provide technology that enables efficient answering of questions related to ESG.

For this purpose, the technology disclosed in this specification is an answer generation system having one or more computers, the computer comprising: a receiving means for receiving questions related to a user's evaluation in an ESG; an acquisition means for acquiring disclosure information previously disclosed by the user; and an answer generation means for automatically generating an answer to the question using the disclosure information acquired by the acquisition means.

Here, the acquisition means may acquire, as the disclosure information, history information that associates questions previously received by the user with answers to those questions.
In addition, the answer generating means may use a plurality of pieces of the history information of the user as the disclosure information when generating an answer to one of the questions.
In addition, the history information may include information on questions and answers to the questions from one evaluation agency that evaluates ESG, and information on questions and answers to the questions from another evaluation agency different from the one evaluation agency.
The system may further comprise an association means for associating the history information with each of the questions, and the answer generating means may generate the answer using the history information associated by the association means.
The acquisition means may acquire IR information of the user as the disclosure information.
The system may further comprise a list display means for displaying a list of the plurality of questions received by the receiving means on a screen.
The list display means may display, in the list, evaluation information relating to an evaluation value in an ESG for each question.
The list display means may display, for each question, disclosure status information indicating whether or not the user has answered the question, in the list.
The list display means may display, for each question in the list, answer non-generation information indicating that the question is one for which the answer generating means cannot generate an answer.
In addition, when the user answers the question or when the user corrects the answer generated by the answer generating means, reference information that the user can refer to may be displayed individually for each question.
The reference information may also include explanation information that explains the question.
In addition, the reference information may include the disclosure information for past questions related to the commentary information.

Furthermore, for this purpose, the technology disclosed in this specification is an answer generation device that is characterized by having a receiving means for receiving questions related to a user's evaluation in an ESG, an acquiring means for acquiring disclosure information previously disclosed by the user, and a transmitting means for transmitting the questions and the disclosure information to an answer generating means that automatically generates answers to the questions.

For this purpose, the technology disclosed in this specification is a program that causes a computer to execute the following functions: accepting questions related to a user's evaluation in an ESG; acquiring disclosure information previously disclosed by the user; and transmitting the questions and the disclosure information to an answer generation means that automatically generates answers to the questions.

The present invention allows for efficient answering of questions regarding ESG.

FIG. 1 is a schematic diagram showing a configuration of an answer generation system according to an embodiment of the present invention. FIG. 2 is a block diagram showing the hardware configuration of devices included in the answer generation system according to the present embodiment. FIG. 2 is a functional block diagram showing an example of a functional configuration of the device. FIG. 2 is a sequence diagram illustrating a process flow in the answer generation system according to the embodiment. 10 is a flowchart showing a basic operation of an answer generating process in the device. 13 is a schematic diagram showing an example of the configuration of a question list display screen displayed on a user terminal. FIG. 10 is a schematic diagram showing an example of the configuration of a question display screen displayed on a user terminal. FIG. 13 is a schematic diagram showing a configuration example of an answer generation screen displayed on a user terminal. FIG. 11 is a schematic diagram showing an example of a configuration in which framework information is displayed on a question display screen displayed on a user terminal. FIG. FIG. 11 is a sequence diagram illustrating the flow of another process in the answer generation system according to the embodiment. FIG. 11 is a sequence diagram illustrating the flow of another process in the answer generation system according to the embodiment. 10 is a flowchart showing another operation of the answer generating process in the device. 11 is a schematic diagram showing an example of the configuration of a reference information display screen displayed on a user terminal. FIG. 13 is a schematic diagram showing an example of the configuration of a related user public information display screen displayed on a user terminal. FIG.

As an example of this embodiment, the following describes how a user who receives a question about ESG (Environment, Society, and Governance) from a rating agency uses this system to automatically generate a response. Specifically, the following describes, with reference to the drawings, an example in which a user who receives a question about CDP, an international environmental non-profit organization (NGO), uses generative AI in this system to automatically generate a response.

The system of this embodiment is adapted to automatically generate answers to questions from various evaluation agencies by using generative AI for natural language, such as ChatGPT. For example, to answer questions from evaluation agencies about ESG, multiple pieces of information about ESG that have already been disclosed by the user are used as a data set. Then, questions from various evaluation agencies are answered using the user's public information (for example, an example of disclosed information), so that correct answers can be generated. And, in the system of this embodiment, by using generative AI, such as ChatGPT, it is possible to generate answers efficiently, reducing time, effort, costs, etc.

The embodiments described below are preferred examples of the present embodiment, and therefore have various technical limitations. However, the scope of the present invention is not limited to these embodiments unless otherwise specified in the following description.

<System Configuration>
FIG. 1 is a diagram showing an example of the configuration of an answer generating system according to the present embodiment.

As shown in FIG. 1, the answer generation system in this embodiment is configured by connecting an answer generation device 10 that automatically generates answers to questions from an evaluation agency and a user terminal 30 used by a user who is requested to answer a question from the evaluation agency via a predetermined network N such as the Internet.

The answer generating device 10 shown in FIG. 1 is a device that stores questions from various rating agencies and information related to ESG disclosed by users, and generates answers to those questions. This answer generating device 10 has a calculation processing function and a communication function, and functions as a server in the answer generating system. In this embodiment, the answer generating device 10 is realized by an electronic device such as a personal computer. Therefore, in the following description, the answer generating device 10 may be referred to as a "server."

The user terminal 30 is a computer operated by a user who is seeking answers to questions from the evaluation agency, and is a device with processing and communication functions. This user terminal 30 is realized, for example, by a desktop or laptop computer, a tablet computer, a smartphone, etc.

Although the network N is configured by the Internet, it is not particularly limited and may be a communication network such as a LAN (Local Area Network) or a WAN (Wide Area Network). The communication lines of the network N may be wired or wireless, and may be a combination of both.

In the example of FIG. 1, only one user terminal 30 is shown, but the number is not particularly limited.
Also, although only one answer generation device 10 is shown, the answer generation device 10 is not limited to a device that operates alone, but may be a distributed server system or a cloud server that cooperates by communicating via a network N. That is, the system of this embodiment can be configured by one or more computers.

<Hardware Configuration>
FIG. 2 is a block diagram showing a hardware configuration of the answer generation device 10 according to the present embodiment.
The answer generation device 10 includes a CPU (Central Processing Unit) 11, a ROM (Read Only Memory) 12, a RAM (Random Access Memory) 13, a bus 14, an input/output interface 15, an output unit 16, an input unit 17, a memory unit 18, and a communication unit 19.

The CPU 11 executes various processes according to programs stored in the ROM 12 or programs loaded from the storage unit 18 to the RAM 13. In other words, the CPU 11 is a processor that executes various processes according to programs and controls the operation of the entire server 1. The programs are, for example, an operating system or firmware for operating each part of the server 1, or application software for implementing the functional blocks described below.

The RAM 13 also stores data and the like necessary for the CPU 11 to execute various processes.
The CPU 11, the ROM 12, and the RAM 13 are interconnected via a bus 14. To this bus 14, an input/output interface 15 is also connected.
An output unit 16 , an input unit 17 , a storage unit 18 , and a communication unit 19 are connected to the input/output interface 15 .

The output unit 16 is composed of a display, a speaker, etc., and outputs various information as images and sounds.
The input unit 17 is composed of a keyboard, a mouse, a microphone, a camera, etc., and inputs various information in response to a received instruction operation. Note that the output unit 16 and the input unit 17 may be integrated into one unit, for example, by a touch panel.

The storage unit 18 stores various data necessary for the server 1 to execute information processing, and is configured, for example, with a hard disk, a DRAM (Dynamic Random Access Memory), and the like.
The communication unit 19 communicates with other devices via a network N including the Internet.

The answer generating device 10 may be provided with a drive to which a removable medium such as a magnetic disk, an optical disk, a magneto-optical disk, or a semiconductor memory is appropriately attached. The drive allows a program read from the removable medium to be installed in the storage unit 18.
Further, the removable media can store various data stored in the storage unit 18 in the same manner as the storage unit 18 .

<Answer generation device 10>
FIG. 3 is a block diagram illustrating an example of the configuration of the answer generating device 10 functioning as a server in this embodiment.
The answer generation device 10 (server) includes an answer generation unit 110 that generates answers, an answer assistance unit 120 that assists in answer generation, a question list generation unit 130 that generates a list of questions, and a question individual display generation unit 140 that displays each question individually. Furthermore, the answer generation device 10 has a reference information display unit 150 that displays reference information, a user public information display unit 160 that displays user public information, and an answer acceptance unit 170 that accepts answers, as well as a question information DB 210 that manages question information, a user public information DB 220 that manages user public information, and a reference information DB 230 that manages reference information.

The answer generation unit 110 is a means having a function of automatically generating answers to questions by using the user public information stored in the user public information DB 220. Then, the answer generation unit 110 transmits the generated answer information to the user terminal 30.
For example, a system that performs natural language processing using large-scale language models (LLM) can be applied to the answer generation unit 110 of this embodiment. A large-scale language model is a natural language processing model that is machine-learned using a large amount of text data. A large-scale language model receives a sentence as input and outputs the sentence. When a large-scale language model is applied to a system for question-and-answering, when a question sentence is input to the large-scale language model, an answer sentence is output from the large-scale language model.

The answer generation unit 110 automatically generates answers using the question answer history information stored in the question answer history information DB 222 as one piece of user public information. That is, in the present embodiment, the answer generation unit 110 automatically generates answers to questions from evaluation institutions in relation to ESG questions using question answer history information including questions and answers about ESG that the user has received in the past from the same or different evaluation institutions as user public information.

In this way, when generating answers to questions from an evaluation agency, answers can be generated more efficiently by using question answer history information, which associates and stores questions that the user has received in the past from the same evaluation agency and their answers, as user public information. Answers can also be generated more efficiently by using question answer history information (for example, an example of history information), which associates and stores similar questions that the user has received in the past from different evaluation agencies and their answers, as user public information.

The answer generation unit 110 also automatically generates an answer using multiple questions and their answers as question answer history information. That is, the answer generation unit 110 automatically generates an answer to one question using multiple pieces of question answer history information. For example, when generating an answer to a question, rather than using only one answer to one question, multiple sets of questions and their answers, such as multiple questions and their answers from the same evaluation agency or questions and their answers from a different evaluation agency, are used.

In this way, instead of using a single question and its answer previously received from the same or a different rating agency, a user can use multiple questions and their answers in response to a single question from a rating agency, thereby generating a more accurate answer that takes into account multiple perspectives.

The answer generation unit 110 also automatically generates answers using the user's IR information stored in the IR information DB 224 (described later) as one piece of user public information. Examples of IR information include securities reports, environmental reports, sustainability reports, and integrated reports. In this way, the answer generation unit 110 of this embodiment generates answers more efficiently by using the user's IR information as user public information when generating answers to questions from the evaluation institution.

The answer generation unit 110 also automatically generates answers using the question answer history information and/or IR information associated by the answer assistance unit 120. In other words, by previously associating the question answer history information and IR information used to generate answers in the answer generation unit 110, answers can be efficiently generated to questions from the evaluation agency.

The answer assistance unit 120 is a means having a function of associating and managing the question answer history information stored in the user public information DB 220 with each question stored in the question information DB 210. Specifically, the questions and their answers included in the question answer history information stored in the question answer history information DB 222 (described later) as one piece of public information are associated with each question stored in the question information DB 210 and managed.

In this case, the question and answer included in the question answer history information associated with each question may be multiple questions and their answers, rather than a single question and its answer as described above.

In other words, when using one question and its answer for a given question, for example, by associating "Question 1.1a-1 of rating agency Q and its answer (Q1.1a-1 = A1.1a-1) for question X," the question identified by Q1.1a-1 and the answer identified by A1.1a-1 will be used in answering question X, and question 1.1a-1 of rating agency Q and its answer can be displayed in association with question X, or an answer to question X can be automatically generated using this question and answer.

In addition, when multiple questions and their answers are used for a given question, for example, by associating "Question X with question 1.1a-1 and its answer (Q1.1a-1 = A1.1a-1), question 1.1a-2 and its answer (R1.1a-2 = A1.1a-2), and question 1.1a-3 and its answer (Z1.1a-3 = A1.1a-3) from rating agency Q and rating agency Z, respectively," in the task of answering question X, the question identified by Q1.1a-1 and the answer identified by A1.1a-1, Three questions and their answers are used: the question identified in R1.1a-2 and the answer identified in A1.1a-2, and the question identified in Z1.1a-3 and the answer identified in A1.1a-3. In association with question X, question 1.1a-2 and its answer from rating agency R, question 1.1a-2 and its answer from rating agency R, and question 1.1a-3 and its answer from rating agency Z can be displayed, or an answer to question X can be automatically generated using these three questions and their answers.

In this way, by previously associating questions from an evaluation agency with appropriate user public information to be used in generating answers (for example, question answer history information that stores a question previously received by the user from the same evaluation agency or a different evaluation agency in association with the answer), it is possible to avoid mistakes in the selection of information to be used, improve the accuracy of the answers, and generate answers efficiently.

The answer assistance unit 120 is also a means having a function of associating and managing, for each question stored in the question information DB 210, the IR information stored in the IR information DB 224 together with the question answer history information stored in the user public information DB 220 or in place of the question answer history information.

In other words, when managing IR information by associating it with the question, for example, if you make an association such as "For question X, use item AAA from the securities report as IR information," item AAA from the securities report will be used when answering question X, and item AAA from the securities report can be displayed in association with question X, or an answer to question X can be automatically generated using item AAA from the securities report.

In this way, by linking in advance IR information in addition to question answer history information as user public information used when generating answers to questions from evaluation institutions, it is possible to improve the accuracy of answers using more appropriate user public information and generate answers efficiently.

The question list generating unit 130 is a means having a function of generating question display information for displaying the question information stored in the question information DB 210 in a list on the user terminal 30 , and presents the generated question display information to the user terminal 30 .
In this way, by displaying a list of questions from the rating agency, the user can easily grasp the questions at a glance and can efficiently and appropriately answer them.

When scoring information indicating an evaluation value to be assigned to an answer to a question stored in the question information DB 210 is stored, the question list generation unit 130 generates question display information that displays the scoring information associated with the question (e.g., an example of evaluation information) in a list together with the question information.
In this way, by displaying the points allocated to the answers in association with the questions from the listed rating agencies, the evaluation value for the answer can be visually grasped, and in the case of questions about ESG, it is possible to increase motivation to answer in order to improve the ESG score, and by prioritizing the order of answers, the answering process can be carried out efficiently and appropriately.

When disclosure status information indicating whether or not a question has been answered is stored for a question stored in the question information DB 210, the question list generation unit 130 generates question display information that displays the disclosure status information associated with the question in a list together with the question information in a visually recognizable manner.
In this way, by displaying the disclosure status in association with the questions from the evaluation organizations displayed in a list, the presence or absence of an answer can be visually and easily grasped at a glance, and answering work can be performed efficiently and appropriately without missing an answer. Also, by displaying only the words "answered" or "not answered" as the disclosure status information and changing the color and size of the text to indicate "answered" or "not answered," the disclosure status information can be visually grasped even more easily.

When answer non-generation information indicating that the answer generation unit 110 cannot automatically generate an answer is stored for a question stored in the question information DB 210, the question list generation unit 130 generates question display information that displays the answer non-generation information associated with the question in a list together with the question information in a visually recognizable state. In other words, the question list generation unit 130 uses the answer non-generation information to display in the list that the question is one for which the answer generation unit 110 cannot generate an answer.
In this way, by displaying the answer non-generation information in association with the questions from the listed rating agencies, the fact that an answer could not be automatically generated can be visually and easily understood at a glance, and answering work can be performed efficiently and appropriately without missing any answers. This answer non-generation information may be characters or may be figures or symbols such as stars or asterisks.

The individual question display generating unit 140 is a means having a function of generating individual question display information for individually displaying questions stored in the question information DB 210 on the user terminal 30, and presents the generated individual question display information to the user terminal 30.

Furthermore, when framework information related to a question is stored for a question stored in the question information DB 210, the question individual display generating unit 140 generates individual question display information including predetermined framework information associated with the question stored in the reference information DB 230, and presents the generated individual question display information on the user terminal 30. This framework information explains the gist of the question using explanations from other evaluation organizations so as to prevent errors in the answer.
In this way, by displaying framework information disclosed by other evaluation institutions for reference for each question from an evaluation institution, it is possible to further deepen understanding of the intent of the question and ensure that answers are error-free.

The reference information display unit 150 is a means having a function of accepting an instruction to display specific reference information associated with a question stored in the reference information DB 230 when reference information related to the question (e.g., an example of reference information) is stored for the question stored in the question information DB 210, and displaying the information on the user terminal 30 together with the individual question display information displayed on the user terminal 30. In other words, the reference information display unit 150 is a means having a function of accepting an instruction to display information that helps understand the question, and displaying the information on the user terminal 30 together with the individual question display information displayed on the user terminal 30. An example of this reference information can be term explanation information that explains a term (e.g., an example of explanation information).

In this way, by displaying reference information for each question from the evaluation agency, the user can refer to the reference information, which makes it easier for the user to understand the question, for example. The user is then assisted in creating their own answer or correcting an automatically generated answer, enabling them to provide an appropriate answer.

The user public information display unit 160 is a means having a function of receiving an instruction to display user public information (e.g., an example of disclosure information) for past questions related to the framework information, and displaying it on the user terminal 30 together with the individual question display information displayed on the user terminal 30. In other words, user public information for past questions related to the framework information is, for example, questions from other evaluation institutions that are similar to the question in question and the user's answer to that question, and the user's public information related to the framework information can be displayed by using question answer history information that associates similar questions received by the user in the past and the answers to those questions and stores them in the user public information DB 220 (question answer history information DB 222 described later).

In this way, by providing, for example, a selection button that displays related public information to the user for each piece of framework information displayed in response to a question from the evaluation agency, it is possible to further deepen understanding of the question.
In addition, the user public information may display similar questions and their answers that the user has received in the past, as well as the full text of user public information regarding the ESG related to the question (disclosure information in the ESG).

The answer receiving unit 170 is a means for registering answer information automatically generated by the answer generating unit 110 or answer information created by the user in the question information DB 210 in association with the question.

The question information DB 210 is a means for storing questions from the evaluation institution. The question information DB 210 stores the latest question information.
The question information DB 210 also serves as a means for storing questions from various rating agencies for each rating agency. In other words, the question information DB 210 does not store questions from a single rating agency, but stores questions from multiple rating agencies.

Before disclosure to the evaluation agency (while the answers are being provided), questions and their answers, including the answers to each question, are stored in this question information DB210 in association with each other. However, after disclosure to the evaluation agency, the newly disclosed questions become question information stored in the question information DB210, and previous questions and their answers become question answer history information stored in the question answer history information DB222 as user public information.

The question information DB 210 also stores, for each question, point information indicating an evaluation value to be assigned to an answer to the question in association with the question.
Furthermore, the question information DB 210 stores disclosure status information indicating whether or not a question has been answered in association with each question.
Furthermore, the question information DB 210 stores answer non-generation information indicating that the answer generator 110 was unable to automatically generate an answer, in association with each question.

The question information DB 210 also stores, for each question, association information with reference information related to the question. In other words, the question information DB 210 stores, for each question, association information that enables extraction of reference information stored in the reference information DB 230. The association of reference information with this question may be performed on a program rather than on a database (DB).

Furthermore, the question information DB210 stores association information with framework information related to the question (e.g., an example of commentary information) for each question. In other words, the question information DB210 stores association information for each question that enables extraction of framework information stored in the reference information DB230. Furthermore, the association with framework information related to a question may also be performed on a program rather than on a database (DB).

The user public information DB220 is a means for storing the user public information disclosed by the user for each user. In this embodiment, this user public information DB220 has a question and answer history information DB222 and an IR information DB224.

The question and answer history information DB222 is a means for storing questions and their answers received by the user in the past in association with each other as part of the user public information. The questions and their answers received by the user in the past are not limited to those from the same rating agency for which the user seeks answers using this system, but also include questions and their answers from other different rating agencies. In other words, instead of storing only past questions and their answers from the same rating agency for which the user sought answers using this system, it stores past questions and their answers from multiple other rating agencies for which the user sought answers using this system.

Furthermore, the questions and their answers stored in the question and answer history information DB 222 may not only cover a single year, but also cover multiple years at multiple evaluation institutions.
Therefore, the question and answer history information DB 222 is composed of a plurality of information groups that store the question and answer history information disclosed by the user to various evaluation institutions as one piece of user public information.

The question and answer history information DB222 may manage the past question and answer history information disclosed by the user collectively for each user, or may manage it separately for each evaluation organization or each year of disclosure.

The IR information DB224 is a means for storing IR information of users as a piece of user public information. This IR (Investor Relations) information is information on the management status, financial status, and business performance trends that a company as a user discloses to investors, and includes, for example, securities reports, environmental reports, sustainability reports, and integrated reports, as mentioned above. Therefore, the IR information DB224 stores the IR information disclosed by the user for each user as a piece of user public information.

The IR information DB224 may manage IR information such as the aforementioned securities reports, environmental reports, sustainability reports, and integrated reports collectively for each user, or may manage each type of information separately.

The reference information DB 230 is a means for storing reference information related to creating answers to questions or correcting automatically generated answers. Examples of reference information include terminology explanation information that explains terms included in a question and framework information related to a question. The reference information DB 230 may collectively manage the terminology explanation information and framework information for each question stored in the question information DB 210, or may manage them separately as a terminology explanation information DB and a framework information DB.

Next, a basic process flow in the answer generation system according to this embodiment will be described.
FIG. 4 is a sequence diagram showing an example of the answer generating process according to the present embodiment.

In step S301, the answer generation device 10 registers the latest questions from the evaluation institutions in the question information DB 210 for each evaluation institution, prior to providing a service to a user.
Furthermore, in step S302, prior to providing the service to the user, the answer generation device 10 registers the user public information already disclosed by the user in the user public information DB 220. More specifically, the answer generation device 10 registers the user public information in the question answer history information DB 222 and the IR information DB 224. This registration of the user public information in the user public information DB 220 may be performed by receiving the user public information from the user, or may be performed by acquiring the user public information via the network N.

In addition, if a user has disclosed answers to an evaluation agency in the past using this system, the question answer history information is registered in the user public information DB220 (question answer history information DB222) as user public information (i.e., question answer history information) provided by the user.

Once the above-mentioned preparations have been completed, the answer generation service provided by this system is implemented, and users can enjoy the service.
To receive this service, the user first accesses the answer generation device 10 using the user terminal 30 and completes the login process.

Then, in step S303, the user terminal 30 accepts an instruction to request (select) display of questions, and transmits the request to the answer generation device 10. In this embodiment, a request for a question list that displays all questions in a list format is transmitted to the answer generation device 10.
In step S304, in response to a question list request from the user terminal 30, the answer generation device 10 generates question display information that displays all questions in a list format, and transmits the generated question display information to the user terminal 30. At this time, if point allocation information, disclosure status information, answer non-generation information, etc. are stored in association with the question information, the answer generation device 10 generates question display information including this information.

In step S305, the user terminal 30 displays the question display information sent from the answer generation device 10 on a display screen (monitor), accepts an instruction to select (click) one question to answer, and sends the instruction to the answer generation device 10.
In step S306, the answer generation device 10 generates individual question display information that displays only the selected question individually in response to the selection of a question on the user terminal 30, and transmits the generated individual question display information to the user terminal 30. At this time, if framework information is stored in association with the question information, the answer generation device 10 generates individual question display information including this information.

In step S307, the user terminal 30 displays the question individual display information transmitted from the answer generation device 10 on a display screen (monitor), and when an instruction to automatically generate an answer to the question is received, the user terminal 30 transmits the request to the answer generation device 10. The user terminal 30 can also receive input of answer information that the user himself answers.
In step S308, in response to a request for automatic generation of an answer from the user terminal 30, the answer generation device 10 automatically generates an answer to the question using the user public information, and transmits the generated answer information to the user terminal 30.

In step S309, the user terminal 30 displays the answer information transmitted from the answer generation device 10 on a display screen (monitor), accepts an instruction to respond (disclose) with this content, and transmits the instruction to the answer generation device 10.
In step S310, in response to an answer (disclosure) instruction on the user terminal 30, the answer generation device 10 registers the answer information generated by the answer generation device 10 (automatically generated) or the answer information entered by the user in the question information DB 210 in association with the question.

If the user wishes to continue answering the question, steps S305 to S310 are repeated. If the user wishes to continue answering the question after completing the answering task, the user accesses the answer generation device 10 again, logs in, and repeats steps S303 to S310.

Next, a detailed flow of the answer generation process in the answer generation device 10 will be described.
FIG. 5 is a flowchart showing an example of a basic operation of the answer generating process according to this embodiment.

As shown in FIG. 5, in step S401, the question list generation unit 130 performs a process of determining whether a question list request for displaying questions in a list format has been accepted (received). If the question list generation unit 130 has accepted a question list request (YES in S401), it proceeds to step S402, and if not (NO in S401), it repeats the determination in step S401.

In step S402, the question list generating unit 130 performs a process of generating question display information for displaying all questions in a list format based on the question information DB 210.
In step S403, the question list generating unit 130 performs a process of transmitting the generated question display information to the user terminal 30 that has made the request.

In addition, in step S404, the question individual display generation unit 140 performs processing to determine whether or not question selection information for selecting a question to be answered has been accepted (received). If the question individual display generation unit 140 has accepted question selection information (YES in S404), it proceeds to step S405, and if not (NO in S404), it repeats the determination in step S404.

In step S405, the question individual display generating unit 140 performs processing for generating individual question display information that individually displays only the selected question.
Furthermore, in step S406, the question individual display generating unit 140 performs processing for transmitting the generated question individual display information to the instructed user terminal 30.

Next, in step S407, the answer generation unit 110 performs processing to determine whether or not it has accepted (received) an instruction (answer generation request) requesting automatic generation of an answer to the question. If the answer generation unit 110 has accepted an answer generation request (YES in S407), it proceeds to step S408, and if not (if the answer generation unit 110 has accepted input of answer information that the user will answer by himself/herself) (NO in S407), it proceeds to step S410.

In step S408, the answer generation unit 110 performs a process of automatically generating an answer to the question using the user public information stored in the user public information DB 220. That is, the answer generation unit 110 automatically generates an answer to the question using the question answer history information stored in the question answer history information DB 222 and the IR information (e.g., securities report, environmental report, sustainability report, integrated report) stored in the IR information DB 224.

In this case, if the answer assistance unit 120 manages the question in association with user public information (i.e., question answer history information and IR information), the answer generation unit 110 performs a process of automatically and efficiently generating an answer to the question using the user public information associated by the answer assistance unit 120.
Furthermore, in step S409, the answer generating unit 110 performs processing for transmitting the generated answer information to the user terminal 30 that has made the request.

In step S410, the answer receiving unit 170 performs a process to determine whether or not it has accepted (received) an instruction to disclose an answer (answer disclosure request). If the answer receiving unit 170 accepts an answer disclosure request (YES in S410), it proceeds to step S411, and if not (NO in S410), it repeats the determination in step S410.

In step S411, the answer receiving unit 170 performs a process to determine whether answer information has been accepted (received). If the answer receiving unit 170 has accepted answer information (YES in S411), the process proceeds to step S412; if not (NO in S411), the process repeats the determination in step S407. That is, the process performs a process to determine whether input of answer information automatically generated by the answer generating unit 110 or input of answer information created by the user has been accepted. Therefore, if an answer disclosure request is accepted without accepting input of answer information by either method, the process returns to step S407 and the determination of whether an answer generation request has been accepted is repeated.

Then, in step S412, the answer receiving unit 170 performs a process of registering the answer information generated by the answer generating unit 110 or the answer information created by the user in the question information DB 210 in association with the question.
Thereafter, the operation of the answer generation device 10 ends (END).

Here, a method for checking the question display information generated in step S402 can be shown in FIG.
FIG. 6 is a schematic diagram showing an example of the configuration of a question list screen 300 displayed on the user terminal 30 on which all questions are displayed in a list format.

In FIG. 6, the question list screen 300 displays a disclosure number column 301 indicating the question number, a score column 302 indicating the evaluation value assigned to the answer to the question, a disclosure status column 303 indicating whether the question has been answered, a question column 304 indicating the content of the question, a difference from last year column 305 indicating whether there have been changes to the previous question, answer non-generation information 306 indicating that an answer cannot be automatically generated, and a "disclose" button 307 to display the questions individually.

Specifically, on the question list screen 300 shown in FIG. 6, five questions shown as "C1.1", "C1.1a", "C1.1b", "C1.1c", and "C1.1d" are listed vertically in the disclosure number column 301 as part of the questions.

In addition, the score column 302 shows a score of "2" for questions with disclosure number C1.1, a score of "5" for questions with disclosure number C1.1a, a score of "7" for questions with disclosure number C1.1b, a score of "9" for questions with disclosure number C1.1c, and a score of "5" for questions with disclosure number C1.1d.

In addition, in the disclosure status column 303, the text "disclosed" indicating that the question has been answered, or "undisclosed" indicating that the question has not been answered, is displayed within an icon image. In this case, to visually emphasize that the question has not been answered, the text "undisclosed" is displayed larger than the text "disclosed," and the size of the horizontal oval surrounding the text is also larger, and the color of the inside of the horizontal oval may be different for "disclosed" and "undisclosed."

In addition, the question field 304 displays, for example, the following questions: "Is there board-level oversight of climate-related issues within the organization?" for disclosure number C1.1; "Identify the position" for disclosure number C1.1a; "Please provide further details regarding the board's oversight of climate-related issues" for disclosure number C1.1b; etc.

Furthermore, the word "none" is displayed in the difference column 305. The difference column 305 is a column that indicates whether or not there are differences in the content of questions or answers between different years, for example.
Furthermore, as answer non-generation information 306, a star-shaped figure is displayed between the difference column 305 and the “Disclose” button 307 for the question with disclosure number C1.1a.

A method for checking the question-individual display information generated in step S405 can be shown in FIG.
FIG. 7 is a schematic diagram showing an example of the configuration of a question-item display screen 310 that is displayed on the user terminal 30 after the “disclose” button 307 is selected (clicked) on the question list screen 300. As shown in FIG.

In FIG. 7, this question individual display screen 310 displays a heading field 311 that simply associates the question disclosure number with its disclosure status, a question display field 312 that individually shows the content of the question along with an English translation, a simple answer field 313 for selectively answering the question with "yes" or "no," an answer input field 314 for providing a detailed explanation of the answer, an "Auto-generate answer" button 315 that requests the automatic generation of an answer, an "Update" button 316 that instructs the disclosure (registration) of the answer, and a "Disclosure hint" button 317 that requests the display of reference information.

7, if the user wishes to answer a question displayed in question display field 312 automatically, he/she selects (clicks) "Auto-generate answer" button 315. On the other hand, if the user wishes to answer the question himself/herself, he/she can click simple answer field 313 to display a "Yes/No" pull-down menu, select one of the desired answers, and directly input the answer in answer input field 314.
When the "Auto-generate answer" button 315 is selected, a message saying "AI is generating answer..." is displayed on the screen until the answer shown in FIG. 8, which will be described later, is displayed.

A method for checking the answer automatically generated in step S408 can be shown in FIG.
FIG. 8 is a schematic diagram showing an example of the configuration of an answer display screen 320 that is displayed on the user terminal 30 after the “Automatically generate answers” button 315 is selected (clicked) on the question individual display screen 310.

In FIG. 8 , the answer “Yes” is automatically entered into the simplified answer field 313 of this answer display screen 320, and the answer “Board level supervision is carried out through the presence of a CSO, the ESG promotion department, and the use of the AAA system” is automatically generated and entered into the answer input field 314.
Furthermore, if the user does not wish to have the answer automatically generated in step S407 and wishes to answer by himself, he can refer to the framework information shown in FIG.

FIG. 9 is a schematic diagram showing an example of the configuration of framework information displayed by scrolling downward the question-item display screen 310 shown in FIG.
In Figure 9, the individual question display screen 310 shows a framework information display field 318 as well as a "View disclosure information" button 319 for requesting the display of user responses to questions from other evaluation institutions that are similar to the question in question.

Next, a flow of displaying reference information that helps understand the gist of a question and input an answer as a secondary process in the answer generation system according to this embodiment will be described with reference to FIGS. 10 and 11. FIG.
10 and 11 are both sequence diagrams showing an example of the reference information display process according to this embodiment.

In step S501, when the user terminal 30 receives an instruction from the user requesting the display of reference information for disclosure (answer), the user terminal 30 transmits the request to the answer generation device 10. In other words, after the question individual display information is displayed on the display screen (monitor) of the user terminal 30 in step S307 described above, the user terminal 30 requests the answer generation device 10 to disclose reference information for answering the question, without immediately requesting automatic generation of an answer.

In step S502, in response to a request for disclosure of reference information from the user terminal 30, the answer generation device 10 transmits the term explanation information included in the question to the user terminal 30 as reference information.
Then, in step S503, the user terminal 30 displays the term explanation information transmitted from the answer generation device 10 on the display screen (monitor).
Therefore, the user can understand the question by referring to the terminology explanation information and decide whether to answer the question himself or to generate an answer automatically (perform step S307).

In addition, in step S601, when the user terminal 30 receives an instruction from the user to request past disclosure information, the user terminal 30 transmits the request to the answer generation device 10. In other words, after the question individual display information is displayed on the display screen (monitor) of the user terminal 30 in the above-mentioned step S307, the user terminal 30 requests the answer generation device 10 for past disclosure information related to the question without immediately requesting automatic generation of an answer.

In step S602, in response to the disclosure request from the user terminal 30, the answer generation device 10 transmits user public information (ESG disclosure information) regarding the ESG related to the question to the user terminal 30 as reference information.
Then, in step S603, the user terminal 30 displays the ESG disclosure information transmitted from the answer generation device 10 on a display screen (monitor).
Therefore, the user refers to the ESG disclosure information to understand the purpose of the question, checks past disclosure information, and decides whether to answer the question himself or automatically generate an answer (perform step S307).

Next, the flow of secondary processing in the answer generation device 10 will be described.
FIG. 12 is a flowchart illustrating an example of a secondary operation of the answer generating process according to this embodiment.

12, in step S701, the reference information display unit 150 performs a process of determining whether a reference display request for displaying reference information related to the question has been accepted (received). If the reference information display unit 150 has accepted a reference display request (YES in S701), the process proceeds to step S702, and if not (NO in S701), the process proceeds to step S801.
In step S702, the reference information display unit 150 performs a process of transmitting reference information related to the question to the user terminal 30 that has made the request.
Thereafter, the operation of the answer generation device 10 ends (END).

In step S801, the user public information display unit 160 performs a process of determining whether a disclosure information request for displaying past disclosure information related to the question has been accepted (received). If the user public information display unit 160 accepts a disclosure information request (YES in S801), the process proceeds to step S802, and if not (NO in S801), the operation of the answer generation device 10 ends (END).
In step S802, the user public information display unit 160 performs a process of transmitting past user public information regarding the ESG related to the question as disclosure information to the user terminal 30 that has made the request.

Here, the method for checking the reference information transmitted in step S702 can be, for example, as shown in FIG.
FIG. 13 is a schematic diagram showing an example of the configuration of a reference information display screen 330 that is displayed on the user terminal 30 after the “Disclosure Hint” button 317 is selected (clicked) on the question individual display screen 310 shown in FIG.

In FIG. 13, a reference information display screen 330 is displayed so as to overlap the question display screen 310, and this reference information display screen 330 is provided with a term explanation display section 321 that explains terms as reference information.

A method for checking the disclosure information transmitted in step S802 can be shown in FIG. 14, for example.
Figure 14 is a schematic diagram showing an example of the configuration of an ESG disclosure information display screen 340 that is displayed on the user terminal 30 after selecting (clicking) the "View disclosure information" button 319 provided in association with the framework information shown in Figure 9, which is displayed by scrolling downward on the question individual display screen 310.

In FIG. 14, the ESG disclosure information display screen 340 has an ESG disclosure information display field 331 following the framework information. This ESG disclosure information display field 331 displays information about ESG to help understand the gist of the question, as well as the user's answers to past questions about ESG that are similar to the question, allowing the user to check past disclosure information about ESG.

The "View Disclosure Information" button 319 selected (clicked) when checking the disclosure information has been changed to a "Close Disclosure Information" button 339 on the ESG disclosure information display screen 340 in which the ESG disclosure information display field 331 is provided. Therefore, by selecting (clicking) the "Close Disclosure Information" button 339, the screen returns to the individual question display screen 310 on which the framework information is displayed, as shown in FIG. 9.

As described above, according to this embodiment, answer sentences are automatically generated in response to ESG-related questions from various rating agencies based on various ESG-related information disclosed by users, making it possible to efficiently generate answers automatically while reducing time, effort, cost, etc. Moreover, since the various information used to generate the answers is disclosed by the users themselves, it is possible to generate answers that are error-free and consistent overall.

In addition, the answers generated using this system ultimately become part of the user's public information, so the more questions you answer using this system, the more public user information you will accumulate, making it possible to efficiently generate highly accurate answers to questions from other evaluation agencies or questions from the same evaluation agency the following year.

The above-mentioned series of processes are merely examples, and are not particularly limited. Therefore, they can be executed by hardware or implemented by software. In other words, it is sufficient for the information processing system to have the functionality to execute the above-mentioned series of processes as a whole, and the type of functional block used to realize this functionality is not particularly limited to the above-mentioned examples.

In addition, in the system configuration example of this embodiment, the answer generating device 10 includes, for example, the answer generating unit 110, but is not limited to this aspect. The answer generating unit 110 may be provided in a server device external to the answer generating device 10. In this case, the answer generating device 10 transmits a question and user public information (an example of disclosure information) to the server device including the answer generating unit 110, and obtains the generated answer from the server device.

As described above, the system of this embodiment automatically generates answer text based on various information about the user that the user discloses, so it is expected that the system will be used not only to answer questions about ESG, but also to automatically generate appropriate answers to a variety of questions.

Here, the answer generating device 10 is an example of an answer generating device. The answer generating unit 110 is an example of a receiving means and an answer generating means. The answer assistance unit 120, the question information DB 210, and the user public information DB 220 are examples of an acquiring means and an associating means. The question list generating unit 130 is an example of a list display means.

Although the present embodiment has been described above, the above-mentioned embodiment is merely an example to facilitate understanding of the present invention, and is not intended to limit the interpretation of the present invention. Furthermore, the present invention is not limited to the above-mentioned embodiment, and any modifications, improvements, etc. that can achieve the object of the present invention are included in the present invention.

10...Answer generation device (server), 30...User terminal, 110...Answer generation unit, 120...Answer assistance unit, 130...Question list generation unit, 140...Question display generation unit, 150...Reference information display unit, 160...User public information display unit, 170...Answer reception unit, 210...Question information DB, 220...User public information DB, 222...Question answer history information DB, 224...IR information DB, 230...Reference information DB

Claims (13)

1. An answer generation system having one or more computers,
The computer includes:
A reception means for receiving a question related to a user's evaluation in the ESG;
An acquisition means for acquiring disclosure information previously disclosed by the user;
an answer generation means for automatically generating an answer to the question using a large-scale language model based on the disclosure information acquired by the acquisition means;
Equipped with
The acquisition means acquires, as the disclosed information, history information that associates questions previously received by the user with answers to the questions, and acquires, as the disclosed information, IR information of the user;
the answer generation means, when the question is input to the large-scale language model, generates an answer to the question using the disclosed information;
An answer generation system comprising: 2. The answer generation system according to claim 1 , wherein the answer generation means uses a plurality of pieces of the history information of the user as the disclosure information when generating an answer to one of the questions. The answer generation system according to claim 2, characterized in that the history information includes information on questions and answers to the questions from one evaluation agency that evaluates ESG, and information on questions and answers to the questions from other evaluation agencies different from the one evaluation agency. a correlation means for correlating the history information with each of the questions,
4. The answer generating system according to claim 1 , wherein the answer generating means generates the answer by using the history information associated by the associating means. The answer generation system according to claim 1, further comprising a list display means for displaying the plurality of questions received by the receiving means in a list on a screen. 6. The answer generating system according to claim 5 , wherein the list display means displays, in the list, evaluation information relating to an evaluation value in an ESG for each question. 6. The answer generating system according to claim 5 , wherein the list display means displays, for each question, disclosure status information indicating whether or not the user has answered the question, in the list. 6. The answer generation system according to claim 5 , wherein the list display means displays, for each question in the list, answer non-generation information indicating that the question is one for which the answer generation means cannot generate an answer. The answer generation system according to claim 1, characterized in that when the user answers the question or when the user modifies the answer generated by the answer generation means, reference information that the user can refer to is displayed individually for each question. 10. The answer generating system according to claim 9 , wherein the reference information includes explanation information that is an explanation of the question. The answer generating system according to claim 10 , wherein the reference information includes the disclosure information for a past question related to the commentary information. A reception means for receiving a question related to a user's evaluation in the ESG;
An acquisition means for acquiring disclosure information previously disclosed by the user;
a transmission means for transmitting the question and the disclosure information to an answer generation means for automatically generating an answer to the question using a large-scale language model based on the disclosure information acquired by the acquisition means ;
Equipped with
The acquisition means acquires, as the disclosed information, history information that associates questions previously received by the user with answers to the questions, and acquires, as the disclosed information, IR information of the user;
The answer generation device, characterized in that, when the question is input to the large-scale language model, the answer generation means generates an answer to the question using the disclosed information . On the computer,
A function of accepting questions related to a user's evaluation in the ESG;
A function of acquiring disclosure information that the user has disclosed in the past;
a function of transmitting the question and the disclosure information to an answer generation means that automatically generates an answer to the question using a large-scale language model based on the acquired disclosure information ;
A program for executing
acquiring, as the disclosed information, history information associating questions previously received by the user with answers to the questions, and acquiring, as the disclosed information, IR information of the user;
The answer generation means generates an answer to a question using the disclosed information when the question is input to the large-scale language model.