WO2023162020A1

WO2023162020A1 - Answer assistance system, answer assistance method, and recording medium

Info

Publication number: WO2023162020A1
Application number: PCT/JP2022/007313
Authority: WO
Inventors: 昌宏 ▲浜▼田; 広樹野村
Original assignee: 日本電気株式会社
Priority date: 2022-02-22
Filing date: 2022-02-22
Publication date: 2023-08-31

Abstract

This answer assistance system comprises an acquisition unit, an estimation unit, and an output unit. The acquisition unit acquires a question made by a questioner in a conference. The estimation unit estimates, from the question acquired by the acquisition unit, an answer candidate for the question by using an answer estimation model that estimates, from a question, an answer candidate for the question. The output unit outputs the answer candidate to a terminal device used for assisting with answering.

Description

Answer support system, answer support method and recording medium

The present invention relates to an answer support system and the like.

At the meeting, answers to questions may be recorded and disseminated. For example, at a general meeting of shareholders, an officer in charge of a field related to the question answers questions from shareholders. Answers to questions posed at shareholders' meetings, for example, may be publicized as indicating corporate policy. Therefore, responses must be made appropriately. In such a case, there may be an adviser who assists the answer by showing the answer draft to the answerer who is on stage. However, the contents of the questions are diverse, and it may be difficult to present appropriate answers in a short time. Therefore, it is desirable to have a technology capable of presenting appropriate answers to questions to respondents. Therefore, techniques for presenting answer candidates to questions are being developed.

The dialogue management server of Patent Document 1 extracts answer candidates to the question from the questioner, and replies to the questioner as an answer to the question.

When the conference system of Patent Document 2 receives information indicating that it is an important comment during an utterance, it transmits data to which information indicating that it is important information is added to the terminals of participants in the conference.

Japanese Patent Application Laid-Open No. 2020-204971 JP 2018-174442 A

The dialogue management server of Patent Document 1 and the conference system of Patent Document 2 may not be able to present an appropriate answer to the question to the respondent.

In order to solve the above problems, the purpose is to provide an answer support system, etc. that makes it easier for respondents to give appropriate answers at meetings.

In order to solve the above problems, the answer support system of the present invention uses an acquisition means for acquiring a question by a questioner at a meeting, and an answer estimation model for estimating answer candidates to the question from the question. An estimation means for estimating an answer candidate to a question from a question and an output means for outputting the answer candidate to a terminal device used for answer support. Then, using an answer estimation model for estimating answer candidates to a question from a question, an answer candidate to the question is estimated from the acquired question, and the answer candidate is output to a terminal device used for answer support.

The recording medium of the present invention includes a process of acquiring a question from a questioner in a meeting, a process of estimating an answer candidate to the question from the acquired question using an answer estimation model for estimating the answer candidate to the question from the question, An answer support program for causing a computer to execute processing for outputting answer candidates is non-temporarily recorded in a terminal device used for answer support.

According to the present invention, it is possible to make it easier for the respondent to give an appropriate answer in the conference.

BRIEF DESCRIPTION OF THE DRAWINGS It is a figure which shows the example of a structure of the conference system of the 1st Embodiment of this invention. FIG. 2 is a diagram schematically showing the flow of questions and answers in the conference system according to the first embodiment of the present invention; FIG. It is a figure which shows the example of the reply support system of the 1st Embodiment of this invention. It is a figure which shows the example of a structure of the adviser terminal device of the 1st Embodiment of this invention. It is a figure which shows the example of the display screen of the 1st Embodiment of this invention. It is a figure which shows the example of the display screen of the 1st Embodiment of this invention. It is a figure which shows the example of the display screen of the 1st Embodiment of this invention. It is a figure which shows the example of the display screen of the 1st Embodiment of this invention. It is a figure which shows the example of the display screen of the 1st Embodiment of this invention. It is a figure which shows the example of the operation|movement flow of the answer support system of the 1st Embodiment of this invention. It is a figure which shows the example of the operational flow of the adviser terminal device of the 1st Embodiment of this invention. It is a figure which shows the example of the operation|movement flow of the answer support system of the 1st Embodiment of this invention. It is a figure which shows the example of a structure of the answer support system of the 2nd Embodiment of this invention. It is a figure which shows the example of the operation|movement flow of the answer support system of the 2nd Embodiment of this invention. It is a figure which shows the example of a structure of other embodiment of this invention.

A first embodiment of the present invention will be described in detail with reference to the drawings. FIG. 1 is a diagram showing an overview of the configuration of a conference system according to this embodiment. The conference system of this embodiment includes, as an example, an answer support system 10, an adviser terminal device 20, and an answerer terminal device 30. FIG.

The conference system is a system that supports respondents who answer questions when answering questions in a conference. A meeting is, for example, a shareholders' meeting. A meeting may be a parliament. The conference is not limited to the above examples.

Fig. 2 is a diagram schematically showing the flow of questions and answers in the conference system. In the example of the conference system in FIG. 2, when a question is asked by a questioner, the answer support system 10 acquires the question asked. The answer support system 10 estimates answer candidates for the question from the question using the answer estimation model, and outputs the estimated answer candidates to the adviser terminal device 20 . An answer estimation model is a trained model that estimates an answer candidate to a question from the question. The answer estimation model will be explained later.

The adviser who operates the adviser terminal device 20 creates an answer by referring to answer candidates. When the answer is created, the adviser terminal device 20 outputs the answer to the respondent terminal device 30 . A mentor is a person who assists a respondent. The adviser assists the respondent in answering the question by showing the respondent a draft answer to the question. An adviser is, for example, an expert in the field of question. For example, at a shareholder meeting, if the question is about the financial situation, the adviser is the financial officer. Examples of advisors are not limited to the above.

The respondent, who is the person who answers in the meeting, refers to the answer displayed on the display device of the respondent terminal device 30, for example, and answers the question. The answer may be a reply. Also, a respondent may be a respondent or a speaker.

If the meeting is a general meeting of shareholders, the questioner is, for example, a shareholder. Also, if the meeting is a general meeting of shareholders, the respondent is, for example, an executive of the company that holds the general meeting of shareholders.

If the meeting is a parliament, the questioner is, for example, a member of the assembly. Also, if the conference is a parliament, the respondent is, for example, a minister, mayor, or other person who speaks in parliament. Meetings for which answer candidates are estimated by the answer support system 10 are not limited to shareholders' meetings and parliaments.

The adviser creates an answer by, for example, operating the adviser terminal device 20 to select an appropriate answer to the question from a plurality of answer candidates indicated as answer candidates. The adviser may refer to the answer candidates to create an answer sentence. Answer candidates also include information related to the answer. Information related to answers is, for example, information for selecting an answer from answer candidates. The information related to the answer may be information for creating the answer.

The answer support system 10 uses an answer estimation model for estimating answer candidates to questions from questions, and estimates answer candidates to questions from questions asked at meetings. The answer support system 10, for example, generates the relationship between the question and the answer given to the question by learning using a machine learning algorithm. The answer support system 10 may estimate answer candidates using an answer estimation model generated outside.

The configuration of the reply support system 10 will be explained. FIG. 3 is a diagram showing an example of the configuration of the reply support system 10. As shown in FIG. The answer support system 10 includes an acquisition unit 11 , an estimation unit 12 , an output unit 13 , a model generation unit 14 and a storage unit 15 .

The acquisition unit 11 acquires questions from questioners in the conference. For example, the acquisition unit 11 acquires text data obtained by converting a voice question of a questioner through voice recognition. When a questioner asks a question by inputting text data into a terminal device, the obtaining unit 11 may obtain the question from the terminal device to which the question was input.

When a conference is being held at multiple venues via a network, the acquisition unit 11 may acquire questions held at each of the multiple venues. If the meeting is an online meeting, the acquiring unit 11 may acquire questions from the terminal device for the online meeting via the network. Further, when each participant of the conference participates in the conference via a terminal device, the acquisition unit 11 may acquire the question from each participant's terminal device. Acquisition unit 11 may acquire information specifying the questioner along with the question. The information specifying the questioner is, for example, the name or the questioner's identification number. The information specifying the questioner is not limited to the above examples.

When the answer support system 10 generates an answer estimation model, the acquisition unit 11 may acquire data that associates questions and answers to the questions. For example, the acquisition unit 11 acquires data that associates a question input to the terminal device and an answer to the question from the terminal device by an operator's operation. Data that associates a question with an answer to the question may be input to the answer support system 10 by the operator's operation. The acquisition unit 11 may store data in which the acquired question and the answer to the question are associated with each other in the storage unit 15 .

The estimating unit 12 uses an answer estimation model for estimating answer candidates to the question from the question, and estimates answer candidates to the question from the question acquired by the acquiring unit 11 . The estimating unit 12 uses, for example, the answer estimating model generated by the model generating unit 14 to estimate answer candidates from the question. As for the answer estimation model, the estimation unit 12 may use an answer estimation model generated by a system external to the answer support system 10 to estimate answer candidates.

As an example, the estimating unit 12 uses, for example, an answer estimating model to extract questions having the same or similar content as the questions acquired by the acquiring unit 11 from data relating questions and answers from past meetings. do. The answer estimation model, for example, converts question text data into feature vectors and calculates the distance between the feature vectors. Then, the answer estimation model extracts questions whose distance between feature vectors is equal to or less than the standard as identical or similar questions. The distance between feature vectors is, for example, the Euclidean distance. What is the distance between feature vectors? It is not limited to the Euclidean distance. An answer estimation model may extract identical or similar questions with implications. The estimating unit 12 refers to data that associates questions and answers, and selects answers associated with the extracted question as answer candidates for the question.

The data that associates the questions and answers asked at the meeting is, for example, the data used to generate the answer estimation model. The data that associates the questions asked in the meeting with the answers may be data different from the data used to generate the answer estimation model. For example, if some of the data that associates questions and answers is used to generate an answer estimation model, data not used to generate the answer estimation model may be used when estimating answer candidates. good. In addition, for example, if the meeting is a general meeting of shareholders, the data linking questions and answers may be held at a general meeting of shareholders held in a different year or at a different company from the one in which the data was used to generate the answer estimation model. data. Further, for example, when the meeting is a parliament, the data that associates questions and answers may be data of a parliament other than the parliament whose data is used to generate the answer estimation model.

The estimating unit 12, for example, calculates the degree of similarity between the question acquired by the acquiring unit 11 and the question included in the data that associates the question and the answer, and extracts the questions whose similarity is equal to or higher than the standard as questions with the same content. do. Then, the estimation unit 12 estimates an answer associated with the extracted question as an answer candidate for the question in the data that associates the question and the answer. The estimating unit 12 may estimate the answer candidates using the order of similarity as the recommended order of the answers. Also, the estimation unit 12 may estimate a predetermined number of answers in descending order of similarity as answer candidates.

As another example, the estimating unit 12 may use an answer estimating model that has learned the relationship between questions and answers in a meeting, and may use a question as an input to estimate an answer.

The estimation unit 12 may group the same or similar questions and estimate an answer candidate for each group. For example, when answering after receiving a plurality of questions in a meeting, the answer candidates may be estimated for each similar question by dividing into groups of similar questions. The estimation unit 12, for example, clusters questions and extracts characteristic questions in each group. Then, the estimation unit 12 inputs the extracted characteristic questions to the answer estimation model to extract answer candidates. Also, the estimation unit 12 may group questions with similar answers after estimating answer candidates for each question.

The estimation unit 12 may estimate an answer suitable for the progress of the conference as an answer candidate in the question-and-answer session of the conference. The answer to the proceedings of the conference is, for example, an answer to avoid receiving additional questions from the questioner who is attending the conference, such as denying the content of the answerer's answer. For example, if the meeting is a shareholders' meeting, topics related to corporate reputation and declining performance are subject to follow-up questions. Therefore, the estimating unit 12 estimates, for example, an answer about improving the company's reputation and business performance as an answer candidate. The estimating unit 12 uses, for example, an answer estimating model to extract answers to questions from data that associates questions and answers to questions, and data that associates scores that indicate the degree of suitability for question-and-answer sessions. Then, the estimation unit 12 estimates, as an answer candidate, an answer whose score, which indicates the degree of suitability for the question-and-answer session, satisfies a preset criterion.

The estimation unit 12 may estimate answer candidates according to the questioner's attributes. The questioner's attribute is, for example, information about the questioner's affiliation and career. When estimating an answer candidate according to the questioner's attribute, the estimating unit 12, for example, estimates an answer that satisfies a preset criterion among the answer candidates as an answer candidate.

For example, the estimating unit 12 extracts the answer to the question from the data that associates the question and the answer to the question, and the data that associates the attributes of the questioner. Then, the estimation unit 12 estimates, as an answer candidate, an answer whose attribute of the questioner matches the attribute of the questioner associated with the answer. The attribute of the questioner further associated with the data that associates the question and the answer to the question is set using, for example, the attribute of the person who asked the question in the past meeting. The questioner's attribute data used as an input when estimating answer candidates is stored in the storage unit 15 in association with, for example, the questioner's identification information. The estimation unit 12 reads the questioner's attribute data from the storage unit 15 based on the questioner's identification information acquired by the acquisition unit 11 . Then, the estimation unit 12 estimates the answer candidate by inputting the read attribute of the questioner into the answer estimation model together with the question.

When the meeting is a general meeting of shareholders, the attributes of the questioner are, for example, any one or more of age, type, number of shares held, holding period of shares, number of times of attendance at the general meeting of shareholders, and number of questions at the general meeting of shareholders. . The type is, for example, information indicating whether the shareholder is an individual investor or an institutional investor. When the meeting is a parliament, the questioner's attributes are, for example, any one or more of affiliated party, post, number of elections, number of questions, age, and committee to which the questioner belongs. The attributes of the questioner are not limited to the above examples.

When estimating an answer candidate according to the questioner's attribute, the estimation unit 12 may estimate an answer candidate to the question using an answer estimation model that estimates the answer candidate from the questioner's attribute and the question. .

The estimation unit 12 may estimate an answer using actual data as part of the answer. For example, in response to a question at a general meeting of shareholders, if an answer is set such as "because sales have decreased by A," the estimation unit 12 calculates sales from financial information data Calculate the amount of decrease in Then, the estimation unit 12 may insert the calculated decrease amount into the part A to estimate the answer candidate. Also, portions of the actual data may be supplemented by advisors.

The estimation unit 12 may estimate answer candidates using different answer estimation models depending on the field of the question. The estimation unit 12 selects an answer estimation model to be used for estimation of answer candidates, for example, based on words included in the question. The relationship between the words included in the question and the answer estimation model is set in advance, for example. Also, the estimation unit 12 may estimate an answer candidate using different answer estimation models according to the questioner's attributes. The relationship between the questioner's attribute and the answer estimation model is set in advance, for example. The estimation unit 12 may classify questions into groups by clustering, and estimate answer candidates using an answer estimation model according to each group.

The output unit 13 outputs the answer candidates estimated by the estimation unit 12 to the terminal device used for answer support. The person who uses the terminal device used for answering assistance is, for example, an adviser who assists the respondent in answering. Also, the person who assists the answer may be the chairperson. Also, the terminal device used for answering support may be the terminal device used by the respondent. The output unit 13 outputs the answer candidates estimated by the estimation unit 12 to the adviser terminal device 20, for example. When there are a plurality of adviser terminal devices 20 , the output unit 13 outputs the answer candidates estimated by the estimation unit 12 to the plurality of adviser terminal devices 20 . The output unit 13 may output the answer candidates estimated by the estimation unit 12 to the respondent terminal device 30 . The output unit 13 may output the answer candidates estimated by the estimation unit 12 to both the adviser terminal device 20 and the respondent terminal device 30 . The output unit 13 may also output answer candidates to a display device (not shown) connected to the answer support system 10 .

The output unit 13 may output answer candidates containing information related to answers. The information related to the answer is, for example, information that the adviser refers to when creating the answer. The output unit 13 outputs, for example, statistical data related to answers as information related to answers. The output unit 13 outputs, for example, financial index data as information related to the answer. The output unit 13 outputs, for example, data on financial index items related to words included in questions or answers as data on financial indexes. For example, the relationship between the words included in the question or the answer and the item of the financial index to be output is set in advance.

The output unit 13 may output information about the questioner as information related to the answer. The information about the questioner is, for example, the questioner's attributes. The questioner's attribute is, for example, at least one or more of the questioner's name, gender, affiliation, number of times of participation in meetings, and question record in meetings. The attributes of the questioner are not limited to the above examples.

When there are multiple advisers, the output unit 13 may output answer candidates to the adviser terminal device 20 of a person suitable as an adviser. A person suitable as an adviser is, for example, a person who is knowledgeable about the content of the question. A person suitable as an adviser may be a person skilled in choosing answers to facilitate proceedings. The output unit 13, for example, outputs answer candidates to the adviser's terminal device 20, which is set in advance for each questioner's attribute. The output unit 13 may output answer candidates to the adviser's terminal device 20 preset according to the word included in the question. The output unit 13 may classify the questions into groups by clustering, and output answer candidates to the adviser terminal devices 20 of advisers set in advance according to the groups.

When outputting answer candidates, the output unit 13 may output a list of multiple questions used to estimate the answer candidates. The output unit 13 outputs, for example, answer candidates for questions selected from among the questions included in the output list of questions.

The output unit 13 may add a conference video to the answer candidates and output them. For example, the output unit 13 adds an image of the questioner to the answer candidate and outputs the answer candidate to the adviser terminal device 20 .
The output unit 13 may add, for example, an image of all attendees or an image of an answerer to the answer candidates and output them. Further, the output unit 13 may add a plurality of images to the answer candidates and output them. As for the conference video, for example, video captured by a camera in the conference hall is input to the answer support system 10 . In addition, when the questioner and the answerer are participating in the conference via the network, images of the questioner and the answerer are input to the answer support system 10 via the network. The output unit 13 adds these input images to the answer candidates and outputs them. Images of the questioner and the answerer may be input to the adviser terminal device 20 without going through the answer support system 10 .

The model generation unit 14 generates an answer estimation model for estimating answer candidates to questions from the questions in the meeting. For example, the model generation unit 14 extracts questions that are the same as or similar to the questions in the meeting, and generates an answer estimation model that uses answers associated with the extracted questions as answer candidates. The model generation unit 14 uses, for example, a natural language processing technique to extract questions having the same or similar content as the question, and generates an answer estimation model that outputs answers associated with the extracted questions as answer candidates.

For example, when an answer is estimated using a different answer estimation model for each question field or questioner's attribute, the model generation unit 14 generates each estimation model using data for each estimation target. may For example, when a different answer estimation model is used for each question attribute, the model generating unit 14 generates an answer estimation model using data relating questions and answers collected for each questioner attribute.

The model generation unit 14 may generate an answer estimation model that has learned the relationship between questions and answers in a meeting. The model generation unit 14 may also generate an answer estimation model that has learned the relationship between questions and answers, using appropriateness of answers to questions as labels. Whether or not the answer to the question is appropriate is set, for example, by determining whether the answer is suitable for question-and-answer sessions by a person in the department holding the meeting. The model generation unit 14 may also group identical or similar questions and generate an answer estimation model for estimating answer candidates for each group.

The model generation unit 14 stores the generated answer estimation model in the storage unit 15. When a plurality of estimated answer models are generated, the model generating unit 14 stores the estimated answer models in the storage unit 15, for example, by adding an identifier for identifying each estimated answer model.

The model generation unit 14 may re-learn the answer estimation model using the data of the questions and answers given at the conference.

The storage unit 15 stores answer estimation models used by the estimation unit 12 to estimate answer candidates. In the case where the answer estimation model extracts the same or similar questions from the data that associates questions and answers in past meetings and estimates answer candidates, the storage unit 15 associates the questions with the answers to the questions. You may save the data obtained. Further, if the answer candidate includes information related to the answer, the storage unit 15 may store the information related to the answer. When an answer estimation model is generated, the storage unit 15 may store data relating questions and answers used for generating the answer estimation model. Further, the storage unit 15 may store the questions acquired by the acquisition unit 11 . In addition, the storage unit 15 may associate and store the question of the questioner, the answer presented by the adviser to the respondent based on the answer candidate, and the answer actually given by the respondent. The answer estimation model used by the estimation unit 12 and data used for estimation of answer candidates may be stored in a storage means other than the storage unit 15 .

Next, the configuration of the adviser terminal device 20 will be described. FIG. 4 is a diagram showing an example of the configuration of the adviser terminal device 20. As shown in FIG. The adviser terminal device 20 includes an answer candidate acquisition unit 21 , an answer candidate output unit 22 , an input unit 23 and an answer output unit 24 .

The answer candidate acquisition unit 21 acquires answer candidates from the answer support system 10. The answer candidate acquisition unit 21 may acquire answer candidates and information related to the answers to the answer candidates from the answer support system 10 . Also, the answer candidate acquisition unit 21 may acquire a conference video from the answer support system 10 .

The answer candidate output unit 22 outputs the answer candidates acquired by the answer candidate acquisition unit 21 to a display device (not shown). The answer candidate output unit 22 may output answer candidates and information related to the answers to the answer candidates. Alternatively, answer candidates and conference video may be output.

The input unit 23 accepts input of answers. The input unit 23 receives, for example, an input of an answer to a question entered by an adviser with reference to answer candidates. The input unit 23 may accept an answer selected from the answer candidates by the adviser as input of the answer to the question.

The reply output unit 24 outputs the reply to the respondent terminal device 30. The answer output unit 24 outputs an answer to the respondent terminal device 30 based on the answer input to the input unit 23, for example.

The respondent terminal device 30 is a terminal device for the respondent to refer to the response prepared by the adviser. The respondent terminal device 30 acquires the reply from the adviser terminal device 20, for example. Then, the respondent terminal device 30 outputs the acquired responses to a display device (not shown). The respondent terminal device 30 may output the voice data of the reply.

For the respondent terminal device 30, for example, a tablet computer or a notebook computer is used. The respondent terminal device 30 may be an audio output device. The respondent terminal device 30 may be an output device that outputs an image to a prompter. Examples of the respondent terminal device 30 are not limited to the above.

FIG. 5 is a diagram showing an example of an estimation result display screen. In the example of the display screen in FIG. 5, questions and candidate answers to the questions are displayed. In the example of the display screen in FIG. 5, an answer entry field and an answer send button are displayed at the bottom of the display screen. In the example of the display screen in FIG. 5, the questions for which answer candidates are estimated are displayed in the question column. In the example of the display screen of FIG. 5, the answer candidates estimated by the estimation unit 12 are displayed in the answer candidate column.

The example of the display screen in FIG. 5 is the display screen of the estimation result when the answer candidate for the question at the general meeting of shareholders is estimated. In the example of the display screen in Fig. 5, in response to the question "Please tell us about measures for continuous growth," "We will continue to make capital investments in priority fields." In addition to existing businesses, we will actively develop new businesses." and "We will continue to invest in research and development to develop new fields." .

The output unit 13 of the response support system 10 outputs image data such as the example of the display screen in FIG. 5 to the adviser terminal device 20, for example. When the answer candidate acquisition unit 21 of the adviser terminal device 20 acquires the image data, the answer candidate output unit 22 outputs image data such as the example of the display screen in FIG. 5 to the display device.

In the example of the display screen of FIG. to output the answer. In the above explanation, the answer support system 10 generates the image data for the display screen. may generate the image data for the display screen. Whether the image data for the display screen is generated by the response support system 10 or the adviser terminal device 20 can be set as appropriate. It is the same in the following description as to which method is used to generate the image data for the display screen.

FIG. 6 is a diagram showing an example of further displaying information about the questioner as information related to the answer in the example of the display screen in FIG. Information about the questioner is, for example, the name and attributes of the questioner. The example of the display screen in FIG. 6 is for a general meeting of shareholders, and the questioner's name, type, number of shares held, and holding period are displayed in the column of questioner data that displays information about the questioner. there is The name of the questioner is, for example, the name of the questioner or the name of the company to which the questioner belongs. The type is, for example, information indicating whether the shareholder is an individual investor or an institutional investor. The number of shares held is, for example, the number of shares owned by the questioner. The holding period is, for example, the period during which the questioner holds the shareholder of the company that is holding the general meeting of shareholders. Also, the question record is, for example, the year in which the questioner asked the question at the general meeting of shareholders in the past. Examples of questioner data are not limited to the above.

FIG. 7 is a diagram showing an example of displaying related data as information related to an answer on the answer candidate display screen. Related data is, for example, information that an adviser refers to when preparing an answer. In the example of the display screen in FIG. 7, as related data, the capital investment amount for each department and research and development expenses for each year are displayed. Examples of related data are not limited to the above.

FIG. 8 is a diagram showing an example of displaying a plurality of questions on the answer candidate display screen. In the example of the display screen in FIG. 8, a plurality of questions asked in the meeting are displayed as a question list. In the example of the display screen in FIG. 8, for example, when any question in the question list is selected, answer candidates for the selected question are displayed in the answer candidate column. In the example of the display screen in FIG. 8, the answer candidates for the question No. 3 in the question list are displayed in the answer candidate column.

FIG. 9 is a diagram showing an example of displaying a video of the conference on the display screen of the estimation result. In the example of the display screen in FIG. 9, for example, an image of the venue of the conference is displayed in the column of the conference image. For example, an image of the questioner is displayed in the conference image column. The video of the conference may be a video of all attendees or a video of the respondent. Also, a plurality of images of the conference may be displayed. Also, the video of the meeting may be input to the adviser terminal device 20 without going through the answer support system 10 .

　In the conference system of this embodiment, the operation of estimating answer candidates and outputting answers based on the answer candidates will be described. FIG. 10 is a diagram showing an example of the operation flow of the reply support system 10 of this embodiment. Also, FIG. 11 is an example of an operational flow of the adviser terminal device 20 .

The operation of the answer support system 10 when estimating answer candidates will be described with reference to FIG. The acquisition unit 11 acquires a question from a questioner in a meeting (step S11). When the question by the questioner is acquired, the estimation unit 12 estimates the answer candidate to the question from the question acquired by the acquisition unit 11 using an answer estimation model for estimating the answer candidate to the question from the question (step S12). When the answer candidate to the question is estimated, the output unit 13 outputs the answer candidate estimated by the estimation unit 12 to the terminal device used for answer support (step S13). The output unit 13 outputs the answer candidates to the adviser terminal device 20, for example. The output unit 13 may output answer candidates to the respondent terminal device 30 .

The operation of the adviser terminal device 20 when outputting an answer based on answer candidates will be described with reference to FIG. When answer candidates are output to the adviser terminal device 20, the answer candidate acquisition unit 21 of the adviser terminal device 20 acquires the answer candidates (step S21). When the answer candidates are acquired, the answer candidate output unit 22 outputs the answer candidates to a display device (not shown) (step S22).

When the answer candidates are displayed on the display device, the adviser operates, for example, the adviser terminal device 20 and inputs an answer based on the answer candidates displayed on the display device.

The input unit 23 acquires an answer input by the adviser's operation (step S23). After obtaining the answer, the answer output unit 24 outputs the answer input by the adviser to the respondent terminal device 30 .

After obtaining the answer, the respondent terminal device 30 outputs the answer to, for example, a display device (not shown). The respondent using the respondent terminal device 30 answers the question by referring to the answer displayed on the display device, for example.

A description will be given of the operation of the model generating unit 14 to generate an estimated answer model when the answer support system 10 generates an estimated answer model. FIG. 12 is a diagram showing an example of an operational flow when the answer support system 10 generates an answer estimation model.

The acquisition unit 11 acquires data that associates a question with an answer to the question (step S31). When the data relating the question and the answer to the question is obtained, the model generation unit 14 generates an answer estimation model for estimating answer candidates to the question from the question (step S32).

After generating the answer estimation model, the model generation unit 14 stores the generated answer estimation model in the storage unit 15 (step S33).

The answer support system 10 of the conference system of the present embodiment estimates answer candidates for the question and outputs the answer candidates to the terminal device of the person related to the answer. For example, the reply support system 10 outputs the reply candidates to the adviser terminal device 20 of the adviser who assists the reply by the reply. The adviser prepares an answer by referring to answer candidates output by the adviser terminal device 20 to the display device, for example. Therefore, by using the answer support system 10, a person related to an answer can create an answer to a question by referring to answer candidates estimated using the answer estimation model. In addition, by referring to the answers obtained from the adviser terminal device 20 via the respondent terminal device 30, the respondent can, for example, appropriately answer the question while on stage. Also, when the answer candidates are output to the respondent terminal device 30, the respondent can refer to the answer candidates and give an answer. As a result, the answer support system 10 makes it easy for the respondent to give an appropriate answer in the conference.

(Second embodiment)
A second embodiment of the present invention will be described in detail with reference to the drawings. FIG. 13 is a diagram showing an example of the configuration of the reply support system 100 of this embodiment. Answer support system 100 includes acquisition unit 101 , estimation unit 102 , and output unit 103 . Acquisition unit 101 acquires a question from a questioner in a conference. The estimating unit 102 estimates answer candidates for the question from the question acquired by the acquiring unit 101 using an answer estimation model for estimating answer candidates for the question from the question. The output unit 103 outputs answer candidates to a terminal device used for answer support.

Here, the acquisition unit 11 of the first embodiment is an example of the acquisition unit 101. Also, the acquiring unit 101 is one aspect of an acquiring unit. The estimator 12 of the first embodiment is an example of the estimator 102 . Also, the estimating unit 102 is one aspect of an estimating means. The output unit 13 of the first embodiment is an example of the output unit 103 . Also, the output unit 103 is one aspect of output means.

The operation of the answer support system 100 will be explained. FIG. 14 shows an example of the operation flow of the reply support system 100. As shown in FIG. The acquisition unit 101 acquires a question from a questioner in a meeting (step S101). When the question is acquired, the estimation unit 102 estimates answer candidates for the question from the question (step S102). When the answer candidates are estimated and the presence or absence of an abnormality is detected, the output unit 103 outputs the answer candidates to the terminal device used for answer support (step S103).

The answer support system 100 of this embodiment estimates answer candidates for a question, and outputs the answer candidates to the terminal device of the person related to the answer. Therefore, by using the answer support system 100, the person related to the answer can refer to the answer candidate on the terminal device and answer the question. As a result, it becomes easier for the respondent to give an appropriate answer in the meeting.

Each process in the answer support system 10 of the first embodiment and the answer support system 100 of the second embodiment can be realized by executing a computer program on a computer. FIG. 15 shows an example of the configuration of a computer 200 that executes a computer program that performs each process in the answer support system 10 of the first embodiment and the answer support system 100 of the second embodiment. The computer 200 includes a CPU (Central Processing Unit) 201 , a memory 202 , a storage device 203 , an input/output I/F (Interface) 204 and a communication I/F 205 .

The CPU 201 reads a computer program for each process from the storage device 203 and executes it. The CPU 201 may be configured by a combination of multiple CPUs. Also, the CPU 201 may be configured by a combination of a CPU and another type of processor. For example, the CPU 201 may be configured by a combination of a CPU and a GPU (Graphics Processing Unit). The memory 202 is composed of a DRAM (Dynamic Random Access Memory) or the like, and temporarily stores computer programs executed by the CPU 201 and data being processed. The storage device 203 stores computer programs executed by the CPU 201 . The memory device 203 is configured by, for example, a nonvolatile semiconductor memory device. Other storage devices such as a hard disk drive may be used as the storage device 203 . The input/output I/F 204 is an interface that receives input from the operator and outputs display data and the like. A communication I/F 205 is an interface that transmits and receives data to and from a terminal device or another information processing device. Further, the adviser terminal device 20 and the respondent terminal device 30 of the first embodiment can also have the same configuration.

Each process in the answer support system 10 of the first embodiment and the answer support system 100 of the second embodiment may be performed by multiple computers connected via a network. For example, the processing related to answer candidate estimation and the processing related to answer estimation model generation may be performed on different computers. The storage unit of the answer support system 10 may be provided in a storage device connected via a network or a storage device managed by a server connected via a network.

The computer program used to execute each process can also be stored and distributed in a non-transitory computer-readable recording medium. As a recording medium, for example, a magnetic tape for data recording or a magnetic disk such as a hard disk can be used. Also, as a recording medium, an optical disc such as a CD-ROM (Compact Disc Read Only Memory) can be used. A nonvolatile semiconductor memory device may be used as a recording medium.

Some or all of the above-described embodiments can also be described in the following supplementary remarks, but are not limited to the following.
[Appendix 1]
an acquisition means for acquiring a question by a questioner at a conference;
an estimation means for estimating answer candidates to a question from the question acquired by the acquisition means using an answer estimation model for estimating answer candidates to the question from the question;
An answer support system comprising: output means for outputting the answer candidate to a terminal device used for answer support.
[Appendix 2]
The output means outputs the answer candidate including information related to the answer.
The response support system described in Supplementary Note 1.
[Appendix 3]
The information related to the answer is information about the questioner,
The response support system described in appendix 2.
[Appendix 4]
The output means outputs the answer candidate to the terminal device used by the person who supports the respondent to support the answer.
The reply support system according to any one of Appendices 1 to 3.
[Appendix 5]
The output means outputs the conference video and the answer candidates to the terminal device.
The response support system described in Supplementary Note 4.
[Appendix 6]
The output means outputs the answer candidate to the terminal device of a person suitable as an answerer of the question, among a plurality of persons who support the answerer.
The reply support system according to appendix 4 or 5.
[Appendix 7]
The estimating means groups related questions to estimate the answer candidates.
The reply support system according to any one of Appendices 1 to 6.
[Appendix 8]
The estimating means estimates answer candidates suitable for the progress of the conference in the question-and-answer session of the conference.
The answer support system according to any one of Appendices 1 to 7.
[Appendix 9]
The estimation means estimates an answer candidate according to the questioner's attribute,
The reply support system according to any one of Appendices 1 to 8.
[Appendix 10]
Further comprising model generation means for generating an answer estimation model for estimating answer candidates to the question from the questions in the meeting,
The reply support system according to any one of Appendices 1 to 9.
[Appendix 11]
captures the question asked by the questioner in the meeting,
Using an answer estimation model for estimating answer candidates for a question from a question, estimating an answer candidate for the question from the obtained question,
outputting the answer candidate to a terminal device used for answer support;
Answer assistance method.
[Appendix 12]
a process of obtaining a question by a questioner in a meeting;
a process of estimating answer candidates to a question from the acquired question using an answer estimation model for estimating answer candidates to the question from the question;
A recording medium for non-temporarily recording, in a terminal device used for answer support, an answer estimation program for causing a computer to execute a process of outputting the answer candidate.

The present invention has been described above using the above-described embodiment as an example. However, the invention is not limited to the embodiments described above. That is, within the scope of the present invention, various aspects that can be understood by those skilled in the art can be applied to the present invention.

10 answer support system 11 acquisition unit 12 estimation unit 13 output unit 14 model generation unit 15 storage unit 20 adviser terminal device 21 answer candidate acquisition unit 22 answer candidate output unit 23 input unit 24 answer output unit 30 answerer terminal device 100 answer support System 101 Acquisition Unit 102 Estimation Unit 103 Output Unit 200 Computer 201 CPU
202 memory 203 storage device 204 input/output I/F
205 Communication I/F

Claims

an acquisition means for acquiring a question by a questioner at a conference;
estimation means for estimating answer candidates to the question from the question acquired by the acquisition means using an answer estimation model for estimating answer candidates to the question from the question;
An answer support system comprising: output means for outputting the answer candidate to a terminal device used for answer support.
The output means outputs the answer candidate including information related to the answer.
The reply support system according to claim 1.
The information related to the answer is information about the questioner,
The answer support system according to claim 2.
The output means outputs the answer candidate to the terminal device used by the person who supports the respondent to support the answer.
An answer support system according to any one of claims 1 to 3.
The output means outputs the conference video and the answer candidates to the terminal device.
The answer support system according to claim 4.
The output means outputs the answer candidate to the terminal device of a person suitable as an answerer of the question, among a plurality of persons who support the answerer.
The reply support system according to claim 4 or 5.
The estimating means groups related questions to estimate the answer candidates.
An answer support system according to any one of claims 1 to 6.
The estimating means estimates answer candidates suitable for the progress of the conference in the question-and-answer session of the conference.
An answer support system according to any one of claims 1 to 7.
The estimation means estimates an answer candidate according to the questioner's attribute,
An answer support system according to any one of claims 1 to 8.
Further comprising a model generation means for generating an answer estimation model for estimating answer candidates to the question from the questions in the meeting,
An answer support system according to any one of claims 1 to 9.
captures the question asked by the questioner in the meeting,
Using an answer estimation model for estimating answer candidates for a question from a question, estimating an answer candidate for the question from the obtained question,
outputting the answer candidate to a terminal device used for answer support;
Answer assistance method.
a process of obtaining a question by a questioner in a meeting;
a process of estimating answer candidates to a question from the acquired question using an answer estimation model for estimating answer candidates to the question from the question;
A recording medium for non-temporarily recording an answer estimation program for causing a computer to execute a process of outputting the answer candidates and a terminal device used for answer support.