Program, method, and information processing device

Abstract

To solve the problem in which there is a need for a technique which makes it even easier to assign tags to a text.SOLUTION: There is provided a program for operating a computer including a computer processor and configured to manage information of a plurality of tags assigned to a text in a storage unit. The program causes the computer processor to perform the steps of: generating an explanatory text describing the tags by providing a large language model with tag information; acquiring a text to which the tags are to be assigned; identifying tags corresponding to the acquired text on the basis of the acquired text and the generated explanatory text of the tags; and presenting the identified tags to a user in association with the acquired text.SELECTED DRAWING: Figure 12

Description

[Technical Field]

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

[0002] Depending on the purpose, tags are attached to text and processed using the tags. For example, by adding tags to classify books, books can be categorized, making it easier to search for books.

[0003] Patent Document 1 describes a platform that accepts questions from questioners and answers from experts via a network. The technology in Patent Document 1 describes assigning tags to questions and identifying candidate answerers based on the tags set for the questions and information set for the answerers. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Patent Publication No. 2022-041304 Summary of the Invention [Problem to be solved by the invention]

[0005] As described above, various processes are performed by assigning tags to text. In this process of assigning tags to text, a human user may assign tags to the text, or a computer may infer candidate tags.

[0006] On the other hand, tags are updated with the times, and new tags may be added. For example, new tags may be added as technology develops, such as artificial intelligence and drones. Also, new business fields may emerge. Tags may be created when a new entity emerges, or new tags may be created when the role of an organization changes.

[0007] This makes it difficult to assign tags to text. For example, when a human user assigns tags, they may not be able to keep track of all newly appearing tags, and as the total number of tags increases, it may become more difficult to assign tags.

[0008] Therefore, there is a need for a technique that makes it even easier to tag text. [Means for solving the problem]

[0009] According to one embodiment of the present disclosure, a program for operating a computer including a computer processor is provided. The program is configured to manage, in a storage unit, information on a plurality of tags assigned to a sentence. The program causes the computer processor to execute the following steps: generating a description describing the tag by providing the tag information to a large-scale language model; acquiring a sentence to which the tag is to be assigned; identifying a tag corresponding to the acquired sentence based on the acquired sentence and the generated tag description; and presenting the identified tag to a user in association with the acquired sentence. [Effects of the Invention]

[0010] According to the present disclosure, it is possible to make it even easier to assign tags to text. [Brief explanation of the drawings]

[0011] [Figure 1] FIG. 1 is a diagram showing the configuration of the system 1. [Figure 2] FIG. 2 is a diagram showing the configuration of the server 20. As shown in FIG. [Figure 3] FIG. 3 is a diagram showing the configuration of the terminal 10. As shown in FIG. [Figure 4]FIG. 4 is a diagram showing the data structure of the questioner database 211. As shown in FIG. [Figure 5] FIG. 5 is a diagram showing the data structure of the respondent database 212. As shown in FIG. [Figure 6] FIG. 6 is a diagram showing the data structure of the tag database 213. As shown in FIG. [Figure 7] FIG. 7 is a diagram showing the data structure of the question and answer history database 214. As shown in FIG. [Figure 8] FIG. 8 is a diagram showing the flow of processing for generating the description of a tag managed in the tag database 213. [Figure 9] FIG. 9 is a diagram showing the flow of processing in which a question is input from a user who asks a question, and an operator user specifies a tag and matches the tag with an answerer. [Figure 10] FIG. 10 is a diagram showing the flow of processing for training a model that estimates tag candidates to be set for a question sentence. [Figure 11] FIG. 11 is a diagram showing the flow of processing for setting the attributes of the respondent user. [Figure 12] FIG. 12 is a diagram showing an example of an operation screen for an operator. DETAILED DESCRIPTION OF THE INVENTION

[0012] Hereinafter, embodiments of the present disclosure will be described with reference to the drawings. In the following description, the same components are denoted by the same reference numerals. The names and functions of the components are also the same. Therefore, detailed descriptions thereof will not be repeated.

[0013] <Outline of embodiment> <1.1 Overall system configuration> FIG. 1 is a diagram showing the configuration of the system 1.

[0014] 1 includes an expert response service server 20, an operator terminal 10, an economic information platform server 91, a large-scale language model service server 93, and a recruitment service server 95. These devices are connected for communication via a network 80.

[0015] In the expert answer service provided by the server 20, multiple experts are registered, and a user who is a questioner posts a question, and a user who is an expert answerer inputs an answer. Here, in the service provided by the server 20, an operator assigns tags to the question text of the questioner. The server 20 extracts candidate answerers for the question text based on the tags assigned to the question text and information such as the career and expertise set by the answerer, and presents the question text to the candidate answerers. When the server 20 receives an answer input from the answerer, it presents the content of the answer to the questioner.

[0016] In this way, the server 20 provides a service that matches a question with a user who can respond to the question by inputting an answer or the like.

[0017] In the illustrated example, one terminal 10 is shown as the terminal used by users of the expert answer service provided by the server 20, but each user operates their own terminal. Terminal 10 is a terminal operated by an operator. Terminal 10A is a terminal operated by the questioner. Terminal 10B is a terminal operated by the answerer.

[0018] Here, the task of recommending tags for a sentence can be achieved by building a model that takes the sentence as input and infers the tag IDs that will likely be assigned to that sentence (the question in the above example). However, the tags assigned to sentences may change over time (for example, due to changes in social issues such as declining birthrates and aging populations, labor shortages, etc., or technological advances such as artificial intelligence technology). In this case, the model must be retrained. Therefore, in order to smoothly provide the service, it is necessary to deal with situations in which new tags are added and the total number of tags changes.

[0019] Therefore, in the following embodiment, we will explain a technology that addresses the above problem by expanding tag data and learning the distance between the text and the expanded tag data. Specifically, we address the above problem from the following two perspectives.

[0020] (1) Tag names alone are insufficient as input for distance learning. To address the above problem, tags are usually given short names, which may result in an insufficient amount of information to be used as input for metric learning.

[0021] (2) When obtaining a model that infers tag candidates for a sentence, it is desirable to prepare negative examples for use in metric learning. If a tag has already been assigned to a sentence, the assigned tag is considered to be a positive example for the sentence. On the other hand, to obtain a model that uses metric learning to infer tag candidates for a sentence, it is considered desirable to have information on negative example tags for the sentence.

[0022] To address the above (1) and (2), the following embodiment describes an example in which a description of tag names is generated and distance learning is performed using a dataset extended by the description of tag names. Furthermore, in the following embodiment, candidate tags to be assigned to a question are presented to an operator, and the operator specifies the tags to be assigned to the question. Therefore, information on tags not assigned by the operator to the question can also be obtained, and tags assigned by the operator can be treated as positive examples, and tags not assigned by the operator can be treated as negative examples.

[0023] In the following example, the expert answer service provided by the server 20 may have a system in which answerers are evaluated if they enter an answer within a predetermined time after the asker posts a question. This shortens the time from when the asker asks a question to when they receive an answer. In this case, by making it easier for an operator to tag the question text in order to match the asker's question with an answerer, the asker can more easily enjoy the value of receiving an answer from the answerer quickly.

[0024] In the following description of the embodiment, an example will be described in which the system 1 assists an operator in assigning tags to a question received from a questioner, as assistance in setting tags to a sentence.

[0025] In this embodiment, each device (terminal device, server, etc.) can also be considered as an information processing device. That is, a collection of devices can be considered as one "information processing device," and system 1 can be formed as a collection of multiple devices. The way in which multiple functions required to realize system 1 according to this embodiment are allocated to one or multiple pieces of hardware can be determined appropriately in consideration of the processing capacity of each piece of hardware and / or the specifications required for system 1.

[0026] The terminal 10 is a device operated by a user. The terminal 10 is realized, for example, as follows. Desktop PCs (Personal Computers), laptop PCs · Handheld devices such as smartphones and tablets Wearable devices worn by users (wristwatches, glasses, etc.) The terminal 10 includes a communication IF (Interface) 12 , an input device 13 , an output device 14 , a memory 15 , a storage 16 , and a processor 19 .

[0027] The communication IF 12 is an interface for inputting and outputting signals so that the terminal 10 can communicate with an external device.

[0028] The input device 13 is a device for receiving input operations from a user (for example, a touch panel, a touch pad, a pointing device such as a mouse, a keyboard, etc.).

[0029] The output device 14 is a device (such as a display or speaker) for presenting information to the user.

[0030] The memory 15 is for temporarily storing programs and data to be processed by the programs, and is a volatile memory such as a DRAM (Dynamic Random Access Memory).

[0031] The storage 16 is for storing data, and is, for example, a flash memory or a hard disk drive (HDD).

[0032] The processor 19 is hardware for executing an instruction set written in a program, and is composed of an arithmetic unit, a register, a peripheral circuit, and the like.

[0033] The server 20 is a device for providing users with an expert answer service in which an answerer answers a question from a questioner as described above.

[0034] The server 20 includes a communication IF 22 , an input / output IF 23 , a memory 25 , a storage 26 , and a processor 29 .

[0035] The communication IF 22 is an interface for inputting and outputting signals so that the server 20 can communicate with external devices.

[0036] The input / output IF 23 functions as an interface with an input device for receiving input operations from the user and an output device for presenting information to the user.

[0037] The memory 25 is for temporarily storing programs and data to be processed by the programs, and is a volatile memory such as a DRAM (Dynamic Random Access Memory).

[0038] The storage 26 is for storing data, and is, for example, a flash memory or a hard disk drive (HDD).

[0039] The processor 29 is hardware for executing an instruction set written in a program, and is composed of an arithmetic unit, registers, peripheral circuits, and the like.

[0040] The economic information platform server 91 distributes economic information such as economic news distributed by the media, and technical information such as technical papers or patent documents. The economic information platform server 91 distributes this information by receiving articles from media that distribute information, or by crawling publicly available information on the Internet, for example.

[0041] The large-scale language model service server 93 is a server that executes language processing tasks using a language model constructed by a learning process including artificial intelligence (AI). An LLM (Large Language Model) is a server that has previously learned a large amount of large-scale data (text data, etc.), for example, web content on the Internet, or a large amount of data stored in a specified database, and can execute various language processing tasks by giving the LLM a task.

[0042] The large-scale language model service server 93 accepts prompt inputs such as text, images, and voice, and generates and responds to the prompts. Examples of LLMs include GPT-3 and GPT-4 developed by OpenAI, and BERT developed by Google. In this embodiment, the large-scale language model service server 93 may use an LLM that has been trained in advance on documents publicly available on the Internet, information distributed by the economic information platform server 91, and the like, as follows: Information such as news articles, technical papers, patent documents, etc. distributed by the economic information platform server 91 The human resources introduction service server 95 provides a service of introducing human resources to perform work, and is a system that provides, for example, a service of matching a business contractor who provides work through outsourcing with a client, or a service of supporting recruitment by employers.

[0043] <1.2 Functional configuration of server 20> 2 is a diagram showing the configuration of the server 20. As shown in FIG. 2, the server 20 functions as a communication unit 201, a storage unit 202, and a control unit 203.

[0044] The communication unit 201 performs processing for the server 20 to communicate with external devices.

[0045] The storage unit 202 stores various data such as a questioner database 211, an answerer database 212, a tag database 213, a question and answer history database 214, and a trained model 215.

[0046] The questioner database 211 is a database of users who are questioners, and will be described in detail later.

[0047] The respondent database 212 is a database of users who are respondents, and will be described in detail later.

[0048] The tag database 213 is a database of tags, which will be described in detail later.

[0049] The question and answer history database 214 is a database that manages answers by answerers to questions by askers, as will be described in detail later.

[0050] The trained model 215 is a trained model used to estimate tags to be assigned to a question entered by a questioner. Specifically, the trained model 215 receives a question and returns a vector of the question. The server 20 obtains the similarity between the vector of the question obtained using the trained model 215 and the vector of the tag using a function that returns the similarity of vectors. That is, the server 20 holds a function that receives a pair of a tag and a question and returns the similarity. Specifically, when a question is entered, the server 20 vectorizes the meaning of the question and recommends tags whose vectors are close to each other in the vector space from among the vectors of tags that have been pre-mapped in the vector space.

[0051] This allows the trained model 215 to identify tags that are determined to have a high degree of similarity to the question entered by the questioner, eliminating the need for the operator to search through all tags and consider which tags to assign.

[0052] The control unit 203 is realized by the processor 29 reading a program stored in the storage unit 202 and executing instructions included in the program. By operating in accordance with the program, the control unit 203 performs functions such as a reception control module 2041, a transmission control module 2042, a user management module 2043, and a tag explanation generation module 2044. A question reception module 2045, a question tag setting module 2046, an answer reception module 2047, a learning processing module 2048, and an answerer tag setting module 2049.

[0053] The reception control module 2041 controls the process by which the server 20 receives signals from external devices in accordance with a communication protocol.

[0054] The transmission control module 2042 controls the process in which the server 20 transmits signals to external devices in accordance with a communication protocol.

[0055] The user management module 2043 is a module for managing information about each user who uses the system 1. Specifically, the user management module 2043 accepts registration of information about users who are questioners, and updates the questioner database 211. The user management module 2043 accepts registration of information about users who are answerers, and updates the answerer database 212.

[0056] The tag description generation module 2044 is a program module that generates descriptions for tags managed in the tag database 213 .

[0057] The tag description generation module 2044 obtains the tag description from the large-scale language model service server 93 by issuing a prompt including an instruction to generate a description of the word indicated in the tag to the large-scale language model service server 93. Specifically, the tag description may be obtained as follows. A prompt including an instruction to create an explanatory text for a specified word (a tag managed in the tag database 213) is given to the server 93 of the large-scale language model service. After searching the internet and services provided by the economic information platform server 91 using the words indicated in the tag, a prompt including the search results and an instruction to summarize the search results is provided to the large-scale language model service server 93. Searching for information for which viewing rights are set for members of an organization such as a business company using the words indicated in the tags, and then providing the search results and a prompt including an instruction to summarize the search results to the large-scale language model service server 93. In the expert response service provided by the server 20, an expert is requested to define the word indicated in the tag, and the response from the expert is used as the tag description. In addition, the response from the expert is summarized by the server 93 of the large-scale language model service and used as the tag description. An example of a tag description generated in this way is shown below.

[0058] Tag name: "Healthcare", Tag description: "Healthcare refers to a wide range of services and activities related to people's health.... It also leads to economic benefits and increased productivity." Tag name: "Electricity", Tag description: "Electricity refers to electrical energy. Electricity is energy generated by the flow of electrons... We need to work to build a sustainable future." The question receiving module 2045 is a program module that receives input of a question from a user who is asking a question.

[0059] The question reception module 2045 receives input of a question from a questioner and updates the question and answer history database 214 .

[0060] The question tag setting module 2046 is a program module that sets tags for questions received from questioners.

[0061] The question tag setting module 2046 sets a tag for the question sentence as follows. By inputting a question to the trained model 215, candidate tags that are estimated are presented to the operator. - Allows operators to specify tags for questions By inputting a question to the trained model 215, the estimated tags are associated with the question. The answer receiving module 2047 is a program module that receives an answer input from an answerer user to a question sentence.

[0062] The answer receiving module 2047 receives answers input from answerers and updates the question and answer history database 214 .

[0063] The learning processing module 2048 is a program module that performs learning processing based on a learning dataset to obtain a trained model.

[0064] The learning processing module 2048 generates a trained model for estimating tags for input question sentences using the question sentences stored in the question and answer history database 214, candidate tags presented to the operator for the question sentences, tags specified by the operator, and tags presented to the operator but not specified by the operator as learning data.

[0065] The learning processing module 2048 learns distance using tags that are positive examples and tags that are negative examples for a question. Specifically, the learning processing module 2048 performs distance learning by fine-tuning a trained model that has been trained in advance. Here, using pairs of questions and tags that are positive examples and pairs of questions and tags that are negative examples, the loss function is defined using squared loss so that the cosine similarity of the embedded representations of positive example pairs approaches 1 and the cosine similarity of the embedded representations of negative example pairs approaches -1.

[0066] The respondent tag setting module 2049 is a program module that sets an attribute tag indicating the attribute of the respondent user to the respondent.

[0067] The answerer tag setting module 2049 refers to the question and answer history database 214 and extracts the attribute tag of the answerer's user based on the tag set in the question sentence to which the answerer answered. The answerer tag setting module 2049 presents the extracted attribute tag candidates to the answerer, accepts an operation to specify the attribute tag from the answerer, sets the answerer's attribute tag, and updates the answerer database 212.

[0068] <1.3 Configuration of Terminal 10> FIG. 3 is a diagram showing the configuration of the terminal 10. As shown in FIG.

[0069] As shown in FIG. 3, the terminal 10 includes multiple antennas (antenna 111, antenna 112), communication units (first communication unit 120, second communication unit 121) corresponding to the respective antennas, an input device 130 (including a touch-sensitive device 131), a display 132, an audio processing unit 140, a microphone 141, a speaker 142, a position information sensor 150, a camera 160, a motion sensor 170, a memory unit 180, and a control unit 190. The terminal 10 also has functions and configurations (e.g., a battery for storing power, a power supply circuit for controlling the supply of power from the battery to each circuit, etc.) that are not specifically shown in FIG. 3. As shown in FIG. 3, the blocks included in the terminal 10 are electrically connected by a bus or the like.

[0070] The antenna 111 emits a signal emitted by the terminal 10 as a radio wave. The antenna 111 also receives a radio wave from space and provides the received signal to the first communication unit 120.

[0071] The antenna 112 emits a signal emitted by the terminal 10 as a radio wave. The antenna 112 also receives a radio wave from space and provides the received signal to the second communication unit 121.

[0072] The first communication unit 120 performs modulation / demodulation processing and the like for transmitting and receiving signals via the antenna 111 so that the terminal 10 can communicate with other wireless devices. The second communication unit 121 performs modulation / demodulation processing and the like for transmitting and receiving signals via the antenna 112 so that the terminal 10 can communicate with other wireless devices. The first communication unit 120 and the second communication unit 121 are communication modules including a tuner, a received signal strength indicator (RSSI) calculation circuit, a cyclic redundancy check (CRC) calculation circuit, a high-frequency circuit, and the like. The first communication unit 120 and the second communication unit 121 perform modulation / demodulation, frequency conversion, and the like for wireless signals transmitted and received by the terminal 10, and provide the received signals to the control unit 190.

[0073] Input device 130 has a mechanism for accepting input operations by a user. Specifically, input device 130 is configured as a touch screen and includes touch-sensitive device 131. Touch-sensitive device 131 accepts input operations by a user of terminal 10. Touch-sensitive device 131 detects the user's touch position on the touch panel, for example, by using a capacitive touch panel. Touch-sensitive device 131 outputs a signal indicating the user's touch position detected by the touch panel to control unit 190 as an input operation.

[0074] The display 132 displays data such as images, videos, and text under the control of the control unit 190. The display 132 is realized by, for example, an LCD, an organic EL display, or the like.

[0075] The audio processing unit 140 modulates and demodulates audio signals. The audio processing unit 140 modulates a signal provided from the microphone 141 and provides the modulated signal to the control unit 190. The audio processing unit 140 also provides the audio signal to the speaker 142. The audio processing unit 140 is realized, for example, by a processor for audio processing. The microphone 141 accepts audio input and provides an audio signal corresponding to the audio input to the audio processing unit 140. The speaker 142 converts the audio signal provided from the audio processing unit 140 into audio and outputs the audio to the outside of the terminal 10.

[0076] The location information sensor 150 is a sensor that detects the location of the terminal 10, and is, for example, a GPS (Global Positioning System) module. The GPS module is a receiving device used in a satellite positioning system. The satellite positioning system receives signals from at least three or four satellites, and detects the current location of the terminal 10 equipped with the GPS module based on the received signals.

[0077] The camera 160 is a device that receives light with a light receiving element and outputs the received light as a captured image. The camera 160 is, for example, a depth camera that can detect the distance from the camera 160 to a subject being photographed.

[0078] The motion sensor 170 includes an acceleration sensor, an angular velocity sensor, etc., and detects the movement of the terminal 10 .

[0079] The storage unit 180 is configured with, for example, a flash memory or the like, and stores data and programs used by the terminal 10. The various types of information stored in the storage unit 180 will be described later.

[0080] The control unit 190 controls the operation of the terminal 10 by reading a program stored in the storage unit 180 and executing instructions included in the program. The control unit 190 is, for example, an application processor. By operating in accordance with the program, the control unit 190 fulfills the functions of an operation reception unit 191, a transmission / reception unit 192, a data processing unit 193, a notification control unit 194, and a storage control unit 195.

[0081] Operation acceptance unit 191 performs processing to accept a user's input operation to an input device such as touch-sensitive device 131. Operation acceptance unit 191 determines the type of operation, such as whether the user's operation is a flick operation, a tap operation, or a drag (swipe) operation, based on information about the coordinates where the user has touched touch-sensitive device 131 with a finger or the like.

[0082] The transmitting / receiving unit 192 performs processing for the terminal 10 to transmit and receive data to and from an external device such as the server 20 in accordance with a communication protocol.

[0083] The data processing unit 193 performs calculations on the data that the terminal 10 has received as input in accordance with a program, and outputs the calculation results to a memory or the like.

[0084] The notification control unit 194 performs processing for displaying a display image on the display 132, processing for outputting sound from the speaker 142, and processing for generating vibrations.

[0085] The storage control unit 195 controls the storage of data in the storage unit 180 .

[0086] A description will be given of various types of information stored in storage unit 180. In one aspect, storage unit 180 stores various types of information such as user information 181, question history information 182, and the like.

[0087] The user information 181 is information about a user who uses the services of the server 20. The user information 181 includes, for example, information such as the user's name, affiliation, and contact information such as an email address.

[0088] The question history information 182 is information indicating the history of questions asked using the services of the server 20.

[0089] <2 Data Structure> 4 is a diagram showing the data structure of the questioner database 211. The questioner database 211 includes an item "questioner user ID," an item "name," an item "email address," an item "organization," an item "department," an item "job title," an item "business area," and an item "questioner score."

[0090] The item "questioner user ID" is information for identifying the user who is the questioner.

[0091] The item "Name" is information about the user's name.

[0092] The item "email address" is information about the email address used to contact the user.

[0093] Specifically, the item "Email address" may include information that identifies the user (email address, name of the user's account in the service of server 20, etc.) in order to accept login from the user to the service provided by server 20.

[0094] The item "organization" is information indicating the organization to which the user belongs.

[0095] The item "Department" is information indicating the department to which the user belongs.

[0096] The item "job title" is information indicating the job title of the user.

[0097] The item "business area" is information about the area of ​​business conducted by the organization to which the user belongs.

[0098] The item "questioner score" is information on an evaluation value determined according to the track record of asking questions as a questioner.

[0099] Specifically, the item "questioner score" may include information on an evaluation value that evaluates the user who is the questioner as follows, with reference to the question and answer history database 214. Number and frequency of questions asked The number of times a question was asked and answered by respondents, the frequency, and the time it took to answer - The amount of information in each question (number of characters, etc.) -Evaluation value of the respondent who answered the question (respondent database 212) 5 is a diagram showing the data structure of the respondent database 212. The respondent database 212 includes an item "respondent user ID", an item "name", an item "email address", an item "organization", an item "department", an item "job title", an item "attribute tag", an item "qualification", and an item "respondent score".

[0100] The item "Answerer User ID" is information for identifying the user who is the answerer.

[0101] The item "Name" is information about the user's name.

[0102] The item "email address" is information about the email address used to contact the user.

[0103] The item "organization" is information indicating the organization to which the user belongs.

[0104] The item "Department" is information indicating the department to which the user belongs.

[0105] The item "job title" is information indicating the job title of the user.

[0106] The item "attribute tag" is information indicating the attribute of the user who is the respondent.

[0107] Specifically, the item "attribute tag" may include the following tag information indicating the respondent's experience and expertise, and multiple tags may be set for the respondent user. Tags that indicate the industry or business area ·Tags that indicate technical content Tags that indicate roles, functions, etc. in an organization The item "Qualification" is information about the qualifications held by the user who is the respondent.

[0108] Specifically, the item "Qualifications" may include the following qualification information that indicates the respondent's experience and expertise: Qualifications recognized as national qualifications Qualifications recognized by industry associations, etc. The item "answerer score" is information on an evaluation value determined according to the record of answers given by the answerer.

[0109] Specifically, the item "answerer score" may refer to the question and answer history database 214 and include information on an evaluation value that evaluates the user who is the questioner as follows. Number of times questions were answered, frequency, and time taken to answer - The amount of information in the answers to each question (number of characters, etc.) - The questioner's rating for the question they answered (Questioner Database 211) 6 is a diagram showing the data structure of the tag database 213. The tag database 213 includes an item "question tag ID," an item "tag," an item "tag registration date," an item "tag use start date," an item "tag description," an item "description update date," an item "vector," and an item "cluster."

[0110] The item "Question tag ID" is information for identifying a tag to be assigned to a question sentence.

[0111] The item "tag" is information indicating the name of a tag.

[0112] Specifically, the item "tag" includes words and keywords that represent the following content: ·Terms that indicate business areas ·Terms that indicate technical content · Words that indicate organizational departments, roles, job duties, etc. The item "tag registration date" is information indicating when the tag was registered in the service provided by the server 20.

[0113] Specifically, the item "tag registration date" includes information about when the administrator of the service of the server 20 registered the tag.

[0114] The item "tag use start date" is information indicating the time when the tag becomes available for use in the service provided by the server 20.

[0115] Specifically, the item "Tag Use Start Date" includes information about the time when an operator can set a tag for a question. For example, there may be cases where the availability of a tag is determined depending on the time as follows: - A term has been newly defined due to a legal amendment, etc., and the period during which that term can be used has been determined. The item "tag description" is information about the description of the tag.

[0116] Specifically, the item "tag description" includes information on the result of processing performed by the tag description generation module 2044 to generate a tag description.

[0117] The item "description update date" is information indicating when the description of the tag was updated.

[0118] Specifically, the item "description update date" includes information on when the tag description generation module 2044 generated the tag description and updated the tag description stored in the tag database 213.

[0119] The item "vector" is a vector obtained by converting the tag description or tag name into an array of numerical values.

[0120] Specifically, the item "vector" holds vectors that represent the meaning of words or sentences, and includes information on vectors obtained by a method such as word2vec.

[0121] The item "cluster" is information for identifying each cluster when tag vectors are classified into clusters.

[0122] Specifically, the item "cluster" may include information on the results of classifying tags into clusters as follows. Classification using the k-means method. Randomly assign a class to each vector data corresponding to each tag, calculate the centroid of the vector for each class, and then reassign each data to the class with the closest centroid, repeating recursively until convergence occurs. 7 is a diagram showing the data structure of the question and answer history database 214. The question and answer history database 214 includes an item "Question ID," an item "Asker User ID," an item "Question Registration Date," an item "Question Content," an item "Question Text Vector," an item "Question Tag Candidate," an item "Operator User ID," an item "Operator-Specified Tag," an item "Tag Setting Date," an item "Answer Candidate," an item "Answer User ID," and an item "Answer Content."

[0123] The item "Question ID" is information that identifies each question.

[0124] The item "questioner user ID" is information for identifying the user who is the questioner.

[0125] The item "Question registration date" is information indicating the time when the service provided by server 20 accepted the registration of a question from a questioner.

[0126] The item "Question content" is information about the question entered by the questioner.

[0127] Specifically, the item "question content" includes information on the content of the question received by the question receiving module 2045 from the questioner.

[0128] The item "Question sentence vector" is a vector obtained by vectorizing the question sentence.

[0129] Specifically, the item "question sentence vector" holds a vector that expresses the meaning of a word or sentence, and includes information on the vector obtained by a method such as word2vec.

[0130] The item "Question tag candidate" is information on candidate tags to be set for the question received from the questioner.

[0131] Specifically, the item "Question tag candidate" includes information on tag candidates identified for a question by the question tag setting module 2046. The tag candidates are presented to the operator along with the question, and assist the operator in specifying tag candidates for the question.

[0132] The item "operator user ID" is information for identifying an operator who is a user who sets a tag to a question.

[0133] The item "operator-specified tag" is information about the tag specified by the operator for the question.

[0134] The item "tag setting date" is information about the time when the operator specified a tag in response to the question.

[0135] The item "answer candidate" is information about the answer candidate who will answer the question.

[0136] Specifically, the item "answer candidate" includes information on answer candidate identified as follows. Based on the degree of match between the tags set by the operator for the question and the respondent's attribute tags, respondents with a high degree of match are prioritized and extracted as candidates. The item "Answerer User ID" is information for identifying the user who is the answerer.

[0137] The item "Answer Content" is information indicating the content of the answer entered by the answerer in response to the question.

[0138] The item "answer date" is information indicating the time when the answer input from the answerer was received.

[0139] <3 operations> FIG. 8 is a diagram showing the flow of processing for generating the description of a tag managed in the tag database 213.

[0140] In step S821, the tag description generation module 2044 of the server 20 reads each tag in the tag database 213. The tag database 213 is configured to manage information on multiple tags assigned to questions entered by a questioner. The tag description generation module 2044 generates a prompt including an instruction to generate a description that explains the tag. The tag description generation module 2044 transmits the generated prompt to the server 93 of the large-scale language model service.

[0141] In step S823, the tag description generation module 2044 of the server 20 receives the tag description generated by the server 93 of the large-scale language model service from the server 93.

[0142] For example, the tag description generation module 2044 sends the following to the server 93 of the large-scale language model service:

[0143] (i) providing a prompt to generate a description that describes the tag;

[0144] (ii) using keywords representing the tags to search the web, including the services provided by the economic information platform server 91, and sending a prompt summarizing the search results; A description for explaining the tag may be generated by at least one of the above.

[0145] The server 20 updates the tag database 213 based on the generated tag description.

[0146] In step S825, the server 20 vectorizes the generated tag description and updates the tag database 213.

[0147] In step S827, the server 20 classifies the tags into clusters based on the proximity of the vectors of the tags. For example, the server 20 classifies the tags by classifying the vectors of the tags into clusters using the k-means algorithm, and updates the tag database 213.

[0148] FIG. 9 is a diagram showing the flow of processing in which a question is input from a user who asks a question, and an operator user specifies a tag and matches the tag with an answerer.

[0149] In step S921, the question receiving module 2045 of the server 20 acquires a question sentence (a sentence to which a tag is to be added; a sentence requesting an answer from an answerer) input by the user (terminal 10A) who is the asker of the question, and updates the question and answer history database 214. The server 20 vectorizes the question sentence and updates the question and answer history database 214.

[0150] In step S923, the question tag setting module 2046 of the server 20 identifies candidate tags to be associated with the question sentence based on the distance between the vector of the question sentence and the vector of the tag.

[0151] For example, the question tag setting module 2046 may refer to the question and answer history database 214 and the tag database 213, and identify tags (tags with similar meanings) whose distance between the question vector obtained from the user asking the question and the tag vector is close within a certain range as candidates for tags to be associated with the question.

[0152] In step S925, the question tag setting module 2046 of the server 20 presents the question text of the questioner user and the identified tag candidates to the operator user.

[0153] In step S911, the terminal 10 displays on the screen the question text of the questioner user and the tag candidates identified by the question tag setting module 2046. The terminal 10 receives, from the operator user, the designation of the tag to be associated with the question text.

[0154] Here, the question tag setting module 2046 of the server 20 may present to the operator user tags whose vectors are close to the question text within a certain distance, and accept from the operator user a specification of the tag to be set for the acquired question text.

[0155] The question tag setting module 2046 may distinguish between tags that are close within a certain distance between the vector of the question text and the vector of the explanation text of the answer and those that are particularly relevant (for example, a predetermined number of tags in order of highest relevance) and those that are not, and present these to the operator user, and accept from the user the specification of tags to be set to the question text.

[0156] The server 20 may store, among the tags presented to the operator, tags designated by the operator's user as positive examples, and tags not designated by the user as negative examples, in association with the sentence obtained from the questioner in the question and answer history database 214.

[0157] The question tag setting module 2046 may display the question text and the identified tag candidates together on the screen of the operator's terminal 10 for the operator user, and may accept an operation to specify a tag from the operator. In response to the operation of the operator user to specify a tag, the question tag setting module 2046 may associate the question text with the specified tag and store them in the question and answer history database 214.

[0158] The question tag setting module 2046 may display on the screen for the operator user, among the tag candidates, tags that are highly relevant to the question sentence, distinguishing them from other tags.

[0159] The question tag setting module 2046 may display candidate tags that are highly relevant to the question sentence on the screen for the operator user in a pre-designated state without the operator user having to perform a designation operation.

[0160] The question tag setting module 2046 refers to the answerer database 212 and provides the operator user with a question tag based on the attribute tag set for the answerer.

[0161] (i) potential respondents associated with the potential tags identified for the question; or

[0162] (ii) Candidates of respondents related to a tag designated by the operator user may be presented.

[0163] The question tag setting module 2046 accepts an operation from the operator user to specify a tag to be presented, and updates the question and answer history database 214 by associating the specified tag with the question text.

[0164] In step S927, the question tag setting module 2046 of the server 20 updates the question and answer history database 214 based on the tag designated by the operator for the question sentence.

[0165] The answer receiving module 2047 extracts answerer candidates based on the tags associated with the question and the attributes of the answerers, by referring to the question and answer history database 214 and the answerer database 212. For example, the higher the degree of match between the question tag and the answerer's attribute tag, the higher the priority of extraction as an answerer candidate.

[0166] The answer receiving module 2047 presents a question to the answerer candidates and receives answer inputs from the answerer candidates.

[0167] In step S929, the answer receiving module 2047 of the server 20 refers to the question and answer history database 214 and presents the answer of the answerer to the questioner.

[0168] FIG. 10 is a diagram showing the flow of processing for training a model that estimates tag candidates to be set for a question sentence.

[0169] In step S1021, the server 20 refers to the question and answer history database 214 and reads out the tags of positive examples and negative examples associated with the question sentence as learning data.

[0170] The server 20 may refer to the question and answer history database 214, and identify tags added by the operator user to sentences obtained from the questioner user as positive examples, and identify tags of clusters different from the tags of the positive examples as negative examples, and use these as a data set for learning.

[0171] In step S1023, the learning processing module 2048 of the server 20 performs learning processing based on the learning dataset to generate a learned model that outputs tags from the tag database 213 for the sentence.

[0172] In step S1025, the server 20 stores the generated trained model 215 in the storage unit 202.

[0173] FIG. 11 is a diagram showing the flow of processing for setting the attributes of the respondent user.

[0174] In step S1121, the answerer tag setting module 2049 of the server 20 refers to the question and answer history database 214 and reads out the tag associated with the question sentence to which the answerer has answered.

[0175] In step S1123, the respondent tag setting module 2049 determines candidate attribute tags to be set to the respondent from among the attribute tags indicating the attributes of the respondent, based on the tags of the question sentence to which the respondent answered. For example, tags of the question sentence to which the respondent answered that appear more frequently than a certain level, or those with the highest appearance frequency, may be set as candidate attribute tags to be set to the respondent.

[0176] In step S1125, the respondent tag setting module 2049 presents candidate attribute tags to the respondent.

[0177] In step S1111, the terminal 10B of the answerer's user displays candidate attribute tags and receives the designation of an attribute tag from the answerer's user.

[0178] In step S1127, the respondent tag setting module 2049 updates the respondent database 212 based on the attribute tag specified by the respondent user.

[0179] <4 Screen example> FIG. 12 is a diagram showing an example of an operation screen for an operator.

[0180] The operation screen 1200 is a screen that accepts an operation from an operator user to specify a tag for a question.

[0181] The tag designation receiving unit 1202 is an operation member that receives tag designation from the operator.

[0182] In the illustrated example, the tag designation receiving unit 1202 displays the tag designated by the user and receives an operation to delete the tag, which corresponds to steps S925 and S911 in FIG.

[0183] The tag candidate display area 1204 is an area that displays candidate tags to be set in the question.

[0184] The tag candidate display area 1204 corresponds to steps S925 and S911 in FIG.

[0185] The tag confirmation operation receiving unit 1206 is an operation member that receives an operation from the user to confirm the tag to be set in the question.

[0186] The tag confirmation operation receiving unit 1206 receives an operation from the user to confirm a tag to a question, extract answerer candidates based on the tag, and recommend answers to the question to the answerer candidates.

[0187] The highly relevant tag candidates 1208 are tags that are highly relevant to the question sentence among the tag candidates identified by the question tag setting module 2046 .

[0188] The highly relevant tag candidates 1208 compare the distance between the vector of the question sentence and the vector of the tag, and display the predetermined number of tags with the closest distance as tags with particularly high relevance in a manner that allows them to be distinguished from other highly relevant tags (tags in the tag candidates 1210). In the example shown in the figure, an operation to delete tag candidates is accepted from the user.

[0189] The tag candidates 1210 are candidates for tags that are extracted as being highly relevant by comparing the distance between the vector of the question sentence and the vector of the tag.

[0190] In the illustrated example, the tag candidate 1210 accepts an operation from the user to delete a tag candidate.

[0191] The question display area 1212 is an area for displaying the question entered by the user who is asking the question.

[0192] The question detail display area 1214 is an area for displaying the details of the question.

[0193] In the illustrated example, the question details display area 1214 displays the following details of the question: Questioner (Questioner Database 211) Questions (Question and Answer History Database 214) - Deadline for responses from respondents The search operation receiving section 1216 is an operating member that receives an operation to search for other questions similar to the current question.

[0194] The search operation receiving unit 1216 searches for other past questions similar to the current question, for example, in the following manner. - Past questions that are closer than a certain distance in vector to the question -Other questions posted in the past have been posted around the same time -Other previous questions from the same user The question switching operation receiving unit 1218 is an operation member that receives an operation to switch the question sentence for which the operator designates a tag to another question sentence.

[0195] The question switching operation receiving unit 1218 The answerer candidate display area 1220 is an area that displays answerer candidates related to the question sentence or tag candidate.

[0196] The respondent candidate display area 1220 corresponds to the processing of step S911 in FIG.

[0197] The details display area 1222 is an area for displaying details of the candidate answerer.

[0198] In the illustrated example, the detailed display area 1222 displays the following candidate respondents. Respondents who have attribute tags that match closely (are similar to) the tag candidates related to the question Respondents who have attribute tags that match (are similar to) the tags specified by the operator for the question <Modification> In addition to the aspects described in the above embodiment, the following may be adopted: The following will be described along the chronological order of experiences from when the user starts using the service.

[0199] (1) The text to which the tag is to be added The above embodiment has been described mainly as an example of adding tags to a question in an expert answer service in which a question from a questioner is matched with an answerer such as an expert, and the answerer answers the question.

[0200] In addition, the invention described in the embodiment can also be applied to examples in which tags are added to various sentences such as the following. Tagging of information provided by the economic information platform server 91 (news releases, IR information of business companies, posts such as blog articles, technical documents such as papers or patent documents) Information provided by the server 95 of the recruitment service (tagging of descriptions of work content for recruitment by outsourcers, tagging of descriptions of job information for recruitment by job seekers, and support for tagging of personnel or job seekers who work on a contract basis (for example, suggesting tag candidates based on self-promotional statements)) As a result, as the amount of text increases and as more tags are added, it becomes more difficult to tag the text manually. However, the invention described in the above embodiment can assist in the tagging process, further encouraging tag-based matching in the service.

[0201] (2) How to generate tag descriptions In the above embodiment, an example has been described in which the tag description generation module 2044 causes the server 93 of the large-scale language model service to generate a description of the tag name.

[0202] In addition to generating a tag description using a language model, it is also possible to search for information on the Internet, services provided by the economic information platform server 91, etc. using the words indicated in the tag, and obtain the tag description based on the search results (for example, if a sentence explaining the tag appears at the top of the search results).

[0203] Alternatively, a search may be performed using the words indicated in the tags for information that is not publicly available and can be viewed internally by the business company, and the description of the tag may be obtained based on the search results.

[0204] (3) Tag classification In the above embodiment, an example has been described in which tags are classified into clusters and question sentences are matched with tags.

[0205] Alternatively, tags may not be classified, and a trained model may be obtained that recommends tags to be assigned to questions based on the distance between the question vector and the tag vector.

[0206] (4) Matching questions with respondents In the above embodiment, an example has been described in which answerer candidates are extracted based on the degree of coincidence between tags set for questions and attribute tags indicating the attributes of answerers.

[0207] Alternatively, the server 93 of the large-scale language model service may generate explanatory text for attribute tags of the respondent's attributes, and the explanatory text for the attribute tags may also be vectorized using a technique such as word2vec to perform matching. For example, candidate respondents may be identified depending on the proximity of the vector of the tag set for the question sentence to the vector of the respondent's attribute tag.

[0208] (5) Users are given benefits in other services based on their scores on the expert answer service. In the above embodiment, an example has been described in which the questioner database 211 and the answerer database 212 manage the scores of questioner users and answerer users.

[0209] In addition, depending on the score of each user, the user may receive benefits in each service provided by, for example, the recruitment service server 95 or the economic information platform server 91. For example, the following benefits may be available. - As a contractor, it increases the chances of receiving a favorable business request. For example, if a user has a score above a certain level, they can receive a specific business request. In recruiting, it increases the chances of receiving a job offer with good conditions. For example, if a job offer is available only to a limited number of users, users with a score above a certain level can apply for the job offer. - In economic information platforms that have a SNS (Social Network Service) function that allows users to comment on news articles and other content and allow third parties to view the comments, it increases the likelihood that the posted comments will be viewed by other users, allowing for monetization. Although several embodiments of the present disclosure have been described above, these embodiments can be embodied in various other forms, and various omissions, substitutions, and modifications can be made without departing from the spirit of the invention. These embodiments and modifications are intended to be included in the scope of the inventions and their equivalents as defined in the claims, as well as in the scope and spirit of the inventions.

[0210] The functions performed by the components described herein may be implemented in circuitry or processing circuitry, including general-purpose processors, application-specific processors, integrated circuits, ASICs (Application Specific Integrated Circuits), a CPU (a Central Processing Unit), conventional circuits, and / or combinations thereof, programmed to perform the described functions. A processor includes transistors and other circuits and is considered to be circuitry or processing circuitry. A processor may also be a programmed processor that executes programs stored in memory.

[0211] In this specification, a circuitry, unit, or means is hardware that is programmed to realize or performs the described functions, which may be any hardware disclosed herein or any hardware known to be programmed to realize or perform the described functions.

[0212] If the hardware is a processor considered to be a type of circuitry, the circuitry, means, or unit is a combination of the hardware and software used to configure the hardware and / or processor.

[0213] <Additional Notes> The matters explained in the above embodiment will be supplemented below.

[0214] (Appendix 1) A program for operating a computer having a computer processor, the program being configured to manage, in a memory unit, information on a plurality of tags to be assigned to sentences, the program causing the computer processor to execute the steps of: generating a description that explains the tag by providing tag information to a large-scale language model; acquiring a sentence to which a tag is to be assigned; identifying a tag corresponding to the acquired sentence based on the acquired sentence and the generated tag description; and presenting the identified tag to a user in association with the acquired sentence.

[0215] (Appendix 2) The program of Appendix 1, wherein in the generating step, the description describing the tag is generated by at least one of providing a prompt to a large-scale language model to generate a description describing the tag, and using keywords representing the tag to search the web, including a platform that distributes economic information, and summarizing the search results.

[0216] (Appendix 3) A program described in any of Appendices 1 to 2, wherein in the generating step, the description of the tag to be generated is vectorized; in the acquiring step, the acquired text is vectorized; in the identifying step, tags whose vectors are close to each other within a certain distance from the vector of the acquired text are identified; and in the presenting step, tags whose vectors are close to each other within a certain distance are presented to the user, and the program accepts from the user the tag to be set to the acquired text.

[0217] (Appendix 4) The program described in Appendix 3 further performs a step of generating a trained model that outputs tags for a sentence using the positive example tags associated with the acquired sentence and the negative example tags as training data, in which, of the presented tags, tags specified by the user are treated as positive examples and tags not specified by the user are treated as negative examples and are stored in a memory unit in association with the acquired sentence.

[0218] (Appendix 5) The program according to Appendix 4, further comprising the steps of: in the generating step, vectorizing the description of the generated tag; classifying the vectors of each generated tag into multiple clusters; setting tags specified by the user for the acquired sentence as positive examples; and setting tags in a cluster different from the positive example tags as negative examples as a training dataset; and generating a trained model that outputs tags for the sentence using the training dataset.

[0219] (Appendix 6) A program described in any of Appendices 1 to 5, wherein in the presentation step, the acquired sentence and the identified tag are displayed on the screen to the user, and an operation to specify a tag is accepted, and the acquired sentence and the specified tag are associated and stored in a memory unit in accordance with the user's operation to specify a tag.

[0220] (Appendix 7) 7. The program according to claim 6, wherein in the presenting step, tags that are highly relevant to the retrieved sentence are displayed on the screen to the user, distinguished from other tags.

[0221] (Appendix 8) A program as described in Appendix 7, in which, in the presenting step, tags that are highly relevant to the acquired sentence are displayed on the screen to the user in a pre-specified state without the user having to perform any operation to specify them.

[0222] (Appendix 9) A program described in any of Appendices 1 to 8, comprising: a step of acquiring, from a user who is the questioner, a sentence that requests an answer from an answerer; a step of accepting from the user an operation to specify a tag to be presented, associating the specified tag with the acquired sentence and storing it in a memory unit; a step of extracting candidate answerers who will answer the acquired sentence based on the tag stored in association with the acquired sentence and the answerer's attribute tag; and a step of accepting input of an answer to the acquired sentence from the candidate answerers.

[0223] (Appendix 10) A step of determining an attribute tag to be set for the respondent based on a tag associated with the sentence to which the respondent replied; and a program as described in Appendix 9.

[0224] (Appendix 11) A program described in any of Appendices 9 to 10, wherein in the presenting step, the program presents to the user at least one of candidate respondents associated with tags identified for the acquired text, or candidate respondents associated with tags specified by the user, based on attribute tags set for the respondents.

[0225] (Appendix 12) A method executed by a computer having a computer processor, configured to manage information on multiple tags assigned to sentences in a memory unit, the method comprising the steps of: generating a description explaining the tag by providing the tag information to a large-scale language model; acquiring a sentence to which a tag is to be assigned; identifying a tag corresponding to the acquired sentence based on the acquired sentence and the generated tag description; and presenting the identified tag to a user in association with the acquired sentence.

[0226] (Appendix 13) An information processing device having a control unit, configured to manage information on multiple tags assigned to sentences in a memory unit, wherein the control unit executes the steps of: generating an explanatory text explaining the tag by providing the tag information to a large-scale language model; acquiring a sentence to which a tag is to be assigned; identifying a tag corresponding to the acquired sentence based on the acquired sentence and the explanatory text of the generated tag; and presenting the identified tag to a user in association with the acquired sentence.

Claims

1. A program for operating a computer including a computer processor, the program being configured to manage, in a storage unit, information on a plurality of tags assigned to a sentence, the program causing the computer processor to: generating a description of the tag by providing tag information to a large-scale language model; obtaining a sentence to which the tag is to be added; Identifying the tag corresponding to the acquired sentence based on the acquired sentence and the generated description of the tag; presenting the identified tags to a user in association with the retrieved sentences; A program that executes.

2. In the generating step, providing prompts to the large scale language model that cause it to generate explanatory text describing the tags; searching the web, including platforms that distribute economic information, using keywords representing the tags, and summarizing the search results; The program according to claim 1 , further comprising: generating an explanatory text for explaining the tag by at least one of the following:

3. In the generating step, the generated tag description is vectorized, In the acquiring step, the acquired sentence is vectorized, In the step of specifying, a tag is specified whose distance between the vector of the acquired sentence and the vector of the tag is close within a certain range; 2. The program according to claim 1, wherein in the presenting step, tags that are close to the user within a certain distance of the vector are presented to the user, and a designation of a tag to be set to the acquired sentence is accepted from the user.

4. Among the presented tags, tags designated by a user are treated as positive examples, and tags not designated by the user are treated as negative examples, which are stored in the storage unit in association with the acquired sentences; The program according to claim 3 , further comprising a step of generating a trained model that outputs tags for sentences using the positive example tags associated with the acquired sentences and the negative example tags as training data.

5. In the generating step, the generated tag descriptions are vectorized, and the generated vectors of the tags are classified into a plurality of clusters; The program according to claim 4, further comprising the step of: setting tags specified by a user for the acquired sentence as positive examples, setting tags of a cluster different from the tags of the positive examples as negative examples, as a training dataset; and generating a trained model that outputs tags for sentences using the training dataset.

6. In the presenting step, the acquired sentence and the identified tag are displayed together on a screen for a user, and an operation of specifying the tag is accepted; The program according to claim 1 , further comprising: storing the acquired sentence and the designated tag in association with each other in a storage unit in response to an operation of the user specifying the tag.

7. 7. The program according to claim 6, wherein in the presenting step, the tag highly relevant to the acquired sentence is displayed on the screen for the user, distinguished from other tags.

8. 8. The program according to claim 7, wherein in the presenting step, tags highly relevant to the acquired sentence are displayed on the screen to the user in a pre-designated state without the user performing an operation to designate them.

9. In the acquiring step, the sentence requesting an answer from a respondent is acquired from a user who is a questioner; In the presenting step, an operation of designating the tag to be presented is accepted from a user, and the designated tag and the acquired sentence are associated with each other and stored in a storage unit; extracting candidate respondents who will respond to the acquired text based on the tag stored in association with the acquired text and attribute tags of the respondents; 2. The program according to claim 1, further comprising: a step of accepting input of a response to the acquired sentence from the answerer candidate.

10. The program according to claim 9 , further comprising: determining the attribute tag to be set for the answerer based on a tag associated with the sentence to which the answerer has replied.

11. In the step of presenting, the user is provided with a response based on the attribute tag set for the respondent. the candidate respondents associated with the tags identified for the retrieved text; or the candidate respondents associated with the tags specified by the user; The program according to claim 9, which presents at least one of the above.

12. A method executed by a computer including a computer processor, the method being configured to manage, in a storage unit, information on a plurality of tags assigned to a sentence, the method comprising: generating a description of the tag by providing tag information to a large-scale language model; obtaining a sentence to which the tag is to be added; Identifying the tag corresponding to the acquired sentence based on the acquired sentence and the generated description of the tag; presenting the identified tags to a user in association with the retrieved sentences; How to perform.

13. An information processing device including a control unit, the information processing device being configured to manage information on a plurality of tags assigned to a sentence in a storage unit, the control unit: generating a description of the tag by providing tag information to a large-scale language model; obtaining a sentence to which the tag is to be added; Identifying the tag corresponding to the acquired sentence based on the acquired sentence and the generated description of the tag; presenting the identified tags to a user in association with the retrieved sentences; An information processing device that executes the above.

Images