JP7704542B2 - Information processing device, information processing method, and information processing program - Google Patents

Description

The present invention relates to an information processing device, an information processing method, and an information processing program.

Information processing technology is known that provides a variety of information to users. The user inputs a search query related to the information they are looking for into an information processing device, and the information processing device searches for the information they are looking for based on the input search query.

JP 2019-153271 A

However, depending on the search query entered by the user, it may not be possible to find the information the user is looking for.

In view of the above problems, the present invention aims to provide users with appropriate information.

In order to solve the above-mentioned problems and achieve the objective, the information processing device according to the present disclosure includes a first acquisition unit that acquires a search query input by a user, a learning unit that generates a learning model that converts the search query, and a conversion unit that converts the search query using the learning model.

According to one aspect of the embodiment, appropriate information can be provided to the user.

FIG. 1 is a diagram illustrating an example of an information processing apparatus according to the first embodiment. FIG. 2 is a diagram illustrating an example of the configuration of the information processing apparatus according to the first embodiment. FIG. 3 is a diagram showing an example of information registered in the first search query database according to the first embodiment. FIG. 4 is a diagram showing an example of information registered in the second search query database according to the first embodiment. FIG. 5 is a diagram showing an example of information registered in the third search query database according to the first embodiment. FIG. 6 is a diagram showing an example of information registered in the Q&A database according to the first embodiment. FIG. 7 is a diagram illustrating an example of a flow of information processing according to the first embodiment. FIG. 8 is a diagram illustrating an example of a hardware configuration.

Below, the information processing device, information processing method, and information processing program according to the present application will be described in detail with reference to the drawings. Note that the information processing device, information processing method, and information processing program according to the present application are not limited to these embodiments. In addition, the same parts in the following embodiments will be denoted by the same reference numerals, and duplicated descriptions will be omitted.

[First embodiment]
1-1. An example of an information processing device
First, an information processing device 10 as an example of an information processing device according to an embodiment will be described with reference to FIG.

FIG. 1 is a diagram showing an example of an information processing device according to the first embodiment. The information processing device 10 shown in FIG. 1 is an information processing device that performs information processing, and is realized by, for example, a server device or a cloud system. For example, the information processing device 10 communicates with terminal devices used by each user via a network N. The network N is various wireless communication networks such as 4G (Generation), 5G, LTE (Long Term Evolution), Wifi (registered trademark), or wireless LAN (Local Area Network), or various wired communication networks.

The terminal device QT1 (hereinafter, sometimes collectively referred to as "terminal device T") is, for example, a PC (Personal Computer), a server device, a smart device such as a smart television, a smartphone, a smart speaker, or a tablet. The terminal device T can communicate with the information processing device 10 via the network N.

The terminal device QT1 may have a screen such as a liquid crystal display with a touch panel function, and may have a function capable of accepting various operations on the content delivered from the information processing device 10, such as tapping, sliding, scrolling, etc., performed by the user with a finger or a stylus. In the example shown in FIG. 1, the terminal device QT1 is assumed to be the terminal device used by the questioner Q1.

The information processing device 10 is a server host computer that provides various services to client terminals. The information processing device 10 distributes content to the terminal device QT1. For example, the information processing device 10 distributes content in which information related to portal sites, game information distribution sites, news sites, auction sites, weather forecast sites, shopping sites, finance (stock prices) sites, route search sites, map providing sites, travel sites, restaurant introduction sites, bulletin board sites, web blogs, etc. is arranged in tiles to the terminal device T. The information processing device 10 of this embodiment provides an information sharing service that enables users to ask and answer questions to each other.

In addition, the information processing device 10 is a server host computer that provides various services to client terminals. For example, the information processing device 10 may be a server that provides information transmission services (hereinafter referred to as information services) such as an information search service, a social networking service (SNS), a news distribution service, an information compilation service, an Internet encyclopedia service, an Internet dictionary service, and a blog service.

The information processing device 10, information processing device 10A, information processing device 10B, and information processing device 10C shown in FIG. 1 are connected via a network N so that they can transmit and receive information to each other. The information processing device 10A is a server that provides a service Y1 (e.g., a question and answer service) which is one of the information services. The information processing device 10B is a server that provides a service Y2 (e.g., an Internet encyclopedia service) which is one of the information services. The information processing device 10C is a server that provides a service Y3 (e.g., an information search service) which is one of the information services. In the example shown in FIG. 1, the four information processing devices, namely, the information processing device 10, the information processing device 10A, the information processing device 10B, and the information processing device 10C, provide different information services, but the number of information processing devices is not limited to four, and may be any other number of information processing devices. That is, there may be only one information processing device 10, and one information processing device 10 may provide any combination of different information services.

For example, in the example shown in FIG. 1, the information processing device 10 trains a learning model to learn the relationship between the search query and the selected search results (step S1). The information processing device 10 accepts an input of a search query from a user (step S2). The information processing device 10 converts the input search query (step S3). The information processing device 10 acquires related information based on the converted search query (step S4). The information processing device 10 displays the acquired related information on the user terminal (step S5).

1-2. Processing of information processing device
Depending on the form of the search query entered by the user, the information desired by the user may not be searchable. Therefore, it is necessary to convert the search query entered by the user into a search query that can search for the information desired by the user.

Therefore, the information processing device 10 generates a learning model for converting search queries, and converts the search query entered by the user using the generated learning model.

For example, the process will be described when a user inputs search query A "What is the difference in performance between tennis rackets made by manufacturer A and manufacturer B?" as a search query. In this case, the information processing device 10 converts the search query into a keyword search query. That is, the information processing device 10 converts search query A into a keyword search query for "manufacturer A," "manufacturer B," "tennis racket," and "performance." After search query A is converted into a keyword search query, the information processing device 10 acquires related information from multiple information services based on the keyword search query. After acquiring the related information, the information processing device 10 displays the acquired related information on the user's terminal device.

This allows the search query entered by the user to be converted into a search query that can obtain information that matches the user's search intent. As a result, it is possible to provide the user with the information that the user intended to obtain by entering the search query.

1-3. Learning process using the first search query database
Here, when converting a search query entered by a user, if the search query can be converted using a learning model that has learned the relationship between the search query and the search results selected by the user from among the search results searched for based on the search query, it is believed that it will be possible to obtain a search query that can obtain search results that are in line with the user's search intent.

Therefore, the information processing device 10 learns the relationship between the search query entered by the user and the search results selected by the user from among the search results corresponding to the search query, and generates a learning model.

For example, the information processing device 10 includes a first search query database 31 (described later) that registers search queries input by a user and selected search results, which mean search results selected by the user from among search results corresponding to the search queries, in association with each other. The information processing device 10 uses the data registered in the first search query database 31 to learn the relationship between the search query input by the user and the search results selected by the user from among search results corresponding to the search queries, and generates a learning model. The information processing device 10 acquires the search query input by the user, inputs the acquired search query into the learning model, and obtains an output of the converted search query. After acquiring the output of the converted search query, the information processing device 10 acquires related information from a plurality of information services using the converted search query. After acquiring the related information, the information processing device 10 displays the acquired related information on the user's terminal device.

This allows the search query entered by the user to be converted into a search query that can obtain information that matches the user's search intent. As a result, it is possible to provide the user with the information that the user intended to obtain by entering the search query.

1-4. Learning process using the second search query database
Here, when converting a search query entered by a user, if the search query can be converted using a learning model that has learned the search query and the related information selected by the user from among related information obtained from multiple information services based on the search query, it is possible to obtain a search query that can obtain related information from multiple information services that is in line with the user's search intent.

Therefore, the information processing device 10 learns the relationship between the search query entered by the user and the related information selected by the user from among related information obtained from multiple information services based on the search query, and generates a learning model.

For example, the information processing device 10 includes a second search query database 32 (described later) that registers a search query input by a user and selected related information, which means related information selected by the user from related information acquired from multiple information services based on the search query, in association with each other. The information processing device 10 uses the data registered in the second search query database 32 to learn the relationship between the search query input by the user and the related information selected by the user from related information acquired from multiple information services based on the search query, and generates a learning model. The information processing device 10 acquires the search query input by the user, inputs the acquired search query into the learning model, and obtains an output of the converted search query. After acquiring the output of the converted search query, the information processing device 10 acquires related information from multiple information services using the converted search query. After acquiring the related information, the information processing device 10 displays the acquired related information on the user's terminal device.

This allows the search query entered by the user to be converted into a search query that can retrieve related information that matches the user's search intent. As a result, it is possible to provide the user with the information that the user intended to obtain by entering the search query.

[1-5. Learning process using the third search query database]
Here, when converting a search query entered by a user, if the search query can be converted using a learning model that has learned the relationship between the search query and the question selected by the user from among related questions obtained from the question and answer service based on the search query, it is possible to obtain a search query that enables related questions that are in line with the user's search intent to be obtained from the question and answer service.

Therefore, the information processing device 10 learns the relationship between the search query entered by the user and the question sentence selected by the user from among related questions obtained from the question and answer service based on the search query, and generates a learning model.

For example, the information processing device 10 includes a third search query database 33 (described later) in which a search query input by a user and a selected question sentence, which means a question sentence selected by the user from among related question sentences acquired from a question and answer service based on the search query, are associated and registered. The information processing device 10 also includes a Q&A database 34 (described later) in which a question sentence input by a user and a set of answer sentences input by other users to the question sentence are associated and registered. The information processing device 10 uses the data registered in the third search query database 33 to learn the relationship between the search query input by the user and the question sentence selected by the user from among related question sentences acquired from a question and answer service based on the search query, and generates a learning model. The information processing device 10 acquires the search query input by the user, inputs the acquired search query into the learning model, and obtains an output of the converted search query. After acquiring the output of the converted search query, the information processing device 10 acquires related question sentences from the question and answer service using the converted search query. When the information processing device 10 acquires the related question text, it displays the related question text on the user's terminal device. Note that the information processing device 10 adds a link to the related question text to access a collection of answers entered by other users, and displays the link on the user's terminal device.

This allows the search query entered by the user to be converted into a search query that can obtain a question that matches the user's search intent. As a result, it is possible to provide the user with the information they wish to obtain by entering a search query.

[1-6. Conversion process]
Here, the information processing device 10 converts the natural language search query into a keyword search query using a learning model for converting a search query.

For example, the process will be described when a user inputs search query A, "What is the difference in performance between tennis rackets made by manufacturer A and manufacturer B?" In this case, the information processing device 10 converts search query A into a search query for the keywords "manufacturer A," "manufacturer B," "tennis racket," and "performance." Based on the search query for the converted keywords, the information processing device 10 acquires related information from multiple information services. After acquiring the related information, the information processing device 10 displays the acquired related information on the user's terminal device.

This allows the system to convert the natural language search query entered by the user into a keyword search query that can retrieve information that matches the user's search intent. As a result, the system can provide the user with the information they intended to obtain by entering a natural language search query.

[1-7. Display of related information]
Here, if, when a user enters a search query, in addition to the search results of a normal search, related information based on the search query can be obtained from multiple information services and displayed on the user's terminal device, it is believed that this can stimulate the use of multiple information services.

Then, the information processing device 10 obtains related information from multiple information services based on the converted search query.

For example, the process will be described when a user inputs a search query A, "What is the difference in performance between tennis rackets made by manufacturer A and manufacturer B?" In this case, the information processing device 10 converts the search query A into a search query for the keywords "manufacturer A," "manufacturer B," "tennis racket," and "performance." Based on the search query of the converted keywords, the information processing device 10 acquires related information from a plurality of information services. For example, the information processing device 10 acquires information on "tennis rackets" made by "manufacturer A" and "manufacturer B" from a shopping site as related information. In addition, the information processing device 10 acquires information on "tennis rackets" made by "manufacturer A" and "manufacturer B" from an auction site as related information. In addition, the information processing device 10 acquires information on "manufacturer A," "manufacturer B," "tennis racket," and "performance" from a bulletin board site as related information. After acquiring the related information, the information processing device 10 displays the acquired related information on the user's terminal device together with other search results.

This allows the search query entered by the user to be converted, related information to be obtained from multiple information services based on the converted search query, and the obtained related information to be displayed on the user's terminal device. This makes it possible to provide the information desired by the user appropriately. It also stimulates the use of multiple information services.

[1-8. Display of questions]
Here, if, when a user inputs a search query, in addition to the search results of a normal search, related questions based on the search query could be obtained from the question and answer service and displayed on the user's terminal device, it would be possible to provide the information the user is looking for, thereby stimulating the use of the question and answer service registered on the portal site.

Therefore, the information processing device 10 obtains related question sentences from the question and answer service based on the converted search query.

For example, the process will be described when a user inputs search query A "What is the difference in performance between tennis rackets made by manufacturer A and manufacturer B?". In this case, the information processing device 10 converts search query A into keyword search query B "manufacturer A", "manufacturer B", "tennis racket", and "performance". Based on the converted keyword search query B, the information processing device 10 acquires a related question from a question and answer service. For example, assume that a user posts question A "What is the difference in performance between tennis rackets made by manufacturer A and manufacturer B" that includes all of keyword search queries B to the question and answer service, and other users have posted answers to question A. In this case, the information processing device 10 acquires question A from the question and answer service. After acquiring question A, the information processing device 10 displays the acquired question A on the user's terminal device together with other search results.

This allows the search query entered by the user to be converted, related questions to be obtained from the question and answer service based on the converted search query, and the obtained questions to be displayed on the user's terminal device. As a result, the information desired by the user can be appropriately provided.

[1-9. Acquiring similar questions using clusters of distributed representations of questions]
Here, when searching for questions, not only questions that contain all of the keywords in the search query are displayed, but also questions that do not contain all of the keywords in the search query but have a similar meaning are displayed.By referring to related questions, users can obtain information that is in line with the search intent of the entered search query.

Therefore, the information processing device 10 clusters the distributed representations corresponding to the text information of the question sentence based on the similarity, calculates the similarity between the distributed representations of the text information of the converted search query and the distributed representations corresponding to the text information of the question sentence, selects a cluster of question sentences related to the search query based on the calculated similarity, and obtains the question sentences included in the selected cluster.

For example, in a question and answer service, it is assumed that a user inputs question B "What is the difference in performance between tennis rackets made by manufacturer C and manufacturer D?", question C "How much do tennis rackets made by manufacturer A and manufacturer B cost?", and question D "What tennis racket manufacturers are there?". Then, it is assumed that answers to question B, question C, and question D are input from other users. In this case, the information processing device 10 calculates distributed representations for question B, question C, and question D. Here, since question B, question C, and question D have similar meanings, the distributed representations are also calculated to be close values. After calculating the distributed representations for question B, question C, and question D, the information processing device 10 classifies question B, question C, and question D into the same cluster A. Here, it is assumed that the information processing device 10 acquires a search query A that is the same as the above-mentioned question A from a user. When the information processing device 10 acquires search query A, it calculates the distributed representation of the acquired search query A. Since question sentence A has a similar meaning to question sentence B, question sentence C, and question sentence D, the distributed representation of search query A, which is the same as question sentence A, is calculated to be close to the distributed representations of question sentence B, question sentence C, and question sentence D. When the information processing device 10 calculates the distributed representation of search query A, it acquires, for example, question sentence B from the questions classified into cluster A. Note that the number of questions acquired from cluster A is not limited to one, and any number of questions may be acquired. When the information processing device 10 acquires question sentence B from cluster A, it causes question sentence B to be displayed on the user's terminal device together with other search results.

This allows the user to be provided with information that matches the search intent of the search query when the user enters the search query. This allows the user to be provided with the appropriate information they are looking for.

2. Configuration of information processing device
An example of the functional configuration of the information processing device 10 described above will be described below. Note that the following description shows an example of the functional configuration of the information processing device 10 that executes information processing at an information service site. Fig. 2 is a diagram showing an example of the configuration of the information processing device according to the first embodiment. As shown in Fig. 2, the information processing device 10 has a communication unit 20, a storage unit 30, and a control unit 40.

The communication unit 20 is realized, for example, by a NIC (Network Interface Card) or the like. The communication unit 20 is connected to the network N by wire or wirelessly, and transmits and receives information between the communication unit 20 and the terminal device T.

The memory unit 30 is realized, for example, by a semiconductor memory element such as a RAM (Random Access Memory) or a flash memory, or a storage device such as a hard disk or an optical disk.

The memory unit 30 also stores a first search query database 31, a second search query database 32, a third search query database 33, a Q&A database 34, a learning model 35, and a distributed representation database 36.

The first search query database 31 stores information about search queries entered by users and search results selected by the users from search results searched based on the search queries. For example, FIG. 3 is a diagram showing an example of information stored in the first search query database 31 according to an embodiment. As shown in FIG. 3, the first search query database 31 stores information having items such as "user ID," "search query," and "selected search result."

Here, the "user ID" is an identifier for identifying a user. Furthermore, the "search query" is information indicating a search query input by a user. Furthermore, the "selected search results" are information indicating search results selected by a user from among search results searched based on the search query input by the user. Note that it is also possible to uniformly acquire search queries from all users without identifying the users, acquire selected search results indicating search results selected by the users from among search results searched based on the acquired search query, and register the acquired search queries and non-selected search queries in the first search query database 31.

For example, in the example shown in FIG. 3, information such as user ID "U1", search query "search query #U1-1", and selected search result "selected search result #U1-1" is registered in the first search query database 31. This information indicates that the user indicated by user ID "U1" entered the search query indicated by search query "search query #U1-1" and selected the information indicated by selected search result "selected search result #U1-1".

In the example shown in FIG. 3, the information to be registered in the first search query database 31 is described using conceptual expressions such as search query #U1 and selected search result #U1-1, but in reality, the text information that constitutes the search query entered by the user and the text information of the selected search result that indicates the search result selected by the user from the search results searched based on the search query will be registered. In addition to the information shown in FIG. 3, any other information related to the search query may be registered in the first search query database 31.

The second search query database 32 stores the search query entered by the user, the related information selected by the user from among related information obtained from multiple information services based on the search query, and the name of the information service from which the related information was obtained. For example, FIG. 4 is a diagram showing an example of information registered in the second search query database 32 according to the first embodiment. As shown in FIG. 4, the second search query database 32 stores information having items such as "user ID," "search query," "selected related information," and "information service name."

For example, in the example shown in FIG. 4, information such as a user ID "U1", a search query "search query #U1-1", and an information service name "information service A" are registered in association with each other. This information indicates that a user indicated by user ID "U1" entered a search query indicated by search query "search query #U1-1", selected related information indicated by non-selected related information "non-selected related information #U1-1", and the selected related information was obtained from an information service indicated by information service name "information service A".

In the example shown in FIG. 4, conceptual expressions such as "search query #U1-1" and "information service A" are described, but in reality, the following will be registered: character information that is the search query entered by the user, character information of related information selected by the user from related information obtained from multiple information services based on the search query, and character information of the name of the information service from which the related information selected by the user was obtained. In addition to the information shown in FIG. 4, any other information related to the search query may be registered in the second search query database 32.

The third search query database 33 stores search queries entered by users and questions selected by users from among questions obtained from the question and answer service based on the search queries. For example, FIG. 5 is a diagram showing an example of information registered in the third search query database 33 according to the first embodiment. As shown in FIG. 5, the third search query database 33 stores information having items such as "user ID," "search query," and "selected question."

For example, in the example shown in FIG. 5, information such as a user ID "U1", a search query "search query #U1-1", and a selected question sentence "selected question sentence #U1-1" are registered in association with each other. Such information indicates that a user identified by user ID "U1" has entered a search query identified by search query "search query #U1-1" and selected a question sentence identified by non-selected question sentence "non-selected question sentence #U1-1".

In the example shown in FIG. 5, conceptual expressions such as "search query #U1-1" and "selected question #U1-1" are described, but in reality, what is registered is the text information of the search query entered by the user, and the text information of the question selected by the user from among related questions obtained from the question and answer service based on the search query. In addition to the information shown in FIG. 5, any other information related to the search query may be registered in the third search query database 33.

The Q&A database 34 stores information about questions entered by users, sets of answers entered by other users to the questions, and answers selected as best answers by the user who entered the question from the sets of answers entered by other users. For example, FIG. 6 is a diagram showing an example of information registered in the Q&A database according to the first embodiment. As shown in FIG. 6, the Q&A database 34 stores information having items such as "question ID," "question," "answer set," and "selection information." Note that a set of answer sets is information that brings together answer sets entered by one or more users in response to a question entered by a user.

In the example shown in FIG. 6, information such as a question ID "Q1", a question "Question #1", a set of answer sentences "Answer Set #1", and selected information "Answer #1-1" are associated and registered. This information indicates that the question ID "Q1" is assigned to a question entered by a user, the question entered by the user is registered in question "Question #1", an answer shown in the set of answer sentences "Answer Set #1" was entered by another user in response to the question entered by the user, and the answer shown in the selected information "Answer #1-1" was selected as the best answer by the user who entered the question.

In the example shown in FIG. 6, conceptual expressions such as "Question #1," "Answer Set #1," and "Answer #1-1" are used, but in reality, the text information of the question entered by the user, the text information of a set of answers entered by other users in response to the question entered by the user, and the text information of the answer selected by the user who entered the question will be registered. In addition to the information shown in FIG. 6, the Q&A database 34 may also register any information related to various questions and answers, such as the user ID of the user who entered the question and the user ID of the user who entered the answer to the entered question.

Unlearned learning models and learned learning models are stored in the learning model 35. As the learned learning models, learning models generated by the learning unit 42, which will be described later, executing a learning process are stored. When the learning unit 42, which will be described later, executes the learning process, the learning process is executed using the unlearned learning models stored in the learning model 35.

In the distributed representation database 36, a question sentence input by a user, a distributed representation of the question sentence, and a cluster classified into the question sentence are registered in association with each other.

The control unit 40 is, for example, a controller, and is realized by a CPU (Central Processing Unit) or an MPU (Micro Processing Unit) executing various programs stored in the storage device of the information processing device 10 using the RAM as a working area. The control unit 40 is also a controller, and may be realized by an integrated circuit such as an ASIC (Application Specific Circuit) or an FPGA (Field Programmable Gate Array).

As shown in FIG. 2, the control unit 40 has an acquisition unit 41, a learning unit 42, a conversion unit 43, a calculation unit 44, a classification unit 45, a reception unit 46, and a provision unit 47.

The acquisition unit 41 includes a first acquisition unit 411, a second acquisition unit 412, and a third acquisition unit 413. The first acquisition unit 411 acquires a search query input by a user. For example, if a user inputs a search query A "What is the difference in performance between tennis rackets from manufacturer A and manufacturer B" into the terminal device T, the first acquisition unit 411 acquires the search query input into the terminal device T via the network N and the communication unit 20.

The second acquisition unit 412 acquires related information from a plurality of information services based on a search query that is converted by the conversion unit 43, which will be described later, from a search query entered by a user. For example, assume that a user enters a search query A, "What is the difference in performance between tennis rackets from manufacturer A and manufacturer B?", as a search query into the terminal device T. In this case, the first acquisition unit 411 acquires the search query A entered by the user, and converts the search query A acquired by the conversion unit 43, which will be described later, into a search query B of keywords, "manufacturer A," "manufacturer B," "tennis racket," and "performance." The second acquisition unit 412 acquires related information from a plurality of information services based on the search query B converted by the conversion unit 43. For example, the second acquisition unit 412 acquires sales information on "tennis rackets" from "manufacturer A" and "manufacturer B" from a shopping site as related information.

The third acquisition unit 413 selects a cluster of question sentences based on the calculation result of the distributed representation of the search query input by the user, and acquires the question sentences included in the selected cluster. That is, after the calculation unit 44 described later calculates the distributed representation of the search query input by the user, the third acquisition unit 413 calculates the similarity between the distributed representation of the search query input by the user and the distributed representation of the cluster of question sentences, and selects a cluster of question sentences with high similarity. After a cluster of question sentences with high similarity is selected, the third acquisition unit 413 acquires the question sentences included in the selected cluster.

The learning unit 42 generates a learning model that converts a search query. For example, the learning unit 42 learns the relationship between the search query registered in the first search query database 31 and the selected search results, and generates a learning model that converts the search query. In this case, for example, when natural sentences included in the character information of the selected search results are input, a learning model that outputs a search query input by a user may be generated. Also, for example, the learning unit 42 learns the relationship between the search query registered in the second search query database 32 and the selected related information, and generates a learning model that converts a search query. In this case, for example, when natural sentences included in the character information of the selected related information are input, a learning model that outputs a search query input by a user may be generated. Also, for example, the learning unit 42 learns the relationship between the search query registered in the third search query database 33 and the selected question sentence, and generates a learning model that converts a search query. In this case, for example, when natural sentences included in the character information of the selected question sentence are input, a learning model that outputs a search query input by a user may be generated. That is, the learning model converts the search query according to the user's behavior after inputting the search query. In other words, the learning model is trained to convert the search query according to the search intent estimated from the user's behavior after inputting the search query.

The learning unit 42 generates a trained learning model by end-to-end deep learning. End-to-end deep learning replaces a machine learning system that required multiple stages of processing from when input data was given until the results were output with one large neural network equipped with multiple layers and modules that perform various processes. In other words, end-to-end deep learning directly learns the correspondence between input and output. By using end-to-end deep learning, there is no need to prepare an intermediate structure specialized for a specific function, and there is no need to consider the problem of problems that arise in upstream processes in natural language processing being postponed to downstream processes.

The conversion unit 43 converts the search query using the learning model. For example, the conversion unit 43 obtains an output of the converted search query by inputting the search query input by the user to the learning model generated by the learning unit 42 learning the relationship between the search query registered in the first search query database 31 and the selected search results. Also, for example, the conversion unit 43 obtains an output of the converted search query by inputting the search query input by the user to the learning model generated by the learning unit 42 learning the relationship between the search query registered in the second search query database 32 and the selected related information. Also, for example, the conversion unit 43 obtains an output of the converted search query by inputting the search query input by the user to the learning model generated by the learning unit 42 learning the relationship between the search query registered in the third search query database 33 and the selected question sentence.

The calculation unit 44 calculates the distributed representation corresponding to the character information. For example, the calculation unit 44 performs morphological analysis on the question sentence, segments it into morphemes, and converts the segmented morphemes into distributed representations. For example, word2vec can be used for converting words into distributed representations. Once all the words constituting the question sentence have been converted into distributed representations, the distributed representation of the question sentence can be obtained, for example, by calculating the average value of the distributed representations of all the words constituting the question sentence. After calculating the distributed representation of the question sentence, the calculation unit 44 registers the distributed representation of the question sentence in the distributed representation database 36 in association with the question sentence. Even if the search query is in natural language, the same process can be performed to obtain the distributed representation of the natural language search query.

The classification unit 45 clusters the distributed representations corresponding to the text information of the question sentence based on the similarity. For example, the classification unit 45 calculates the cosine similarity between the distributed representations corresponding to the text information of the question sentence and the distributed representations corresponding to the text information of another question sentence. Note that the classification unit 45 may calculate the similarity between the distributed representations based on any index other than the cosine similarity as long as the index is applicable as a distance measure between vectors. For example, the classification unit 45 may calculate the value of a predetermined distance function such as the Euclidean distance between the distributed representations, the distance in a non-Euclidean space such as a hyperbolic space, the Manhattan distance, the Mahalanobis distance, etc.

After calculating the similarity of the distributed representations corresponding to the text information of the question sentence, the classification unit 45 clusters the distributed representations of the question sentence based on the similarity of the distributed representations. After clustering the distributed representations of the question sentence, the classification unit 45 registers the clusters of the distributed representations of the question sentence in the distributed representation database 36 in association with the question sentence and the distributed representations of the question sentence. For example, K-means clustering can be used to cluster the distributed representations of the question sentence. K-means clustering classifies data into a given number of clusters (K) using the average value of the clusters.

The reception unit 46 accepts registration of questions and answers from users. For example, when the reception unit 46 accepts a question sentence from a user's terminal device, it registers the accepted sentence as a question in the Q&A database 34. In addition, when the reception unit 46 accepts an answer sentence from a terminal device used by an answerer, it associates the accepted sentence with the question ID of the corresponding question sentence and registers it in the Q&A database 34. This makes it possible to accept new questions and answers posted by users and update the Q&A database 34.

The providing unit 47 provides at least either the related information acquired by the second acquiring unit 412 or the question acquired by the third acquiring unit 413 to the user who input the search query. That is, the providing unit 47 may display the related information acquired by the second acquiring unit 412, or may display the question acquired by the third acquiring unit 413. The providing unit 47 may also display both the related information acquired by the second acquiring unit 412 and the question acquired by the third acquiring unit 413 on the user's terminal device.

3. Processing Procedure
Next, the procedure of information processing by the information processing device 10 according to the first embodiment will be described with reference to FIG. 7. FIG. 7 is a flowchart showing an example of the flow of information processing according to the first embodiment. For example, the information processing device 10 acquires a search query input by a user (step S101). Then, the information processing device 10 converts the acquired search query (step S102). Then, the information processing device 10 acquires related information from a plurality of information services (step S103). Then, the information processing device 10 displays the acquired related information on the terminal device of the user who input the search query (step S104).

4. Hardware Configuration
The information processing device 10 according to the embodiment described above is realized by a computer 1000 having a configuration as shown in Fig. 8, for example. Fig. 8 is a diagram showing an example of a hardware configuration. The computer 1000 is connected to an output device 1010 and an input device 1020, and has a configuration in which a calculation device 1030, a primary storage device 1040, a secondary storage device 1050, an output IF (Interface) 1060, an input IF 1070, and a network IF 1080 are connected by a bus 1090.

The arithmetic device 1030 operates based on programs stored in the primary storage device 1040 and the secondary storage device 1050, programs read from the input device 1020, and the like, and executes various processes. The primary storage device 1040 is a memory device, such as a RAM, that primarily stores data used by the arithmetic device 1030 for various calculations. The secondary storage device 1050 is a storage device in which data used by the arithmetic device 1030 for various calculations and various databases are registered, and is realized by a ROM (Read Only Memory), HDD (Hard Disk Drive), flash memory, etc.

The output IF 1060 is an interface for transmitting information to be output to an output device 1010 that outputs various types of information, such as a monitor or a printer, and is realized, for example, by a connector conforming to a standard such as USB (Universal Serial Bus), DVI (Digital Visual Interface), or HDMI (registered trademark) (High Definition Multimedia Interface). The input IF 1070 is an interface for receiving information from various input devices 1020, such as a mouse, keyboard, and scanner, and is realized, for example, by a USB.

The input device 1020 may be a device that reads information from, for example, an optical recording medium such as a CD (Compact Disc), a DVD (Digital Versatile Disc), or a PD (Phase change rewritable Disk), a magneto-optical recording medium such as an MO (Magneto-Optical disk), a tape medium, a magnetic recording medium, or a semiconductor memory. The input device 1020 may also be an external storage medium such as a USB memory.

The network IF 1080 receives data from other devices via the network N and sends it to the computing device 1030, and also transmits data generated by the computing device 1030 to other devices via the network N.

The arithmetic unit 1030 controls the output device 1010 and the input device 1020 via the output IF 1060 and the input IF 1070. For example, the arithmetic unit 1030 loads a program from the input device 1020 or the secondary storage device 1050 onto the primary storage device 1040 and executes the loaded program.

For example, when the computer 1000 functions as the information processing device 10, the arithmetic unit 1030 of the computer 1000 realizes the functions of the control unit 40 by executing a program loaded onto the primary storage device 1040.

5. Effects
The information processing device 10 according to the present disclosure includes a first acquisition unit 411 that acquires a search query input by a user, a learning unit 42 that generates a learning model for converting the search query, and a conversion unit 43 that converts the search query using the learning model.

With this configuration, the search query entered by the user is converted using a learning model, making it possible to provide the user with appropriate information.

The learning unit 42 also learns the relationship between the search query entered by the user and the search results selected by the user from among the search results corresponding to the search query, and generates a learning model.

With this configuration, it is possible to generate a learning model that can convert a search query entered by a user into a search query that can obtain search results that match the user's search intent. As a result, it is possible to provide appropriate information to the user.

The learning unit 42 also learns the relationship between the search query entered by the user and the related information selected by the user from among related information obtained from multiple information services based on the search query, and generates a learning model.

With this configuration, it is possible to generate a learning model that can convert a search query entered by a user into a search query that can obtain related information that matches the user's search intent. As a result, it is possible to provide appropriate information to the user.

In addition, the learning unit 42 learns the relationship between the search query entered by the user and the question sentence selected by the user from among related questions obtained from the question and answer service based on the search query, and generates a learning model.

With this configuration, it is possible to generate a learning model that can convert a search query entered by a user into a search query that can obtain a question that matches the user's search intent. As a result, it is possible to provide appropriate information to the user.

The conversion unit 43 also uses the learning model to convert natural language search queries into keyword search queries.

With this configuration, a search query in natural language entered by a user can be converted into a keyword search query. This makes it possible to provide the user with appropriate information.

The information processing device 10 according to the present disclosure also includes a second acquisition unit 412 that acquires related information from among a plurality of information services based on the search query converted by the conversion unit 43.

With this configuration, a search query entered by a user can be converted into a search query that can search for related information that matches the user's search intent, and related information can be obtained from multiple information services. This makes it possible to provide the user with appropriate information.

The second acquisition unit 412 also acquires a question sentence as related information from a question and answer service as an information service based on the search query converted by the conversion unit 43.

With this configuration, a search query entered by a user is converted into a search query that can search for a question that matches the user's search intent, and the question can be obtained as related information from a question and answer service as an information service. This makes it possible to provide the user with appropriate information.

The information processing device 10 according to the present disclosure also includes a calculation unit 44 that calculates distributed representations corresponding to character information, a classification unit 45 that clusters question sentences based on the similarity of the distributed representations corresponding to the question sentences, and a third acquisition unit 413 that selects a cluster of question sentences based on the calculation result of the distributed representation of the search query input by the user and acquires the question sentences included in the selected cluster.

With this configuration, a cluster of questions can be selected based on the results of computing the distributed representation of the search query entered by the user, and the questions contained in the selected cluster can be obtained. This makes it possible to provide the user with appropriate information.

The information processing method according to the present disclosure includes the steps of acquiring a search query input by a user, converting the search query, acquiring related information from a plurality of information services based on the converted search query, and displaying the acquired related information.

With this configuration, a search query entered by a user can be converted into a search query that can search for information that matches the user's search intent, and related information can be obtained from multiple information services. This makes it possible to provide the user with appropriate information.

The information processing program disclosed herein causes a computer to execute the steps of acquiring a search query entered by a user, converting the search query, acquiring related information from multiple information services based on the converted search query, and displaying the acquired related information.

The above describes the embodiments of the present application in detail with reference to the drawings, but this is merely an example, and the present invention can be embodied in other forms with various modifications and improvements based on the knowledge of those skilled in the art, including the forms described in the disclosure section of the invention.

The above-mentioned "section, module, unit" can be read as "means" or "circuit." For example, an acquisition unit can be read as an acquisition means or an acquisition circuit.

REFERENCE SIGNS LIST 10 Information processing device 20 Communication unit 30 Memory unit 31 First search query database 32 Second search query database 33 Third search query database 34 Q&A database 35 Learning model 36 Distributed representation database 40 Control unit 41 Acquisition unit 411 First acquisition unit 412 Second acquisition unit 413 Third acquisition unit 42 Learning unit 43 Conversion unit 44 Calculation unit 45 Classification unit 46 Reception unit 47 Provision unit N Network T Terminal device

Claims (9)

A first acquisition unit that acquires a search query input by a user;
a learning unit that generates a learning model that outputs the search query when the user inputs a natural sentence contained in character information of a search result selected by the user from among search results searched based on the search query ;
a conversion unit that converts the search query input by the user into a search query corresponding to a search intent estimated from the user's behavior after inputting the search query, using the learning model;
An information processing device comprising: The information processing device according to claim 1 , wherein the learning unit generates a learning model that outputs a search query when a natural sentence contained in character information of related information selected by a user from related information obtained from a plurality of information services based on a search query input by the user is input. The information processing device according to claim 1 , wherein the learning unit generates a learning model that outputs a search query when a natural sentence contained in character information of a question selected by a user from related questions obtained from a question and answer service based on a search query input by the user is input. The conversion unit converts a natural language search query into a keyword search query using the learning model.
The information processing device according to claim 1 . and a second acquisition unit that acquires related information from a plurality of information services based on the search query converted by the conversion unit.
The information processing device according to claim 1 . the second acquisition unit acquires a question sentence as related information from a question and answer service as an information service based on the search query converted by the conversion unit;
The information processing device according to claim 5 . A calculation unit that calculates a distributed representation corresponding to character information;
a classification unit that clusters the questions based on the similarity of the distributed representations corresponding to the questions;
and a third acquisition unit that selects a cluster of questions related to the search query based on a calculation result of the distributed representation of the converted search query, and acquires the questions included in the selected cluster.
The information processing device according to claim 1 . 1. A computer-implemented information processing method, comprising:
A first acquisition step of acquiring a search query input by a user;
a learning process for generating a learning model that outputs the search query when the user inputs a natural sentence contained in character information of a search result selected by the user from among search results searched based on the search query;
a conversion step of converting the search query input by the user into a search query corresponding to a search intent estimated from the user's behavior after inputting the search query , using the learning model;
An information processing method comprising: A first acquisition step of acquiring a search query input by a user;
a learning procedure for generating a learning model that outputs the search query when the user inputs a natural sentence contained in character information of a search result selected by the user from among search results searched based on the search query;
a conversion step of converting the search query input by the user into a search query corresponding to a search intent estimated from the user's behavior after inputting the search query, using the learning model;
An information processing program that causes a computer to execute the above.

Images