Presentation device and program

Abstract

To provide information desired by a user from preliminarily prepared information.SOLUTION: A presentation device comprises: a first acquisition unit which acquires a question from a user; a first search unit which searches for a plurality of categories of the question acquired by the first acquisition unit with the question acquired by the first acquisition unit as a search condition on the basis of association information where a question, an answer to the question, and a category of the question are associated with each other; a first presentation unit which presents the plurality of searched categories; a second acquisition unit which acquires the category selected by the user from the plurality of presented categories; a second search unit which searches the association information for an answer with the question acquired by the first acquisition unit as a search condition; a third search unit which searches the association information for an answer with the question acquired by the first acquisition unit and the category acquired by the second acquisition unit as a search condition; and a second presentation unit which presents the answer searched by the second search unit and the answer searched by the third search unit to the user.SELECTED DRAWING: Figure 2

Description

[Technical field]

[0001] The present invention relates to a presentation device and a program. [Background technology]

[0002] It is common for companies and the like to use web pages to provide users with information about the products and services they offer. Users may make inquiries to companies and the like after seeing the information posted on a web page. It would take a lot of effort for companies and the like to have operators respond to all inquiries from users. To reduce the effort of responding to user inquiries, companies and the like may set up an FAQ section on their web pages. This FAQ section lists frequently asked questions and their answers (hereinafter sometimes referred to as FAQ). If the user checks the FAQ section and no longer needs to make an inquiry, they do not make an inquiry to the company and the like, thereby reducing the effort of the company and the like.

[0003] Here, if one were to prepare FAQs that cover all possible inquiries from users, the number would be enormous. Also, it is not easy for users to find the information they want from the vast number of FAQs. For this reason, companies and the like could acquire inquiries from users and provide the users with FAQs corresponding to the acquired questions. When the number of FAQs to be presented to the user exceeds a certain range, the number of FAQs to be presented to the user is narrowed down by asking the user questions. This allows the user to easily find the information they want. Patent Document 1, for example, can be cited as an example of a document that discloses such a technology. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] JP 2015-36945 A Summary of the Invention [Problem to be solved by the invention]

[0005] Patent Document 1 discloses a question and answer control program capable of searching for an FAQ that matches the intent of a user's question. According to this technology, natural language written by a user is analyzed, and an FAQ that matches the intent of the user's question is searched for according to the analysis result. If the natural language written by a user can be correctly analyzed using this conventional technology, it may be possible to provide a suitable answer. However, a user does not necessarily write natural language that can be correctly analyzed by a computer. If a computer cannot correctly analyze the natural language written by a user, there is a problem in that it is difficult to provide an FAQ that matches the intent of the user's question.

[0006] Therefore, an object of the present invention is to provide a user with information that he or she desires from information prepared in advance. [Means for solving the problem]

[0007] [1] One aspect of the present invention is a presentation device including: a first acquisition unit that acquires a question from a user; a first search unit that searches for a plurality of categories of the question acquired by the first acquisition unit based on correspondence information in which a question, an answer to the question, and a category of the question are associated with each other, using the question acquired by the first acquisition unit as a search condition; a first presentation unit that presents the plurality of searched categories; a second acquisition unit that acquires a selected category selected by the user from the plurality of presented categories; a second search unit that searches for an answer from the correspondence information using the question acquired by the first acquisition unit as a search condition; a third search unit that searches for an answer from the correspondence information using the question acquired by the first acquisition unit and the category acquired by the second acquisition unit as search conditions; and a second presentation unit that presents the answer searched for by the second search unit and the answer searched for by the third search unit to the user.

[0008] [2] Also, one aspect of the present invention is that in the presentation device described in [1], the second search unit and the third search unit use different search algorithms to search for an answer from the corresponding information.

[0009] [3] Moreover, in one aspect of the present invention, in the presentation device described in [1] or [2], the second search unit searches for an answer from the corresponding information based on a probabilistic similarity between the question acquired by the first acquisition unit and each of a plurality of questions present in the corresponding information.

[0010] [4] Furthermore, in one aspect of the present invention, in the presentation device described in any one of [1] to [3], the third search unit searches for an answer from the correspondence information based on a cosine similarity between the question vector acquired by the first acquisition unit and each vector of a plurality of questions present in the correspondence information.

[0011] [5] Moreover, one aspect of the present invention is a presentation device according to any one of [1] to [4], wherein the second search unit and the third search unit each search for a plurality of answers from the corresponding information, and the second presentation unit presents the plurality of results searched by the second search unit and the plurality of results searched by the third search unit separately and independently.

[0012] [6] Moreover, one aspect of the present invention is a presentation device according to any one of [1] to [5], further comprising a third acquisition unit that acquires information regarding the accuracy of the information presented on the second presentation unit from the user, and the first search unit, the second search unit, and the third search unit each learn based on the information acquired by the third acquisition unit.

[0013] [7] Also, one aspect of the present invention is a program that causes a computer to execute the following steps: a first acquisition step of acquiring a question from a user; a first search step of searching a plurality of categories of the question acquired by the first acquisition step using the question acquired by the first acquisition step as a search condition based on correspondence information in which a question, an answer to the question, and a category of the question are associated with each other; a first presentation step of presenting the plurality of searched categories; a second acquisition step of acquiring a selected category selected by the user from the plurality of presented categories; a second search step of searching for an answer from the correspondence information using the question acquired by the first acquisition step as a search condition; a third search step of searching for an answer from the correspondence information using the question acquired by the first acquisition step and the category acquired by the second acquisition step as search conditions; and a second presentation step of presenting the answer searched for by the second search step and the answer searched for by the third search step to the user. Effect of the Invention

[0014] According to the present invention, it is possible to provide a presentation device and a program capable of providing an FAQ that matches the intent of a user's question. [Brief description of the drawings]

[0015] [Figure 1] 1 is a block diagram of a system according to a first embodiment. [Diagram 2] 2 is a functional configuration diagram showing an example of a functional configuration of the presentation device according to the first embodiment. FIG. [Diagram 3] FIG. 4 is a diagram for explaining an example of a learning stage of a search unit according to the first embodiment. [Figure 4] FIG. 2 is a sequence diagram for explaining an example of a processing flow of the system according to the first embodiment. [Diagram 5] FIG. 4 is a functional configuration diagram showing an example of the functional configuration of a third search unit according to the first embodiment. [Figure 6]1 is a diagram for explaining an example of a screen configuration presented by the presentation device according to the first embodiment at the time of startup. FIG. [Figure 7] 1 is a diagram for explaining an example of a screen configuration when a question is input and presented by the presentation device according to the first embodiment. FIG. [Figure 8] 4 is a diagram for explaining an example of a screen configuration when a category is selected and presented by the presentation device according to the first embodiment. FIG. [Figure 9] 10 is a diagram for explaining an example of a screen configuration when a search result presented by the presentation device according to the first embodiment is selected. FIG. [Figure 10] FIG. 11 is a functional configuration diagram showing an example of the functional configuration of a presentation device according to a second embodiment. [Figure 11] FIG. 11 is a diagram for explaining an example of information stored in a chatbot log according to the second embodiment. [Figure 12] FIG. 11 is a diagram for explaining an example of a learning stage of the presentation device according to the second embodiment. [Figure 13] FIG. 11 is a diagram showing an example of a processing flow using a chatbot log 92 by the learning comprehensive engine 31 according to the second embodiment. [Figure 14] FIG. 11 is a sequence diagram for explaining an example of a processing flow of a system having a feedback function from a user according to the second embodiment. [Figure 15] FIG. 11 is a diagram for explaining an example of a screen configuration when a question is solved and is presented by the presentation device according to the second embodiment. [Figure 16] FIG. 11 is a first diagram for explaining an example of a screen configuration presented by the presentation device according to the second embodiment when a question remains unsolved. [Figure 17] FIG. 11 is a second diagram for explaining an example of a screen configuration presented by the presentation device according to the second embodiment when the question is unsolved. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0016] [Embodiment 1] The presentation device and the program according to the present embodiment will be described in detail below with reference to the accompanying drawings, showing preferred embodiments. Note that the present embodiment is not limited to these embodiments, and includes various modifications or improvements. In other words, the components described below include those that a person skilled in the art can easily imagine, and those that are substantially the same, and the components described below can be appropriately combined. In addition, the present embodiment allows various omissions, substitutions, or modifications of the components within the scope of the gist of the present invention.

[0017] First, the configuration of a system according to the first embodiment will be described with reference to FIGS.

[0018] FIG. 1 is a block diagram of a system according to the first embodiment. An example of the system 1 will be described with reference to the figure. The system 1 includes a plurality of terminal devices 10 and a presentation device 20. A terminal device 10-1 and a terminal device 10-2 included in the system 1 are examples of the plurality of terminal devices 10. The plurality of terminal devices 10 and at least one presentation device 20 are connected to each other via a network NW. The network NW is a predetermined communication network configured using lines. An example of the network NW is the Internet. The predetermined communication network NW is an open network, and devices connected to the communication network NW can communicate with each other using a predetermined protocol. In the Internet layer of the communication network NW, communication is performed using the Internet Protocol (IP) or the like.

[0019] In the illustrated example, the terminal device 10 and the presentation device 20 are connected to each other via a network NW. However, the present embodiment is not limited to this example, and for example, the terminal device 10 and the presentation device 20 may be directly connected to each other and configured as a single device.

[0020] The terminal device 10 is operated by a user. Based on the user's operation, the terminal device 10 transmits question information QI to the presentation device 20 via the network NW. The question information QI includes information related to a question the user asks the system. The question information QI includes, for example, words and sentences using natural language. The question information QI includes, for example, character data input by the user by operating an input device such as a keyboard or a touch panel, and voice data input by operating a voice acquisition device such as a microphone. The terminal device 10 may be, for example, a device such as a personal computer, a tablet computer, or a smartphone.

[0021] The presentation device 20 acquires question information QI from each of the multiple terminal devices 10. The presentation device 20 searches for answer information AI to be presented to the user based on each acquired question information QI. The search for answer information AI is performed by referring to correspondence information stored in a storage device (not shown). The correspondence information is a collection of information in which a question, an answer to the question, and a category to which the question belongs are associated with each other. The answer information AI may include at least one piece of information in the correspondence information. More preferably, the answer information AI may include information on a set of multiple questions in the correspondence information and the answers to the questions. The presentation device 20 transmits the answer information AI to the terminal device 10 that has transmitted the question information QI.

[0022] The correspondence information may correspond to, for example, information on the date of an answer to a question, a title or a number identifying the question and the answer to the question, an identification number identifying the category of the question, etc. Furthermore, the correspondence information may further correspond to information on the content of an interaction between the user and the system, an identification number identifying the interaction, subcategories that are further hierarchical categories, and identification numbers identifying the subcategories, etc.

[0023] The answer information AI may include, for example, a question searched for by referring to the corresponding information, an answer to the question, and a URL (Uniform Resource Locator) indicating an address where this information is stored. The answer information AI does not necessarily need to include information about the question, but may include, for example, an answer to the question. The answer information AI may further include information about the audio that reads out the answer, information about an image used to supplementarily explain the answer, and the like.

[0024] The terminal device 10 acquires answer information AI corresponding to the transmitted question information QI from the presentation device 20. The terminal device 10 presents the answer information AI acquired from the presentation device 20 to the user. The answer information AI is presented, for example, using words or sentences using a natural language. The answer information AI may be presented in a visual manner using text data or in an auditory manner using voice data. Furthermore, the answer information AI may be presented in a manner other than a visual or auditory manner.

[0025] FIG. 2 is a functional configuration diagram showing an example of the functional configuration of the presentation device according to the first embodiment. An example of the functional configuration of the presentation device 20 will be described with reference to the same figure. The presentation device 20 includes a first acquisition unit 21, a first search unit 22, a first presentation unit 23, a second acquisition unit 24, a second search unit 25, a third search unit 26, and a second presentation unit 27. Each of these functional units is realized, for example, by using a computer and software. Also, each functional unit may be realized by using an electronic circuit as necessary. Furthermore, each functional unit does not have to be included in a single device, and the presentation device 20 may be configured by a plurality of devices.

[0026] The first acquisition unit 21 acquires question information QI transmitted from the terminal device 10. The question information QI transmitted from the terminal device 10 is, in other words, information acquired based on an operation from a user. The first acquisition unit 21 transmits the acquired question information QI to the first search unit 22, the second search unit 25, and the third search unit 26.

[0027] The first search unit 22 acquires the question information QI from the first acquisition unit 21. The first search unit 22 searches the corresponding information using the acquired question information QI as a search condition, thereby searching for information on the category to which the question information QI belongs (hereinafter referred to as category information CI) from among the categories included in the corresponding information. The first search unit 22 searches for a plurality of categories in descending order of similarity as the category of the question information QI. In other words, it can be said that the category information CI includes information on a plurality of categories. The first search unit 22 transmits the category information CI acquired by the search to the first presentation unit 23.

[0028] The first presentation unit 23 acquires the category information CI from the first search unit 22. The first presentation unit 23 presents the acquired category information CI to the terminal device 10 that has transmitted the question information QI. That is, the first presentation unit 23 presents a plurality of categories to the user. Here, the method of presenting a plurality of categories is not limited to an example in which the first presentation unit 23 directly presents the category information CI to the terminal device 10. For example, the presentation device 20 may transmit the category information CI to the terminal device 10, and the terminal device 10 may present the category information CI to the user by a suitable method according to the transmitted category information CI. The user selects a category that is closest to the question to which the user wants an answer from among the presented plurality of categories. Specifically, the user operates the terminal device 10 to select one of the categories and transmits the information to the presentation device 20 as the selected category information SCI.

[0029] The second acquisition unit 24 acquires a category selected by the user from among the multiple categories presented by the first presentation unit 23. Specifically, the second acquisition unit 24 acquires selected category information SCI transmitted from the terminal device 10. The selected category information SCI includes information on the category selected by the user from the category information CI. The second acquisition unit 24 transmits the acquired selected category information SCI to the third search unit 26. Here, the user may not be able to select a category closest to the question to which the user wishes to receive an answer. A case in which the user cannot select a category is, for example, a case in which the presented options do not include a corresponding category. In such a case, the user may transmit information indicating that a corresponding category does not exist as the selected category information SCI. In this case, all categories in the correspondence information may be treated as having been selected.

[0030] Here, the search according to this embodiment will be described. The search according to this embodiment includes an indirect search for an answer desired by the user and a direct search. In the case of indirect search, a plurality of categories estimated to correspond to a question from the user are first presented, the user is asked to select a category that is closest to the question from the plurality of categories, and an answer belonging to the selected category is searched for. That is, in the case of indirect search, answer information AI is searched for using selected category information SCI and question information QI as search conditions. In the case of direct search, an answer corresponding to a question from the user is directly searched for from among corresponding information. That is, in the case of direct search, answer information AI is searched for using question information QI as a search condition without relying on selected category information SCI.

[0031] The third search unit 26 performs an indirect search. That is, the third search unit 26 can also be said to be an indirect search unit. The third search unit 26 acquires question information QI from the first acquisition unit 21. Furthermore, the third search unit 26 acquires selected category information SCI from the second acquisition unit 24. The third search unit 26 searches for a plurality of results, that is, indirect search answer information AI2, from among the corresponding information using the acquired question information QI and selected category information SCI as search conditions. The indirect search answer information AI2 is answer information AI obtained when searching indirectly. The third search unit 26 transmits the indirect search answer information AI2 obtained as a result of the search to the second presentation unit 27. Here, the search may be based on a machine learning method or a rule base.

[0032] The second search unit 25 performs a direct search. In other words, the second search unit 25 can also be said to be a direct search unit. The second search unit 25 acquires question information QI from the first acquisition unit 21. The second search unit 25 searches for multiple results, i.e., direct search answer information AI1, from among the corresponding information using the acquired question information QI as a search condition. The direct search answer information AI1 is answer information AI obtained when searching directly. The second search unit 25 transmits the direct search answer information AI1 obtained as a search result to the second presentation unit 27.

[0033] The second presentation unit 27 receives the direct search answer information AI1 from the second search unit 25. The second presentation unit 27 acquires the indirect search answer information AI2 from the third search unit 26. The second presentation unit 27 presents the direct search answer information AI1 and the indirect search answer information AI2 as answer information AI to the terminal device 10 that has transmitted the question information QI. Here, the method of presenting the answer information AI is not limited to the example in which the second presentation unit 27 presents the answer information AI directly to the terminal device 10. For example, the presentation device 20 may transmit the answer information AI to the terminal device 10, and the terminal device 10 may present the category information CI to the user by a suitable method according to the transmitted category information CI.

[0034] In addition, the second search unit 25 and the third search unit 26 may search for an answer to a question from among the correspondence information using different search algorithms. For example, the second search unit 25 may determine the correspondence relationship between a plurality of pieces of information included in the correspondence information and the question information QI using an algorithm using probabilistic similarity. In that case, the second search unit 25 searches for answer information AI from among the correspondence information based on the relationship determined by the probabilistic similarity. The third search unit 26 may determine by an algorithm using cosine similarity between a vector of information included in the correspondence information and a vector of the question information QI. In that case, the third search unit 26 searches for answer information AI from among the correspondence information based on the determined relationship. However, the present embodiment is not limited to this example, and for example, the second search unit 25 and the third search unit 26 may search using the same algorithm. In addition, the second search unit 25 may search based on an algorithm using cosine similarity. Furthermore, the third search unit 26 may search based on an algorithm using probabilistic similarity.

[0035] FIG. 3 is a diagram for explaining an example of learning of the search unit according to the embodiment. With reference to the figure, an example of learning of the first search unit 22, the second search unit 25, and the third search unit 26 will be explained. The learning unit 30 includes a learning comprehensive engine 31 and a correspondence information database 32. Each of these functional units is realized, for example, by using a computer and software. Each functional unit may also be realized by using an electronic circuit as necessary. Each functional unit does not have to be included in a single device, and may be in a form in which the learning unit 30 is configured by a plurality of devices. Each functional unit may also include a storage means such as a semiconductor memory or a magnetic hard disk device as necessary.

[0036] The learning comprehensive engine 31 transmits DB control information DBCI to the corresponding information database 32. The corresponding information database 32 receives the DB control information DBCI from the learning comprehensive engine 31. The corresponding information database 32 transmits answer information AI and category information CI to the learning comprehensive engine 31 according to the received DB control information DBCI. The learning comprehensive engine 31 receives the answer information AI and category information CI from the corresponding information database 32. That is, the learning comprehensive engine 31 issues an instruction to the corresponding information database 32 and acquires information necessary for learning. The learning comprehensive engine 31 trains the first search unit 22, the second search unit 25, and the third search unit 26 based on the received answer information AI and category information CI. Here, the learning by the learning comprehensive engine 31 may be performed separately and independently for each search unit. The learning comprehensive engine 31 transmits update information UI based on the learning to the first search unit 22, the second search unit 25, and the third search unit 26, and updates each search unit. When the third search unit 26 performs a rule-based search, the learning comprehensive engine 31 does not necessarily need to update the third search unit 26. The learning of each search unit by the learning comprehensive engine 31 may be performed, for example, when a predetermined condition is satisfied, or may be performed based on an operation by an administrator.

[0037] FIG. 4 is a sequence diagram for explaining an example of a processing flow of the system according to the embodiment. An example of the processing flow of the system will be explained with reference to the same figure. In the figure, a user operates the terminal device 10 to transmit a question to the presentation device 20 via the network NW (step S41). The presentation device 20 searches for a plurality of category information CI from among the categories included in the correspondence information by using the transmitted question as a search condition. Next, the presentation device 20 presents a plurality of category information to the terminal device 10 (step S42). The user operates the terminal device 10 to select a category to which the transmitted question is determined to belong from among the plurality of presented categories via the network NW (step S43).

[0038] The presentation device 20 searches for direct search answer information AI1 from among the corresponding information using only the question information QI as a search condition, regardless of the selected category information SCI. Furthermore, the presentation device 20 searches for indirect search answer information AI2 from among the corresponding information using the selected category information SCI and the question information QI as search conditions. The presentation device 20 presents an answer to the terminal device 10 based on the direct search answer information AI1 and the indirect search answer information AI2 (step S44).

[0039] 5 is a diagram for explaining an example of a functional configuration of the third search unit according to the embodiment. With reference to the same figure, an example of the functional configuration of the third search unit will be explained. The third search unit 26 includes a word segmentation processing unit 261, a vectorization processing unit 262, and an inference unit 263.

[0040] The word segmentation processor 261 acquires the question information QI from the first acquisition unit 21. The word segmentation processor 261 divides the acquired question information QI, for example, into words for each morpheme. That is, the word segmentation processor 261 breaks down a sentence included in the question information QI into words as preprocessing for the vectorization processor 262. The word segmentation processor 261 transmits morpheme information MI, which is information about at least one word broken down into morphemes, to the vectorization processor 262.

[0041] The vectorization processing unit 262 receives the morpheme information MI from the word segmentation processing unit 261. The vectorization processing unit 262 converts each of the multiple words included in the received morpheme information MI into a vector. Converting words into vectors means converting the words included in the morpheme information MI into a number sequence so that they can be easily processed. For example, converting words into vectors may involve determining vectors corresponding to words by machine learning. That is, the vectorization processing unit 262 may convert words into vectors as a preprocessing for the inference unit 263. The vectorization processing unit 262 transmits vector information VI, which is information about the vectorized morpheme information MI, to the inference unit 263.

[0042] The inference unit 263 receives the vector information VI from the vectorization processing unit 262, and receives the selected category information SCI from the second acquisition unit 24. The inference unit 263 performs inference on the indirect search answer information AI2 from among the corresponding information, using the received vector information VI and selected category information SCI as conditions. Inference means performing a search by machine learning. The inference unit 263 transmits the indirect search answer information AI2 obtained by inference to the second presentation unit 27.

[0043] In addition, when the vectorization processing unit 262 is realized by a machine learning method, the vectorization processing unit 262 can appropriately vectorize learned words, but may not be able to appropriately vectorize unlearned words. When a word that cannot be appropriately vectorized (hereinafter, referred to as an unlearned word) is included in the question information QI, the question information QI including the unlearned word does not need to be decomposed into words for each morpheme by the word segmentation processing unit 261 and then vectorized by the vectorization processing unit 262. When vectorization is not performed, the inference unit 263 may search for the question information QI from the corresponding information using, for example, the number of words commonly included in the question information QI and the information in the corresponding information as a search condition.

[0044] Next, an example of a screen display according to the first embodiment will be described with reference to FIGS.

[0045] 6 is a diagram for explaining an example of a screen display at the time of startup according to the embodiment. With reference to the same figure, an example of a screen display at the time of startup will be explained. A startup screen display 50 has a suggestion side icon 51, a startup speech bubble 52, a text information input section 53, and an execution button 54 as a screen configuration.

[0046] The presentation side icon 51, for example, indicates the sender of the message. By indicating the sender of the message, even when multiple users send messages, it becomes possible to identify who sent the message. When a user operates the terminal device 10 to send question information QI to the presentation device 20, the user and the presentation device 20 may exchange information one-to-one. In such a case, the sender of the speech bubble is not confused, so the presentation side icon 51 does not necessarily need to be displayed.

[0047] The start-up speech bubble 52 is a message from the presentation device 20 to the user. The start-up speech bubble 52 clearly indicates that the user can ask the presentation device 20 a question. The start-up speech bubble 52 is displayed when the presentation device 20 is accessed. The term "access" refers to connecting to the presentation device 20, and includes, for example, writing or reading data to the presentation device 20. In the illustrated example, the start-up speech bubble 52 contains the text "Please enter the content of your inquiry." Note that, if it is recognizable that a question can be asked to the presentation device 20, the start-up speech bubble 52 does not necessarily need to be displayed.

[0048] The text information input section 53 is a text box used for the user to input a question. In the illustrated example, the text information input section 53 contains a sentence "cannot log in." That is, in the illustrated example, the user is asking a question to the effect that he or she is in trouble because he or she is unable to log in. Note that the text information input section 53 may contain a description that allows the user to recognize that he or she can ask a question to the presentation device 20 before the user inputs a question, for example.

[0049] The execute button 54 is a button element operated by the user. When the execute button 54 is operated by the user, the question input to the text information input unit 53 is transmitted to the presentation device 20. The execute button 54 may be shaped in a manner that allows the user to recognize that transmission is possible, such as an "airplane" mark.

[0050] 7 is a diagram for explaining an example of a screen display when a question is input according to the embodiment. With reference to the same figure, an example of the screen display when a question is input will be explained. A screen display 60 when inputting text information further has an input text balloon 61 and a category information balloon 65 as a screen configuration.

[0051] The input text bubble 61 is a question input to the text information input unit 53. In the illustrated example, the input text bubble 61 contains the text "I can't log in." That is, the example shown in FIG. 7 shows a screen display that is displayed after the user presses the execute button 54 in the example shown in FIG. 6. The text "I can't log in" is acquired as question information QI by the first acquisition unit 21. The presentation device 20 searches for category information CI corresponding to the question information QI, and displays the searched category information CI in the category information bubble 65.

[0052] The category information balloon 65 has, as its components, category buttons 62a to 62f (hereinafter, when the category buttons are not to be distinguished from one another, they may be simply referred to as category buttons 62), other category buttons 63, and an unknown button 64. The multiple category buttons 62 are preferably displayed in order of increasing similarity with the question information QI. In the illustrated example, the first to sixth category buttons are displayed in order of increasing similarity with the question information QI. It is sufficient that at least one category button 62 is displayed. Here, if the number of category buttons 62 as options increases, it may become a burden for the user to select a category. Therefore, it is preferable that the presentation device 20 displays about six category buttons 62, which is a number that does not make it a burden for the user to select a category.

[0053] The user selects a category to which the user's question is thought to belong from among the multiple displayed category buttons 62. The presentation device 20 acquires information on the selected category as selected category information SCI. If the user determines that the category to which the user's question is thought to belong is not included in the multiple displayed category buttons 62, the user may select the other category button 63. When the other category button 63 is selected, the presentation device 20 may display category buttons 62 of further categories (categories with the seventh or higher similarity in the illustrated example). The presentation device 20 may repeat the process of displaying further category buttons until a category button or the unknown button 64 is selected.

[0054] Furthermore, if the user cannot determine the category to which the question he or she asks likely belongs, the user may select the unknown button 64. When the unknown button 64 is selected, the second acquisition unit 24 may acquire, as selected category information SCI, that the category is unknown. Furthermore, when the unknown button 64 is selected, the second acquisition unit 24 may acquire, for example, all categories included in the correspondence information as selected category information SCI. By acquiring all categories as selected category information SCI, all categories included in the correspondence information are transmitted to the third search unit 26 as selected category information SCI.

[0055] Note that the term "selection" refers to, for example, the user operating the terminal device 10 to select a button. The term "selection" may also refer to, for example, clicking any of the button elements displayed in the category information balloon 65. The term "selection" may also refer to inputting a word or sentence, etc., into the text information input unit 53, which clearly indicates that any of the multiple category buttons 62, the other category button 63, or the unknown button 64 is to be selected, and transmitting the word or sentence to the presentation device 20 using the execute button 54.

[0056] Fig. 8 is a diagram for explaining an example of a screen display when a category is selected according to the embodiment. With reference to the same figure, an example of the screen display when a category is selected will be explained. The example shown in Fig. 8 is an example of a screen displayed after the user selects the first category in the screen display shown in Fig. 7. The category selection screen display 70 further has an input category bubble 71 and a search result bubble 77 as a screen configuration.

[0057] The input category bubble 71 displays information about the category selected by the user. The input category bubble 71 is for the user to confirm the selection. For example, if the category selected by the user is explicitly displayed using the category information bubble 65, the presentation device 20 does not necessarily need to display the input category bubble 71.

[0058] The search result bubble 77 has an indirect search result button group 74, a direct search result button group 75, and an other search result button 76 as components. The indirect search result button group 74 and the direct search result button group 75 may each have a plurality of search result buttons that display answer information AI. In the illustrated example, the indirect search result button group 74 and the direct search result button group 75 each have first to fifth search result buttons. The plurality of search result buttons may be displayed in order of the degree of similarity between the question information QI and the answer information AI. Each search result button may display the title of the answer information AI. At least one search result button is displayed, and for example, about five buttons may be displayed so as not to burden the user with selection. In addition, when explaining an example of an embodiment, the number of search result buttons is set to five in accordance with the figure.

[0059] The indirect search result button group 74 displays the indirect search answer information AI2 obtained by the third search unit 26 performing an indirect search. In the illustrated example, the indirect search result button group 74 includes an indirect search result button 72a to an indirect search result button 72e as the indirect search answer information AI2. In addition, the direct search result button group 75 displays the direct search answer information AI1 obtained by the second search unit 25 performing a direct search. In the illustrated example, the indirect search result button group 74 includes an indirect search result button 72a to an indirect search result button 72e as the indirect search answer information AI2. In addition, in the illustrated example, the indirect search result button group 74 is displayed above the direct search result button group 75. This is because the indirect search, in which a search is performed using a category selected by the user, is considered to be more likely to be an answer to the user's question. However, this embodiment is not limited to this example, and the direct search result button group 75 may be displayed above the indirect search result button group 74. Furthermore, the components of the indirect search result button group 74 (indirect search result button 72a to indirect search result button 72e) and the components of the direct search result button group 75 (direct search result button 73a to direct search result button 73e) may be displayed alternately.

[0060] The other search results button 76 is a button to be selected when the answer information AI desired by the user is not present in the indirect search result button group 74 and the direct search result button group 75. When the user selects the other search results button 76, the search result buttons with the sixth to tenth highest similarities, respectively, are further displayed.

[0061] FIG. 9 is a diagram for explaining an example of a screen display when a search result is selected according to the embodiment. With reference to the same figure, an example of a screen display when a search result is selected will be described. The example shown in FIG. 9 is an example of a screen displayed after a user selects any of the button elements displayed in the search result bubble 77 in the screen display shown in FIG. 8. A search result selection screen display 80 further includes an answer information bubble 81 and an answer content bubble 82 as a screen configuration. Specifically, the user selects a search result button that is considered to be closest to the question asked by the user from among the multiple button elements displayed in the search result bubble 77 (i.e., a search result button that is determined to include answer information AI to question information QI). The title of the answer information AI included in the selected search result button is displayed in the answer information bubble 81. Note that the answer information bubble 81 is not necessarily required to be displayed since it is simply for the user to confirm the selection. Thereafter, the answer information AI included in the selected search result button is displayed in the answer content bubble 82.

[0062] [Summary of embodiment 1] According to the embodiment described above, the presentation device 20 includes a first acquisition unit 21 that acquires question information QI from a user, a first search unit 22 that searches for category information CI from the correspondence information using the question information QI acquired by the first acquisition unit 21 as a search condition based on correspondence information in which a question, an answer to the question, and a category of the question are associated with each other, a first presentation unit that presents the category information CI to the user, a second acquisition unit 24 that acquires selected category information SCI selected by the user, a second search unit 25 that searches for direct search answer information AI1 from the correspondence information using the question information QI acquired by the second acquisition unit 24 as a search condition, a third search unit 26 that searches for indirect search answer information AI2 from the correspondence information using the question information QI acquired by the second acquisition unit 24 and the selected category information SCI acquired by the second acquisition unit 24 as search conditions, and a second presentation unit 27 that presents the direct search answer information AI1 and the indirect search answer information AI2 to the user. That is, the presentation device 20 presents to the user indirect search answer information AI2, which includes the category selected by the user in the search conditions, and direct search answer information AI1, which does not include the category selected by the user in the search conditions. Here, if answers are searched only within the range of categories selected by the user, answers belonging to different categories may not be presented, and the answer desired by the user may not be presented. According to this embodiment, even if the user selects a category that is incorrect from the purpose of the question information QI for category information CI based on the question information QI, the direct search answer information AI1, which is not dependent on the category, is further presented. Therefore, the presentation device 20 according to this embodiment can provide an FAQ that matches the intention of the user's question.

[0063] According to the above-described embodiment, in the presentation device 20, the second search unit 25 searches for direct search answer information AI1 from among the corresponding information using the question information QI acquired by the second acquisition unit 24 as a search condition, and the third search unit 26 searches for indirect search answer information AI2 from among the corresponding information using the question information QI acquired by the second acquisition unit 24 and the selected category information SCI acquired by the second acquisition unit 24 as search conditions. The second search unit 25 and the third search unit 26 search for answer information AI from among the corresponding information using different search algorithms. That is, according to this embodiment, the second search unit 25 and the third search unit 26 perform searches using different algorithms. By performing searches using different algorithms, a wider range of search results can be obtained than when a search is performed using the same algorithm. Therefore, by obtaining a wider range of search results, the presentation device 20 according to this embodiment can provide FAQs that match the intention of the user's question.

[0064] According to the above-described embodiment, in the presentation device 20, the second search unit 25 searches for the direct search answer information AI1 from among the corresponding information based on an algorithm using probabilistic similarity, using the question information QI acquired by the second acquisition unit 24 as a search condition. The second search unit 25 searches for the answer information AI without relying on the selected category information SCI, so the search targets are enormous. It is not easy to search for the answer information AI to be provided to the user from among the enormous search targets. Therefore, the second search unit 25 can efficiently search for the answer information AI suitable for the user from among the enormous search targets by determining the respective relationships between the multiple pieces of information included in the corresponding information and the question information QI using probabilistic similarity. Therefore, the presentation device 20 according to this embodiment can provide an FAQ that matches the intention of the user's question.

[0065] According to the above-described embodiment, in the presentation device 20, the third search unit 26 searches for the indirect search answer information AI2 from among the corresponding information based on an algorithm using cosine similarity, using the question information QI acquired by the second acquisition unit 24 and the selected category information SCI acquired by the second acquisition unit 24 as search conditions. The third search unit 26 determines the relationship between each of the multiple pieces of information included in the corresponding information and the question information QI by an algorithm using cosine similarity between the vector of the information included in the corresponding information and the vector of the question information QI. That is, the third search unit 26 can determine the relationship by placing importance on the meaning of the words included in the corresponding information and the question information QI through a search using the cosine similarity of the vectors. Therefore, the presentation device 20 according to this embodiment can provide an FAQ that matches the intention of the user's question.

[0066] According to the above-described embodiment, in the search result selection screen display 80 of the presentation device 20, the direct search answer information AI1 searched by the second search unit 25 and the indirect search answer information AI2 searched by the third search unit 26 are presented separately and independently by the indirect search result button group 74 and the direct search result button group 75. The indirect search result including the selected category information SCI in the search conditions is more likely to include the answer information AI desired by the user than the direct search result not including the selected category information SCI in the search conditions. That is, by presenting the indirect search result button group 74 and the direct search result button group 75 separately and independently, the user can easily select the answer information AI desired by the user. Therefore, the presentation device 20 according to this embodiment can provide an FAQ that matches the intention of the user's question.

[0067] [Embodiment 2] Next, a second embodiment will be described. The second embodiment is different from the first embodiment in that it further has a function of learning or relearning based on feedback (FB) from a user. In the following description of the second embodiment, the description of the functions already described in the first embodiment may be omitted. First, the configuration of the system according to the second embodiment will be described with reference to Figs. 10 to 14.

[0068] FIG. 10 is a diagram for explaining an example of a presentation device having a user feedback function according to the second embodiment. With reference to the same figure, an example of a presentation device 20A having a user feedback function will be explained. In the configuration of the presentation device 20A, the same components as those of the presentation device 20 may be denoted by the same reference numerals and explanations thereof may be omitted. The presentation device 20A differs from the presentation device 20 in that it further includes a third acquisition unit 91 and a chatbot log 92. Each of these functional units is realized, for example, by using a computer and software. In addition, each functional unit may be realized by using an electronic circuit as necessary. Furthermore, each functional unit does not have to be included in a single device, and may be in a form in which the presentation device 20A is configured from a plurality of devices. In addition, each functional unit may include a storage means such as a semiconductor memory or a magnetic hard disk device as necessary.

[0069] The third acquisition unit 91 acquires feedback information FI transmitted from the terminal device 10. The feedback information FI transmitted from the terminal device 10 is information acquired based on an operation from a user. The feedback information FI includes information related to the answer information AI presented to the user. The third acquisition unit 91 transmits the acquired feedback information FI to the chatbot log 92. The feedback information FI includes, for example, information related to whether the answer information AI presented to the user is an answer to the question information QI (hereinafter, may be referred to as correct / incorrect information EI). The correct / incorrect information EI may be, for example, one-bit information of 0 and 1. The feedback information FI may include, for example, selected category information SCI to which the user has determined that the question information QI belongs, or may include answer information AI selected by the user (hereinafter, may be referred to as selected answer information SAI).

[0070] The chatbot log 92 receives and stores the feedback information FI from the third acquisition unit 91. In the illustrated example, the chatbot log 92 is a database separate and independent from the correspondence information database 32.

[0071] FIG. 11 is a diagram for explaining an example of information stored in a chatbot log according to the second embodiment. With reference to the figure, an example of information stored in the chatbot log 92 will be explained. The chatbot log 92 stores question information QI, selected category information SCI, selected answer information SAI, and correct / incorrect information EI, which are included in feedback information FI. That is, the chatbot log 92 stores information acquired based on a user's operation. The information stored in the chatbot log 92 may be used, for example, to train each search, or may be used for viewing by an administrator as performance report information.

[0072] In the illustrated example, the question information QI, the selected category information SCI, the selected answer information SAI, and the correct / incorrect information EI are stored together in the chatbot log 92. However, the present embodiment is not limited to this example. The question information QI, the selected category information SCI, the selected answer information SAI, and the correct / incorrect information EI may be stored separately in a plurality of databases. Specifically, the question information QI, the selected category information SCI, and the selected answer information SAI may be stored in the chatbot log 92, and the correct / incorrect information EI may be stored in a feedback DB (not shown). By storing the question information QI, the selected category information SCI, the selected answer information SAI, and the correct / incorrect information EI in different databases, the question information QI, the selected category information SCI, and the selected answer information SAI can be preferably normalized.

[0073] FIG. 12 is a diagram for explaining an example of learning based on feedback from a user according to an embodiment. With reference to the same figure, an example of learning based on feedback from a user will be described. In the configuration of the learning unit 30A having an FB, the same components as those of the learning unit 30 may be denoted by the same reference numerals and description thereof may be omitted. The learning unit 30A having an FB differs from the learning unit 30 in that it further includes a chatbot log 92. The learning unit 30A having an FB further includes the chatbot log 92, thereby causing the first search unit 22, the second search unit 25, and the third search unit 26 to learn.

[0074] The learning comprehensive engine 31 transmits DB control information DBCI to the corresponding information database 32 and the chatbot log 92. The corresponding information database 32 and the chatbot log 92 each receive DB control information DBCI from the learning comprehensive engine 31. The corresponding information database 32 transmits answer information AI and category information CI to the learning comprehensive engine 31 according to the DB control information DBCI. The chatbot log 92 transmits question information QI, selected category information SCI, selected answer information SAI, and correct / incorrect information EI to the learning comprehensive engine 31 according to the DB control information DBCI. The learning comprehensive engine 31 receives answer information AI and category information CI from the corresponding information database 32. Furthermore, the learning comprehensive engine 31 receives question information QI, selected category information SCI, selected answer information SAI, and correct / incorrect information EI. That is, the learning comprehensive engine 31 issues instructions to the corresponding information database 32 and the chatbot log 92 to obtain information necessary for learning. The comprehensive engine for learning 31 performs learning based on the received answer information AI and category information CI. The comprehensive engine for learning 31 performs learning based on the received question information QI, selected category information SCI, selected answer information SAI, and correct / incorrect information EI. The comprehensive engine for learning 31 transmits update information UI based on the learning to the first search unit 22, the second search unit 25, and the third search unit 26, and updates each search unit. Note that when the third search unit 26 performs a search based on a rule base, the comprehensive engine for learning 31 does not necessarily need to update the third search unit 26.

[0075] 13 is a diagram showing an example of a processing flow using the chatbot log 92 by the learning comprehensive engine 31 according to embodiment 2. With reference to the figure, an example of a processing flow using the chatbot log 92 by the learning comprehensive engine 31 according to embodiment 2 will be described. The learning comprehensive engine 31 performs processing based on, for example, a predetermined condition or an operation by an administrator.

[0076] The learning comprehensive engine 31 judges whether the answer information AI presented to the user is an answer to the question information QI based on the correct / incorrect information EI (step S901). If the answer information AI presented to the user is an answer to the question information QI (step S901; Yes), the learning comprehensive engine 31 learns based on the question information QI, the selected category information SCI, and the selected answer information SAI (step S902). If the answer information AI presented to the user is not an answer to the question information QI (step S901; No), the administrator associates the correct category information CI and answer information AI with the user's question information QI (step S903). The learning comprehensive engine 31 learns based on the associated question information QI, category information CI, and answer information AI (step S904). If the user has not answered whether the answer information AI presented to the user is an answer to the question information QI (step S901; unknown), the processes of steps S903 and S904 are performed. Note that, if the user has not answered, the administrator does not necessarily need to associate the category information CI and the answer information AI with the user's question information QI.

[0077] Fig. 14 is a sequence diagram for explaining an example of a process flow of a system having a user feedback function according to an embodiment. With reference to the same figure, an example of a process flow of a system having a user feedback function will be explained. The process illustrated in Fig. 12 includes a feedback transmission process (step S94) in addition to the process illustrated in Fig. 4. In the explanation given with reference to the same figure, the same processes as those in Fig. 4 may be denoted by the same reference numerals and explanations thereof may be omitted.

[0078] 4, the presentation device 20 presents an answer to the terminal device 10 (step S44). The user operates the terminal device 10 to transmit feedback on the presented answer via the network NW (step S94). Here, the feedback transmitted by the user is feedback information FI.

[0079] FIG. 15 is a diagram for explaining an example of a screen display when a question is solved according to the embodiment. With reference to the same figure, an example of a screen display when a question is solved will be explained. A screen display 100 when a question is solved is different from a screen display 80 when a search result is selected in that the screen display 100 further includes a question solved button 101, a question unsolved button 102, a question solved speech bubble 103, and an answer completed speech bubble 104 as a screen configuration. In the illustrated example, the question solved button 101 and the question unsolved button 102 are shown in a form included in the answer speech bubble 82. However, the present embodiment is not limited to this example, and the question solved button 101 and the question unsolved button 102 may be included in a speech bubble displayed separately and independently from the answer speech bubble 82, or may be displayed outside the speech bubble.

[0080] When the question solution button 101 is selected, content clearly indicating that the user's question has been solved is displayed on the terminal device 10 as a question solution bubble 103. In the illustrated example, the word "good" is written in the question solution bubble 103. After that, an answer completion bubble 104 is displayed. Note that the question solution bubble 103 is not necessarily required to be displayed, since it is merely for the user to confirm the content of the selection.

[0081] 16 is a first diagram for explaining an example of a screen display when a question is unresolved according to the embodiment. The first diagram of the screen display when a question is unresolved will be explained with reference to the same figure. The first unresolved question screen display 110 differs from the search result selection screen display 80 in that it further includes a question solved button 101, an unresolved question button 102, an unresolved question balloon 111, a plurality of preliminary search result balloons 117, and a balloon 118 of a selected preliminary search result as a screen configuration.

[0082] When the unsolved question button 102 is selected, content indicating that the question is not solved is displayed on the terminal device 10 as an unsolved question balloon 111. After that, a plurality of preliminary search result balloons 117 are displayed on the terminal device 10. Note that the unsolved question balloon 111 is merely for the user to confirm the content of the selection, and does not necessarily have to be displayed.

[0083] The multiple preliminary search result bubbles 117 may include, for example, a preliminary indirect search result button group 114, a preliminary direct search result button group 115, and another preliminary search result button 116, which are separate and independent. In addition to the answer information AI not selected by the user in the screen display of FIG. 9, the preliminary indirect search result button group 114 and the preliminary direct search result button group 115 may display answer information AI instead of the selected answer information AI. Specifically, when the indirect search answer information AI2 having the highest similarity between the question information QI and the answer information AI is selected, the four indirect search answer information AI2 that were not selected and the indirect search answer information AI2 having the sixth highest similarity are displayed in the indirect search button. Also, since the user has not selected the direct search answer information AI1 from the preliminary direct search result button group 115, the preliminary direct search result button group 115 displays the direct search answer information AI1 similar to the direct search result button group 75 displayed in FIG. 9. In the illustrated example, the preliminary indirect search result button group 114 and the preliminary direct search result button group 115 each have five buttons as components. However, this embodiment is not limited to this example. The preliminary indirect search result button group 114 and the preliminary direct search result button group 115 only need to have a number of buttons as components that do not burden the user when selecting. Specifically, the preliminary indirect search result button group 114 and the preliminary direct search result button group 115 may have, for example, three or four buttons each as components, or the preliminary indirect search result button group 114 may have more buttons as components than the preliminary direct search result button group 115.

[0084] The user selects a search result button that is determined to include answer information AI to question information QI from among multiple preliminary search result bubbles 117. The input preliminary search result bubble 118 displays the title of the answer information AI included in the selected search result button, etc. Note that the selected preliminary search result bubble 118 does not necessarily have to be displayed, since it is simply for the user to confirm the input contents. Thereafter, a new answer content bubble 82 is displayed. The new answer content bubble 82 displays the answer information AI included in the selected search result button. Note that the new answer content bubble 82 may further display to the user a question solved button 101 and a question unsolved button 102 for the answer.

[0085] FIG. 17 is a second diagram for explaining an example of a screen display when a question is unsolved according to the embodiment. With reference to the same figure, the second diagram of the screen display when a question is unsolved will be explained. The second question unsolved screen display 120 is similar to the first question unsolved screen display 110 in that it further has a screen configuration including a question solved button 101, a question unsolved button 102, an unsolved question balloon 111, a plurality of preliminary search result balloons 117, and a balloon 118 of a selected preliminary search result. The second question unsolved screen display 120 is different from the first question unsolved screen display 110 in that the answer information AI having the sixth to tenth highest similarity between the question information QI and the answer information AI is displayed in the preliminary indirect search result button group 114 and the preliminary direct search result button group 115. By displaying the answer information AI with the sixth to tenth highest similarity between the question information QI and the answer information AI, the user can easily display answer information AI other than the answer information AI presented in the indirect search result button group 74 and the direct search result button group 75. [Summary of embodiment 2]

[0086] According to the above-described embodiment, the presentation device 20A may further include a third acquisition unit 91 and a chatbot log 92 in addition to the functional configuration of the presentation device 20. By further including the third acquisition unit 91 and the chatbot log 92, the presentation device 20A can learn based on the chatbot log 92 created from feedback from a user. The presentation device 20A can present answer information AI desired by the user by using the chatbot log 92 for learning. Therefore, the present invention can provide an FAQ that matches the intention of the user's question.

[0087] In addition, all or part of the functions of each unit of the presentation device 20 and the presentation device 20A in the above-mentioned embodiment may be realized by recording a program for realizing these functions in a computer-readable recording medium, reading the program recorded in the recording medium into a computer system, and executing it. Note that the "computer system" here includes hardware such as an OS and peripheral devices.

[0088] In addition, the term "computer-readable recording medium" refers to portable media such as flexible disks, optical magnetic disks, ROMs, and CD-ROMs, and storage units such as hard disks built into computer systems. Furthermore, the term "computer-readable recording medium" may also include those that dynamically hold a program for a short period of time, such as a communication line when transmitting a program via a network such as the Internet or a communication line such as a telephone line, and those that hold a program for a certain period of time, such as a volatile memory inside a computer system that serves as a server or client in such cases. Furthermore, the above program may be one that realizes part of the above-mentioned functions, or may be one that can realize the above-mentioned functions in combination with a program already recorded in the computer system.

[0089] Although an embodiment of the present invention has been described above, the present invention is not limited to the above embodiment, and various modifications can be made without departing from the spirit of the present invention. [Explanation of symbols]

[0090] 1...system, 10...terminal device, 20...presentation device, 21...first acquisition unit, 22...first search unit, 23...first presentation unit, 24...second acquisition unit, 25...second search unit, 26...third search unit, 27...second presentation unit, AI1...direct search answer information, AI2...indirect search answer information, 32...corresponding information database, 50...startup screen display, 51...presentation side icon, 52...startup speech bubble, 53...text information input unit, 54...execute button, 60...text information input screen display, 61...input text speech bubble, 62...category button, 63...other category button, 64...unknown button, 65...category information speech bubble, 70...category selection screen display, 71...input category speech bubble, 74...indirect search result button group, 75...direct search Result button group, 76...other search result button 76, 77...search result bubble, 80...screen display when search result is selected, 81...answer information bubble, 82...answer content bubble, 20A...presentation device, 91...third acquisition unit, 92...chatbot log, 100...screen display when question is resolved, 101...question resolved button, 102...question unresolved button, 103...question resolved bubble, 104...answer end bubble, 110...screen display when first question is unresolved, 111...question unresolved bubble, 114...preliminary indirect search result button group, 115...preliminary direct search result button group, 116...other preliminary search result button, 117...multiple preliminary search result bubbles, 118...bubble of input preliminary search result, screen display when second question is unresolved 120

Claims

1. A first acquisition unit that acquires a question from a user; a first search unit that searches a plurality of categories of the question acquired by the first acquisition unit using the question acquired by the first acquisition unit as a search condition based on correspondence information in which a question, an answer to the question, and a category of the question are associated with each other; a first presentation unit that presents the searched categories; a second acquisition unit that acquires a category selected by the user from among a plurality of presented categories; a second search unit that searches for an answer from the correspondence information using the question acquired by the first acquisition unit as a search condition; a third search unit that searches for an answer from the corresponding information using the question acquired by the first acquisition unit and the category acquired by the second acquisition unit as search conditions; a second presentation unit that presents to the user the answer searched for by the second search unit and the answer searched for by the third search unit; A presentation device comprising:

2. The second search unit and the third search unit use different search algorithms to search for an answer from the corresponding information. The presentation device according to claim 1 .

3. The second search unit searches for an answer from the corresponding information based on a probabilistic similarity between the question acquired by the first acquisition unit and each of a plurality of questions present in the corresponding information. The presentation device according to claim 1 or 2.

4. The third search unit searches for an answer from the correspondence information based on a cosine similarity between the vector of the question acquired by the first acquisition unit and vectors of each of a plurality of questions present in the correspondence information. The presentation device according to claim 1 or 2.

5. the second search unit and the third search unit each search for a plurality of answers from the correspondence information; The second presentation unit presents a plurality of results searched by the second search unit and a plurality of results searched by the third search unit separately and independently. The presentation device according to claim 1 or 2.

6. A third acquisition unit that acquires information regarding the correctness of the information presented on the second presentation unit from the user, The first search unit, the second search unit, and the third search unit each learn based on the information acquired by the third acquisition unit. The presentation device according to claim 1 or 2.

7. On the computer, A first acquisition step of acquiring a question from a user; a first search step of searching a plurality of categories of the question acquired in the first acquisition step using the question acquired in the first acquisition step as a search condition based on correspondence information in which a question, an answer to the question, and a category of the question are associated with each other; a first presentation step of presenting the searched categories; a second acquisition step of acquiring a category selected by the user from among the plurality of categories presented; a second search step of searching for an answer from the corresponding information using the question acquired in the first acquisition step as a search condition; a third search step of searching for an answer from the corresponding information using the question acquired in the first acquisition step and the category acquired in the second acquisition step as search conditions; a second presentation step of presenting to the user the answer found by the second search step and the answer found by the third search step; A program that executes the following.

Images