Information processing device, information processing method, and information processing program

Abstract

To increase convenience in performing maintenance of a knowledge database.SOLUTION: An information processing apparatus 100 performs maintenance of a knowledge database 132 managing questions and answers and tag information in association with each other, and comprises: a use history database 134 that manages use history information indicating information on a user's use history of the knowledge database 132; a processing unit 120 that generates, based on the use history information, candidate information indicating candidates for addition or update of at least either one of the questions, answers, or tag information in the knowledge database 132; an output unit (maintenance candidate presentation unit) 150 that outputs the generated candidate information; an input unit (maintenance candidate correction unit) 160 that receives, from a manager, input of correction on the output candidate information; and a correction database 133 that manages correction information indicating the correction on the candidate information input from the manager.SELECTED DRAWING: Figure 4

Description

[Technical field]

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

[0002] In recent years, an increasing number of companies are introducing chatbots to deal with labor shortages and reduce costs. Chatbots simulate human dialogue using a messenger or chat-like interface. For example, when a user, such as a company's customer or employee, inputs a question to a chatbot, the chatbot automatically outputs an answer to the question.

[0003] To realize automatic answers by a chatbot, not only is it necessary to create a knowledge database that associates questions and answers, but it is also necessary to perform appropriate maintenance such as registering new questions and answers in the knowledge database and updating existing questions and answers. For example, Patent Document 1 discloses that a computer system that automatically responds to questions from a terminal connected to a network includes a storage that stores document data and registration data, and executes a step of extracting candidate data to be registered in the storage from the document data, a step of classifying and scoring the registration candidate data into one of the registration data, and a step of displaying the scoring result. In this way, Patent Document 1 automatically provides support for determining whether to add a new FAQ to an existing FAQ (Frequently Asked Questions) where additional question items are expected. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] JP 2019-36210 A Summary of the Invention [Problem to be solved by the invention]

[0005] However, Patent Document 1 does not mention, for example, a case in which the new FAQ is inappropriate, and is insufficient in convenience for an administrator who maintains the knowledge database.

[0006] An object of the present disclosure is to provide a technique for improving convenience for an administrator who maintains a knowledge database. [Means for solving the problem]

[0007] An information processing device according to one aspect of the present disclosure is an information processing device that maintains a knowledge database that manages questions, answers, and tag information in association with each other, and is equipped with a processing unit that manages usage history information indicating information related to a user's usage history of the knowledge database, and generates candidate information indicating candidates for adding or updating the questions or the tag information in the knowledge database based on the usage history information, an output unit that outputs the generated candidate information, an input unit that receives corrections to the output candidate information from an administrator, and a correction database that manages correction information indicating corrections to the candidate information input by the administrator.

[0008] An information processing method according to one aspect of the present disclosure is an information processing method for maintaining a knowledge database that manages questions, answers, and tag information in association with each other, and generates candidate information indicating candidates for adding or updating the questions or tag information in the knowledge database based on usage history information indicating information regarding a user's usage history of the knowledge database managed in the usage history database, outputs the generated candidate information, and when an administrator inputs a correction to the output candidate information, manages correction information indicating the correction to the candidate information input by the administrator in a correction database.

[0009] An information processing program according to one aspect of the present disclosure causes a computer that controls an information processing device that maintains a knowledge database that manages questions, answers, and tag information in association with each other to execute the following process: generate candidate information indicating candidates for adding or updating at least one of the questions, the answers, or the tag information in the knowledge database based on usage history information indicating information related to a user's usage history of the knowledge database managed in the usage history database; output the generated candidate information; and, when a correction to the output candidate information is input by an administrator, manage the correction information indicating the correction to the candidate information input by the administrator in a correction database.

[0010] Furthermore, these comprehensive or specific aspects may be realized by a system, an apparatus, a method, an integrated circuit, a computer program, or a recording medium, or may be realized by any combination of a system, an apparatus, a method, an integrated circuit, a computer program, and a recording medium. Effect of the Invention

[0011] According to the present disclosure, it is possible to improve convenience when maintaining a knowledge database. [Brief description of the drawings]

[0012] [Figure 1] Figure 1 shows the processing operation screen for adding a new QA, where Figure 1(a) is the entrance screen, Figure 1(b) is the QA maintenance screen, Figure 1(c) is the list screen of newly added QA candidates, and Figure 1(d) is the screen for adding a new QA. [Diagram 2] Figure 2 shows the processing operation screen for updating existing QAs, where Figure 2(a) is the entrance screen, Figure 2(b) is the QA maintenance screen, Figure 2(c) is the list of update QA candidates screen, and Figure 2(d) is the QA update screen. [Diagram 3] Figure 3 shows the screens for processing operations to add and update tag information, where Figure 3(a) is the QA maintenance screen, Figure 3(b) is the tag candidate list screen, Figure 3(c) is the manual correction screen, and Figure 3(d) is the correction result screen. [Figure 4] FIG. 4 is a block diagram of an embodiment of an information processing device according to the present disclosure. [Diagram 5] FIG. 5 is an example of a knowledge database. [Figure 6] FIG. 6 is an example of a correction database. [Figure 7] FIG. 7 is a conceptual diagram conceptually showing the procedure of the maintenance process of the knowledge database executed by the information processing device. [Figure 8] FIG. 8 is a flowchart including a procedure for maintaining a knowledge database by analyzing usage history. [Figure 9] FIG. 9 is a flowchart including a procedure for maintaining the knowledge database by correcting tag candidates. [Figure 10] Figure 10 is a flowchart including the steps for training an AI to make administrator corrections to tag addition candidates. [Figure 11] FIG. 11 is a block diagram showing an example of a hardware configuration of an information processing device according to the present disclosure. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0013] Hereinafter, the embodiments of the present disclosure will be described in detail with reference to the drawings as appropriate. However, more detailed description than necessary may be omitted. For example, detailed description of already well-known matters and duplicated description of substantially the same configuration may be omitted. This is to avoid the following description becoming unnecessarily redundant and to facilitate understanding by those skilled in the art. Note that the accompanying drawings and the following description are provided to enable those skilled in the art to fully understand the present disclosure, and are not intended to limit the subject matter described in the claims.

[0014] (Present embodiment) The information processing device of this embodiment performs maintenance of an automatic answering system such as a chatbot. To realize automatic answers by a chatbot, it is necessary not only to create a knowledge database that associates questions and answers (QA), but also to perform appropriate maintenance such as registering new questions and answers in the knowledge database and updating existing questions and answers. The information processing device of the present disclosure has been developed to achieve appropriate maintenance.

[0015] Maintenance of the knowledge database includes, for example, processes such as adding or updating questions, and adding or updating tag information. The information processing device can generate candidate information that is a candidate for performing such maintenance processes based on usage history information of users who have used the knowledge database. In particular, the information processing device of this embodiment accepts corrections to such candidate information from the device administrator and manages the corrections as correction information, thereby improving the convenience of maintenance.

[0016] Each of Fig. 1 to Fig. 3 shows a display screen displayed on an administrator terminal of an administrator who manages a knowledge database by operating an information processing device of this embodiment. The administrator uses the information processing device of this embodiment for maintenance of the knowledge database, and each of the display screens of Fig. 1 to Fig. 3 is a screen displayed corresponding to a different maintenance process.

[0017] FIG. 1 shows a display screen presented to an administrator when an information processing device performs a usage history analysis based on the usage history information of a knowledge database by a user and adds a new QA. Usage history analysis includes, for example, trend analysis and satisfaction evaluation analysis. Trend analysis is a method of analyzing trends in data with emphasis on changes in the time series of data, i.e., changes in the time series of data. Trend analysis in this example is used to obtain more appropriate question, answer, and tag information as a result of analyzing questions, answers, and tag information present in the usage history information. Satisfaction evaluation analysis is a method of deriving more appropriate data based on the satisfaction level obtained from the user. Satisfaction evaluation in this example is used to obtain more appropriate question, answer, and tag information as a result of analyzing the user's satisfaction level from the usage history information.

[0018] FIG. 1(a) shows the entrance screen that the information processing device first displays when an administrator accesses the device, and displays two button items, "QA Data" and "QA Maintenance," under the title "QA Database." "QA Data" is a button for accessing individual QAs. "QA Maintenance" is a button that the administrator operates when performing maintenance on the knowledge database, and by operating this button, the maintenance function of the information processing device of this embodiment can be used.

[0019] Figure 1(b) shows the QA maintenance screen that is displayed after the administrator operates "QA maintenance" in Figure 1(a). The QA maintenance screen displays three buttons: "Newly added QA candidate list," "Updated QA candidate list," and "Tag candidate list." "Newly added QA candidate list" is a button that accesses a list of new QA candidates that should be added as a result of trend analysis. "Updated QA candidate list" is a button that accesses a list of QA candidates that should be updated by performing trend analysis on the user's usage history of the knowledge database. Details of how to operate "Updated QA candidate list" will be described later using Figure 2. "Tag candidate list" is a button that accesses a list of candidates for tag information that should be added by performing trend analysis on the user's usage history of the knowledge database. Details of how to operate "Tag candidate list" will be described later using Figure 3.

[0020] FIG. 1(c) shows a newly added QA candidate list screen that is displayed after the administrator operates the "Newly added QA candidate list" in FIG. 1(b), and shows a list of candidates for QAs to be added under the title "Newly added QA candidate". In this example, questions such as "I want to keep a dog that doesn't bark" and "Do I need to take a walk every day?" are listed as additional candidate information. These additional candidate information are automatically created and displayed by the information processing device based on the user's usage history information. For example, if many users enter search keywords such as "dog that doesn't bark" in the chatbot's search field within a short period of time, the information processing device creates and displays the question "I want to keep a dog that doesn't bark" that includes these keywords.

[0021] Figure 1(d) shows the screen for adding a new Q&A that appears after the administrator operates "I want to have a dog that doesn't bark" in Figure 1(c). The screen for adding a new Q&A shows two items: "Question candidates," which are candidates for new questions to be added, and "Reasons for being selected as a candidate," which are the reasons why the question was selected as a candidate.

[0022] In this example, the word "doesn't bark" does not exist in the QA of the knowledge database among the search keywords "dog that doesn't bark," which are the user's usage history information. The question "I want to keep a dog that doesn't bark" in Figure 1(c) was automatically created by the information processing device based on the search keywords. First, the display screen displays "I want to keep a dog that doesn't bark" as a question candidate, and displays the reason for its selection as a candidate, "There is no question that corresponds to the trending word 'doesn't bark'." This reason is the result of trend analysis that reflects the time-series changes in trending word data.

[0023] In this example, the administrator determines that the displayed question, "I want to keep a dog that doesn't bark," is not an appropriate question candidate, and operates his own terminal to manually correct "I want to keep a dog that doesn't bark" to "Are there any dogs that don't bark?" The administrator wishes to add this corrected question, so he presses the displayed "Add" button to add the corrected question to the knowledge database.

[0024] Note that Figure 1 shows an example of listing candidate questions in a Q&A and adding a new question, but it is also possible to list candidate answers in a Q&A and add a new answer.

[0025] Figure 2 shows the display screen presented to the administrator when the information processing device performs a usage history analysis based on the usage history information of the knowledge database by the user and updates the existing QA. Figure 2(a) is the entrance screen that the information processing device first displays when the administrator accesses it, and is the same entrance screen as Figure 1(a). Figure 2(b) is the QA maintenance screen that is displayed after the administrator operates "QA maintenance" in Figure 2(a), and is the same QA maintenance screen as Figure 1(b).

[0026] FIG. 2(c) is an update QA candidate list screen that is displayed after the administrator operates the "update QA candidate list" in FIG. 2(b). The update QA candidate list screen shows a list of candidate QAs to be updated under the title "update QA candidate list." "Update" here includes correction and deletion. In this example, questions such as "What should I pay attention to when keeping a pet for the first time?", "What is grooming?", "I want to keep a dog that doesn't bark," and "Do I need to take it for a walk every day?" are listed as update candidate information. This update candidate information is automatically created and displayed by the information processing device based on the user's usage history information.

[0027] Figure 2(d) is the Q&A update screen that is displayed after the administrator operates "What should I pay attention to when keeping a pet for the first time?" in Figure 2(c). The Q&A update screen displays two items: the "Answer" and the "Reason for being nominated," which is the reason why the answer was selected as a candidate. The answer here, "Prepare the appropriate environment for keeping the pet. Necessary items are..." is an answer that exists in the knowledge database. However, an analysis of user satisfaction evaluations shows that this answer has not received a satisfactory rating. Therefore, the reason why the answer was selected as a candidate is, "I was not satisfied with the answer. Please consider revising it."

[0028] If the administrator wishes to correct this answer, they can use their own device to correct the answer to "Prepare the breeding environment. Necessary items are..." and press the "Update" button to update the answer in the knowledge database.

[0029] Note that while Figure 2 shows an example of listing candidate questions in the Q&A and updating existing answers, it is also possible to list candidate questions in the Q&A and update existing questions.

[0030] Fig. 3 shows a display screen presented to an administrator when an information processing device performs a usage history analysis based on the usage history information of a knowledge database by a user and adds and updates tag information. The entrance screens corresponding to Figs. 1(a) and 2(a) are omitted from the illustration. Fig. 3(a) shows a QA maintenance screen that is displayed after the administrator operates "QA maintenance" on the omitted entrance screen, and is the same QA maintenance screen as Figs. 1(b) and 2(b).

[0031] FIG. 3(b) is a tag candidate list screen that is displayed after the administrator operates the "tag candidate list" in FIG. 3(a). The tag candidate list screen shows a list of combinations of questions and tags, which are tag information related to the questions, under the title "Question Tag". This list is a list of combinations for which tags should be added or for which tags should be updated. In this example, after the information processing device automatically displays the combination of the question "Indoors..." and the tag "Dog", the administrator operates his / her own terminal to add the tag "Indoors". Also, after the information processing device displays the combination of the question "Pet..." and the tag "Hotel", the administrator operates his / her own terminal to add the tag "Dog".

[0032] Figure 3(c) shows the manual correction screen that is displayed after the administrator operates the combination of the question "Pets..." and the tag "Dog, Hotel" in Figure 3(b). The manual correction screen displays the two items "Question" and "Tags" selected in Figure 3(b). In this example, after the question "What kind of dogs can stay at pet hotels?" selected in Figure 3(b) and the tag "Dog, Hotel" are displayed, the administrator operates his / her own terminal to update the tag "Dog, Hotel" to the tag "Dog, Pet Hotel". If the administrator wishes to make a correction using this candidate, he / she can press the "Correct" button to correct and update the tag in the knowledge database.

[0033] Fig. 3(d) is a correction result screen that is displayed after the administrator updates the tag in Fig. 3(c). The correction result screen shows a list of combinations of questions and updated tags, which are the updated tag information, under the title "Question Tag". In this example, the information processing device displays the combination of the question "Indoors..." and the tag "Dog", as well as the question "Pet..." and the updated tag combination of "Dog, Pet Hotel". The updated tag "Dog, Pet Hotel" will also be displayed in the list of candidate tag information when the question "Pet..." is asked in the future.

[0034] 4 is a block diagram of an embodiment of an information processing device. The information processing device 100 includes a QA creation unit 110, a processing unit 120, a dictionary database 131, a knowledge database 132, a correction database 133, a usage history database 134, a maintenance candidate presentation unit 150, a maintenance candidate correction unit 160, a learning unit 170, a question reception unit 141, a response display unit 142, a satisfaction evaluation reception unit 143, and a QA search unit 180. The information processing device 100 is incorporated in, for example, a server device, and is connected to each of the administrator terminal 10, the user terminal 20, and the developer terminal 30 via a network such as the Internet.

[0035] The processes performed by the QA creation unit 110, the processing unit 120, the maintenance candidate presentation unit 150, the maintenance candidate correction unit 160, the learning unit 170, the question reception unit 141, the response display unit 142, the satisfaction evaluation reception unit 143, and the QA search unit 180 may be realized by the processor 1001 shown in FIG. 11 operating in cooperation with the memory 1002, the storage 1003, and / or the communication device 1006. Therefore, the processes described mainly with these blocks can be interpreted as processes mainly with the processor 1001. The dictionary database 131, the knowledge database 132, the correction database 133, and the usage history database 134 may be configured in the storage 1003 shown in FIG. 11. The information processing device may receive (accept) input information from each of the administrator terminal 10, the user terminal 20, and the developer terminal 30 through the communication device 1006 and the network shown in FIG. 11. The information processing device may transmit (output) output information to each of the administrator terminal 10, the user terminal 20, and the developer terminal 30 via the communication device 1006 and the network shown in FIG.

[0036] FIG. 5 shows an example of a knowledge database 132 that manages questions, answers, and tag information in association with each other, and the database is shown in a matrix format. One QA, which is one row of the matrix, is composed of a question, an answer, a tag (tag information), and other additional information, which are associated with each other. As shown in this figure, it is possible to assign multiple tags to one QA. It is also possible to assign the same tag to multiple QAs, such as tag 1. n QAs make up the knowledge database of one information processing device 100.

[0037] 6 is an example of a correction database that manages correction information indicating corrections to candidate information input from the administrator's administrator terminal 10, and shows the database in a matrix format. One correction item, which is one row of the matrix, is composed of a word before correction, its type, the correction type, and a word after correction, which are related to each other.

[0038] FIG. 7 conceptually illustrates an information processing method implemented by the information processing device 100, that is, a maintenance process for the knowledge database 132, and in particular the steps (1) to (6) of adding and updating tag information.

[0039] (1) Assume that the knowledge database 132 stores, for example, the question shown in the figure, “What is the difference between the new coronavirus and influenza?” and an answer to that question, that is, a QA that is a set of question and answer.

[0040] (2) The automatic tag generation unit 122 automatically generates additional tag candidates, which are candidate information, in response to the question in (1). In this example, the tags "COVID-19", "Influenza", and "Difference" are additionally generated in response to the question in (1). The maintenance candidate presentation unit 150 outputs the tag candidates generated by the automatic tag generation unit 122 to the administrator terminal 10, and the screen in FIG. 3(b) is displayed on the administrator terminal 10.

[0041] (3) The administrator manually operates the administrator terminal 10 to correct the tag that was added and generated in (2). In this example, "novel coronavirus" is updated to "COVID-19" and "difference" is deleted. The maintenance candidate correction unit 160 accepts the corrections entered from the administrator terminal 10 of the administrator, sends the correction information to the knowledge database 132, and updates the QA in the knowledge database 132. Furthermore, the maintenance candidate correction unit 160 also sends the correction information to the correction database 133 as shown in FIG. 7, and the correction database 133 manages the correction information.

[0042] (4) After the process of (3), the automatic tag generation unit 122 generates additional tags "COVID-19," "Influenza," and "Difference" in response to the same question as in (1) above. Although the question is the same as in (1), at this point the correction made by the administrator in (3) is not reflected. In the conventional maintenance method, this is output as is to the administrator terminal 10, so that the administrator must make the correction in (3) again, which increases the maintenance burden on the administrator.

[0043] (5) Meanwhile, in the information processing device 100 of this embodiment, the usage history analysis unit 121, which is further responsible for analyzing the usage history, updates the tag candidates, which are candidate information, based on the usage history information in the usage history database 134 and the correction information in the correction database 133. In particular, the usage history analysis unit 121 refers to the correction information in the correction database 133 registered in (3) and reflects it in the tag update. In this example, as in (3), for the tag automatically added in (4), "novel coronavirus" is updated to "COVID-19" and "difference" is deleted. The maintenance candidate presentation unit 150 outputs the tag candidates thus updated to the administrator terminal 10, and the screen of FIG. 3(b) is displayed on the administrator terminal 10.

[0044] (6) The usage history analysis unit 121 also reflects the processing of (5) on other questions. For example, in the case of a question similar to (1), "Are masks necessary for COVID-19?", without processing of (5), the automatic tag generation unit 122 is considered to additionally generate tags "COVID-19" and "mask." In this embodiment, the usage history analysis unit 121 takes into consideration the update from "COVID-19" to "COVID-19" in (5), and updates (corrects) the tag "COVID-19" additionally generated by the automatic tag generation unit 122 for that question to "COVID-19." This correction may be displayed on the administrator terminal 10 to request confirmation from the administrator, or may be automatically corrected without confirmation.

[0045] As described above, the information processing device 100 according to the present embodiment can present candidates for questions, answers, and tag information to be newly added or updated to the administrator based on the usage history analysis. For example, the information processing device 100 presents candidates for questions, answers, and tag information to be newly added or updated to the administrator based on rapidly increasing words reflecting trend analysis. This allows the administrator to add or update questions, answers, and tag information after checking the candidates that are automatically output, thereby making it possible to reduce the burden of maintenance. Note that rapidly increasing words are words that are frequently asked by users during a certain period of time. For example, a word that is searched for in a certain period of time twice or more times the average number of words that have appeared in the past period of time is considered to be a rapidly increasing word.

[0046] Furthermore, the information processing device 100 of this embodiment records corrections (additions, updates) made by the administrator to the presented candidates in the correction database 133. Therefore, when proposing new tag candidates for a Q&A related to a question similar to the previous question received later from a user, the correction database 133 can be referenced and the previously made corrections can be reflected and presented. This makes it possible to easily reflect the previous corrections in Q&A related to new questions that may arise in the future, and makes it possible to suppress an increase in the maintenance burden on the administrator.

[0047] As described above, the information processing device 100 according to the present embodiment can improve the convenience when maintaining the knowledge database 132.

[0048] 8 is a flowchart including procedures for performing maintenance of the knowledge database 132 by analyzing usage history in an information processing method implemented by a computer 1000 (see FIG. 11) such as a processor 1001 (see FIG. 11) that manages control processing of the information processing device 100 of this embodiment. An information processing program that causes the computer 1000 to execute each procedure is stored in the memory 1002 or storage 1003 (see FIG. 11) of the information processing device 100.

[0049] An administrator operates the administrator terminal 10, and in response to the operation, the QA creating unit 110 creates a QA, which is a set of a question and an answer, and registers the QA in advance in the knowledge database 132 (step S11).

[0050] A user operates the user terminal 20, and the question receiving unit 141 receives the question and transmits it to the QA search unit 180. In response to the user's question from the question receiving unit 141, the QA search unit 180 searches for an answer by referring to the knowledge database 132, transmits the searched answer to the response display unit 142, and registers the exchange of the question and the answer in the usage history database 134. As a result, the usage history of the information processing device 100 is accumulated in the usage history database 134 (step S12).

[0051] The usage history analysis unit 121 of the processing unit 120 acquires QA by referring to the knowledge database 132, and acquires usage history by referring to the usage history database 134 (step S13). Furthermore, the usage history analysis unit 121 creates candidate information for questions and answers to be used for maintenance based on trend analysis and satisfaction evaluation analysis (step S14). The trend analysis and satisfaction evaluation analysis are specific examples of usage history analysis, where the trend analysis is performed based on the usage history information in the usage history database 134, and the satisfaction evaluation analysis is performed based on the user's satisfaction evaluation received by the satisfaction evaluation receiving unit 143.

[0052] The usage history analysis unit 121 determines whether the same QA already exists (step S15). If the same QA does not already exist (step S15: NO), the maintenance candidate presentation unit 150 presents candidate information to the manager terminal 10 (step S16). If the same QA exists (step S15: YES), the process ends.

[0053] When the maintenance candidate correction unit 160 receives a correction operation by the administrator from the administrator terminal 10 in response to the presentation of the candidate information on the administrator terminal 10 (step S17: YES), the maintenance candidate correction unit 160 judges whether the administrator corrected or rejected the candidate information (step S18). If the correction was rejected, the process ends (step S19: YES). If the correction was not rejected (step S19: NO), the maintenance candidate correction unit 160 reflects the correction in the knowledge database 132 and registers the correction information of the correction in the correction database 133 (step S20).

[0054] After creating the candidate information in step S14, the maintenance candidate presentation unit 150 presents the candidate information to the administrator terminal 10 in step S16, and requests the administrator's confirmation. However, for example, the administrator can operate the administrator terminal 10 to set the information processing device 100 to omit confirmation of the candidate information. In such a setting, the usage history analysis unit 121 adds a question or answer to the knowledge database 132 or changes the question or answer based on the candidate information. This makes it possible to add a question or answer to the knowledge database 132 or change the question or answer without the administrator having to confirm the candidate information, thereby saving the administrator the trouble of confirmation.

[0055] Furthermore, the usage history analysis unit 121 does not necessarily need to generate candidate information every time the usage history is acquired in step S13. That is, the usage history analysis unit 121 may execute a process of generating candidate information periodically or irregularly. When candidate information is generated in the generation process, the maintenance candidate presentation unit 150 may notify the administrator of information indicating that the candidate information has been generated via the administrator terminal 10. This allows the administrator to check the candidate information only when notified, which reduces unnecessary checking work by the administrator and prevents omissions in checking.

[0056] 9 is a flowchart including procedures for performing maintenance of the knowledge database 132 by correcting tag candidates, which are candidate information, in an information processing method implemented by a computer 1000 (see FIG. 11) such as a processor 1001 (see FIG. 11) that manages control processing of the information processing device 100 of the present disclosure. An information processing program that causes the computer 1000 to execute each procedure is stored in the memory 1002 or storage 1003 (see FIG. 11) of the information processing device 100.

[0057] An administrator operates the administrator terminal 10, and in response to the operation, the QA creating unit 110 creates a QA, which is a set of a question and an answer, and registers the QA in advance in the knowledge database 132 (step S31).

[0058] The automatic tag generation unit 122 of the processing unit 120 acquires the QA by referring to the knowledge database 132 (step S32). Next, the automatic tag generation unit 122 acquires synonyms or homonyms by referring to the dictionary database 131 (step S33). Furthermore, the automatic tag generation unit 122 generates tag information, which is candidate information, based on the QA acquired from the knowledge database 132 and the synonyms or homonyms acquired from the dictionary database 131 (step S34). The automatic tag generation unit 122 generates candidates for adding or updating tag information based on the dictionary database 131, making it possible to add or update technical terms, compound words, and the like that are difficult to add or update.

[0059] The automatic tag generation unit 122 determines whether an identical tag already exists (step S35). If the identical tag does not already exist (step S35: NO), the maintenance candidate presentation unit 150 presents additional tag candidates to the manager terminal 10 (step S36). If the identical tag exists (step S35: YES), the process ends.

[0060] When the maintenance candidate correction unit 160 receives a correction operation by the administrator from the administrator terminal 10 in response to the presentation of the tag candidate in the administrator terminal 10 (step S37: YES), the maintenance candidate correction unit 160 judges whether the administrator corrected or rejected the candidate information (step S38). If the correction is rejected, the process ends (step S39: YES). If the correction is not rejected (step S39: NO), the maintenance candidate correction unit 160 reflects the correction in the knowledge database 132 and registers the correction information of the correction in the correction database 133 (step S40).

[0061] 10 is a flowchart including a procedure for making an AI learn corrections made by an administrator to tag addition candidates (in the case of automatic tag addition correction) in the information processing method executed by the information processing device 100 of the present disclosure. This procedure is a version in which, after the procedure in FIG. 9, the automatic tag generation unit 122 generates tag information that is the candidate information in steps S32 to S34 based on learning by the learning unit 170.

[0062] The maintenance candidate correction unit 160 detects that the manager has corrected the candidate (step S51). If the correction is not to be registered (step S52: NO), the process ends. If the correction is to be registered (step S52: YES), the learning unit 170 performs AI learning on the content of the correction (step S53), and the maintenance candidate correction unit 160 reflects the correction in the knowledge database 132 (step S54). The automatic tag generation unit 122 of the processing unit 120 reflects the learning result (step S55).

[0063] In this procedure, in step S53, the learning unit 170 learns the pattern of the relationship between the question and tag information based on the knowledge database 132. Based on this learning, the automatic tag generation unit 122 generates candidates for adding or updating tag information based on the learned pattern. Since the automatic tag generation unit 122 generates candidate information based on the pattern learned by the learning unit 170, it becomes possible to provide more appropriate tag information the next time candidate information is presented.

[0064] After the procedure of Fig. 8, the usage history analysis unit 121 may create candidate information for the questions and answers of steps S13 and S14 based on learning by the learning unit 170, using a procedure similar to that of Fig. 10. That is, the learning unit 170 learns the pattern of corrections to the candidate information based on the correction information managed in the correction database 133. In response to this learning, the usage history analysis unit 121 generates candidate information based on the learned pattern and the usage history information. Since the usage history analysis unit 121 generates candidate information based on the pattern learned by the learning unit 170, it becomes possible to provide more appropriate candidate information the next time candidate information is presented.

[0065] Furthermore, the processing unit 120 may record, as information for the developer, the candidate information that the administrator did not adopt for the addition or change of the knowledge database 132. The candidate information that was not adopted for the addition or change is stored in the storage 1003 (see FIG. 11). The developer can operate the developer terminal 30 to check the candidate information that was not adopted for the addition or change, analyze whether the maintenance of the knowledge database 132 is being performed appropriately, and consider improvements so that more appropriate candidates are presented.

[0066] Furthermore, the maintenance candidate presentation unit 150 functioning as an output unit may output usage history information related to the generation of the candidate information together with the candidate information when outputting the candidate information to the manager terminal 10. This allows the manager to check the actual usage history and manage the validity of the candidate information.

[0067] Furthermore, when the maintenance candidate presentation unit 150 outputs candidate information indicating additional candidates for an answer to a question to the administrator terminal 10, the maintenance candidate presentation unit 150 also outputs the results of searching the question using a specified search service. The results include the URL of the search results and an API (Application Programming Interface) provided by the search service. This enables the administrator to determine the validity of the additional candidates for an answer to the question.

[0068] <Hardware configuration> FIG. 11 is a block diagram showing an example of a hardware configuration of an information processing device 100 according to the present disclosure.

[0069] The information processing device 100 may be configured as a computer 1000 shown in Fig. 11. In this case, the processing of the blocks included in the information processing device 100 described above may be realized by the computer 1000 executing a computer program.

[0070] The computer 1000 includes a processor 1001, a memory 1002, a storage 1003, an input device 1004, an output device 1005, a communication device 1006, a graphics processing unit (GPU) 1007, a reading device 1008, and a bus 1009. The processor 1001, the memory 1002, the storage 1003, the input device 1004, the output device 1005, the communication device 1006, the GPU 1007, and the reading device 1008 are connected to the bus 1009, and can transmit and receive data in both directions via the bus 1009.

[0071] The processor 1001 is a device that executes a computer program stored in the memory 1002 or the storage 1003 and realizes the processing of the above-mentioned blocks. Examples of the processor 1001 include a CPU (Central Processing Unit), an MPU (Micro Processing Unit), a controller, an LSI (large scale integration), an ASIC (Application Specific Integrated Circuit), a PLD (Programmable Logic Device), and an FPGA (Field-Programmable Gate Array).

[0072] The memory 1002 is a device that stores computer programs and data that are operated by the computer 1000. The memory 1002 may include a read-only memory (ROM) and a random access memory (RAM).

[0073] The storage 1003 is configured as a non-volatile storage medium, and is a device that stores computer programs and data handled by the computer 1000. Examples of the storage 1003 include a hard disk drive (HDD), a solid state drive (SSD), and a flash memory.

[0074] The input device 1004 is a device that accepts data to be input to the processor 1001. Examples of the input device 1004 include a keyboard, a mouse, a touch pad, and a microphone.

[0075] The output device 1005 is a device that outputs data generated by the processor 1001. Examples of the output device 1005 include a display and a speaker.

[0076] The communication device 1006 is a device that transmits and receives data to and from other devices, such as a server (not shown) or terminals 10, 20, and 30, via a communication network. The communication device 1006 may include a transmitting unit that transmits data and a receiving unit that receives data. The communication device 1006 may support either wired communication or wireless communication. An example of wired communication is Ethernet (registered trademark). Examples of wireless communication are IEEE802.11, Bluetooth, LTE (Long Term Evolution), 4G, and 5G.

[0077] The GPU 1007 is a device that processes image rendering at high speed. The GPU 1007 may be used for AI processing. For example, the GPU 1007 may be used for learning processing of a learning model and / or generating processing of candidate information using a learning model.

[0078] The reader 1008 is a device that reads data from a recording medium such as a digital versatile disk read only memory (DVD-ROM) or a universal serial bus (USB) memory.

[0079] <Summary of this disclosure> As described above, the information processing device 100, which maintains the knowledge database 132 that manages questions, answers, and tag information in association with each other, includes a usage history database 134 that manages usage history information indicating information related to a user's usage history of the knowledge database 132, a processing unit 120 that generates candidate information indicating candidates for adding or updating at least one of the questions, answers, or tag information in the knowledge database 132 based on the usage history information, an output unit (maintenance candidate presentation unit) 150 that outputs the generated candidate information, an input unit (maintenance candidate correction unit) 160 where corrections to the outputted candidate information are input by an administrator, and a correction database 133 that manages correction information indicating corrections to the candidate information input by the administrator.

[0080] This allows the administrator to correct the candidate information indicating the candidates for adding or updating questions or tag information, and to manage the correction information indicating the correction, thereby improving the convenience of maintaining the knowledge database 132.

[0081] The information processing device 100 further includes a learning unit 170 that learns correction patterns of candidate information based on the correction information managed in the correction database 133, and the processing unit 120 generates candidate information based on the learned patterns and usage history information.

[0082] As a result, the processing unit 120 generates candidate information based on the pattern learned by the learning unit 170, and therefore, the next time candidate information is presented, it becomes possible to provide more appropriate candidate information.

[0083] When the setting is such that confirmation of candidate information by the administrator is omitted, the processing unit 120 adds or modifies a question or answer in the knowledge database 132 based on the candidate information.

[0084] This makes it possible to add a question or answer to or change a question or answer in the knowledge database 132 without the administrator having to check the candidate information, thereby saving the administrator the trouble of checking.

[0085] The processing unit 120 executes a process for generating candidate information on a regular or irregular basis, and when candidate information is generated in the generation process, notifies the administrator of information indicating that the candidate information has been generated.

[0086] This allows the administrator to check the candidate information only when notified, thereby reducing unnecessary checking work by the administrator and preventing oversights in checking.

[0087] The processing unit 120 records the candidate information that the administrator did not adopt for the addition or modification of the knowledge database 132 as information for the developer.

[0088] This allows the developer to check the candidate information, analyze whether the knowledge database 132 is being properly maintained, and consider improvements so that more appropriate candidates are presented.

[0089] The information processing device 100 further includes a learning unit 170 that learns a pattern of the relationship between the question and the tag information based on the knowledge database 132, and the processing unit 120 generates candidates for adding or updating the tag information based on the learned pattern.

[0090] As a result, the processing unit 120 generates candidate information based on the pattern learned by the learning unit 170, and therefore, the next time candidate information is presented, it becomes possible to provide more appropriate tag information.

[0091] The information processing device 100 further includes a dictionary database 131 that includes at least one of synonyms and homonyms, and the processing unit 120 generates candidates for adding or updating tag information based on the dictionary database 131.

[0092] As a result, the processing unit 120 generates candidates for adding or updating tag information based on the dictionary database 131 that includes at least one of synonyms and homonyms, making it possible to add or update technical terms, compound words, etc. that are difficult to add or update.

[0093] The output unit 150 outputs the candidate information together with the usage history information related to the generation of the candidate information.

[0094] This enables the administrator to check the actual usage history and manage the validity of the candidate information.

[0095] When outputting candidate information indicating additional candidates for the answer, the output unit 150 also outputs the result of searching the question using a predetermined search service.

[0096] This allows the administrator to determine the validity of additional candidate answers to the question.

[0097] An information processing method for maintaining a knowledge database 132 that manages questions, answers, and tag information in association with each other generates candidate information indicating candidates for adding or updating at least one of questions, answers, or tag information in the knowledge database based on usage history information indicating information related to a user's usage history of the knowledge database 132 managed in a usage history database 134, outputs the generated candidate information, and when an administrator inputs a correction to the output candidate information, manages in a correction database 133 the correction information indicating the correction to the candidate information input by the administrator.

[0098] This allows the administrator to correct the candidate information indicating the candidates for adding or updating questions or tag information, and to manage the correction information indicating the correction, thereby improving the convenience of maintaining the knowledge database 132.

[0099] The information processing program causes a computer that controls an information processing device that maintains a knowledge database 132 that manages questions, answers, and tag information in association with each other to execute the following process: generate candidate information indicating candidates for adding or updating at least one of questions, answers, or tag information in the knowledge database based on usage history information indicating information related to a user's usage history of the knowledge database 132 managed in a usage history database 134, output the generated candidate information, and when an administrator inputs a correction to the output candidate information, manage in a correction database 133 the correction information indicating the correction to the candidate information input by the administrator.

[0100] This allows the administrator to correct the candidate information indicating the candidates for adding or updating questions or tag information, and to manage the correction information indicating the correction, thereby improving the convenience of maintaining the knowledge database 132.

[0101] Although the embodiments have been described above with reference to the accompanying drawings, the present disclosure is not limited to such examples. It is clear that a person skilled in the art can conceive of various modifications, corrections, substitutions, additions, deletions, and equivalents within the scope of the claims, and it is understood that these also belong to the technical scope of the present disclosure. In addition, the components in the above-mentioned embodiments may be arbitrarily combined within the scope of the invention. [Industrial Applicability]

[0102] According to the present disclosure, corrections to candidate information indicating candidates for adding or updating tag information are input by an administrator, and the correction database manages the correction information indicating corrections to the input candidate information, thereby improving convenience when maintaining the knowledge database. [Explanation of symbols]

[0103] 10 Administrator terminal 20 User terminal 30 Developer Terminal 100 Information processing device 110 QA Creation Department 120 Processing section 121 Usage History Analysis Unit 122 Automatic tag generation section 131 Dictionary Database 132 Knowledge Database 133 Correction Database 134 Usage History Database 141 Question Reception Department 142 Response display unit 143 Satisfaction Evaluation Reception Department 150 Maintenance candidate presentation unit (output unit) 160 Maintenance candidate correction section (input section) 170 Learning Department 180 QA Search Department

Claims

1. An information processing device that maintains a knowledge database that associates and manages questions, answers, and tag information, comprising: a usage history database for managing usage history information indicating information related to a user's usage history of the knowledge database; a processing unit that generates candidate information indicating candidates for adding or updating at least one of the question, the answer, and the tag information in the knowledge database based on the usage history information; an output unit that outputs the generated candidate information; an input unit into which a correction to the output candidate information is input by an administrator; a correction database for managing correction information indicating corrections to the candidate information input by the administrator, the processing unit, when the administrator is set to omit confirmation of the candidate information, adds the question or the answer to the knowledge database or changes the question or the answer based on the candidate information. Information processing device.

2. An information processing device that maintains a knowledge database that associates and manages questions, answers, and tag information, comprising: a usage history database for managing usage history information indicating information related to a user's usage history of the knowledge database; a processing unit that generates candidate information indicating candidates for adding or updating at least one of the question, the answer, and the tag information in the knowledge database based on the usage history information; an output unit that outputs the generated candidate information; an input unit into which a correction to the output candidate information is input by an administrator; a correction database for managing correction information indicating corrections to the candidate information input by the administrator, the processing unit executes a process of generating the candidate information periodically or irregularly, and when the candidate information is generated in the generation process, notifies the administrator of information indicating that the candidate information has been generated. Information processing device.

3. An information processing device that maintains a knowledge database that associates and manages questions, answers, and tag information, comprising: a usage history database for managing usage history information indicating information related to a user's usage history of the knowledge database; a processing unit that generates candidate information indicating candidates for adding or updating at least one of the question, the answer, and the tag information in the knowledge database based on the usage history information; an output unit that outputs the generated candidate information; an input unit into which a correction to the output candidate information is input by an administrator; a correction database for managing correction information indicating corrections to the candidate information input by the administrator; a learning unit that learns a pattern of a relationship between the question and the tag information based on the knowledge database, The processing unit generates candidates for adding or updating the tag information based on the learned pattern. Information processing device.

4. An information processing device that maintains a knowledge database that associates and manages questions, answers, and tag information, comprising: a usage history database for managing usage history information indicating information related to a user's usage history of the knowledge database; a processing unit that generates candidate information indicating candidates for adding or updating at least one of the question, the answer, and the tag information in the knowledge database based on the usage history information; an output unit that outputs the generated candidate information; an input unit into which a correction to the output candidate information is input by an administrator; a correction database for managing correction information indicating corrections to the candidate information input by the administrator, When outputting candidate information indicating additional candidates for the answer, the output unit also outputs a search result for the question using a predetermined search service. Information processing device.

5. a learning unit that learns a correction pattern of the candidate information based on the correction information managed in the correction database, The processing unit generates the candidate information based on the learned pattern and the usage history information. The information processing device according to claim 1 .

6. the processing unit records the candidate information that the administrator did not adopt for the addition or modification of the knowledge database as information for developers. The information processing device according to claim 1 .

7. a dictionary database including at least one of synonyms and antonyms; The processing unit generates candidates for adding or updating the tag information based on the dictionary database. The information processing device according to claim 1 .

8. The output unit outputs, together with the candidate information, the usage history information related to generation of the candidate information. The information processing device according to claim 1 .

9. An information processing method in which an information processing device having a processor and a memory maintains a knowledge database in which questions, answers, and tag information are associated with each other and managed, comprising: The processor in cooperation with the memory: generating candidate information indicating candidates for addition or update of at least one of the question, the answer, or the tag information in the knowledge database based on usage history information indicating information regarding the user's usage history of the knowledge database managed in a usage history database; Outputting the generated candidate information; when a correction to the output candidate information is input from an administrator, correction information indicating the correction to the candidate information input from the administrator is managed in a correction database; If the setting is such that confirmation of the candidate information by the administrator is omitted, adding the question or the answer to the knowledge database or modifying the question or the answer based on the candidate information. Information processing methods.

10. An information processing method in which an information processing device having a processor and a memory maintains a knowledge database in which questions, answers, and tag information are associated with each other and managed, comprising: The processor in cooperation with the memory: generating candidate information indicating candidates for addition or update of at least one of the question, the answer, or the tag information in the knowledge database based on usage history information indicating information regarding the user's usage history of the knowledge database managed in a usage history database; Outputting the generated candidate information; when a correction to the output candidate information is input from an administrator, correction information indicating the correction to the candidate information input from the administrator is managed in a correction database; executing a process for generating the candidate information on a regular or irregular basis, and when the candidate information is generated in the process, notifying the administrator of information indicating that the candidate information has been generated; Information processing methods.

11. An information processing method in which an information processing device having a processor and a memory maintains a knowledge database in which questions, answers, and tag information are associated with each other and managed, comprising: The processor in cooperation with the memory: generating candidate information indicating candidates for addition or update of at least one of the question, the answer, or the tag information in the knowledge database based on usage history information indicating information regarding the user's usage history of the knowledge database managed in a usage history database; Outputting the generated candidate information; when a correction to the output candidate information is input from an administrator, correction information indicating the correction to the candidate information input from the administrator is managed in a correction database; learning a pattern of a relationship between the question and the tag information based on the knowledge database; generating candidates for adding or updating the tag information based on the learned patterns; Information processing methods.

12. An information processing method in which an information processing device having a processor and a memory maintains a knowledge database in which questions, answers, and tag information are associated with each other and managed, comprising: The processor in cooperation with the memory: generating candidate information indicating candidates for addition or update of at least one of the question, the answer, or the tag information in the knowledge database based on usage history information indicating information regarding the user's usage history of the knowledge database managed in a usage history database; Outputting the generated candidate information; when a correction to the output candidate information is input from an administrator, correction information indicating the correction to the candidate information input from the administrator is managed in a correction database; When outputting candidate information indicating additional candidates for the answer, a search result for the question using a predetermined search service is also output. Information processing methods.

13. A computer controls an information processing device that maintains a knowledge database that associates and manages questions, answers, and tag information, generating candidate information indicating candidates for addition or update of at least one of the question, the answer, or the tag information in the knowledge database based on usage history information indicating information regarding the user's usage history of the knowledge database managed in a usage history database; Outputting the generated candidate information; when a correction to the output candidate information is input from an administrator, correction information indicating the correction to the candidate information input from the administrator is managed in a correction database; If the setting is such that confirmation of the candidate information by the administrator is omitted, adding the question or the answer to the knowledge database or modifying the question or the answer based on the candidate information. An information processing program that executes processing.

14. A computer that controls an information processing device that maintains a knowledge database that associates and manages questions, answers, and tag information, comprising: generating candidate information indicating candidates for addition or update of at least one of the question, the answer, or the tag information in the knowledge database based on usage history information indicating information regarding the user's usage history of the knowledge database managed in a usage history database; Outputting the generated candidate information; when a correction to the output candidate information is input from an administrator, correction information indicating the correction to the candidate information input from the administrator is managed in a correction database; executing a process for generating the candidate information on a regular or irregular basis, and when the candidate information is generated in the process, notifying the administrator of information indicating that the candidate information has been generated; An information processing program that executes processing.

15. A computer that controls an information processing device that maintains a knowledge database that associates and manages questions, answers, and tag information, comprising: generating candidate information indicating candidates for addition or update of at least one of the question, the answer, or the tag information in the knowledge database based on usage history information indicating information regarding the user's usage history of the knowledge database managed in a usage history database; Outputting the generated candidate information; when a correction to the output candidate information is input from an administrator, correction information indicating the correction to the candidate information input from the administrator is managed in a correction database; learning a pattern of a relationship between the question and the tag information based on the knowledge database; generating candidates for adding or updating the tag information based on the learned patterns; An information processing program that executes processing.

16. A computer that controls an information processing device that maintains a knowledge database that associates and manages questions, answers, and tag information, comprising: generating candidate information indicating candidates for addition or update of at least one of the question, the answer, or the tag information in the knowledge database based on usage history information indicating information regarding the user's usage history of the knowledge database managed in a usage history database; Outputting the generated candidate information; when a correction to the output candidate information is input from an administrator, correction information indicating the correction to the candidate information input from the administrator is managed in a correction database; When outputting candidate information indicating additional candidates for the answer, a search result for the question using a predetermined search service is also output. An information processing program that executes processing.

Images