JP7369920B2 - servers and computer programs - Google Patents

Description

The present invention relates to a server and a computer program that implement knowledge management.

It is said that the average business person spends about 150 hours a year searching for things. If you work 240 days a year (8 hours a day), the percentage of time you spend looking for things is 150 hours/1920 hours = 7.8%.

According to the 2016 Private Sector Salary Statistics Survey (National Tax Agency), the average annual salary for salaried workers is approximately 4.22 million yen. Assuming that the amount required to hire a business person is twice the annual salary, the average annual salary x 2 = approximately 8.5 million yen.

Therefore, the annual cost of searching for items per business person is estimated to be approximately 8.5 million yen x 7.8% = approximately 660,000 yen. For a company with 1,000 employees, the cost of searching for items for all employees is estimated to be approximately 660,000 yen x 1,000 people = approximately 660 million yen. For a company with 5,000 employees, the cost of searching for items for all employees is estimated to be approximately 660,000 yen x 5,000 people = approximately 3.3 billion yen.

As described above, the cost that business people spend searching for items is relatively large, and the larger the company, the greater the cost. Therefore, a system is needed to efficiently search for information. Intra-company search (enterprise search) has conventionally been known as a mechanism for searching for information within a company. Additionally, there is a known method for increasing the efficiency of document searches by tagging documents (for example, see Patent Document 1).

Japanese Patent Application Publication No. 2018-195305 Japanese Patent Application Publication No. 2016-110607

https://www.traina.ai/solution/textmining/, retrieved June 17, 2019 https://www.traina.ai/solution/smartknowledge/, retrieved June 17, 2019

However, in the conventional document search technology as described in Patent Document 1, the search target is limited to documents. Additionally, enterprise search has the following characteristics: (1) There is little awareness of SEO, making it difficult to properly tag documents, and (2) Appropriate answers to what you are looking for are often obtained from humans. .

The present invention was made in view of these problems, and its purpose is to provide an enterprise search that can efficiently search for documents and people in a unified manner, and that can promote knowledge sharing and activation of communication. It is in.

One aspect of the present invention relates to a server. This server uses a means to acquire information via a network from an information source that holds information within the company, and by analyzing the acquired information, it can identify the person associated with the information. means for associating with the tags associated with the tags, means for receiving search requests from user terminals via a network, and means for generating search results including information on persons associated with tags that match the search content of the received search requests. , means for providing the user's terminal via a network with an interface for promoting interaction with a person included in the generated search results, and acquiring the content of the interaction performed via the provided interface. and a means for generating a search result for another search request using a model generated or updated by learning the content of the acquired interaction using a machine learning algorithm. .

Note that the present invention also includes any combination of the above-mentioned constituent elements, and mutual substitution of constituent elements and expressions of the present invention among devices, methods, systems, computer programs, recording media storing computer programs, etc. It is effective as an aspect of

According to the present invention, in enterprise search, documents and people can be searched efficiently and centrally, and knowledge sharing and communication can be activated.

FIG. 1 is a schematic diagram for explaining a knowledge management system according to an embodiment. 2 is a schematic diagram showing the system configuration of the knowledge management system of FIG. 1. FIG. 2 is a hardware configuration diagram of the knowledge data bank server in FIG. 1. FIG. 2 is a block diagram showing the functions and configuration of the knowledge data bank server of FIG. 1. FIG. FIG. 5 is a data structure diagram showing an example of the document information holding section of FIG. 4. FIG. FIG. 5 is a data structure diagram showing an example of the employee information holding section of FIG. 4. FIG. 2 is a flowchart showing the flow of a series of processes in the knowledge data bank server of FIG. 1. FIG. FIG. 3 is a representative screen diagram of a unified search screen displayed on a display of a user's terminal. FIG. 5 is a data structure diagram showing an example of a dialogue content holding unit in FIG. 4; FIG. 3 is a representative screen diagram of a chat room screen displayed on a display of a user's terminal. FIG. 7 is a representative screen diagram of another unified search screen displayed on the display of the user's terminal.

Hereinafter, the same or equivalent components, members, and processes shown in each drawing will be denoted by the same reference numerals, and redundant explanations will be omitted as appropriate. Further, in each drawing, some members that are not important for the explanation are omitted.

FIG. 1 is a schematic diagram for explaining a knowledge management system 2 according to an embodiment. In an enterprise search, the knowledge management system 2 makes it possible to identify a person through a tag by directly or indirectly associating a tag attached to a document within a company with a person in the company. As a result, the knowledge management system 2 can not only simply search for documents, but also promote knowledge sharing and communication with people within the company.

The knowledge management system 2 includes a portal site server 4 that manages a company's portal site, a file server 6 of the company, a mail server 8 of the company, and a directory server that manages basic employee information and access rights information of the company. 10 and a knowledge data bank server 102.

The knowledge data bank server 102 collects information from information sources that hold information within a company, such as a portal site server 4, a file server 6, a mail server 8, and a directory server 10, via a network such as the Internet (not shown). automatically retrieved. More specifically, the knowledge data bank server 102 crawls the portal site server 4, file server 6, and mail server 8, and extracts text from the collected documents. The knowledge data bank server 102 acquires basic employee information and access right information from the directory server 10.

The knowledge data bank server 102 automatically analyzes the collected information, adds tags to it, and converts it into knowledge. The knowledge data bank server 102 tags documents by inputting the raw text of the collected documents into a predetermined language processing engine. The language processing engine may be realized using, for example, the language analysis engine "Benten" provided by Nomura Research Institute, Inc. and the text data analysis system "TRAINA Text Mining" equipped with it (see Non-Patent Document 1). .

The knowledge data bank server 102 handles information that identifies a person, such as a person's name, appearing in a document as a tag (hereinafter referred to as a person name tag; tags other than person name tags are referred to as non-person name tags). The knowledge data bank server 102 uses personal name tags to attach non-human name tags to people in the company, and also automatically generates a social graph among the people in the company.

A user 12, who may be a person within a company (such as an employee), sends a search request specifying a search keyword to the knowledge database server 102 from his or her own terminal. The knowledge database server 102 extracts tags from search keywords, and identifies documents having the extracted tags, people associated with the documents, people having the extracted tags, and the like. The knowledge database server 102 generates a search result including information on the specified document and the specified person, and sends it to the user 12's terminal. The user 12 can learn about related documents and related experts through the search results.

The terminal of the user 12 is a stationary terminal such as a desktop PC, or a portable terminal such as a smartphone, a tablet terminal, or a laptop PC. An application program that enables enterprise search in the knowledge management system 2 is installed on the user 12's terminal. When this application program is executed by the terminal of the user 12, the terminal realizes various functions. Alternatively, the enterprise search in the knowledge management system 2 may be provided using the browser function of the user's 12 terminal.

The search results sent to the user 12's terminal are provided with a flow line (such as a link or button) to an interface for promoting interaction with the person included in the search result. When the user 12 desires to interact with a person included in the search results, the knowledge database server 102 generates an interface to implement the virtual community 14 and invites the person to the virtual community via the network.

The knowledge data bank server 102 automatically acquires the content of the dialogue that took place in the virtual community 14 and inputs it to the machine learning algorithm. The knowledge data bank server 102 uses the results of this machine learning to automatically register unknown words and similar words in the dictionary of the language processing engine. Automatic learning of dialogue content may be realized using, for example, "TRAINA Smart Knowledge" provided by Nomura Research Institute, Inc. (see Non-Patent Document 2).

FIG. 2 is a schematic diagram showing the system configuration of the knowledge management system 2. As shown in FIG. A directory server 10, a portal site server 4, and a file server 6 are provided within an in-house network 16. A crawl server 18 provided in the same in-house network 16 crawls the portal site server 4 using HTTP(S), and crawls the file server 6 using the CIFS (Common Internet File System) protocol. Crawling is performed to obtain employee information and authority information from the directory server 10 using LDAP (Lightweight Directory Access Protocol) or the like. The crawl server 18 transmits the collected text information, employee information, and authority information to the knowledge database server 102 outside the internal network 16 using HTTPS. The knowledge data bank server 102 crawls the mail server 8 outside the in-house network 16 using POP, IMAP, HTTP(S), Graph API, or the like. The knowledge data bank server 102 receives a search request using HTTP(S) or the like from a terminal of a user 12 within the company network 16.

Note that in other embodiments, the mail server 8 may be located in the company network 16, or at least one of the portal site server 4, file server 6, directory server 10, and crawl server 18 may be located in the company network 16. It can be outside. Further, although FIGS. 1 and 2 illustrate a case in which each server consists of one server for convenience, there is no limit to the number of servers.

FIG. 3 is a hardware configuration diagram of the knowledge data bank server 102 of FIG. 1. The knowledge data bank server 102 includes a memory 110, a processor 112, a communication interface 114, a display 116, and an input interface 118. These elements are each connected to a bus 120 and communicate with each other via the bus 120.

Memory 110 is a storage area for storing data and programs. Data and programs may be stored permanently or temporarily in memory 110. The processor 112 implements various functions of the knowledge data bank server 102 by executing programs stored in the memory 110. The communication interface 114 is an interface for transmitting and receiving data to and from the outside of the knowledge data bank server 102. The communication interface 114 is connected to a network and exchanges data with the crawl server 18 and the terminal of the user 12 via the network. Display 116 is a device for displaying various information. The input interface 118 is a device for receiving input from the administrator of the knowledge management system 2.

FIG. 4 is a block diagram showing the functions and configuration of the knowledge data bank server 102 of FIG. 1. In terms of hardware, each block shown here can be realized by elements and mechanical devices such as a computer's CPU, and in terms of software, it can be realized by computer programs, etc., but here, we will explain how they are realized by their cooperation. It depicts the functional blocks that will be implemented. Therefore, those skilled in the art who have been exposed to this specification will understand that these functional blocks can be realized in various ways by combining hardware and software.

The knowledge data bank server 102 includes an in-company information collection section 130, a document tagging section 132, a person tagging section 134, a social graph generation section 136, a search request reception section 138, a search processing section 140, and an automatic search processing section 140. Answer generating section 142, search result providing section 144, community generating section 146, learning section 148, raw data holding section 150, document information holding section 152, employee information holding section 154, dialogue content holding section 155 and.

The company information collection unit 130 acquires company information from the portal site server 4, file server 6, mail server 8, and directory server 10 via the network. The in-company information collection unit 130 acquires documents (portal sites, files, etc.) collected by the crawl server 18 and their text information from the crawl server 18 via the network. The in-company information collection unit 130 acquires documents (emails, etc.) and their text information from the mail server 8 via the network. The in-company information collection unit 130 stores the acquired document and its text information in the raw data holding unit 150.

The document tagging unit 132 analyzes the document acquired by the in-house information collection unit 130 and associates the person associated with the document with the non-person name tag attached to the document. The document tagging unit 132 processes the text information and metadata of the document held in the raw data holding unit 150 using the language processing engine described above to extract human name tags and non-personal name tags of the document. The document tagging unit 132 attaches (tags) the extracted personal name tag and non-personal name tag to the document. Specifically, the document tagging unit 132 registers information for identifying a document and a personal name tag and a non-personal name tag extracted from the text information of the document in the document information holding unit 152 in association with each other. The document tagging unit 132 automatically generates a summary from the text information of the document, associates the generated summary with the document, and registers it in the document information storage unit 152.

FIG. 5 is a data structure diagram showing an example of the document information holding unit 152 of FIG. 4. As shown in FIG. The document information holding unit 152 stores the document name of the document, the path that specifies the location where the document is stored, the person name tag extracted from the text information of the document, and the non-standard information extracted from the text information of the document. A person name tag and a summary of the document are stored in association with each other. In the document information holding unit 152, it can be said that a personal name tag, that is, a person, and a non-personal name tag are associated with each other via a document.

Returning to FIG. 4, the person tagging unit 134 attaches a non-person name tag to a person within the company. The person tagging unit 134 identifies employees of the company based on employee information acquired from the directory server 10. The person tagging unit 134 extracts, for each employee, the non-person name tag that is associated with the employee (the corresponding person name tag) in the document information holding unit 152, and stores the extracted non-person name tag in the employee information holding unit. Register at 154.

FIG. 6 is a data structure diagram showing an example of the employee information holding unit 154 of FIG. 4. As shown in FIG. The employee information holding unit 154 stores an employee ID that identifies the employee, the name of the employee, a non-person name tag extracted for the employee, the specialized field of the employee, and related employee information that identifies employees related to the employee. The ID and the ID are stored in association with each other.

The person tagging unit 134 counts non-person name tags extracted for a certain employee. When the number of non-human name tags exceeds a predetermined threshold, the person tagging unit 134 determines the field corresponding to the non-human name tag as the field of expertise of the employee. For example, if 20 non-human name tags "voice recognition" are extracted from the document information holding unit 152 for the employee name "Nomura ●●" with the employee ID "A001" and the threshold value is 15, the person tagging unit 134 determines that “Nomura ●●”’s specialty is “speech recognition” and registers it in the employee information holding unit 154 as such. In addition to the above, the employee's specialty area may be determined by employee information obtained from the directory server 10, or may be manually registered by the administrator. Alternatively, an employee's area of expertise may be determined using an algorithm such as machine learning.

Note that by referring to the employee information holding unit 154, the user 12 can know what kind of non-personal name tag is attached to him or herself.

Returning to FIG. 4, the social graph generation unit 136 generates a social graph between employees by referring to the document information holding unit 152. If a plurality of person name tags are registered for the same document in the document information holding unit 152, the social graph generation unit 136 determines that the employees specified by these person name tags are related. The social graph generation unit 136 registers the determined relationship between the employees in the employee information storage unit 154 as a related employee ID. For example, if the social graph generation unit 136 determines that the employee name “Nomura ●●” and the employee name “Tanaka ●●” are related in a certain document, the social graph generation unit 136 generates the employee information holding unit 154 The employee ID of "Tanaka ●●" is registered in the related employee ID of "Nomura ●●", and the employee ID of "Nomura ●●" is registered in the related employee ID of "Tanaka ●●". In addition to the above, the social graph may be determined based on employee information obtained from the directory server 10, or may be manually registered by the administrator.

The search request receiving unit 138 receives a search request including search contents such as a search keyword and a search sentence from the terminal of the user 12 via the network. The search content includes information expressing what the user 12 wants to search in text.

The search processing unit 140 extracts non-person name tags that match the search content by analyzing the search content included in the search request received by the search request receiving unit 138. This analysis and extraction may be realized by a technique similar to that used by the document tagging unit 132. The search processing unit 140 uses the extracted non-person name tag to obtain corresponding information from three sources: (1) document information holding unit 152, (2) employee information holding unit 154, and (3) automatically generated answers. and generate search results.

The search processing unit 140 refers to the document information holding unit 152 and identifies an entry having the extracted non-person name tag. The search processing unit 140 includes the document name, path, person name tag, and summary contained in the specified entry in the search result. At this time, the person name tag is converted to the name of the corresponding employee.

The search processing unit 140 refers to the employee information holding unit 154, and if there is an employee who has a specialized area corresponding to the extracted non-person name tag, the search processing unit 140 identifies the employee and includes the employee in the search results. For example, the search processing unit 140 refers to the employee information holding unit 154 in FIG. Identify it as a home and include it in search results.

The search processing unit 140 includes answers generated by an automatic answer generation unit 142, which will be described later, in the search results.

The automatic answer generation unit 142 automatically generates information corresponding to the non-person name tag extracted by the search processing unit 140. The automatic answer generation unit 142 generates, for example, a summary of past conversations related to the extracted non-person name tag. The automatic answer generation unit 142 generates an appropriate answer corresponding to the non-person name tag using a model that is generated and updated by machine learning that receives as input the contents of a conversation in a chat room, which will be described later. This model is a model that takes, for example, a non-person name tag as input and outputs past answers or summaries thereof. The automatic answer generation unit 142 may refer to the dialogue content holding unit 155 to identify participants in past dialogues corresponding to the output answers and summaries.

For example, assume that there is a search request using voice recognition as a keyword, and a community is formed and a dialogue is held between the requestor and an expert. A learning unit 148, which will be described later, acquires this dialogue and updates the model by mechanically learning the contents of the acquired dialogue. Next, when there is another search request using speech recognition as a keyword, the automatic answer generation unit 142 inputs the non-human name tag "speech recognition" into the updated model to retrieve information about past conversations or dialogues regarding speech recognition. A summary can be output. The automatic answer generation unit 142 passes this past interaction or its summary to the search processing unit 140, which includes it in the search results for another search request.

The search result providing unit 144 transmits the search results generated by the search processing unit 140 to the requesting terminal via the network.

The community generating unit 146 provides the user 12's terminal via the network with an interface for promoting interaction with employees included in the search results provided by the search result providing unit 144. The search results provided by the search result providing unit 144 correspond to documents tagged with non-human names that match the search content of the search request and employee information associated with the documents on the terminal of the user 12. It is provided in the form of a unified search screen (described later) that is displayed with the This unified search screen is configured to allow the user 12 to select an employee with whom he/she wishes to interact from among the displayed employees. The community generation unit 146 accepts employee selections through the unified search screen and generates a chat room (described later) that allows interaction with the selected employees. The community generation unit 146 invites the selected employees to the chat room.

The learning unit 148 acquires the content of the dialogue performed via the interface provided by the community generating unit 146 and registers it in the dialogue content holding unit 155. The learning unit 148 generates or updates the model used by the automatic answer generation unit 142 by mechanically learning the content of the acquired dialogue.

FIG. 9 is a data structure diagram showing an example of the dialogue content holding unit 155 of FIG. 4. As shown in FIG. The conversation content holding unit 155 stores a chat ID that specifies a chat in a past chat room, a start time and an end time of the chat, a participant ID that specifies a participant of the chat, and the content of the chat. A non-person name tag related to the chat is associated and held. This non-human name tag may be a non-human name tag extracted from the search content of the search that caused the establishment of the chat room. For example, assume that there is a search request using voice recognition as a keyword, and a chat is held in a chat room with the requester and an expert as participants. In this case, the employee ID of the requester and the employee ID of the expert are registered in the participant ID of the dialogue content holding unit 155, and "voice recognition" is registered in the non-person name tag.

Note that in this embodiment, a case has been described in which the automatic answer generation unit 142 automatically generates an answer using a model, but the present invention is not limited to this. In another embodiment, instead of automatic generation using a model, the automatic answer generation unit 142 refers to the conversation content storage unit 155 and uses the chat content corresponding to the non-person name tag extracted by the search processing unit 140. and the participant ID may be specified. In this case, the automatic answer generation unit 142 may generate a summary of the content of the specified chat using a model generated or updated by machine learning.

The operation of the knowledge data bank server 102 with the above configuration will be explained.
FIG. 7 is a flowchart showing the flow of a series of processes in the knowledge data bank server 102 of FIG. The knowledge data bank server 102 collects information within the company (S702). The knowledge data bank server 102 analyzes the collected information (S704). The knowledge data bank server 102 tags the document with a non-personal name tag and a personal name tag based on the analysis result (S706). The knowledge data bank server 102 tags the person with a non-person name tag based on the result of step S706 (S708). The knowledge data bank server 102 receives a search request from a user's terminal (S710). The knowledge data bank server 102 determines a non-person name tag that matches the search content (S712). The knowledge data bank server 102 identifies the person associated with the determined non-person name tag (S714). The knowledge database server 102 generates a search result including information on the identified person (S716). The knowledge data bank server 102 transmits the generated search results to the requesting terminal (S718). The knowledge data bank server 102 accepts a request to create a chat room specifying a person (S720). The knowledge data bank server 102 creates a chat room (S722). The knowledge data bank server 102 sends an invitation to the chat room to the specified person via the network (S724). The knowledge data bank server 102 acquires the content of the conversation in the chat room (S726). The knowledge data bank server 102 mechanically learns the content of the acquired dialogue (S728).

FIG. 8 is a representative screen diagram of a unified search screen 800 displayed on the display of the user's 12 terminal. The unified search screen 800 accepts search requests from the user 12 and presents search results to the user 12. The unified search screen 800 includes a search content input area 802, a search button 804, an extraction tag display area 806, a first search result 808, a second search result 810, a third search result 812, and a fourth search result. 814 and an invite button 816.

Search content input area 802 and search button 804 are used to generate a search request. The user 12 enters desired search content in the form of keywords or sentences in the search content input area 802, and presses the search button 804. When the user 12's terminal detects that the search button 804 has been pressed, it acquires the information entered in the search content input area 802 as the search content, generates a search request including the acquired search content, and sends the knowledge via the network. It is sent to the data bank server 102.

The extraction tag display area 806, the first search result 808, the second search result 810, the third search result 812, and the fourth search result 814 are areas that display search results for the transmitted search request. The extracted tag display area 806 is an area that estimates and displays tags related to the content input by the user 12, and particularly displays non-personal name tags extracted from the search content input in the search content input area 802. The extracted tag display area 806 accepts the selection of a non-personal name tag by the user 12. The user 12 selects a desired non-human name tag from among the non-human name tags displayed in the extraction tag display area 806. The terminal of the user 12 or the knowledge data bank server 102 that has received the selection narrows down the search results based on the selected non-person name tag. In the example of FIG. 8, six non-human name tags ("speech recognition", "speech synthesis", "deep learning", "AI", "big data", "statistical language processing", and "statistical language processing") displayed in the extraction tag display area 806 are shown in the example of FIG. ”) (“speech recognition,” “AI,” “big data”) are selected by the user 12, and the search results (first Search results 808, second search results 810, third search results 812, and fourth search results 814) are displayed.

The first search result 808 includes an image 818 of the document identified as a result of the search, a viewing authority presence/absence 820 indicating whether the user 12 has viewing authority for the document, and a document name 822 of the document. It includes a summary 824 of the document, a non-person name tag 826 attached to the document, an employee name 828 of the employee identified by the person name tag extracted from the document, and a check box 830. The presence or absence of viewing authority 820 may be determined from information about the user 12 obtained from the directory server 10. In the example of FIG. 8, since the user 12 has viewing authority for the specified document "Study on Speech Recognition," the document name 822 is displayed in a hyperlink format that includes a link to the document. When the user 12 presses down on the hyperlink, the document "Speech Recognition Survey" is retrieved and displayed.

The second search result 810 displays documents that are identified as a result of the search but for which the user 12 does not have permission to view, in a manner different from the first search result 808. In the second search result 810, the image 832 is replaced with a dummy image such as "?", and the presence/absence of viewing authority 834 indicates that there is no viewing authority. Document name 836 does not include a link. Elements other than these, that is, a summary 838, a non-person name tag 840, an employee name 842, and a check box 844 are displayed in the same manner as the first search result 808.

Normally, the user 12 does not have permission to view, for example, other people's e-mails (not only e-mails but also some materials that the user 12 does not have permission to view, and the present invention can be applied to these as well). Therefore, conventional systems do not allow users to search for e-mails of other employees of the same company. In this embodiment, although full-text searches of other people's e-mails are prohibited, personal name tags, non-person name tags, and summaries extracted from the e-mails are made searchable and displayed in the search results. Therefore, by maintaining the confidentiality of secret information and allowing such secret information to be "caught" in the search, a more accurate search can be achieved. Note that the configuration may be such that personal name tags, non-personal name tags, and summaries extracted from e-mails or materials are searchable, and the search results for personal name tags and non-personal name tags are displayed without displaying the search results for summaries.

Note that in the first search result 808 and the second search result 810, the employee names 828 and 842 are displayed as "creator", respectively. This is because the corresponding person name tag is extracted from the document creator information included in the document's metadata. When a person name tag is extracted from the text information of a document, a different expression than "author" may be used, such as "related party".

Further, in this embodiment, a case will be described in which an employee identified by a person name tag attached to a document identified as a result of a search is displayed together with the document in the search results, but the present invention is not limited to this. For example, the knowledge data bank server 102 refers to the employee information holding unit 154 and determines whether the user 12 who requested the search and the employee identified by the person name tag attached to the document identified as a result of the search are related employees. Employees may be included in the search results only if they are related via ID. In this case, since only employees who are so-called "friends" of the user 12 are displayed in the search results, it is possible to increase the probability that the user will accept the invitation to chat.

The third search result 812 corresponds to an expert (expert) extracted from the employee information holding unit 154. The third search result 812 includes a profile 846 that displays the name and contact information of the expert identified as a result of the search, a non-human name tag 848 attached to the expert, an expert name 850 of the expert, and a check box 852. , has.

The fourth search result 814 corresponds to the answer generated by the automatic answer generation unit 142. The fourth search result 814 is displayed in the form of a chatbot 854 teaching a summary 856 of past related interactions. The fourth search result 814 displays a non-person name tag 857 corresponding to a summary 856 of past related conversations, participant names 858 of participants in past related conversations, and check boxes 860. When the automatic answer generation unit 142 generates the fourth search result 814, the non-person name tag 857 and the participant name 858 may be specified by referring to the conversation content storage unit 155.

The user 12 selects the employees (Suzuki ●●, Tanaka ●●, Nomura ●●, Sasaki ●●, Select the employee you would like to consult from among the following employees (Sato●●) and check the corresponding checkbox. When the user 12 presses the invite button 816, the terminal accepts employees whose checkboxes are checked as employees that the user 12 would like to invite to chat. The terminal transmits a chat room generation request including information on employees whom the user 12 would like to invite to chat to the knowledge data bank server 102. Upon receiving the chat room creation request, the knowledge data bank server 102 creates a chat room and sends an e-mail containing a link to the chat room to the invited employee. The process itself for creating a chat room and inviting users to the chat room is well known (for example, see Patent Document 2), and therefore will not be described in detail in this specification.

FIG. 10 is a representative screen diagram of the chat room screen 862 displayed on the display of the user 12's terminal. When the user 12 presses the invite button 816 on the unified search screen 800 in FIG. 8, the screen transitions to a chat room screen 862. The chat room screen 862 includes a participant list area 864 that displays a list of participants in the chat room, a chat display area 866 that displays chat content posted by participants, and a chat area where chat content that the participants want to post is input. It has a content input area 868 and a send button 870.

The knowledge data bank server 102 automatically acquires the chat (dialogue) in the chat room screen 862. The knowledge data bank server 102 may perform a process of providing incentives such as points to those invited participants who provide answers via chat. This makes it possible to further activate communication.

In addition, for documents that the user does not have permission to view, such as the document displayed in the second search result 810 in FIG. You can request access to. Furthermore, in Figure 8, chat room (virtual room) invitation candidates are displayed in a selectable manner, but in addition to this display, each chat room invitation candidate's number of chat room invitations at that point, The configuration may also be such that at least one of the number of participations and the number of comments made in the chat room is displayed. In addition, in the example of FIG. 8, although there are no cases where chat room invitation candidates are displayed duplicately, there are cases where duplicate candidates are displayed, and the duplicately displayed candidates are displayed selectable. When the user selects a candidate that has been displayed in duplicate, a process may be performed in which all of the multiple display candidates that have been displayed in duplicate are selected.

In the embodiments described above, examples of the holding unit are a hard disk or a semiconductor memory. Further, based on the description of this specification, each part may be implemented by a CPU (not shown), an installed application program module, a system program module, a semiconductor memory that temporarily stores the contents of data read from a hard disk, etc. Those skilled in the art will understand that this can be done.

According to the knowledge data bank server 102 according to the present embodiment, it is possible to centrally search for documents and people within a company. This allows users to obtain deeper, more accurate, and more useful knowledge about what they are looking for or what they want to know, not only through documents (explicit knowledge) but also through communication with people (live information, tacit knowledge).

Furthermore, the knowledge data bank server 102 according to the present embodiment provides a flow line to a virtual community such as a chat room based on the search results. Therefore, communication between people can be further activated.

In the present embodiment, the contents of the dialogue conducted in such a virtual community are automatically acquired and learned, and the results are reflected in future searches as answers by the bot. Therefore, the accuracy and accuracy of future searches can be improved by using the content of past conversations.

The configuration and operation of the knowledge data bank server 102 according to the embodiment have been described above. This embodiment is merely an example, and those skilled in the art will understand that various modifications can be made to the combinations of each component and each process, and that such modifications are also within the scope of the present invention.

In the embodiment, a case has been described in which the search request receiving unit 138 receives a search request including a search keyword or a search sentence as the search content, but the present invention is not limited to this. For example, the search request accepting unit 138 may accept any document uploaded from the terminal of the user 12 as a search request. In this case, the search request receiving unit 138 processes the entire document as the search content and extracts the non-personal name tag from it. As a result, documents arbitrarily uploaded by the user 12 are automatically searched and similar documents and experts in related fields are presented on a screen such as the unified search screen 800, as well as on a chat room screen 862. A virtual community like this is automatically formed. This allows the user 12 to easily collect and examine in-house knowledge about the document he or she has uploaded, and also to discuss the document with experts in the virtual community.

In the embodiment, the chat room screen 862 has been described as an example of an interface for promoting dialogue with people included in the search results, but the interface is not limited to this, and can be used, for example, by e-mail, instant message, telephone conference, video conference, etc. Other interfaces may also be employed.

In the embodiment, a case has been described in which the user 12 invites the selected employee to the chat room on the unified search screen 800, but the invention is not limited to this. For example, the knowledge data bank server 102 may refer to the employee information holding unit 154. Chat room participants may be automatically selected. For example, the knowledge data bank server 102 acquires the non-human name tag selected in the extraction tag display area 806 of the central search screen 800, identifies the employee to whom the acquired non-human name tag is attached in the employee information holding unit 154, Processing may be executed to invite the identified employee to the chat room. The knowledge data bank server 102 may invite related employees of the identified employee. Alternatively, the knowledge data bank server 102 may invite the employee selected by the user 12 on the unified search screen 800 and its related employees to the chat room.

At this time, if attributes such as position and affiliation are assigned to employees, a process may be performed to exclude employees with specific attributes from being invited (or may be excluded from invitation display). For example, employees in positions above a certain level may be excluded from invitations. Furthermore, the knowledge data bank server 102 may be configured so that employees themselves can automatically decline invitations to chat rooms.
Alternatively, if attributes such as position and affiliation are assigned to employees, a process may be performed in which employees with specific attributes are invited (or may be included in the invitation display). For example, a process may be performed to invite only employees in the department to which the user belongs.

In the embodiment, means may be provided to monitor the personal name tag and non-personal name tag that are automatically added to documents by the document tagging unit 132. The monitoring means excludes tags generated by the document tagging unit 132 according to predetermined tag exclusion rules. The tag exclusion rule may be defined by, for example, personal information, prohibited words, abusive language, compliance, numerical value, amount of money, user definition, etc.

FIG. 11 is a representative screen diagram of another unified search screen 900 displayed on the display of the user's 12 terminal. The unified search screen 900 displays statistical information 902 of the search results. The statistical information 902 includes a company department 904 and a bar graph 906 showing the total number of employees and documents that match the non-person name tag selected in the extraction tag display area 806 among the employees and documents belonging to the department. , and the total 908 are displayed in association with each other. When the user 12 presses down on a depleted department among the corporate departments 904, a screen is displayed for checking details of employees and documents belonging to that department that match the non-person name tag selected in the extraction tag display area 806. (For example, a screen similar to the unified search screen 800 in FIG. 8) is displayed. Furthermore, if you specify the level of an individual employee, detailed information about that employee will be displayed.

2 Knowledge Management System, 4 Portal Site Server, 6 File Server, 8 Mail Server, 10 Directory Server, 12 User, 102 Knowledge Data Bank Server.

Claims (5)

A means for acquiring information via a network from an information source that holds information within the company;
means for associating a person associated with the information with a tag attached to the information by analyzing the acquired information;
means for receiving a search request from a user's terminal via a network;
means for generating search results including information on a person associated with a tag that matches the search content of the received search request;
means for providing the user's terminal via a network with an interface for facilitating interaction with a person included in the generated search results;
means for acquiring the content of the interaction made via the provided interface;
Equipped with
The generating means generates a search result for another search request using a model generated or updated by learning the content of the acquired interaction using a machine learning algorithm.
server. The associating means determines that the field of expertise of a person associated with a tag corresponding to a certain field whose number exceeds a predetermined threshold is the field;
2. The server according to claim 1, wherein the generating means includes in the search results a person who has a specialized field corresponding to a tag that matches the search content . The search result includes information on a document associated with a tag that matches the search content and information on a corresponding person,
The generating means searches for documents that the user has permission to view and documents that the user does not have permission to view;
When the search results include information on a document that the user does not have permission to view, the providing means displays the information on the document on the screen displaying the search results, but does not display the content of the document. The server according to claim 1 , which does not allow the user to browse . The screen displaying the search results displays information on the document and information on the person in association with each other ,
4. The server according to claim 3 , wherein the screen is configured to allow the user to select a person with whom he/she wishes to interact from among the displayed persons. A computer program stored in a recording medium of a user's terminal, which when executed by a computer of the terminal, in the terminal,
the ability to send search requests to a server over the network;
As a search result for a submitted search request, a screen is displayed on the display that displays information on documents with tags that match the search content of the search request in association with information on people associated with the documents. and the function to
a function of accepting a person selected by the user from among the people displayed on the screen as a person with whom the user wants to have a dialogue;
A function for displaying an interface for interacting with a selected person via a network on a display;
Sending the content of the interaction performed via the displayed interface to the server to generate or update a model used when generating search results for another search request by learning using a machine learning algorithm. Function and
A computer program that makes this possible .

Images