Information processing system, control method and program for information processing device

Abstract

This invention provides an information processing system, a control method for an information processing device, and a program for information processing devices, which enable users within an organization to efficiently access and utilize necessary information according to their viewing permissions. [Solution] The information processing system 100 includes a linkage processing unit 31 that acquires target information from each of the business applications 3a to 3c, a viewing permission acquisition unit 32 that acquires viewing permission information indicating the viewing permission set for each user for the target information, an information management unit 33 that manages the target information and viewing permission information in association, a user request acquisition unit 34 that acquires requests from users, a search target determination unit 35 that determines the search target information that will be the scope of the search for the request based on the viewing permission information and target information set for the user who made the request, a request-related information extraction unit 36 ​​that extracts request-related information related to the request from the search target information, and a response generation unit 37 that performs processing to generate a response to the request to be presented to the user based on the request-related information.

Description

Technical Field

[0005] , ,

[0001] The present invention relates to an information processing system, a control method for an information processing apparatus, and a program.

Background Art

[0002] Conventionally, in an organization such as a company, software, web services, etc. for supporting business operations have been used. For example, Patent Document 1 describes a technology related to this type of business support.

[0003] Patent Document 1 relates to a help information display system that displays help information for assisting a user's operation on a user terminal operated by the user. The help information display system of Patent Document 1 includes a reception unit that receives a user's selection operation for querying help information, a display control unit that displays a help screen including help information for assisting the user's operation, and a help information storage unit that stores help information and information on a business screen associated with the help information for each piece of help information. When the reception unit receives a user's selection operation, the display control unit acquires the help information associated with the business screen displayed on the user terminal from the help information storage unit and displays a help screen including the acquired help information.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] In organizations such as companies, various business applications such as messaging applications, cloud storage services, scheduling applications, and workflow applications are used in parallel. These business applications are often provided by different service providers, making it difficult to centralize business-related information. However, when information necessary for internal business operations is scattered across multiple business applications, new employees find it difficult to locate the information they need, which can lead to slower adaptation to the organization and lower employee retention rates. Furthermore, inquiries about necessary business information may concentrate on specific individuals or departments that centrally manage this information, potentially leading to a shortage of resources needed for their actual work. While existing FAQ systems can address this information scattering, they have limitations when the information needed changes over time and is scattered across multiple business applications. Moreover, building an FAQ system itself requires considerable effort. Therefore, organizing information accumulated within an organization using language models such as LLMs is a viable option. However, organizational information often has access restrictions in place to mitigate information leakage risks, and language models such as LLMs may present information that should not be disclosed to users without access permissions when providing answers.

[0006] This invention has been made in view of such circumstances, and aims to provide an information processing system, an information processing device control method, and a program that enable users belonging to an organization to efficiently access and utilize necessary information according to their viewing permissions, even when the organization's information is scattered across multiple business applications. [Means for solving the problem]

[0007] To achieve the above objective, one aspect of the present invention is an information processing system comprising: a linkage processing means that links with a plurality of business applications and acquires target information from each of the business applications; a viewing permission acquisition means that acquires viewing permission information indicating viewing permission set for each user with respect to the target information; an information management means that manages the target information and the viewing permission information in association; a user request acquisition means that acquires requests from the user; a search target determination means that determines search target information that will be the scope of the search for the request based on the viewing permission information set for the user who made the request and the target information; a request-related information extraction means that extracts request-related information related to the request from the search target information; and a response generation means that performs processing to generate a response to the request to be presented to the user based on the request-related information.

[0008] Furthermore, one aspect of the present invention is a control method for an information processing device, which includes: a linkage processing step that links with a plurality of business applications and acquires target information from each of the business applications; a viewing permission acquisition step that acquires viewing permission information indicating viewing permissions set for each user with respect to the target information; an information management step that manages the target information and the viewing permission information in association; a user request acquisition step that acquires a request from the user; a search target determination step that determines search target information that will be the scope of the search for the request based on the viewing permission information set for the user who made the request and the target information; a request-related information extraction step that extracts request-related information related to the request from the search target information; and a response generation step that performs processing to generate a response to the request to be presented to the user based on the request-related information.

[0009] Furthermore, one aspect of the present invention is a program that causes a computer to execute the following steps: a linkage processing step that links with a plurality of business applications and acquires target information from each of the business applications; a viewing permission acquisition step that acquires viewing permission information indicating viewing permissions set for each user with respect to the target information; an information management step that manages the target information and the viewing permission information in association; a user request acquisition step that acquires a request from the user; a search target determination step that determines the search target information that will be the scope of the search for the request based on the viewing permission information set for the user who made the request and the target information; a request related information extraction step that extracts request related information related to the request from the search target information; and a response generation step that performs processing to generate a response to the request to be presented to the user based on the request related information. [Effects of the Invention]

[0010] According to the present invention, even when organizational information is scattered across multiple business applications, it is possible to provide an information processing system, a control method for an information processing device, and a program that enable users belonging to an organization to efficiently access and utilize the necessary information according to their viewing permissions. [Brief explanation of the drawing]

[0011] [Figure 1] This figure shows an information processing system according to one embodiment of the present invention. [Figure 2] This is a block diagram showing the hardware configuration of the information processing device according to this embodiment. [Figure 3] This is a functional block diagram showing an example of the functional configuration of the information processing device according to this embodiment. [Figure 4] This flowchart shows an example of the process for registering target information into a search database. [Figure 5] This flowchart shows an example of the process flow for generating responses to user requests in an information processing device. [Figure 6]This flowchart shows an example of the process flow for generating responses to user requests on a user terminal. [Figure 7] This figure shows an example of a request settings screen where a user enters a request. [Figure 8] This figure shows an example of a request settings screen where a response is provided. [Figure 9] This figure shows an example of a request settings screen with filters applied to the provided responses. [Figure 10] This figure shows an example of a request settings screen where responses are presented with the set filters applied. [Figure 11] This is a functional block diagram showing an example of the functional configuration of an information processing device related to a modified example. [Modes for carrying out the invention]

[0012] One embodiment of the present invention will be described below with reference to the drawings.

[0013] <System Configuration> First, let's describe the overall system configuration. Figure 1 is a diagram showing an information processing system 100 according to one embodiment of the present invention.

[0014] The information processing system 100 of this embodiment works in conjunction with multiple business applications 3a to 3c used by the organization, aggregates and manages business-related information obtained from each of the multiple business applications 3a to 3c, and provides a service that provides business-related information to users belonging to the organization.

[0015] The business applications 3a to 3c are software, cloud services, etc. used by users belonging to an organization. The business applications 3a to 3c are communication tools that can conduct chats between users, messaging applications such as SNS (Social Networking Service), cloud storage services that can share data between users, schedule applications that manage user schedules, workflow applications that manage business approvals, progress, etc. For example, external software and web services such as Slack (registered trademark), Notion, Google Drive, Google Workspace (registered trademark), Salesforce (registered trademark), Box (registered trademark), Outlook (registered trademark), etc. correspond to the business applications 3a to 3c. Although three business applications 3a to 3c are illustrated, the number of business applications 3a to 3c may be two or four or more.

[0016] The user is, for example, an employee, a manager, etc. who belongs to an organization that uses the information processing system 100 and performs the business of the organization. The user uses the information processing system 100 through the user terminal 2. The user terminal 2 is composed of a smartphone, a tablet, a personal computer, etc. Although only one user terminal 2 is illustrated in FIG. 1, it is assumed that each user terminal 2 of a plurality of users belonging to the organization uses the information processing system 100 in parallel.

[0017] The information regarding the business provided by the information processing system 100 is generated by the information processing device 1, which is a server that transmits and receives various information to and from the user terminal 2 via a communication network such as the Internet. The information processing device 1 of the present embodiment generates an answer to a user's request using the information of the organization that aggregates and manages and the language model 5. The language model 5 is a generative AI such as a large language model (LLM: Large Language Model).

[0018] <Hardware Configuration> Next, an example of the hardware constituting the information processing device 1 will be described. Figure 2 is a block diagram showing the hardware configuration of the information processing device 1 according to this embodiment. The information processing device 1 includes a CPU (Central Processing Unit) 11 as a processor, a ROM (Read Only Memory) 12, a RAM (Random Access Memory) 13, a bus 14, an input / output interface 15, an output unit 16, an input unit 17, a storage unit 18, a communication unit 19, and a drive 20.

[0019] The CPU 11 executes various processes according to the program stored in the ROM 12 or the program loaded from the storage unit 18 into the RAM 13. The RAM 13 also stores data necessary for the CPU 11 to execute various processes. The CPU 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output interface 15 is also connected to this bus 14.

[0020] The input / output interface 15 is connected to an output unit 16, an input unit 17, a storage unit 18, a communication unit 19, and a drive 20. The output unit 16 consists of a display, speakers, etc., and outputs various information as images and sounds. The input unit 17 consists of a keyboard, mouse, etc., and inputs various information. The storage unit 18 consists of a hard disk, DRAM (Dynamic Random Access Memory), etc., and stores various data. The communication unit 19 communicates with other devices via a network, including the Internet.

[0021] The drive 20 is appropriately equipped with removable media 21, which may consist of a magnetic disk, optical disk, magneto-optical disk, or semiconductor memory. Programs read from the removable media 21 by the drive 20 are installed in the storage unit 18 as needed. The removable media 21 can also store various types of data stored in the storage unit 18, just like the storage unit 18.

[0022] The hardware configuration described here is merely an example. The computer described in this embodiment, including the information processing device 1, may have the same configuration as that shown in Figure 2, or it may have a different configuration. Furthermore, the computer may consist of two or more computers. The user terminal 2 in Figure 1 is, for example, a smartphone, tablet, or personal computer having a configuration similar to the hardware configuration shown in Figure 2.

[0023] <Functional configuration> Next, the functional configuration of the information processing device 1 will be described. Figure 3 is a functional block diagram showing an example of the functional configuration of the information processing device 1 according to this embodiment.

[0024] As shown in Figure 3, the information processing device 1 includes a linkage processing unit 31, a viewing permission acquisition unit 32, an information management unit 33, a user request acquisition unit 34, a search target determination unit 35, a request-related information extraction unit 36, a response generation unit 37, an evaluation reception unit 38, and an evaluation reflection unit 39, all of which are functional units executed on the processor (CPU 11).

[0025] The collaboration processing unit 31 is a collaboration processing means that collaborates with multiple business applications 3a to 3c and obtains target information from each of the business applications 3a to 3c. The collaboration processing unit 31 collaborates with multiple business applications 3a to 3c, for example, through an API (Application Programming Interface).

[0026] The target information acquired by the integration processing unit 31 includes, for example, conversation logs such as messages, conversations, and posts; documents, files, and attached data; events set in the scheduling application; calendars, tasks, and tickets; and workflow execution history such as approvals and reviews in the workflow application.

[0027] The target information acquired by the collaborative processing unit 31 is structured and registered in the search database 90, which is constructed for Retrieval Augmented Generation (RAG). The search database 90 is a source of information referenced when generating responses using the language model 5 in response to user requests, and it holds the target information in a searchable format.

[0028] This section describes an example of building a RAG search database 90 as a source of FAQ candidates for inquiries within an organization. The integration processing unit 31 integrates with business applications 3a-3c such as Slack, Box, and Outlook to obtain the target information necessary for creating FAQ candidates.

[0029] The integration processing unit 31 extracts posts from the "Frequently Asked Questions" channel in Slack and the corresponding threads as target information for creating FAQs. The search target determination unit 35 analyzes text from business manuals (PDFs) in Box and extracts it as target information, as well as inquiry and response history from Outlook email history. The integration processing unit 31 structures this extracted target information to make it easy to search and registers it as "FAQ candidates" in the search database 90.

[0030] The access permission acquisition unit 32 is an access permission acquisition means that acquires access permission information indicating the access permissions set for each user for the target information. Access permission information is, for example, information that identifies the scope of the target information that a user can access, and includes the group and affiliation information set for the user. Access permission information is set for each user belonging to an organization. Therefore, for a given target information, users in a specific department or position may be able to access it, but users in other departments or who do not hold a specific position may not be able to access it.

[0031] The timing and method of obtaining access permission information by the access permission acquisition unit 32 are not particularly limited. For example, the access permission acquisition unit 32 may acquire access permission information when linking with multiple business applications 3a to 3c using OAuth, or it may accept user-specific access permission settings from the organization's administrator as access permission information.

[0032] The information management unit 33 is an information management means that manages target information in association with access permission information. For example, the information management unit 33 adds access permission metadata (ACL: Access Control List) to each target information registered in the search database 90.

[0033] The user request acquisition unit 34 is a user request acquisition means that acquires requests from users. The user request acquisition unit 34 acquires requests entered by the user from the user terminal 2 via a communication network such as the Internet. In this embodiment, requests made by the user are made in a conversational format via a chatbot.

[0034] The types of requests made by users include, for example, questions about certain information or technologies, as well as the creation of FAQs, parts lists, and materials used for work such as presentations. Requests are made, for example, in the form of a chatbot conversation. Requests can be accompanied by images, files in specified data formats, audio data, etc.

[0035] The user request acquisition unit 34 of this embodiment acquires filter settings as condition information specified by the user. Filter information is information that specifies the range to be searched. Filter information includes application filters, file format filters, time period filters, organization filters, user filters, in-app filters, and custom filters.

[0036] The application filter is a condition that specifies the business applications 3a-3c in which the target information is stored. The file format filter is a condition that narrows the search range by the file format of the target information. The period filter is a condition that narrows the search range by the date the target information was updated, and you can select the start date and end date. The organization filter is a condition that narrows the search range by the group, such as the organization or department to which the poster of the target information belongs. The user filter is a condition that narrows the search range by the poster of the target information. A poster is, for example, a user belonging to an organization. The in-app filter is a condition that further narrows the search range within business applications 3a-3c, such as projects or workspaces. The custom filter is a condition that users can register in JSON (JavaScript Object Notation) format, which allows for filtering such as excluding noisy information sources.

[0037] The search target determination unit 35 is a search target determination means that determines the search target information that will be the scope of the search for the request, based on the viewing permission information and target information set for the user who made the request. If filter information has been set by the user, the search target determination unit 35 determines the search target information to satisfy the filter information.

[0038] The search target determination unit 35 includes target information for which viewing permissions have been set. For example, the search target determination unit 35 includes conversation logs in the search target information even if the user who made the request did not participate, if the user has viewing permissions for those logs.

[0039] The request-related information extraction unit 36 ​​is a request-related information extraction means that extracts request-related information from the search target information. The request-related information extraction unit 36 ​​is implemented, for example, using an application such as a search engine. Furthermore, if images or files are attached to the request, similar target information is extracted as request-related information. Note that the search engine may utilize the language model 5. Whether or not the information is similar is determined based on the context of the request text, the names of the attached images or files, image analysis, file content analysis, etc.

[0040] The response generation unit 37 is a response generation means that processes information to generate a response to a request, which is then presented to the user. The response generation unit 37 outputs a prompt to the language model 5 instructing it to generate a response to the user's request based on the request-related information extracted from the search target information, and then retrieves the response result from the language model 5. The response generation unit 37 then presents the retrieved response result to the user.

[0041] In this embodiment, if the response generation unit 37 does not find any target information for providing an answer, it presents (suggests) a question to the user. The question may be obtained by the response generation unit 37 sending a prompt to the language model 5 instructing it to create a question, or the question may be selected according to conditions that are set in advance using a rule-based system.

[0042] The format of the response generated by the response generation unit 37 corresponds to the user's request. For example, if the request is a question about technology used in business, the user will be presented with the answer to the question along with a link to the document on which the answer is based. If the request is for the creation of a bill of materials based on images such as drawings, the user will be presented with the bill of materials created by the language model 5. Images such as drawings may be attached to the user when making a request, or they may be included in the search target information registered in the search database 90. If the request is for contract difference extraction, the difference between the contract to be verified and the contract to be compared will be output as the response.

[0043] Furthermore, the response generation unit 37 can also generate an automated version of some business workflows as a response based on the question text included in the request. Alternatively, it may determine from the question text of the request whether a series of business workflows such as scheduling, leave application, and approval are involved, and connect to the appropriate dedicated chatbot for each type of task. Examples of dedicated chatbots for each type of task include project analysis, inventory management, ordering support, expense reimbursement, and legal checks. It may also determine from the question text of the request whether to perform a search in the search database 90, connect to another linked dedicated chatbot for a different task, or execute a workflow.

[0044] Furthermore, users can configure settings such as language, number of responses, and response format for the chatbot they are using on the home screen. The number of responses setting specifies the number of bullet points presented in the response. The response format setting allows users to choose whether the response is in natural language or a bulleted list.

[0045] The evaluation reception unit 38 is an evaluation reception means that receives user evaluations of the generated answers. The evaluation reception unit 38 displays, for example, a feedback input unit on the user terminal 2 for the user to input their views on the answers, along with the answers generated by the answer generation unit 37. The feedback input unit may be, for example, feedback buttons for the user to select their feelings such as satisfaction or dissatisfaction with the answers, or it may be text for pointing out specific issues. The evaluation reception unit 38 obtains user feedback in accordance with the operations on the user terminal 2.

[0046] The evaluation reflection unit 39 acquires feedback entered by the user and performs evaluation reflection processing to reflect the feedback in the request-related information extraction process (search process) of the request-related information extraction unit 36, the response generation process for the response to the request of the response generation unit 37, etc. For example, in the evaluation reflection processing, the priority of extracting request-related information by the request-related information extraction unit 36 ​​is adjusted. In addition, in the evaluation reflection processing, weighting or structural modification of the information to be searched may be performed based on the feedback.

[0047] The evaluation and feedback process will be explained using an example where a user requests the extraction and identification of risks in contract clauses. Suppose the response provided in response to the user's request points out that "there are deficiencies in the confidentiality clause," and a revised version is presented to the user along with the response. In this case, suppose the user reviews the response and finds that there are no actual deficiencies, so they discard the revised version and adopt the original clause. The evaluation reception unit 38 receives this series of operations log as feedback. This operation log also includes information such as which outputs were rejected.

[0048] The evaluation reflection unit 39 incorporates feedback into the request-related information extraction process by re-ranking the search results of the search database 90 using a different, simpler machine learning model. This ensures that the feedback is reflected in subsequent request-related information extraction processes. This allows the feedback to influence the search results of the search database 90, thereby improving extraction accuracy. Alternatively, the evaluation reflection unit 39 modifies the prompts output to the language model 5 based on the feedback from the evaluation reception unit 38, lowering the risk detection threshold for similar clauses and skipping or only providing reference information for those clauses. This ensures that the feedback is reflected in subsequent response generation processes. In this way, the accuracy of responses is improved by logging the natural in-work operations of users using the information processing system 100.

[0049] <Flowchart of the process for registering target information in the search database> Referring to Figure 4, the flow of the process for registering target information to the search database 90 by the linkage processing unit 31, the access rights acquisition unit 32, and the information management unit 33 will be explained. Figure 4 is a flowchart showing an example of the flow of the process for registering target information to the search database 90.

[0050] In step S1, the integration processing unit 31 interacts with the target business applications 3a to 3c via the communication network and obtains target information to be registered in the search database 90.

[0051] In step S2, the integration processing unit 31 registers the target information obtained from business applications 3a to 3c in the search database 90.

[0052] In step S3, the access permission acquisition unit 32 acquires access permission information in which access permissions for each user are set for the target information.

[0053] In step S4, the information management unit 33 adds access permission metadata to the target information registered in the search database 90 and associates the target information with the access permission information.

[0054] Through the above series of processes, the target information, along with access permission information, is registered in the search database 90. The timing of this series of processes is not particularly limited. The registration process may be executed when a request is made by the user, may be executed by batch processing at a predetermined time, or may be updated in real time.

[0055] <Response generation process flow> Next, we will explain the process of generating responses to user requests by the user request acquisition unit 34, the search target determination unit 35, the request-related information extraction unit 36, the response generation unit 37, the evaluation reception unit 38, and the evaluation reflection unit 39.

[0056] Figure 5 is a flowchart showing an example of the flow of the response generation process and feedback reflection process in the information processing device 1 in response to a user's request.

[0057] In step S11, the user request acquisition unit 34 acquires the request details entered by the user from the user terminal 2.

[0058] In step S12, the search target determination unit 35 refers to the search database 90 and determines the search target information based on the access permission information associated with the target information.

[0059] In step S13, the request-related information extraction unit 36 ​​searches for and extracts request-related information for the request, using the search target information determined by the search target determination unit 35 as the search scope.

[0060] In step S14, the response generation unit 37 outputs a prompt to the language model 5 that includes an instruction to generate a response based on the extracted request-related information, and retrieves the response to the request from the language model 5 and presents it to the user.

[0061] In step S15, if the user request acquisition unit 34 receives an instruction to apply filter information indicating that a filter has been set by the user who was presented with the response results, the process proceeds to step S16 (step S15; Yes). In step S16, the search target determination unit 35 applies the conditions included in the filter information to the search target information, and the search target information is narrowed down. After the processing in step S16, the process returns to step S13, and the search and extraction of request-related information and the response generation process are performed sequentially based on the updated range of search target information.

[0062] In step S15, if the user request acquisition unit 34 determines that the user who was presented with the response result did not set a filter, the process proceeds to step S17 (step S15; No).

[0063] In step S17, if the evaluation reception unit 38 receives feedback from the user via the user terminal 2 regarding the response, the process proceeds to step S18 (step S17; Yes). In step S18, the evaluation reflection unit 39 executes a process to reflect the received feedback, and this flow ends. In step S17, if the evaluation reception unit 38 does not receive feedback, this flow ends (step S17; No). If the user makes a new request, the process proceeds to step S11.

[0064] Next, we will explain the response generation process on the user terminal 2 side. Figure 6 is a flowchart showing an example of the flow of the response generation process and feedback reflection process in response to the user's request on user terminal 2.

[0065] In step S21, user terminal 2 receives input from the user request. User terminal 2 receives the user request via text input or voice input in a thread on the chatbot screen that uses the information processing system 100.

[0066] In step S22, user terminal 2 transmits the request received from the user to information processing device 1. For example, user terminal 2 transmits the conversation history on the thread on user terminal 2 to information processing device 1 as the user's request. This process initiates the process of step S11 in Figure 5.

[0067] In step S23, the user terminal 2 displays the response to the request received from the information processing device 1. For example, the user terminal 2 displays the response in a conversational format on a thread. The response to the request received from the information processing device 1 is the one presented to the user by the process in step S14 of Figure 5.

[0068] In step S24, if the user has set a filter, the process proceeds to step S25 (step S24; Yes). In step S25, the user terminal 2 sends filter information indicating the content of the filter set by the user, along with the request on the user's thread. Following this process, the process in step S16 of Figure 5 is executed. After the process in step S25, the process returns to step S23, and processes such as receiving the response from the information processing device 1 after the filter has been applied are performed sequentially.

[0069] If no filter settings are made by the user on user terminal 2 in step S24, the process proceeds to step S26 (step S24; No).

[0070] In step S26, if user terminal 2 receives feedback input from the user regarding the answer, the process proceeds to step S27 (step S26; Yes). In step S27, user terminal 2 executes the process of sending the feedback input from the user to information processing device 1, and this flow ends. This process executes the process of step S18 in Figure 5.

[0071] If no feedback is received from the user regarding the response in step S26, this flow terminates (step S26; No). If the user submits a new request, the process proceeds to step S21.

[0072] <Request settings screen displayed on the user's terminal> Next, we will explain an example of the request settings screen displayed on user terminal 2, referring to Figures 7 to 10. The request settings screen is displayed when, for example, the user launches an application installed on user terminal 2 and selects a chatbot. The user enters their request in a conversational format with the chatbot.

[0073] Figure 7 shows an example of a request setting screen where a user enters a request. The request setting screen shown in Figure 7 displays a user request display 101 that displays the user's request, a user request input field 102 where the user enters the request, and a response display 110 that displays the answer to the request. In the example in Figure 7, the user request display 101 displays the name of the user who made the request, along with the message "Tell me about NapAnt's services." This request was entered through the user request input field 102. At this point, the response display 110 displays "Searching for information..." because the process described in Figure 5 is being executed on the information processing device 1 and the answer is being generated. The answer is displayed via streaming and eventually transitions to the request setting screen in Figure 8. For example, the generation of the answer starts in about 3 seconds, and the completed answer is displayed within 10 seconds.

[0074] Figure 8 shows an example of a request setting screen where answers are presented. The request setting screen shown in Figure 8 displays a user request display 101, a user request input field 102, an answer display 110, along with a filter setting section 111 and a search button 112. The answer display 110 displays the answers sent from the information processing device 1. In this example, the answer display 110 displays three answers along with links 115a to 115c to the data that led to each answer. The user can access the source of the answer by selecting the link 115a to 115c corresponding to the answer.

[0075] The filter setting unit 111 is a setting input unit where the user sets conditions for the information to be searched. In this example, the user can set conditions such as the type of business application 3a to 3c to be used, the file type, the start date of the period, and the end date of the period as filter information. The filter setting unit 111 allows the user to set conditions in a pull-down format. When the user sets the filter conditions and selects the search button 112, the request setting screen in Figure 8 transitions to the request setting screen in Figure 9.

[0076] Figure 9 shows an example of a request settings screen with filters applied to the provided answers. The request settings screen shown in Figure 9 displays a user request display 101, a user request input field 102, an answer display 110, along with a filter setting section 111 and a search button 112. In the filter setting section 111 of the request settings screen shown in Figure 9, Google Drive is selected as the business application 3a~3c and PDF is selected as the file format, as these are the conditions entered by the user. The start date of the period is set to 2023 / 01 / 01, and the end date of the period is set to today, when the search was performed. Below the search button 112, "Searching for information..." is displayed because answers based on the new range of search target information to which the filter has been applied are in the process of being generated. When new answers are generated, the request settings screen in Figure 9 transitions to the request settings screen in Figure 10.

[0077] Figure 10 shows an example of a request setting screen where answers reflecting the set filters are displayed. The request setting screen shown in Figure 10 displays a user request input field 102, an answer display 110, a filter setting section 111, and a search button 112. Three answers are displayed in the answer display 110 of the setting screen shown in Figure 10. Of the three answers, links 115d and 115e of the data that led to each answer are displayed for two of the answers. In addition, since the last answer is beyond the display range, the user can select the expansion section 116 to display the rest of the answer. The link for the third answer can be confirmed by operating this expansion section 116 to display the rest.

[0078] As described above, the information processing system 100 of this embodiment includes a linkage processing unit 31 that cooperates with multiple business applications 3a to 3c and acquires target information from each of the business applications 3a to 3c; a viewing permission acquisition unit 32 that acquires viewing permission information indicating the viewing permission set for each user with respect to the target information; an information management unit 33 that manages the target information and viewing permission information in association; a user request acquisition unit 34 that acquires requests from users; a search target determination unit 35 that determines the search target information that will be the scope of the search for the request based on the viewing permission information and target information set for the user who made the request; a request-related information extraction unit 36 ​​that extracts request-related information related to the request from the search target information; and a response generation unit 37 that performs processing to generate a response to the request to be presented to the user based on the request-related information.

[0079] Furthermore, the control method for the information processing device 1 of this embodiment includes a cooperation processing step that cooperates with a plurality of business applications 3a to 3c and acquires target information from each of the business applications 3a to 3c; a viewing permission acquisition step that acquires viewing permission information indicating the viewing permission set for each user for the target information; an information management step that manages the target information and viewing permission information in association; a user request acquisition step that acquires a request from a user; a search target determination step that determines the search target information that will be the scope of the search for the request based on the viewing permission information and target information set for the user who made the request; a request-related information extraction step that extracts request-related information related to the request from the search target information; and a response generation step that performs processing to generate a response to be presented to the user in response to the request based on the request-related information.

[0080] Furthermore, the computer program of this embodiment cooperates with multiple business applications 3a to 3c and causes the computer to execute the following steps: a cooperation processing step to acquire target information from each of the business applications 3a to 3c; a viewing permission acquisition step to acquire viewing permission information indicating the viewing permission set for each user for the target information; an information management step to manage the target information and viewing permission information in association; a user request acquisition step to acquire a request from a user; a search target determination step to determine the search target information that will be the scope of the search for the request based on the viewing permission information and target information set for the user who made the request; a request-related information extraction step to extract request-related information related to the request from the search target information; and a response generation step to perform processing to generate a response to the request to be presented to the user based on the request-related information.

[0081] In this way, with the control method or program for the information processing system 100 and the information processing device 1 configured, the generation of responses to user requests is performed based on request-related information extracted from searchable information set to a scope corresponding to the viewing permissions of the user who made the request. As a result, information for which viewing permissions are not set will not be presented to the user, effectively reducing the risk of information leakage such as permission violations and external leaks. Furthermore, since target information for which the user does not have viewing permissions is dynamically excluded from the searchable information according to the user's viewing permissions, the user's affiliation and position are automatically reflected in the searchable information. Contextual control based on affiliation and position is also automatically possible, enabling the presentation of responses appropriate to the user who made the request. Moreover, target information related to business operations dispersed across multiple business applications 3a to 3c can be extracted and aggregated into the search database 90. Since the extraction of target information can be automated, manual file uploading can be eliminated. An automated FAQ system is also possible, improving business efficiency. The percentage of users who can resolve issues themselves increases, the occurrence of inquiries about target information between users is reduced, and the overall business efficiency of the organization is improved.

[0082] Furthermore, in this embodiment, the target information includes conversation logs between users using business applications 3a to 3c. The collaboration processing unit 31 acquires the conversation logs as target information, and the search target determination unit 35 includes conversation logs in the search target information even if the user who made the request did not participate, if they have viewing permission. Chat conversations contain useful information, but when the information is scattered across multiple business applications 3a to 3c, it is difficult to utilize it effectively due to user viewing permission issues. In this regard, with the configuration of this embodiment, the scope of the search target information is set based on viewing permission even if the user did not participate, thus avoiding situations where conversation logs are excluded from the search target information because the user who made the request did not participate, and enabling the generation of appropriate responses according to the user's viewing permission.

[0083] In this embodiment, the user request acquisition unit 34 acquires condition information indicating the conditions for the search target information specified by the user, and the search target determination unit 35 determines the search target information to satisfy the condition information. Since the search database 90 contains a large amount of target information, there is a risk that the necessary target information may not be accessible, and an appropriate answer may not be generated. In this regard, as in this embodiment, by setting condition information using a filter function, the information accessed can be restricted and accuracy can be improved. Therefore, the user's intent can be better reflected in the range of search target information that serves as the source of the answer, and a more appropriate answer can be generated. Also, as in this embodiment, when the user makes a request in free text format, there is a risk that the search accuracy and answer accuracy will decrease, but the range of information to be targeted can be narrowed down by the condition information set by the filter function, and the answer can be controlled in detail.

[0084] Furthermore, in this embodiment, the condition information includes at least one of the following: the period of the target information, the file format, the creator, and the department in the business. By using specific condition information such as the period of the target information, the file format, the creator, and the department in the business, the scope of the search target information can be narrowed, thereby shortening and improving the efficiency of the extraction time (search time) of the target information, and ensuring that request-related information is extracted from an appropriate range. For example, by limiting the creator, information posted by creators who output incorrect information can be excluded from the scope of the search target information.

[0085] Furthermore, the information processing system 100 of this embodiment further includes an evaluation receiving unit 38 that receives user evaluations of the generated answers, and an evaluation reflection unit 39 that reflects the evaluations from the evaluation receiving unit in the determination of search target information, the generation of answers, or both. As a result, feedback is accumulated as users continue to use the information processing system 100 to provide feedback. By reflecting the accumulated feedback in the determination of search target information and the generation of answers, the accuracy of more appropriate answers can be improved.

[0086] <Variation> Next, a modified example with a configuration different from the information processing system 100 of the above embodiment will be described. In the following description, components common to or similar to those of the above embodiment may be denoted by the same reference numerals, and detailed descriptions may be omitted.

[0087] In conventional generative AI applications, mechanisms for proactively taking action based on the "current situation" in business operations are underdeveloped, and optimizing results using a graphical understanding of interpersonal and business relationships remains a technical challenge. Therefore, in the modified example, structured and unstructured data generated across multiple business applications 3a to 3c may be linked and integrated in real time, and knowledge graphs and relationship graphs may be constructed based on the user's authority, relationships, behavioral history, and business context. The knowledge graphs and relationship graphs are then presented to the user.

[0088] The modified information processing system 100 integrates target information acquired from multiple business applications 3a to 3c to construct a knowledge graph and a human relationship graph. Figure 11 is a functional block diagram showing an example of the functional configuration of the information processing device 1a related to the modified example.

[0089] As shown in Figure 11, the information processing device 1a includes a linkage processing unit 31, a viewing permission acquisition unit 32, an information management unit 33, a user request acquisition unit 34, a search target determination unit 35, a request-related information extraction unit 36, a response generation unit 37, an evaluation reception unit 38, an evaluation reflection unit 39, an analysis processing unit 40, a status monitoring unit 41, and a proposal processing unit 42.

[0090] The modified information processing device 1a further includes an analysis processing unit 40, a status monitoring unit 41, and a suggestion processing unit 42, in addition to the functional configuration of the information processing device 1 of the above embodiment.

[0091] The analysis processing unit 40 will now be described. The analysis processing unit 40 constructs a knowledge graph and a human relationship graph based on target information obtained from multiple business applications 3a to 3c. Analysis by the analysis processing unit 40 is performed based on graphing, structuring, or both of the target information. Based on the graphed or structured information, the analysis processing unit 40 generates inferences as evaluations, such as the extraction of future trends and issues. For example, the analysis processing unit 40 extracts knowledge entities and relationships from the target information registered in the search database 90 based on documents, utterances, and project affiliations. It then dynamically evaluates the frequency, influence, and complementary relationships between users. The evaluation by the analysis processing unit 40 may be performed using the language model 5 or by a rule-based algorithm. The analysis processing unit 40 also enables more advanced control of access permissions using dynamic situations and human relationships.

[0092] The analysis results from the analysis processing unit 40 are stored in the search database 90 and managed so that they can be viewed by the user. The timing of the analysis performed by the analysis processing unit 40 is not particularly limited and may be performed at the user's request.

[0093] The status monitoring unit 41 will now be described. The status monitoring unit 41 monitors the user activity history on business applications 3a to 3c and detects predetermined abnormal, stagnant, or insufficient situations. For example, the status monitoring unit 41 detects abnormalities such as "progress stagnation," "insufficient communication," and "review not yet handled" from an event stream that collects, distributes, and processes event data occurring in real time.

[0094] Furthermore, the status monitoring unit 41 may detect predetermined abnormal, stagnant, or deficient situations that take context into account using the language model 5. For example, the status monitoring unit 41 may use the language model 5 to extract conversations, documents, and people related to predetermined abnormal, stagnant, or deficient situations within the context window. The status monitoring unit 41 may also extract user behavior patterns by tracking this information in chronological order. The information extracted by the status monitoring unit 41 is registered in the search database 90.

[0095] The proposal processing unit 42 will now be described. The proposal processing unit 42 uses generative AI such as the language model 5 to propose appropriate people, information, and actions for resolving the situation. The solution information includes the designation of appropriate people for resolving the situation, the information necessary for the resolution, and the actions to resolve the situation. The proposal processing unit 42 generates the solution information based on the evaluation results of the analysis processing unit 40 registered in the search database 90 and the predetermined abnormal, stagnant, or deficient situations that were detected.

[0096] For example, the suggestion processing unit 42 refers to information such as relationship graphs registered in the search database 90 for the language model 5 and prompts the generation of solution information to resolve the detected abnormal, stagnant, or deficient situation. The solution information presents the user with appropriate people, information, and actions to resolve the situation. In the solution information, for example, based on the relationship graph and behavioral context, an "appropriate person" who can resolve the abnormal, stagnant, or deficient situation, such as "someone who can help" or "similar past cases," is explicitly designated.

[0097] The suggestion processing unit 42, for example, interacts with business applications 3a-3c (e.g., Slack) and presents to the user in natural language on those business applications 3a-3c that "This action is currently active." The suggestion processing unit 42 may also automatically send a message, create a document, or initiate a workflow based on the user's selection after being presented with the solution information.

[0098] In another modified version, the evaluation reception unit 38 collects and records feedback information as a log, such as whether the proposal from the proposal processing unit 42 was accepted, rejected, or modified. The evaluation reflection unit 39 then performs model control or search weighting updates to improve the generation accuracy for subsequent generations. For example, the evaluation reflection unit 39 uses the feedback information to perform prompt control, search reranking of the search database 90, etc., to improve the proposal accuracy for subsequent generations.

[0099] As described above, the modified information processing system 100, in addition to the configuration of the above embodiment, further includes a situation monitoring unit 41 that monitors the user's behavior history on business applications 3a to 3c and detects predetermined abnormal, stagnant, or insufficient situations, and a suggestion processing unit 42 that uses the analysis results of the target information to generate solution information for resolving the predetermined abnormal, stagnant, or insufficient situations detected by the situation monitoring unit 41, and presents this information to the user. This enables automatic detection of situations based on the user's behavior and relationships, and allows for real-time or rapid identification of abnormal, stagnant, or insufficient situations. It can also automatically suggest "what to do now" to the user on business applications 3a to 3c and provide assistance in execution. It can also perform sales support such as detecting stalls in business negotiations and having the response generation unit 37 suggest and introduce appropriate superiors or past successes, contract work such as presenting similar past cases and reviewers for contracts that have not yet been reviewed, and onboarding support such as presenting the most frequently asked questions and internal mentor candidates to new employees.

[0100] Furthermore, in the modified version, the evaluation reception unit 38 records whether the proposal is accepted or rejected, and the evaluation reflection unit 39 performs model control or search weighting updates to improve the accuracy of proposal generation in subsequent versions. As a result, from the perspective of outcome learning, accuracy can be effectively and naturally improved from the work log as static training data.

[0101] Although embodiments and modifications of the present invention have been described above, the present invention is not limited to the embodiments and modifications described above, and any modifications, improvements, etc. that can achieve the objectives of the present invention are included in the present invention.

[0102] Furthermore, the series of processes described above can be executed by hardware or by software. In other words, the functional configuration described above is merely illustrative and not particularly limiting. That is, it is sufficient that the information processing device 1 is equipped with the functionality to execute the series of processes described above as a whole, and the type of functional block used to realize this functionality is not particularly limited to the example above. Also, the location of the functional block is not particularly limited and can be arbitrary. For example, the functional block of the information processing device 1 may be transferred to another device, etc. Conversely, the functional block of another device may be transferred to a server, etc. Also, a single functional block may be composed of hardware alone, software alone, or a combination of both.

[0103] When a series of processes are executed by software, the programs that make up that software are installed on a computer or other device from a network or storage medium. The computer may be a computer built into dedicated hardware. Alternatively, the computer may be a computer capable of performing various functions by installing various programs, such as a server, a general-purpose smartphone, or a personal computer.

[0104] Such recording media containing programs may consist not only of removable media (not shown) distributed separately from the main device to provide the programs, but also of recording media provided pre-installed in the main device. Since programs can be distributed via a network, the recording media may be installed on or accessible from a computer connected to or capable of connecting to a network.

[0105] In this specification, the step of describing a program to be recorded on a recording medium includes not only processes that are performed chronologically in that order, but also processes that are not necessarily performed chronologically, but are executed in parallel or individually. Furthermore, in this specification, the term "system" refers to an overall system composed of multiple devices, means, etc. [Explanation of symbols]

[0106] 1.1a Information Processing Device 2 User terminals 3a-3c Business Applications 5 Language Models 31. Interoperability Processing Unit 32 Access Permission Acquisition Section 33 Information Management Department 34. User Request Acquisition Department 35 Search Target Determination Unit 36 Request-Related Information Extraction Unit 37 Answer generation part 38 Evaluation Reception Department 39 Evaluation Reflection Section 90 Search Databases 100 Information Processing Systems

Claims

1. A linkage processing means that interacts with multiple business applications and acquires target information from each of the said business applications, A means for obtaining viewing permission information that indicates the viewing permission set for each user with respect to the aforementioned target information, Information management means for managing the aforementioned target information and the aforementioned access permission information in association with each other, A user request acquisition means for acquiring requests from the aforementioned user, A search target determination means that determines the search target information that will be the scope of the search for the request based on the viewing permission information set for the user who made the request and the target information, A request-related information extraction means for extracting request-related information related to the request from the search target information, A response generation means that performs processing to generate a response to the request to be presented to the user based on the request-related information, An information processing system equipped with the following features.

2. The aforementioned target information includes: This includes conversation logs between users using the aforementioned business application, The aforementioned cooperative processing means is The aforementioned conversation log is acquired as the target information, The search target determination means is, If the user who made the request does not participate in the conversation log, it will also be included in the search information if the user has the necessary viewing permissions. The information processing system according to claim 1.

3. The aforementioned user request acquisition means is Obtain condition information indicating the conditions for the search target information specified by the user, The search target determination means is, The search target information is determined to satisfy the aforementioned condition information. The information processing system according to claim 1.

4. The aforementioned conditional information includes: The aforementioned information includes at least one of the following: period, file format, creator, and department within the business. The information processing system according to claim 3.

5. An evaluation receiving means for receiving user evaluations of the generated response, An evaluation reflection means that reflects the evaluation of the evaluation receiving means in the determination of the search target information, the generation of the response, or both; It also has, An information processing system according to any one of claims 1 to 4.

6. A status monitoring unit monitors the user's activity history on the aforementioned business application and detects predetermined abnormal, stagnant, or insufficient situations. A proposal processing unit generates resolution information for resolving predetermined abnormalities, stagnation, or shortages detected by the status monitoring unit, using the analysis results of the target information to create a large-scale language model, and presents it to the user. The information processing system according to claim 1, further comprising:

7. A linkage processing step that involves coordinating with multiple business applications and obtaining target information from each of the said business applications, A step to obtain viewing permission information that indicates the viewing permissions set for each user with respect to the aforementioned target information, An information management step that manages the aforementioned target information and the aforementioned access permission information in association with each other, A user request acquisition step, which involves obtaining a request from the aforementioned user, A search target determination step, which determines the search target information that will be the scope of the search for the request based on the viewing permission information set for the user who made the request and the target information, A request-related information extraction step, which extracts request-related information related to the request from the search target information, A response generation step that performs processing to generate a response to the request to be presented to the user based on the request-related information, A control method for an information processing device, including the device itself.

8. A linkage processing step that involves coordinating with multiple business applications and obtaining target information from each of the said business applications, A step to obtain viewing permission information that indicates the viewing permissions set for each user with respect to the aforementioned target information, An information management step that manages the aforementioned target information and the aforementioned access permission information in association with each other, A user request acquisition step, which involves obtaining a request from the aforementioned user, A search target determination step, which determines the search target information that will be the scope of the search for the request based on the viewing permission information set for the user who made the request and the target information, A request-related information extraction step, which extracts request-related information related to the request from the search target information, A response generation step that performs processing to generate a response to the request to be presented to the user based on the request-related information, A program that causes a computer to execute something.

Images