Document utilization service provision system, method, and program
The document utilization service system addresses credibility issues in generative AI chatbots by integrating a cloud server to display answers with relevant document portions and keywords, allowing users to verify the AI-generated content's reliability.
Patent Information
- Authority / Receiving Office
- JP · JP
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2024-09-20
- Publication Date
- 2026-04-02
AI Technical Summary
Users of chatbots utilizing generative AI face challenges in correctly judging the credibility of answers due to the AI-generated content.
A document utilization service provision system that integrates with a cloud server equipped with generative AI, providing a user interface to display answers alongside the relevant document portions and search keywords, enabling users to verify the credibility of the answers.
Enables users to accurately assess the credibility of AI-generated answers by displaying the source document sections and search keywords, thereby enhancing trust in the answer generation process.
Smart Images

Figure 2026056714000001_ABST
Abstract
Description
Technical Field
[0001] Embodiments of the present invention relate to a document utilization service providing system, method, and program.
Background Art
[0002] In recent years, dialogue systems such as chatbots have begun to spread. In addition, generative AI (Generative AI), which uses deep learning to learn a vast amount of training data (text, images, voices, etc.) on the web and generates new content based on the patterns, has attracted attention.
[0003] By applying a chatbot using this generative AI, for example, based on a large number of documents accumulated in various departments within a company as the source, a system can be constructed that, when an employee simply asks a question in natural language, can clearly summarize the information the employee wants to know and answer it.
Prior Art Documents
Patent Documents
[0004]
Patent Document 1
Patent Document 2
Summary of the Invention
Problems to be Solved by the Invention
[0005] However, unlike a scenario-based chatbot where answers are prepared in advance, in the case of a chatbot using generative AI, since the AI creates the answers, some questioners feel uneasy about the credibility of the answers, and it has been difficult for the questioners to correctly judge the credibility of the answers.
[0006] One embodiment of the present invention provides a document utilization service provision system, method, and program that enables questioners to correctly judge the credibility of answers in a chatbot utilizing generation AI. [Means for solving the problem]
[0007] According to the embodiment, the document utilization service provision system is a system that provides a service to support the utilization of documents by cooperating with a cloud server having a generating AI that accepts questions in a dialogue format and answers the questions based on pre-provided documents. The system comprises a user interface unit, a dialogue mediation unit, a screen editing unit, and a document storage unit. The user interface unit transmits screen data for the user interface, which includes a first area for inputting questions, a second area for displaying questions and answers in chronological order in a way that suggests a dialogue, and a third area for displaying the relevant part of the document that served as the basis for the answer, to the user terminal and receives input information on the screen from the user terminal. The dialogue mediation unit transmits the questions entered in the first area of the screen on the user terminal to the generating AI on the cloud server and receives the answers to the questions from the generating AI on the cloud server. The screen editing unit places the answers in the second area of the screen. The document storage unit stores the documents. The dialogue mediation unit receives the search keywords used to search for the relevant portion of the document that formed the basis of the answer, and index information indicating the relevant portion of the document that formed the basis of the answer, along with the answer, from the generating AI on the cloud server. The screen editing unit places the search keywords along with the answer in the second area of the screen, retrieves the relevant portion of the document indicated by the index information from the document storage unit and places it in the third area of the screen, and sets the search keywords within the relevant portion of the document to be highlighted. [Brief explanation of the drawing]
[0008] [Figure 1] A diagram showing one example configuration of a document utilization service provision system according to the embodiment. [Figure 2] This figure shows an example of a user interface screen (usage phase) displayed on the browser of a user terminal that accesses the document utilization service provision system of the embodiment. [Figure 3] Figure 2 shows an example of a screen that appears in the browser in response to a keyword being entered on the screen. [Figure 4] Figure 3 shows an example of a screen that appears in the browser when a question candidate is selected on the screen. [Figure 5] Figure 4 shows an example of a screen that appears in the browser in response to an operation performed on the document name display area on the screen. [Figure 6] A flowchart illustrating the operating procedure of the document utilization service provision system according to the embodiment. [Modes for carrying out the invention]
[0009] The embodiments will be described below with reference to the drawings.
[0010] Figure 1 shows an example configuration of a document utilization service provision system 1 according to an embodiment. Figure 1 also shows how the document utilization service provision system 1 interacts with the cloud server 2 and how it communicates with the user terminal 3.
[0011] The document utilization service provision system 1 is a system that provides services to support the utilization of documents in cooperation with a cloud server 2 having a generation AI 21. The document utilization service provision system 1 is built on an information processing device equipped with a CPU 100 and a storage device 200. The document utilization service provision system 1 may also provide services to user terminals 3 as a cloud server.
[0012] Cloud Server 2 is an information processing device that provides cloud services such as SaaS (Software as a Service), IaaS (Infrastructure as a Service), and PaaS (Platform as a Service). Here, it is assumed that Cloud Server 2 has a Generative AI 21 that accepts questions in an interactive format and answers questions based on pre-provided documents, and that it provides a service that leases the functionality of this Generative AI 21. Cloud Server 2 processes and stores the given documents as a knowledge base 22. For example, Cloud Server 2 divides the document into predetermined units such as sections, paragraphs, and character counts, and creates a search index for each predetermined unit. This search index is also information that allows the Document Utilization Service Provisioning System 1 to retrieve the relevant part from the documents stored in the document storage unit 51, which will be described later. The pre-provided documents that are stored as the knowledge base 22 are, for example, a large amount of documents accumulated by a company that uses the Document Utilization Service Provisioning System 1.
[0013] User terminal 3 is an information processing device such as a PC (Personal Computer) or smartphone capable of running a browser 31. The user accesses the document utilization service provision system 1 via the browser 31 on user terminal 3.
[0014] The document utilization service provision system 1 includes a user interface unit 11, a screen editing unit 12, a knowledge registration unit 13, a dialogue mediation unit 14, and a data unit for the document storage unit 51. These processing units are implemented by the CPU 100 executing programs stored in the storage device 200. These processing units may also be implemented by hardware such as electrical circuits. The data unit is created within the storage device 200.
[0015] The user interface unit 11 communicates with the browser 31 of the user terminal 3, for example, using HTTPS (HyperText Transfer Protocol over transport Security). Specifically, in response to a request from the browser 31, the user interface unit 11 sends a screen for the user interface, created as an HTML (HyperText Markup Language) file, to the browser 31. The user interface unit 11 also receives user input information from the browser 31 on the screen for the user interface on the user terminal 3.
[0016] The screen editing unit 12 edits the user interface screen (HTML file) that should be sent to the browser 31 of the user terminal 3 by the user interface unit 11. The screen editing unit 12 determines the user interface screen to be presented to the user according to the user input information received from the browser 31 by the user interface unit 11. In other words, the document utilization service provision system 1 can appropriately transition the screens displayed in the browser 31 according to the user input by having the user interface unit 11 and the screen editing unit 12 work together.
[0017] In cases where a user accesses the document utilization service provision system 1 via the browser 31 on user terminal 3, there are broadly two phases: a preparation phase in which a document is registered, and a utilization phase in which answers are obtained by asking questions based on the document as a source. The user can select between the preparation phase and the utilization phase, for example, by inputting on a user interface screen displayed on the browser 31. The access destination (URL: Uniform Resource Locator) to the document utilization service provision system 1 may also be changed between the preparation phase and the utilization phase.
[0018] In the preparation phase, the knowledge registration unit 13 stores the user's document received by the user interface unit 11 from the browser 31 in the document storage unit 51 and registers it in the cloud server 2. The transmission and reception of data between the cloud server 2 by the knowledge registration unit 13 and the dialogue mediation unit 14 described later is performed by communication compliant with, for example, TCP (Transmission Control Protocol) / IP (Internet Protocol) under the control of a communication unit that controls communication via the Internet, which is not shown in FIG. 1.
[0019] The document utilized by the document utilization service providing system 1 of the present embodiment to receive a registration request from a user is, for example, a document accessible from a Web page publicly available on an intranet for employees by the company to which the user belongs. The document utilization service providing system 1 receives, separately from the registration of the document, a list of addresses of documents embedded in the aforementioned Web page from the browser 31 of the user terminal 3. The knowledge registration unit 13 stores this address in the document storage unit 51 together with the document.
[0020] Further, the document utilization service providing system 1 receives, as optional registration information, the storage address of the document in, for example, a file server on the intranet where the original of the document publicly available on the aforementioned Web page is stored. When the storage address of the original is transmitted from the browser 31 of the user terminal 3 together with the document, the knowledge registration unit 13 also stores this storage address of the original in the document storage unit 51.
[0021] In the utilization phase, the dialogue mediation unit 14 transmits the input information of the user received by the user interface unit 11 from the browser 31, specifically, the question, to the generation AI 21 of the cloud server 2. When the dialogue mediation unit 14 receives an answer generated for the question from the generation AI 21 of the cloud server 2, it transmits the answer to the screen editing unit 12.
[0022] This answer includes the keywords used by the generating AI 21 to search the document, and index information (the aforementioned search index) of the relevant parts of the document (predetermined units such as sections and paragraphs) that were found using these keywords and that formed the basis of the answer. Based on this attached information, the screen editing unit 12 edits a user interface screen for presenting the generating AI 21's answer in a format that allows the user to correctly judge its credibility. Details of the user interface screen that the document utilization service provision system 1 of this embodiment provides to the user for presenting the generating AI 21's answer will be described later.
[0023] Furthermore, some users may not know what questions to ask to obtain the information they want. For such users, Generation AI21 has a function that generates and returns suggested questions related to a given keyword. In other words, users can also enter keywords instead of questions.
[0024] The dialogue mediation unit 14 transmits user input information received by the user interface unit 11 from the browser 31 to the generating AI 21 on the cloud server 2, regardless of whether it is a question or a keyword. The dialogue mediation unit 14 also transmits the answer or question candidate received from the generating AI 21 to the screen editing unit 12. The screen editing unit 12 determines whether it is an answer or a question candidate, for example, based on the accompanying information. If the screen editing unit 12 determines that it is a question candidate, it edits a user interface screen for presenting the question candidate generated by the generating AI 21 in a form that allows the user to input the question candidate as a question. Details of the user interface screen that the document utilization service provision system 1 of this embodiment provides to the user for presenting the question candidate from the generating AI 21 will also be described later.
[0025] Figure 2 shows an example of a user interface screen 300 (usage phase) displayed on the browser 31 of a user terminal 3 that accesses the document utilization service provision system 1. As mentioned above, this screen 300 is edited by the screen editing unit 12 of the document utilization service provision system 1. Hereafter, all user interface screens 300 displayed on the browser 31 of the user terminal 3 are edited by the screen editing unit 12 based on user input information and information from the generated AI 21.
[0026] As shown in Figure 2, screen 300 includes a search term input area a1, a chat area a2, and a source document area a3.
[0027] The search term input area a1 is where the user enters their question. The search term input area a1 also contains a free word input area a11 and a submit button a12.
[0028] The free-word input area a11 is an area where the user enters keywords or questions using a keyboard or other device. The send button a12 is a button that, when operated with a pointing device such as a mouse, instructs the system to send the keywords or questions entered in the free-word input area a11 to the document utilization service provision system 1.
[0029] Furthermore, when the generating AI 21 generates and returns a question candidate, the screen 300 displayed in the browser 31 has the question candidate set as an option in the free word input area a11 of the search term input area a1. In other words, the image editing unit 12 edits screen 300 (HTML file) so that the question candidate is set as an option in the free word input area a11 of the search term input area a1.
[0030] Chat area a2 is an area where the questions (including keywords) entered and sent by the user in the free word input area a11, and the responses from the generating AI21, are displayed chronologically in a way that resembles a dialogue.
[0031] The source document area a3 is the area where the relevant portion of the document that formed the basis of the generated AI21's answer is displayed. The display of the relevant portion of the document in this source document area a3 will be described later.
[0032] Figure 3 shows an example of the screen 300 displayed in the browser 31 of the user terminal 3 after the keyword "optional service" is entered into the free word input area a11 of the search term input area a1 on the screen 300 of Figure 2, and the send button a12 of the same search term input area a1 is pressed.
[0033] Here, in response to the input of the keyword "optional service," the generating AI21 of cloud server 2 will: "What exactly are the optional services?" • "What are the main industries that offer optional services?" • "What are the advantages of optional services?" Let's assume that the following question candidates are generated. The method by which Generation AI21 generates question candidates related to the keyword input is not considered here.
[0034] The keyword "optional service" entered in the free word input area a11 will be displayed in the chat area a2 (b1) according to the settings in the HTML file, upon operation of the send button a12. Consequently, the free word input area a11 will become blank, allowing the user to enter text.
[0035] When a question candidate is sent from the generating AI 21, the screen editing unit 12 displays the question candidate as an option in the search text input area a1 (b2). The user may choose to use the question candidate displayed as an option, or may enter a separate question in the free word input area a11.
[0036] Figure 4 shows an example of the screen 300 that appears on the browser 31 of the user terminal 3 after selecting the question candidate "What exactly are optional services?" on screen 300 of Figure 3, entering it into the free word input area a11, and operating the send button a12.
[0037] As described above, in response to the operation of the send button a12, the question "What exactly are optional services?" entered in the free word input area a11 is displayed in the chat area a2 (c1).
[0038] Here, we assume that the generating AI 21 on cloud server 2 generated the answer shown by symbol C2 in Figure 4 in response to this question. This answer is not merely an excerpt from a document, but is generated by generating AI 21 based on one or more documents. The method by which generating AI 21 generates the answer to the question is not considered here. The generating AI 21's answer is accompanied by the keywords that generating AI 21 used to search the documents, and index information of the relevant parts of the documents that were searched using these keywords and that served as the basis for the answer.
[0039] When the screen editing unit 12 places the response (c2) of the generating AI 21 into the chat area a2, it attaches keywords, which are supplementary information, to the response along with an object such as "Keywords used for search:" (c3). Note that when the screen editing unit 12 places the response (c2) of the generating AI 21 into the chat area a2, it means that the HTML file is edited so that the response (c2) of the generating AI 21 is displayed in the chat area a2 when screen 300 is displayed on the browser 31 of the user terminal 3.
[0040] This allows the user to first understand that the document that formed the basis for the answer to the question, "What exactly are optional services?" was searched using the keywords "optional services," "definition," "content," and "details." The method by which Generative AI21 determines the search keywords for the document from the question is not discussed here.
[0041] Furthermore, the screen editing unit 12 retrieves the relevant portion of the document corresponding to the index information attached to the response of the generating AI 21 from the document storage unit 51 and places it in the source document area a3 of screen 300 (c4[c4-1, c4-2, c4-3]). The placement of this relevant portion of the document in the source document area a3 is also achieved by editing the HTML file for screen 300.
[0042] When the screen editing unit 12 places the relevant portion of the document into the source document area a3, it searches for the search keyword within that portion of the document. If the search keyword is found, the screen editing unit 12 sets the location of the search keyword in the relevant portion of the document to be highlighted (c5[c5-1, c5-2, c5-3]). This highlighting setting is also achieved by editing the HTML file for screen 300.
[0043] This allows users to refer to the relevant portion of the document that served as the basis for the answer (c2) displayed in the chat area a2 in the source document area a3, and to easily identify the sections within that document that contain the search keywords through highlighting.
[0044] Thus, the document utilization service provision system 1 of this embodiment allows users to refer to the document on which the generated AI 21's answer is based, along with the search keywords highlighted, enabling users to correctly judge the credibility of the generated AI 21's answer.
[0045] Furthermore, as mentioned above, the document storage unit 51 stores not only the document itself, but also an address that allows employees to access that document from a web page on the company network. When the screen editing unit 12 places the relevant portion of the document in the source document area a3, it places the document name (c6) and embeds the aforementioned address in association with the location where the document name is placed. The document name is the file name of the document stored as a file in the document storage unit 51.
[0046] This allows the user to open a document and view the entire document by manipulating the document name (c6) on the screen 300 displayed by the browser 31 of the user terminal 3 using a pointing device such as a mouse. Figure 5 shows an example in which the document (d1) is displayed in a separate window by manipulating the display area of the document name (c6) on the screen 300.
[0047] Furthermore, as mentioned above, the document storage unit 51 stores the storage address of the original document as arbitrary registration information. If the storage address of the original document is stored, the screen editing unit 12 places an object such as (Latest Display) (c7 in Figure 4) after the document name (c6 in Figure 4), and embeds the aforementioned storage address in association with the placement of this (Latest Display) object.
[0048] For example, if there is a delay in updating a document to the company's internal webpage after an update, the document displayed by manipulating the document name with a mouse or other pointing device will be the pre-update version. By using the "Show Latest" function with a mouse or other pointing device, users can always access the latest version of the document.
[0049] Figure 6 is a flowchart showing the operation procedure of the document utilization service provision system 1 of this embodiment.
[0050] The document utilization service provider system 1 sends an HTML file of the user interface screen 300 (initial screen) to the browser 31 of the user terminal 3 (S101). After sending the HTML file, the document utilization service provider system 1 waits to receive a keyword or question from the browser 31 of the user terminal 3 (S102). If neither a keyword nor a question is received (S102: NO), the document utilization service provider system 1 determines whether a timeout has been detected (S103). If a timeout is detected (S103: YES), the document utilization service provider system 1 terminates processing. If a timeout is not detected (S103: NO), the document utilization service provider system 1 continues to wait.
[0051] If a keyword or question is received (S102: YES), the document utilization service provider system 1 sends the received keyword or question to the generating AI 21 of the cloud server 2 (S104). The document utilization service provider system 1 then receives a candidate question related to the keyword or an answer to the question from the generating AI 21 of the cloud server 2 (S105).
[0052] If the cloud server 2 receives a candidate question from the generating AI 21 (S106: NO), the document utilization service provider system 1 sends an HTML file of the screen containing the candidate question to the browser 31 of the user terminal 3 (S107). In this case, the document utilization service provider system 1 returns to step S102 and waits to receive a keyword or question from the browser 31 of the user terminal 3.
[0053] On the other hand, if the system receives a response from the AI 21 generated by the cloud server 2 (S106:YES), the document utilization service provider system 1 first searches for the search keyword in the relevant section of the source document and sets the highlighting (S108). The search keyword is attached to the response as supplementary information. Then, the document utilization service provider system 1 sends the received response and an HTML file of the user interface screen 300, which contains the relevant section of the source document with the search keyword highlighted, to the browser 31 of the user terminal 3 (S109), and terminates the process.
[0054] Incidentally, the document utilization service provision system 1 of this embodiment can be used not only as a tool for searching documents, but also as a tool for summarizing documents by linking with the AI 21 generated by the cloud server 2. Thus, the document utilization service provision system 1 of this embodiment is not simply a tool for searching documents.
[0055] As described above, in the document utilization service provision system 1 of the embodiment, the relevant portion of the document that served as the basis for the generated AI 21's answer is displayed along with the answer, and the search keywords within that portion of the document are highlighted.
[0056] As a result, the document utilization service provision system 1 of the embodiment enables the user to correctly judge the credibility of the answers generated by the AI 21.
[0057] While several embodiments of the present invention have been described, these embodiments are presented as examples only and are not intended to limit the scope of the invention. These novel embodiments can be carried out in a variety of other forms, and various omissions, substitutions, and modifications can be made without departing from the spirit of the invention. These embodiments and their variations are included in the scope and spirit of the invention, as well as in the claims and their equivalents. [Explanation of Symbols]
[0058] 1...Document utilization service provision system, 2...Cloud server, 3...User terminal, 11...User interface section, 12...Screen editing section, 13...Knowledge registration section, 14...Dialogue mediation section, 21...Generative AI, 22...Knowledge base, 31...Browser, 51...Document storage section, 100...CPU, 200...Storage device, a1...Search term input area, a2...Chat area, a3...Source document area, a11...Free word input area, a12...Send button.
Claims
1. A system that provides a service to support the use of documents by collaborating with a cloud server having a generative AI that accepts questions in an interactive format and answers those questions based on pre-provided documents, A user interface unit transmits data of a user interface screen to a user terminal, which includes a first area for inputting a question, a second area for displaying the question and its answer in chronological order in a way that suggests a dialogue, and a third area for displaying the relevant portion of the document that served as the basis for the answer, and receives input information on the screen from the user terminal. A dialogue mediation unit that transmits the question entered in the first area of the screen on the user terminal to the generating AI on the cloud server, and receives the answer to the question from the generating AI on the cloud server, A screen editing unit that places the aforementioned answer in the second area of the screen, A document storage unit for storing the aforementioned documents, It is equipped with, The dialogue mediation unit receives, along with the answer, the search keywords used to search for the relevant portion of the document that formed the basis of the answer, and index information indicating the relevant portion of the document that formed the basis of the answer, from the generating AI on the cloud server. The aforementioned screen editing department, The search keywords are placed together with the answer in the second area of the screen. The system retrieves the relevant portion of the document indicated by the index information from the document storage unit and places it in the third area of the screen, and sets the system to highlight the search keyword within the relevant portion of the document. A system for providing document utilization services.
2. The first area of the screen allows input of keywords for causing the AI to generate the question, The dialogue mediation unit can receive candidate questions generated based on the keywords from the generating AI on the cloud server. When the screen editing unit receives the candidate questions from the generating AI on the cloud server, it places the candidate questions as options in the first area of the screen. A document utilization service provision system according to claim 1.
3. The document storage unit stores address information along with the document for acquiring and displaying the entire document, which is the portion of which is displayed in the third area of the screen, on the user terminal. The screen editing unit places a string representing the document name in the third area along with the corresponding portion of the document, associating it with the address information. A document utilization service provision system according to claim 1.
4. The aforementioned screen editing unit places operation buttons on the screen for inputting a pre-set string into the first area. A document utilization service provision system according to claim 1.
5. The highlighting of the search keywords in the relevant section of the aforementioned document is called highlighting. A document utilization service provision system according to claim 1.
6. A method implemented by a system that provides a service to support the use of documents, in cooperation with a cloud server having a generative AI that accepts questions in an interactive format and answers the questions based on pre-provided documents, The system transmits screen data for a user interface, which includes a first area for inputting a question, a second area for displaying the question and its answer in chronological order in a way that suggests a dialogue, and a third area for displaying the relevant portion of the document that served as the basis for the answer, to the user terminal, and receives input information on the screen from the user terminal. The user terminal transmits the question entered in the first area of the screen to the generating AI on the cloud server, and receives the answer to the question from the generating AI on the cloud server. The answer is placed in the second area of the screen. The search keywords used to search the relevant portion of the document that formed the basis of the aforementioned answer, and the index information indicating the relevant portion of the document that formed the basis of the aforementioned answer, are received from the generating AI on the cloud server along with the aforementioned answer. The search keywords are placed together with the answer in the second area of the screen. The relevant portion of the document indicated by the index information is placed in the third area of the screen, and settings are made to highlight the search keyword within the relevant portion of the document. method.
7. A computer for building a system that provides a service to support the use of documents, in cooperation with a cloud server having a generative AI that accepts questions in an interactive format and answers those questions based on pre-provided documents, A user interface unit transmits data of a user interface screen to a user terminal, which includes a first area for inputting a question, a second area for displaying the question and its answer in chronological order in a way that suggests a dialogue, and a third area for displaying the relevant portion of the document that served as the basis for the answer, and receives input information on the screen from the user terminal. A dialogue mediation unit that transmits the question entered in the first area of the screen on the user terminal to the generating AI on the cloud server, and receives the answer to the question from the generating AI on the cloud server. A screen editing unit that places the aforementioned answer in the second area of the screen, Document storage unit for storing the aforementioned documents, It is a program that makes it function as such. The dialogue mediation unit receives, along with the answer, the search keywords used to search for the relevant portion of the document that formed the basis of the answer, and index information indicating the relevant portion of the document that formed the basis of the answer, from the generating AI on the cloud server. The aforementioned screen editing department, The search keywords are placed together with the answer in the second area of the screen. The system retrieves the relevant portion of the document indicated by the index information from the document storage unit and places it in the third area of the screen, and sets the system to highlight the search keyword within the relevant portion of the document. program.
Citation Information
Patent Citations
Document output system, document output method, and document output program
JP2020144674A
Interface providing device, interface providing method, and program
JP2021131594A