Information processing system, information processing device, information processing method, and program
The information processing system addresses the limitation of existing document generation techniques by using a conversational AI agent and machine learning to create and modify document structures, ensuring user-specific document generation and refinement.
Patent Information
- Authority / Receiving Office
- JP · JP
- Patent Type
- Applications
- Current Assignee / Owner
- RICOH CO LTD
- Filing Date
- 2024-12-19
- Publication Date
- 2026-07-01
AI Technical Summary
Existing document generation techniques, such as those described in Patent Document 1, do not consider document modification, limiting their effectiveness in generating documents tailored to user specifications.
An information processing system utilizing a conversational AI agent that interacts with users to create and modify document structures based on user input, employing machine learning models like Large Language Models (LLM) to generate and refine document content, and a graph database to provide information for modifications.
Enables the generation of documents that accurately reflect user intent by allowing for interactive document structure creation and modification, resulting in a final document that meets user-specific requirements.
Smart Images

Figure 2026109068000001_ABST
Abstract
Description
Technical Field
[0004] ,
[0006] , , , , , , , , , , , ,
[0005] , [Figure 1] , ,
[0007] ,
[0001] The present invention relates to an information processing system, an information processing apparatus, an information processing method, and a program.
Background Art
[0002] Conventionally, techniques for generating new documents using existing document data have been studied. For example, Patent Document 1 discloses a technique for generating the gist of a patent document related to a new invention using existing patent documents.
Summary of the Invention
Problems to be Solved by the Invention
[0003] However, the technique of Patent Document 1 can generate patent documents but does not consider document modification.
[0004] This figure shows an example of the configuration of an information processing system according to an embodiment of the present invention. [Figure 2] This figure shows an example of the hardware configuration of the information processing device 10 in an embodiment of the present invention. [Figure 3] This figure shows an example of the functional configuration of an information processing system in an embodiment of the present invention. [Figure 4] This is a sequence diagram illustrating an example of a processing procedure performed by the information processing device 10. [Figure 5] This is a sequence diagram illustrating an example of a processing procedure performed by the information processing device 10. [Figure 6] This is a sequence diagram illustrating an example of a processing procedure performed by the information processing device 10. [Figure 7] This figure shows an example of how the dialogue screen will look. [Figure 8] This figure shows an example of the display of the dialogue screen after entering summary information. [Figure 9] This figure shows an example of how to display a document structure. [Figure 10] This figure shows an example of the interactive screen displayed after entering correction instructions for the document structure. [Figure 11] This figure shows an example of the revised document structure. [Figure 12] This figure shows an example of a message indicating that no changes to the document structure are necessary. [Figure 13] This figure shows an example of how the target draft is displayed. [Figure 14] This diagram illustrates the specification of the scope of revisions to the target draft. [Figure 15] This figure shows an example of the interactive screen displayed after entering the correction details for the correction range. [Figure 16] This figure shows an example of how an information map is displayed. [Figure 17] This figure shows an example of a state where a node in the information map is selected. [Figure 18] This figure shows the first example of operations on an information map. [Figure 19]This figure shows a second example of operations on an information map. [Figure 20] This figure shows an example of how the revised draft will appear. [Figure 21] This figure shows an example of a message indicating that no correction is needed. [Figure 22] This figure shows an example of the interactive screen displayed at the end of document generation. [Modes for carrying out the invention]
[0008] Embodiments of the present invention will be described below with reference to the drawings. Figure 1 is a diagram showing an example of the configuration of an information processing system in an embodiment of the present invention. In Figure 1, one or more terminals 20 are connected to the information processing device 10 via a network such as a LAN (Local Area Network) or the Internet.
[0009] The information processing device 10 is one or more computers having software (hereinafter referred to as "agents") that interacts with the user. An agent is an example of a conversational AI, and is a personified virtual entity that the user sees as a conversation partner. Interaction with the conversational AI (conversation partner) means that when the user inputs a message, a response is output in response. In this embodiment, the agent assists the user in creating a document desired by the user based on the interaction with the user. For example, the agent receives information from the user indicating an outline of the document to be created (for example, information indicating what kind of document the user wants to create) and generates the structure of the document (for example, chapters or table of contents, etc.; hereinafter referred to as "document structure"). The agent presents the generated document structure to the user and modifies the document structure according to the user's instructions. When the user determines that no further modifications are needed to the document structure, the agent automatically generates a document according to the document structure using a pre-prepared set of document data. The first version of the document does not necessarily contain the content the user intended. Therefore, the agent accepts modification instructions for the generated document from the user. The agent obtains information related to the correction instructions from a set of document data and presents the obtained information to the user. The agent then makes corrections to the generated document based on the information selected by the user from the presented information set and the correction instructions. In this embodiment, "document" refers to an electronically recorded document, but its data format is not limited to a specific one. The agent may also be called an automated response means, AI agent, digital clone, personalized AI, AI assistant, automated response AI, conversation partner, AI chatbot, companion, concierge, or virtual dialogue interface. The agent may also be a virtual human displayed as a conversation partner on the screen of terminal 20 as a 3D avatar modeled after a person.
[0010] The terminal 20 is a device that functions as a user interface of the information processing system. For example, a PC (Personal Computer), a smartphone, a tablet terminal, or the like may be used as the terminal 20. The terminal 20 receives an input from a user and transmits information (hereinafter referred to as "input information") corresponding to the input to the information processing apparatus 10. The terminal 20 also receives and displays information generated for the input information from the information processing apparatus 10.
[0011] In addition, in the present embodiment, it is assumed that the information processing system is operated in a certain company (hereinafter referred to as "Company X"). Therefore, users who can access the information processing apparatus 10 are persons belonging to Company X, such as employees of Company X. However, the service provided by the information processing apparatus 10 may be generally made public as a cloud service.
[0012] FIG. 2 is a diagram showing a hardware configuration example of the information processing apparatus 10 in the embodiment of the present invention. As shown in FIG. 2, the information processing apparatus 10 is constructed by a computer. As shown in FIG. 2, it includes a CPU (Central Processing Unit) 101, a ROM (Read Only Memory) 102, a RAM (Random Access Memory) 103, an HD (Hard Disk) 104, an HDD (Hard Disk Drive) controller 105, a display 106, an external device connection I / F (Interface) 108, a network I / F 109, a data bus 110, a keyboard 111, a pointing device 112, a DVD-RW (Digital Versatile Disk Rewritable) drive 114, and a media I / F 116.
[0013] Of these, the CPU 101 controls the operation of the entire information processing unit 10. The ROM 102 stores programs used to drive the CPU 101, such as the IPL (Initial Program Loader). The RAM 103 is used as the work area for the CPU 101. The HD 104 stores various data such as programs. The HDD controller 105 controls the reading or writing of various data to the HD 104 according to the control of the CPU 101. The display 106 displays various information such as cursors, menus, windows, characters, or images. The external device connection I / F 108 is an interface for connecting various external devices. In this case, external devices include, for example, USB (Universal Serial Bus) memory and printers. The network I / F 109 is an interface for data communication using a communication network. The data bus 110 is an address bus and data bus, etc., for electrically connecting each component such as the CPU 101 shown in Figure 2.
[0014] The keyboard 111 is a type of input means equipped with multiple keys for inputting characters, numbers, and various instructions. The pointing device 112 is a type of input means for selecting and executing various instructions, selecting processing targets, and moving the cursor. The DVD-RW drive 114 controls the reading or writing of various data to the DVD-RW 113, which is an example of a removable recording medium. Note that it is not limited to DVD-RW, but may also be DVD-R, etc. The media I / F 116 controls the reading or writing (storage) of data to the recording medium 115, such as flash memory.
[0015] Figure 3 shows an example of the functional configuration of an information processing system in an embodiment of the present invention. In Figure 3, the information processing device 10 has, as functional units constituting an agent, a reception unit 11, a document generation unit 12, a conversion unit 13, a search unit 14, an AI control unit 15, a modified usage information acquisition unit 16, and a display control unit 17. The information processing device 10 also has, as storage units constituting an agent, a data storage unit 121 and a graph storage unit 122. Furthermore, the information processing device 10 has an AI 150.
[0016] Each part of the agent as a functional unit and the AI 150 are realized by processing that one or more programs installed on the information processing device 10 cause the CPU 101 to execute. Each part of the agent as a storage unit can be realized using, for example, the HD 104 or a storage device that can be connected to the information processing device 10 via a network.
[0017] AI150 is a machine learning model (e.g., a neural network) trained to take text as input and generate text corresponding to that text (hereinafter referred to as the "prompt") (hereinafter referred to as the "response"). For example, AI150 generates the text with the highest probability of occurrence as a response to a prompt, based on the learning results. For example, a generative AI using a Large Language Mode (LLM) may be used as AI150. LLM is a machine learning model trained on natural language processing using a large amount of text data, and is used in many NLP tasks such as generating responses to specific questions, automatic text generation, text summarization, translation, and sentiment analysis. It can also be used in various applications such as education, entertainment, customer service, and product development. In this embodiment, the prompt is text containing a message entered by the user. Note that the information processing device 10 does not necessarily have AI150. In this case, a generative AI publicly available on the internet or elsewhere may be used as AI150.
[0018] Here, machine learning is a technique for enabling computers to acquire human-like learning abilities. It refers to a technique in which a computer autonomously generates algorithms necessary for judgment, such as data identification, from pre-inputted training data, and applies these algorithms to new data to make predictions. The learning method for machine learning can be supervised learning, unsupervised learning, semi-supervised learning, reinforcement learning, or deep learning, or a combination of these learning methods; the learning method for machine learning is not restricted.
[0019] The reception unit 11 receives input from the user. For example, the reception unit 11 receives summary information from the user that shows an outline of the document the user desires. The reception unit 11 also receives correction instructions for the document structure generated by the document generation unit 12 based on the summary information. The reception unit 11 also receives instructions from the user to generate a document based on the document structure specified by the user (the revised document structure if there are correction instructions, or the original document structure if there are no correction instructions). The reception unit 11 also receives correction instructions, etc., for the document generated by the document generation unit 12 in response to the document generation instructions. More precisely, the user input is made to the terminal 20. Therefore, the reception unit 11 receives information from the terminal 20 corresponding to the input received by the terminal 20.
[0020] The document generation unit 12 generates a document with a predetermined document structure using predetermined document data based on interaction with the user, and also modifies the document based on interaction with the user. The predetermined document structure is, for example, a document structure specified by the user. In this embodiment, the document generation unit 12 first generates a document structure based on summary information received by the reception unit 11, and generates a document with that document structure when the user instructs it to generate a document based on that document structure. The document generation unit 12 also makes modifications to the generated document structure or the document based on the user's modification instructions for the generated document. In this embodiment, a modification instruction for a document includes specifying the scope of modification and specifying the content of the modification. The scope of modification refers to the range of the document to be modified, and the content of the modification is information indicating how to modify the scope of modification.
[0021] The conversion unit 13 converts a string representing the document structure into a vector that represents the meaning of the string using multi-dimensional numerical values (hereinafter referred to as a "semantic vector"). The semantic vector can be generated using natural language processing such as BERT.
[0022] The search unit 14 searches a pre-prepared set of document data for document data related to a string indicating a document structure specified by the user. The set of document data is stored in the data storage unit 121. For example, the data storage unit 121 pre-stores document data related to various operations of company X. Each employee of company X may upload document data to the data storage unit 121 at any time. For each such document data, the data storage unit 121 stores the document data and the semantic vector for each chunk of the document data. A chunk is a part of the document data obtained by dividing the document data into predetermined units. The units for dividing the document data may be the number of characters, the number of sentences, or semantic units (e.g., paragraphs), and are not limited to predetermined ones. Hereinafter, the semantic vector of each chunk will be referred to as a "chunk vector". In addition, the set of document data may be stored in the data storage unit 121, separated into folders by field, etc.
[0023] The search unit 14 calculates the similarity between the semantic vector of the string representing the document structure and the chunk vector of each chunk related to the document data for each document data, and identifies the chunk related to the chunk vector with the highest similarity for the document data (hereinafter referred to as "similar chunk"). The search unit 14 compares the similarity of the similar chunks for each document data and extracts the top N similar chunks. Therefore, N document data are effectively extracted. Cosine similarity may be used to evaluate the similarity between vectors, or other metrics may be used.
[0024] The AI control unit 15 responds to requests from the document generation unit 12 by sending prompts to the AI 150 in response to those requests and receiving responses from the AI 150 to those prompts. For example, the AI control unit 15 sends a prompt to the AI 150 indicating an instruction to generate a document structure based on summary information and receives a response (document structure) to that prompt. The AI control unit 15 also sends a prompt to the AI 150 indicating an instruction to make corrections in response to a correction request for the document structure and receives a response (corrected document structure) in response to that prompt. The AI control unit 15 also sends a prompt to the AI 150 indicating an instruction to generate a document based on a string indicating the document structure and document data retrieved by the search unit 14 and receives a response (document) in response to that prompt. The AI control unit 15 also sends a prompt to the AI 150 indicating an instruction to make corrections in response to a correction request for the generated document and receives a response (corrected document) in response to that prompt.
[0025] When the Modification Information Acquisition Unit 16 receives a modification instruction for a document generated by the Document Generation Unit 12, it acquires a group of information extracted from the set of document data stored in the Data Storage Unit 121, which is related to the document and the modification instruction, from the Graph Storage Unit 122 as a candidate for information to be used for modification. The Graph Storage Unit 122 stores the information extracted from the set of document data stored in the Data Storage Unit 121 in the form of a graph database. Specifically, the Graph Storage Unit 122 stores graph data in which the information extracted from the set of document data stored in the Data Storage Unit 121 is used as nodes, and the relationships between such information are used as edges. The Modification Information Acquisition Unit 16 acquires the group of information related to the document and the modification instruction by searching the graph related to the document and the modification instruction from the graph data. Hereinafter, this graph will be referred to as the "Information Map". The Information Map is displayed on the Terminal 20. The Document Generation Unit 12 performs modifications to the document based on the information selected by the user from the group of information related to the Information Map and the modification instruction. The information that constitutes the set of information extracted from the document data set may be words or sequences of words contained in the document data set, or it may be knowledge or concepts contained in the document data set. The set of information from the document data set can be extracted based on known methods for constructing graph databases.
[0026] The display control unit 17 displays various screens on the terminal 20 by transmitting information generated by the agent in response to user input. For example, the display control unit 17 displays the document structure and document generated by the document generation unit 12, and the information map acquired by the modification and usage information acquisition unit 16 on the terminal 20. In this case, the terminal 20 may display the screen using a browser.
[0027] On the other hand, terminal 20 has a reception unit 21, a communication unit 22, and a display control unit 23. Each of these units is realized by a program installed on terminal 20 that causes the terminal 20's CPU to execute a process.
[0028] The reception unit 21 receives user operations on the terminal 20.
[0029] The communication unit 22 controls communication with the information processing device 10.
[0030] The display control unit 23 controls the display of screens, etc. (for example, the dialogue screen 510 described later, the document structure and document generated by the document generation unit 12, and the information map acquired by the modification and usage information acquisition unit 16) based on the information (display data), etc. received from the information processing device 10.
[0031] The following describes the processing steps performed by the information processing device 10. Figures 4 to 6 are sequence diagrams illustrating an example of the processing steps performed by the information processing device 10.
[0032] The user first enters summary information about the document they intend to create (hereinafter referred to as the "target document") into the dialogue screen displayed on terminal 20. The summary information is entered, for example, in text format written in natural language.
[0033] Figure 7 shows an example of a dialogue screen display. The dialogue screen 510 shown in Figure 7 includes a dialogue display area 511 and a message input area 512. The dialogue display area 511 is the area where the content of the dialogue between the agent and the user is displayed. In the initial state, the dialogue display area 511 displays a greeting message g1 ("Is there anything I can help you with regarding your work?") prompting the user to enter a message. The message input area 512 is the area for receiving messages from the user and includes a send icon 5121.
[0034] When a user enters a message containing summary information into the message input area 512 and clicks the send icon 5121, the terminal 20 sends the entered summary information to the information processing device 10 (S101). At this time, the display content of the dialogue screen 510 changes according to the click of the send icon 5121, as shown in Figure 8.
[0035] Figure 8 shows an example of the display of the dialogue screen after the input of summary information. In Figure 8, the same parts as in Figure 7 are denoted by the same reference numerals, and their explanations are omitted. In the dialogue display area 511 of the dialogue screen 510 in Figure 8, the message m1 ("I want to create a planning specification for product A") has been added. Message m1 is the message that was entered as summary information. Thus, the summary information can be anything as long as it indicates what kind of document the user wants to create. However, it is thought that the more specific the summary information is, the higher the likelihood that the user will obtain the desired document sooner.
[0036] When the reception unit 11 of the information processing device 10 receives the summary information, it inputs the summary information to the document generation unit 12 (S102). The document generation unit 12 inputs the input summary information to the AI control unit 15 (S103).
[0037] The AI control unit 15 generates a prompt containing information instructing the generation of a document structure based on the summary information, and sends the prompt to the AI 150 (S104). The prompt is, for example, the following text:
[0038] <Example of a prompt> Please generate a document structure that meets the following requirements. Document structure refers to the chapters or table of contents of the document.
[0039] {Overview Information} <Example of prompts> Note that {Summary Information} will be replaced with the summary information entered into the AI control unit 15.
[0040] The AI150 generates text (a string indicating the document structure) in response to a prompt based on the learned parameters and sends the text to the AI control unit 15 (S105).
[0041] When the AI control unit 15 receives the text, it outputs the document structure indicated by the text to the document generation unit 12 (S106). The document generation unit 12 stores the document structure output from the AI control unit 15 as the "latest document structure" in the RAM 103 or HD 104, etc., and inputs the document structure to the display control unit 17 (S107). The display control unit 17 displays the document structure on the dialogue screen 510 that the terminal 20 is displaying (S108). Specifically, the display control unit 17 generates display data for the document structure and transmits the display data to the terminal 20 in order to display the document structure.
[0042] Figure 9 shows an example of a document structure display. In Figure 9, the same reference numerals are used for parts identical to those in Figure 8, and their explanations are omitted. The dialogue display area 511 of the dialogue screen 510 shown in Figure 9 has a response r1 from the agent added to it. Response r1 includes a string indicating the document structure.
[0043] Next, the user enters a message indicating whether or not modifications are needed for the document structure into the message input area 512 of the dialogue screen 510, and then clicks the send icon 5121. For example, if modifications are needed, the user enters a message indicating that modifications are required; if modifications are not needed, the user enters a message indicating that modifications are not needed.
[0044] In response to the input of the message, terminal 20 transmits the input message to information processing device 10 (S109). For example, if a correction instruction is input, the display content of the dialogue screen 510 changes as shown in Figure 8.
[0045] Figure 10 shows an example of the display of the dialogue screen after inputting instructions for modifying the document structure. In Figure 10, the same parts as in Figure 9 are denoted by the same reference numerals, and their explanations are omitted. In the dialogue display area 511 of the dialogue screen 510 shown in Figure 10, a message m2 ("Add market information after the objective") indicating instructions for modifying the document structure has been added.
[0046] When the receiving unit 11 of the information processing device 10 receives the message, it inputs the message to the document generation unit 12 (S110). The document generation unit 12 branches the processing based on whether the message is a correction instruction or not. Specifically, if the message is a correction instruction, steps S121 to S128 are executed.
[0047] In step S121, the document generation unit 12 inputs the received message (correction instruction) and the stored "latest document structure" to the AI control unit 15 (S121).
[0048] The AI control unit 15 generates a prompt containing information instructing the AI 150 to make corrections to the input "latest document structure" according to the input correction instructions, and sends the prompt to the AI 150 (S122). The prompt is, for example, the following text:
[0049] <Example of a prompt> The current document structure is as follows:
[0050] {latest document structure} Please make the following corrections to this document structure according to the instructions provided.
[0051] {Correction instructions} <Example of prompts> Note that "latest document structure" is applied to {latest document structure}, and the correction instructions entered into the AI control unit 15 are applied to {correction instructions}.
[0052] Based on the learned parameters, AI150 generates text (a string indicating the revised document structure) in response to the prompt and sends the text to AI control unit 15 (S123).
[0053] When the AI control unit 15 receives the text, it outputs the revised document structure indicated by the text to the document generation unit 12 (S124). The document generation unit 12 stores the document structure output from the AI control unit 15 as the "latest document structure" in the RAM 103 or HD 104, etc., and inputs the document structure to the display control unit 17 (S125). The display control unit 17 displays the document structure on the terminal 20 (S126). Specifically, the display control unit 17 generates display data for the document structure and transmits the display data to the terminal 20 in order to display the document structure.
[0054] Figure 11 shows an example of the display of the revised document structure. In Figure 11, the same parts as in Figure 10 are denoted by the same reference numerals, and their explanations are omitted. In the dialogue display area 511 of the dialogue screen 510 in Figure 11, response r2 has been added. Response r2 includes the document structure that has been revised based on the revision instructions. In this document structure, "Market Information" has been added after "Purpose" as per the user's revision instructions. Note that in Figure 11, the greeting message g1 is excluded from the display target by scrolling up, etc., due to the addition of response r2. In subsequent dialogue screens 510, existing responses or messages are excluded from the display target in response to the addition of new responses or messages.
[0055] Next, the user enters a message indicating whether or not modifications are needed to the document structure into the message input area 512 of the dialogue screen 510 and clicks the send icon 5121. The terminal 20 then sends the entered message to the information processing device 10 (S127). When the receiving unit 11 of the information processing device 10 receives the message, it inputs the message into the document generation unit 12 (S128). If the message is a modification instruction, the document generation unit 12 repeats steps S121 onwards. If the message indicates that modifications are not needed, the process proceeds to step S131 in Figure 5.
[0056] Figure 12 shows an example of inputting a message indicating that no modifications are needed to the document structure. In Figure 12, the same parts as in Figure 11 are denoted by the same reference numerals, and their explanations are omitted. In the dialogue display area 511 of the dialogue screen 510 shown in Figure 12, message m3 ("It's good") has been added. Message m3 is an example of a message indicating that no modifications are needed to the document structure shown in response r2.
[0057] Whether or not a user message indicates that no correction is needed may be determined based on whether or not a predetermined keyword is included in the message, or by querying AI150.
[0058] If a message indicating that no modifications are necessary for the initial document structure (the document structure indicated by response r1) is entered, such as message m3, steps S121 to S128 are not executed, and the process proceeds to step S131 in Figure 5.
[0059] Hereafter, a document structure that the user has determined does not require modification will be referred to as the "final document structure."
[0060] Steps S131 to S139 in Figure 5 are loop processes that are repeated for each item that is the smallest unit in the final document structure (hereinafter referred to as the "minimum item"). The minimum item in the document structure refers to the item corresponding to the lowest level in the hierarchical structure of the document structure. For example, in a document structure with a three-tiered structure that includes one or more major items (e.g., chapters), each major item includes one or more sub-items (e.g., sections), and each sub-item includes one or more minor items (e.g., sub-items), the minor items correspond to the minimum items. Alternatively, in the case of a document structure that does not have a hierarchical structure (a document structure in which each item is in a parallel relationship), such as the document structure included in response r2 in Figure 12, each item becomes the minimum item. Each minimum item is processed in the order of its appearance in the document structure (for example, from the beginning in the case of a table of contents). Hereinafter, the minimum items that are processed in this loop process will be referred to as "target items".
[0061] In step S131, a string indicating the target item is input to the conversion unit 13. The string indicating the target item is, for example, the item name of the target item (for example, the title). If the document structure has a hierarchical structure, the string indicating the target item also includes the item names of each ancestor of the target item (for example, the title). The item names of each ancestor of the target item refer to the item names corresponding to each node that has the target item as a descendant node when the hierarchical structure of the document structure is represented as a graph. For example, if the target item is a sub-item in the above three levels, the string indicating the target item is a string formed by concatenating the item name of the target item, the item name of the medium item containing the target item, and the item name of the major item containing the medium item. Hereinafter, the string indicating the target item will be referred to as the "target item text". The target item text can be said to be a string indicating the document structure. More precisely, the target item text can also be said to be a string obtained by dividing the string indicating the document structure based on the document structure.
[0062] Next, the conversion unit 13 converts the input target item text into a semantic vector to generate a vector that represents the meaning of the target item text (hereinafter referred to as the "target item vector") (S132). Subsequently, the conversion unit 13 inputs the target item vector and the target item text to the search unit 14 (S133).
[0063] The search unit 14 compares the input target item vector with the chunk vectors stored in the data storage unit 121 for each document data and each chunk, thereby identifying similar chunks for each document data and extracting some similar chunks that have a relatively high similarity to the target item text (S134). The search unit 14 generates search results for each extracted similar chunk, including related document information for that similar chunk. The related document information for a certain similar chunk includes the chunk ID of that similar chunk, and the document ID and document name of the document data to which that similar chunk belongs. Subsequently, the search unit 14 outputs the search results to the document generation unit 12 (S135).
[0064] The document generation unit 12 inputs the search results and target item text to the AI control unit 15 (S136).
[0065] The AI control unit 15 sends a prompt to the AI 150 containing information indicating instructions for generating a document based on the input search results and target item text (S137). The prompt may be, for example, the following text:
[0066] <Example of a prompt> The title of the document you want to generate is as follows:
[0067] {Target item text} Please generate a document for this title using the following document data as a reference.
[0068] {A set of similar chunks related to the search results} <Example of prompts> The target item text is assigned to {target item text}, and the set of similar chunks related to the search results from the search unit 14 is assigned to {set of similar chunks related to the search results}.
[0069] AI150 generates text corresponding to the prompt based on learned parameters and sends the text to AI control unit 15 (S138). This text is a draft of the document corresponding to the prompt (hereinafter referred to as the "target item draft"). When AI control unit 15 receives the target item draft, it outputs the target item draft to document generation unit 12 (S139).
[0070] When steps S131 to S139 are executed for all minimum items of the final document configuration, the document generation unit 12 stores the target item drafts generated for each minimum item in a text (hereinafter referred to as "target draft") RAM 102 or HD 104, etc., arranged (connected) in the order corresponding to the final document configuration, and inputs the target draft to the display control unit 17 (S140). The target draft has the meaning of a draft of the target document generated based on the final document configuration. The display control unit 17 displays the input target draft on the terminal 20 (S141). Specifically, the display control unit 17 generates display data for the target draft and transmits this display data to the terminal 20 in order to display the target draft.
[0071] Figure 13 shows an example of the display of the target draft. In Figure 13, the same reference numerals are used for parts identical to those in Figure 12, and their explanations are omitted. The dialogue screen 510 shown in Figure 13 has been expanded horizontally (widthwise), and a target draft display area 513 has been added to the left side.
[0072] The target draft display area 513 is the area where the target draft is displayed. Figure 13 shows an example where the agent's response r3 is displayed along with the target draft. Response r3 is a message prompting the user to review the target draft and suggest corrections as needed.
[0073] The user refers to the target draft displayed in the target draft display area 513 and considers whether any revisions are necessary. If there are parts that need revision, the user issues a revision instruction. In issuing a revision instruction, the user first specifies the range that includes the relevant part (hereinafter referred to as the "revision range").
[0074] Figure 14 is a diagram illustrating the specification of the scope of revisions to the target draft. In Figure 14, the same reference numerals are used for parts identical to those in Figure 13, and their explanations are omitted.
[0075] Figure 14 shows an example where the area enclosed by the dashed line p1 is designated as the modification area in the target draft display area 513. The modification area p1 may be specified by dragging the mouse over the text to be included in the modification area p1, or by other operations.
[0076] In response to the user's specification of the modification range, terminal 20 transmits information indicating the modification range (hereinafter referred to as "modification range information") to information processing device 10 (S151 in Figure 6). The modification range information indicates the location of the modification range in the target draft. The location of the modification range in the target draft may, for example, be the number of characters at the start and end positions of the modification range from the beginning of the target draft.
[0077] When the receiving unit 11 of the information processing device 10 receives the correction range information, it inputs the correction range information to the document generation unit 12 (S152).
[0078] Next, when the user enters a message indicating the content of the corrections to the correction range into the message input area 512 of the dialogue screen 510 and clicks the send icon 5121, the terminal 20 sends the correction content indicated in the message to the information processing device 10 (S153). At this time, the display content of the dialogue screen 510 changes as shown in Figure 15.
[0079] Figure 15 shows an example of the display of the dialogue screen after inputting the modification content for the modification range. In Figure 15, the same parts as in Figure 14 are denoted by the same reference numerals, and their explanations are omitted. In the dialogue display area 511 of the dialogue screen 510 shown in Figure 15, a message m4 ("I want to add supporting information for the specified area") indicating the modification content for the modification range has been added.
[0080] When the receiving unit 11 of the information processing device 10 receives the modification content, it inputs the modification content to the document generation unit 12 (S154). The document generation unit 12 inputs the modification content and the portion of the target draft stored in RAM 102 or HD 104, etc., that is included in the modification range indicated by the modification range information (hereinafter referred to as the "modified portion") to the modification usage information acquisition unit 16 (S155). The modification usage information acquisition unit 16 performs a search on the graph storage unit 122 using a query based on the modification content and the modified portion, and acquires an information map (information group) corresponding to the query (S156). The query can be generated using an existing query language such as Cypher. The information map is a graph that represents the relationships between nodes, with the information corresponding to the query being used as nodes from the knowledge-based information contained in the document data stored in the data storage unit 121. Subsequently, the modification usage information acquisition unit 16 inputs the information map to the display control unit 17 (S157). The display control unit 17 displays the information map, etc., on the dialogue screen 510 currently displayed on the terminal 20 (S158). Specifically, the display control unit 17 generates display data for the information map, etc., and transmits the display data to the terminal 20 in order to display the information map, etc.
[0081] Figure 16 shows an example of the display of an information map. In Figure 16, the same reference numerals are used for parts identical to those in Figure 15, and their explanations are omitted. In the dialogue display area 511 of the dialogue screen 510 shown in Figure 16, a response r4 and an information map M1 have been added. The information map M1 is the information map displayed by the display control unit 17 in step S158. The response r4 is a message from the agent displayed by the display control unit 17 along with the information map in step S158. The response r4 is a message prompting the user to select information (nodes) to be added as supporting information from the group of information (group of nodes) related to the information map M1.
[0082] The user selects one or more nodes from the group of nodes in information map M1 that correspond to the information to be used to modify the modified part (in this case, information to be added as supporting information for the modified part).
[0083] Figure 17 shows an example of a state in which nodes in the information map are selected. In Figure 17, the same reference numerals are used as in Figure 16, and their explanations are omitted. In Figure 17, the four nodes selected in information map M1 are represented in black.
[0084] Furthermore, to assist user selection, the information map M1 may be made capable of operations as shown in Figures 18 and 19.
[0085] Figure 18 shows a first example of operation on the information map. In the information map M1 shown in Figure 18, supplementary information (annotations) about each node is attached to that node. In the example in Figure 18, the supplementary information is a string indicating the name of the folder (the name of the information category) in which each piece of information is stored in the data storage unit 121. In Figure 18, it appears that the supplementary information is displayed for all nodes simultaneously, but for example, the supplementary information may be displayed only for nodes that the user has hovered over with the mouse.
[0086] Figure 19 shows a second example of operation on the information map. In the information map M1 shown in Figure 19, a list of check buttons (hereinafter referred to as "checklist L1") is added for each folder in the data storage unit 121 to which the information corresponding to any node belongs. When any check button in checklist L1 is unchecked, the node corresponding to the information belonging to the folder corresponding to that check button is hidden. As a result, the user can narrow down the selection candidates.
[0087] Once the user has finished selecting a node and the confirmation button b1 is pressed, terminal 20 sends a list of information IDs (identification information) corresponding to the selected node (hereinafter referred to as the "selected information ID list") to the information processing device 10 (S159).
[0088] When the receiving unit 11 of the information processing device 10 receives the list of selected information IDs, it inputs the list of selected information IDs to the modified usage information acquisition unit 16 (S160). The modified usage information acquisition unit 16 acquires information (text information) corresponding to each ID included in the list of selected information IDs from the graph storage unit 122 and inputs a list of such information (hereinafter referred to as the "list of selected information") to the document generation unit 12 (S161). The document generation unit 12 inputs the modified portion related to the modified range information received in step S152, the modified content received in step S154, and the list of selected information to the AI control unit 15 (S162).
[0089] The AI control unit 15 generates a prompt as correction instruction information that instructs the AI 150 to make corrections according to the content of the correction based on the selection information list for the part to be corrected, and sends the prompt to the AI 150 (S163).
[0090] The prompt might look like this:
[0091] <Example of a prompt> The documents to be corrected are as follows:
[0092] {Modified part} Please make the following corrections to this document.
[0093] {Modification details} Please use the following information when making corrections.
[0094] {List of Selected Information} <Example of prompts> Note that {modified portion} will be replaced with the modified portion, {modified content} will be replaced with the modified content, and {list of selected information} will be replaced with the list of selected information.
[0095] Based on the learned parameters, AI150 generates text in response to the prompt (a document with corrected parts corresponding to the corrected content) and sends the text (hereinafter referred to as "corrected result") to AI control unit 15 (S164). AI control unit 15 outputs the corrected result to document generation unit 12 (S165).
[0096] The document generation unit 12 modifies the target draft based on the modification results (S166). Specifically, the document generation unit 12 replaces the modified portion of the target draft with the modification results. Next, the document generation unit 12 inputs the modified target draft to the display control unit 17 (S167). The display control unit 17 displays the modified target draft, etc., on the dialogue screen 510 displayed on the terminal 20 (S168). Specifically, the display control unit 17 generates display data for the modified target draft, etc., and transmits this display data to the terminal 20 in order to display the modified target draft, etc.
[0097] Figure 20 shows an example of how the revised target draft is displayed. In Figure 20, the same reference numerals are used for parts identical to those in Figure 17, and their explanations are omitted. In the dialogue screen 510 shown in Figure 20, the content displayed in the target draft display area 513 has been replaced with the revised target draft. In addition, a response r5 from the agent indicating that revisions have been made has been added to the dialogue display area 511.
[0098] If the user requires further modifications, steps S151 to S168 in Figure 6 are repeated.
[0099] If the user determines that no further modifications are needed, they enter a message indicating that no modifications are necessary in the message input area 512 of the dialogue screen 510 and click the send icon 5121. As a result, the display content of the dialogue screen 510 changes as shown in Figure 21.
[0100] Figure 21 shows an example of a message indicating that no correction is needed. In Figure 21, the same parts as in Figure 20 are denoted by the same reference numerals, and their explanations are omitted. In Figure 21, as an example of a message indicating that no correction is needed, the message m5 "This content is OK" has been added to the interactive display area 511.
[0101] In response to such a message input, terminal 20 transmits the message to information processing device 10 (S171). Upon receiving the message, the receiving unit 11 of the information processing device 10 inputs the message to the document generation unit 12 (S172). In response to the message input, the document generation unit 12 converts the target draft into a predetermined format (e.g., PDF format), saves the converted target draft as the target document to a storage area accessible to the user, and then inputs a response including link information (e.g., URL) of the location where the target document is saved to the display control unit 17 (S173). The display control unit 17 displays the input response on the dialogue screen 510 displayed on terminal 20 (S174). Specifically, the display control unit 17 generates display data for the response and transmits the display data to terminal 20 in order to display the response.
[0102] Figure 22 shows an example of the display of the dialogue screen at the end of document generation. In Figure 22, the same reference numerals are used for parts identical to those in Figure 21, and their descriptions are omitted. In the dialogue display area 511 of the dialogue screen 510 shown in Figure 22, a response r6 has been added. Response r6 includes a message indicating that document generation is complete and a link L1 with link information to the target document. The user can download the target document by clicking link L1.
[0103] As described above, according to this embodiment, a document is generated based on a document structure generated by the document generation unit 12 based on the summary information, and which the user has determined does not require modification (i.e., a document structure specified by the user). Furthermore, a document is generated that reflects the user's modification instructions. Therefore, the user can make the desired modifications to the document.
[0104] In addition, the summary information above may specify the output data format. The output data format refers to the data format of the target document, such as pptx, xlsx, or docx. In this case, in step S104 of Figure 4, the document generation unit 12 may send a prompt to the AI 150 that also includes an instruction to generate a document structure with a layout that matches the output data format. By doing so, a document structure that matches the output format can be proposed to the user.
[0105] Furthermore, the generation of the target draft may allow for the creation and insertion of diagrams and charts.
[0106] Alternatively, a machine learning model may be trained for each user to learn the correspondence between summary information and the document structure adopted (specified) by the user, and this machine learning model may be used to generate a document structure corresponding to the summary information. Or, if multiple types of agents are defined, such training may be performed for each agent. Each of the multiple types of agents has a different set of document data to be searched in step S134 of Figure 5 (for example, a set of document data distinguished by folder in the data storage unit 121). For example, the set of document data may be divided by specialized field, and an agent may be assigned to each division. In this way, each agent will generate a different response based on different document data for the same input. In this case, when the user inputs summary information, they only need to select which agent to have perform document generation.
[0107] Alternatively, a machine learning model may be trained to recognize the correspondence between selected information (selected information) in the information map for each user or agent, and this machine learning model may be used to select the information sets that make up the information map.
[0108] Additionally, the source file for the document the user wants to create may be specified as or as part of the summary information.
[0109] Alternatively, an information map may be generated for each agent (i.e., for each different data source). In this case, multiple information maps may be displayed simultaneously, or it may be possible to switch between them on a per-agent basis.
[0110] Furthermore, while the above example describes a case where the document structure is generated based on summary information entered by the user, the document structure may also be selected by the user from one or more pre-prepared templates. In this case, the document structure corresponding to the selected template will be the document structure specified by the user. Such templates may be prepared, for example, for each type of document used in business operations.
[0111] Alternatively, a document structure created by the user may be designated as the target for use by the user. In this case, the user only needs to input the document structure instead of the summary information. Therefore, in steps S101 and S102 of Figure 4, the document structure is transmitted from the terminal 20 to the document generation unit 12 instead of the summary information. Following step S102, the document generation unit 12 can then execute steps S131 onwards in Figure 5 based on the document structure.
[0112] Furthermore, the information processing device 10 is not limited to a general-purpose server computer, as long as it is a device equipped with communication capabilities. The information processing device 10 may be, for example, an output device such as a PJ (Projector), IWB (Interactive White Board: an electronic whiteboard with the ability to communicate with each other), or digital signage, a HUD (Head Up Display) device, industrial machinery, imaging devices, sound collection devices, medical equipment, networked home appliances, a notebook PC (Personal Computer), a mobile phone, a smartphone, a tablet device, a game console, a PDA (Personal Digital Assistant), a digital camera, a wearable PC, or a desktop PC.
[0113] Furthermore, each function of this embodiment can be realized by one or more processing circuits. Hereinafter, "processing circuit" as used herein includes processors programmed to execute each function by software, such as processors implemented by electronic circuits, as well as devices such as ASICs (Application Specific Integrated Circuits), DSPs (digital signal processors), FPGAs (field programmable gate arrays), and conventional circuit modules designed to execute the functions described above.
[0114] Furthermore, the apparatus in this embodiment represents only one of several computing environments for carrying out the embodiments disclosed herein.
[0115] In one embodiment, the information processing device 10 includes a plurality of computing devices, such as a server cluster. The plurality of computing devices are configured to communicate with each other via any type of communication link, including a network or shared memory, and perform the processing disclosed herein. Similarly, terminal T may include a plurality of computing devices configured to communicate with each other.
[0116] Although embodiments of the present invention have been described in detail above, the present invention is not limited to these specific embodiments, and various modifications and changes are possible within the scope of the gist of the present invention as described in the claims.
[0117] Examples of the present invention are as follows: <1> A document generation unit that generates a document having a predetermined document structure using predetermined document data, A display control unit that displays the aforementioned document, It has, The display control unit, in response to a modification instruction for the document, displays a group of information extracted from the document data set that is related to the modification instruction. The document generation unit performs modifications to the document based on the modification instructions and the information selected from the information group. An information processing system characterized by the following: <2> The document generation unit generates a document structure based on summary information indicating the outline of the document, and generates a document having the document structure when the user instructs it to do so. Characterized by <1> The information processing system described above. <3> A search unit that searches for document data related to a string indicating the document structure from the aforementioned predetermined set of document data. It has, The document generation unit generates the document based on the retrieved document data. Characterized by <1> or <2> The information processing system described above. <4> The document generation unit generates the document based on the interaction with the user and modifies the document based on the interaction with the user. Characterized by <1> ~ <3> The information processing system described in any of the following. <5> The display control unit displays the information group using a graph in which the information constituting the information group is represented as nodes and the relationships between the information are represented as edges. Characterized by <1> ~ <4> The information processing system described in any of the following. <6> A document generation unit that generates a document having a predetermined document structure using predetermined document data, A display control unit that displays the aforementioned document, It has, The display control unit, in response to a modification instruction for the document, displays a group of information extracted from the document data set that is related to the modification instruction. The document generation unit performs modifications to the document based on the modification instructions and the information selected from the information group. An information processing device characterized by the following: <7> A document generation procedure that generates a document having a predetermined document structure using predetermined document data, A first display control procedure for displaying the aforementioned document, A second display control procedure that, in response to a correction instruction for the said document, displays a group of information extracted from the set of document data, which is related to the said correction instruction. A document modification procedure for performing modifications on the document based on the modification instructions and information selected from the information group, An information processing method characterized by a computer executing the following. <8> A document generation procedure that generates a document having a predetermined document structure using predetermined document data, A first display control procedure for displaying the aforementioned document, A second display control procedure that, in response to a correction instruction for the said document, displays a group of information extracted from the set of document data, which is related to the said correction instruction. A document modification procedure for performing modifications on the document based on the modification instructions and information selected from the information group, A program that causes a computer to execute something. [Explanation of Symbols]
[0118] 10 Information Processing Devices 11 Reception Department 12 Document Generation Unit 13 Conversion section 14 Search Section 15 AI control section 16 Modified usage information acquisition unit 17 Display Control Unit 20 devices 21 Reception Department 22 Communications Department 23 Display Control Unit 121 Data Storage Unit 122 Graph memory unit 150 AI [Prior art documents] [Patent Documents]
[0119] [Patent Document 1] Japanese Patent Publication No. 2023-007268
Claims
1. A document generation unit that generates a document having a predetermined document structure using predetermined document data, A display control unit that displays the aforementioned document, It has, The display control unit, in response to a modification instruction for the document, displays a group of information extracted from the document data set that is related to the modification instruction. The document generation unit performs modifications to the document based on the modification instructions and the information selected from the information group. An information processing system characterized by the following:
2. The document generation unit generates a document structure based on summary information indicating the outline of the document, and generates a document having the document structure when the user instructs it to do so. The information processing system according to feature 1.
3. A search unit that searches for document data related to a string indicating the document structure from the aforementioned predetermined set of document data. It has, The document generation unit generates the document based on the retrieved document data. The information processing system according to feature 1.
4. The document generation unit generates the document based on the interaction with the user and modifies the document based on the interaction with the user. The information processing system according to feature 1.
5. The display control unit displays the information group using a graph in which the information constituting the information group is represented as nodes and the relationships between the information are represented as edges. The information processing system according to feature 1.
6. A document generation unit that generates a document having a predetermined document structure using predetermined document data, A display control unit that displays the aforementioned document, It has, The display control unit, in response to a modification instruction for the document, displays a group of information extracted from the document data set that is related to the modification instruction. The document generation unit performs modifications to the document based on the modification instructions and the information selected from the information group. An information processing device characterized by the following:
7. A document generation procedure that generates a document having a predetermined document structure using predetermined document data, A first display control procedure for displaying the aforementioned document, A second display control procedure that, in response to a correction instruction for the document, displays a group of information extracted from the document data, which is related to the correction instruction; A document modification procedure for performing modifications on the document based on the modification instructions and information selected from the information group, An information processing method characterized by a computer executing the following.
8. A document generation procedure that generates a document having a predetermined document structure using predetermined document data, A first display control procedure for displaying the aforementioned document, A second display control procedure that, in response to a correction instruction for the document, displays a group of information extracted from the document data, which is related to the correction instruction; A document modification procedure for performing modifications on the document based on the modification instructions and information selected from the information group, A program that causes a computer to execute something.