Automatic document generation method using hearing by generative ai

Abstract

The present invention relates to a system for hearing necessary information from a user by using a generative AI and automatically generating a document. This system generates a question on the basis of a prompt and a schema, and collects information while dynamically changing the question in response to the user's answer. The collected information is structured according to the schema and finally output as a document in JSON format or the like. Thus, it is possible to efficiently prepare a document without requiring the intervention of an expert.

Description

Automatic Document Generation Method Using Generative AI for Hearing

[0009] , ,

[0008] , ,

[0007] , ,

[0010]

[0001] The present invention belongs to the field of artificial intelligence (AI) technology, and particularly relates to a method of automatically generating a document by hearing necessary information from a user using generative AI.

[0002] Currently, when creating specialized documents such as business descriptions and patent specifications, hearing by experts is required. However, experts' time is limited, and hearing takes time and cost. In addition, knowledge that has not been documented remains personal as tacit knowledge and is difficult to share across the organization.

[0003] WO / 2023 / 032100 proposes a document generation system based on user input, but there are still issues regarding efficient information collection and hearing.

[0004] WO / 2023 / 032100

[0005]

[0006] Non-Patent Document 1:

[0007] The purpose of the present invention is to reduce the time and cost required for document creation by efficiently collecting necessary information from a user using generative AI and automatically generating a document as an alternative to hearing by experts.

[0008] In the present invention, by presenting a prompt and a document schema to generative AI, generative AI asks the user for necessary information in an asynchronous chat format based on the schema. The question content is automatically changed according to the user's answer, and hearing is performed so that all necessary items are filled. The collected information is output in JSON format structured along the schema.

[0009] According to the present invention, it is possible to efficiently collect necessary information from a user without the intervention of an expert and automatically generate a document. As a result, the time and cost required for document creation can be significantly reduced. In addition, by using an asynchronous chat format, the user can provide information at a time convenient for themselves.

[0010] Figure 1 is a flowchart of the automated document generation system according to the present invention, illustrating the process from when the generating AI generates questions based on prompts and documents, when it obtains user responses and updates the documents, to when it finally completes the documents.

[0011] Figure 2 is a block diagram of the automated document generation system of the present invention, illustrating how the system processes prompts, schemas, and documents, and how the generating AI interacts with users through a user interface, handling questions and verifying documents.

[0012] This invention relates to a system that collects information from users and automatically generates documents.

[0013] The system's operation and configuration will be described based on Figures 1 and 2.

[0014] Figure 1 shows a flowchart of an automated document generation system using generation AI.

[0015] The system (code 210) first generates an empty document (code 213) based on the schema (code 212).

[0016] This schema defines the template and structure of the generated documents and includes the necessary information fields.

[0017] The generating AI (code 214) automatically generates the questions necessary to fill in the empty document based on the prompt (code 211) and presents them to the user via the user interface (code 220).

[0018] The user enters their answer to a question through a chat interface (code 221), and that answer is analyzed by a generating AI (code 214) before proceeding to the next step.

[0019] Based on the user's response, the system (code 210) determines whether the document (code 213) needs to be updated.

[0020] If no update is needed, the next question is generated; if necessary, the document (code 213) is updated.

[0021] Document updates are performed automatically according to the schema (code 212), and the necessary fields are filled in.

[0022] Once all the necessary information has been collected, the document is finally rendered and output in a format that the user can review.

[0023] Figure 2 shows a block diagram of the system (reference numeral 210).

[0024] The system is structured around a generator AI module (code 214) and processes prompts (code 211), schemas (code 212), and documents (code 213).

[0025] The prompt (code 211) defines the information to be collected from the user and forms the basis for the generating AI (code 214) to generate questions.

[0026] A schema (code 212) defines the structure and template of a document (code 213) and determines what information is placed where and how.

[0027] The generated document (code 213) is automatically constructed based on the user's responses.

[0028] The generating AI (code 214) generates questions based on the prompt (code 211) and schema (code 212), analyzes the user's response, and updates the document (code 213).

[0029] This allows for efficient information collection and the generation of the final document (code 213).

[0030] The user interface (indicated by 220) is a means for the user to interact with the system.

[0031] The user answers questions through a chat interface (code 221) and reviews the generated document (code 213) using a document viewer (code 222).

[0032] The generating AI (code 214) analyzes information from the user and dynamically updates the document (code 213), enabling efficient document generation without the need for expert intervention.

[0033] Example 1: When a user wants to create a patent specification, they provide the generation AI with the schema and prompts for the patent specification. The generation AI asks questions such as "What is the title of the invention?" and "What problem does the invention aim to solve?" to collect necessary information from the user. The collected information is output in JSON format and can be used as a draft of the patent specification.

[0034] Example 2: When creating a business process description for use in internal controls, the generation AI is provided with the schema for the business process description. The generation AI collects information through questions such as "What kind of business process should we organize?" and "How does that business process begin?", and ultimately generates the business process description.

[0035] This invention can be used in various industrial fields that require document creation. Specifically, it can be applied to the automatic generation of documents that previously required expert interviews, such as internal control operational descriptions, patent application specifications, contracts, and reports.

[0036] 101: Creating an empty document (the process of generating an empty document based on a schema)

[0037] 102: Question generation (a process in which the AI ​​automatically generates questions based on prompts)

[0038] 103: User responses (information entered by the user in response to a question)

[0039] 104: Determining whether an update is needed (The generating AI analyzes the user's response and determines whether the document needs to be updated.)

[0040] 105: Document updates (the process of updating documents based on user responses)

[0041] 106: Document rendering (the process of generating the final document after all the information has been collected)

[0042] 107: Final generated document (completed document created by the generation AI)

[0043] 108: Schema (structure and template for generating documents)

[0044] 109: Prompt (instructions and underlying information for generating questions)

[0045] 210: Entire automatic document generation system

[0046] 211: Prompt (information for generating questions)

[0047] 212: Schema (elements defining the structure and template of the document)

[0048] 213: Document (generated document)

[0049] 214: Generation AI (AI module that generates questions, analyzes user answers, and updates the document)

[0050] 220: User interface (interface for interaction between the system and the user)

[0051] 221: Chat interface (interface for the user to answer questions)

[0052] 222: Document viewer (interface for viewing the generated document)

[0053]

[0054]

Claims

An automated document generation method characterized by presenting prompts and schemas to a generating AI, having the generating AI ask the user for necessary information based on the schema, automatically modifying the questions according to the user's answers while collecting information, and finally generating a structured document based on the schema. An automated document generation method according to claim 1, characterized in that the generating AI sequentially updates each item of the document based on information collected from the user. An automated document generation method according to claim 1 or 2, characterized in that the user interface exchanges information in a chat format and includes a document viewer that allows the generated document to be viewed in real time.

Images