KR-20260063079-A - TRANSFORMER-BASED ELECTRONIC DISCLOSURE DOCUMENT SERVICE PLATFORM APPARATUS AND METHOD

Abstract

The present invention relates to a transformer-based electronic disclosure document service platform device and method, wherein the device comprises: a data collection unit that collects internet information of said company consisting of electronic disclosure information of said company and SNS information, news information or stock post information; an electronic disclosure learning unit that learns a transformer-based electronic disclosure analysis model using electronic disclosure data extracted based on said company's electronic disclosure information; a corporate management data generation unit that generates corporate analysis data consisting of management activities and management results by inputting said company's internet information into a corporate analysis model; a corporate analysis data learning unit that tunes said transformer-based electronic disclosure analysis model using said corporate analysis data; and an annotation electronic disclosure document generation unit that automatically detects abnormal data based on the user's level of understanding and interest through said transformer-based electronic disclosure analysis model, adds annotations to said electronic disclosure documents to generate an annotation electronic disclosure document, said electronic disclosure documents determine the cause of the annotation based on said management activities and determine the solution to the annotation based on said management results, and includes a real-time question-and-answer assistant.

Inventors

• 김지완
• 박민호
• 배교연
• 김영표
• 이재승
• 이경원
• 오효원
• 김완수

Assignees

• (주) 사이냅소프트

Dates

Publication Date

20260507

Application Date

20241030

Claims (8)

1. A data collection unit that collects internet information of the said company, consisting of the company's electronic disclosure information, SNS information, news information, or stock post information; An electronic disclosure learning unit that learns a transformer-based electronic disclosure analysis model using electronic disclosure data extracted based on the electronic disclosure information of the above-mentioned company; A corporate management data generation unit that inputs internet information of the above-mentioned company into a corporate analysis model to generate corporate analysis data consisting of management activities and management results; A corporate analysis data learning unit that tunes the transformer-based electronic disclosure analysis model using the above corporate analysis data; and A transformer-based electronic disclosure document service platform device comprising an annotation electronic disclosure document generation unit that automatically detects abnormal data based on the user's level of understanding and interest through the above transformer-based electronic disclosure analysis model, adds annotations to the electronic disclosure document to generate an annotation electronic disclosure document, and the above electronic disclosure document determines the cause of the annotation based on the above management activity and determines the solution to the annotation based on the above management result, and includes a real-time question-and-answer assistant.
2. In paragraph 1, the above electronic disclosure learning department A transformer-based electronic disclosure document service platform device characterized by preprocessing the normal format of the above electronic disclosure information to generate pairs of corporate-independent disclosure terms and corporate-dependent disclosure information, and converting the pairs of corporate-independent disclosure terms and corporate-dependent disclosure information into embedding vectors, performing NER (Named Entity Recognition) analysis, performing question-and-answer analysis, or performing summary analysis to extract the above electronic disclosure data.
3. In paragraph 1, the above-mentioned corporate analysis data generation unit A transformer-based electronic disclosure document service platform device characterized by collecting unstructured documents composed of the above-mentioned SNS (Social Networking Service) information, the above-mentioned news information, or the above-mentioned stock post information, and performing preprocessing to remove hateful expressions from the above-mentioned unstructured documents through the above-mentioned corporate analysis model to generate sentiment-based corporate analysis data as first corporate management information.
4. In paragraph 3, the above-mentioned corporate analysis data generation unit A transformer-based electronic disclosure document service platform device characterized by generating management-based corporate analysis data into second corporate management information by performing preprocessing to extract management activities and management results from the unstructured documents through the corporate analysis model.
5. In paragraph 1, the above-mentioned corporate analysis data learning unit A transformer-based electronic disclosure document service platform device characterized by tuning the transformer-based electronic disclosure analysis model by converting each of the first and second corporate management information derived from the above corporate analysis data into an embedding vector, performing NER (Named Entity Recognition) analysis, or performing summary analysis.
6. In paragraph 1, the annotation electronic disclosure document generation unit A transformer-based electronic disclosure document service platform device characterized by determining the level of understanding and the level of interest based on the user's settings and adjusting the level of understanding and the level of interest by analyzing the user's questions and answers through the real-time question-and-answer assistant.
7. In paragraph 1, the annotation electronic disclosure document generation unit A transformer-based electronic disclosure document service platform device characterized by automatically detecting abnormal data within the electronic disclosure document of the company through the transformer-based electronic disclosure analysis model and reflecting it in the annotated electronic disclosure document, wherein the annotated electronic disclosure document analyzes the cause of the outlier based on the management activities and presents a solution for the outlier based on the management results.
8. In a transformer-based electronic disclosure document service platform method performed on an electronic disclosure document service platform device, A data collection step for collecting internet information of the above-mentioned company, consisting of the company's electronic disclosure information, social media information, news information, or stock post information; An electronic disclosure learning step for training a Transformer-based electronic disclosure analysis model using electronic disclosure data extracted based on the electronic disclosure information of the above-mentioned company; A corporate management data generation step that inputs the internet information of the above-mentioned company into a corporate analysis model to generate corporate analysis data consisting of management activities and management results; A corporate analysis data learning step for tuning the transformer-based electronic disclosure analysis model using the corporate analysis data; and It includes the step of automatically detecting abnormal data based on the user's level of understanding and interest through the above-mentioned transformer-based electronic disclosure analysis model, adding annotations to the electronic disclosure document, and generating an anomaly detection annotated electronic disclosure document. A method for a transformer-based electronic disclosure document service platform that includes a real-time question and answer assistant, wherein the above-mentioned annotation electronic disclosure document determines the cause of an outlier based on the above-mentioned management activities and determines the solution to the outlier based on the above-mentioned management results.

Description

Transformer-Based Electronic Disclosure Document Service Platform Apparatus and Method The present invention relates to electronic disclosure document service platform technology, and more specifically, to a transformer-based electronic disclosure document service platform device and method capable of providing an annotated electronic disclosure document by collecting corporate internet information to generate corporate analysis data and performing annotation on the electronic disclosure document while considering the user's level of understanding and interest. The Data Analysis, Recognition, and Transfer System (DART) is a corporate information disclosure platform operated by the Financial Supervisory Service that transparently provides companies' financial information, key management matters, and disclosure materials to investors and the general public. Currently, many companies are publishing information through electronic disclosure systems following digitalization, and the number of users and views utilizing these systems is on an upward trend every year. However, while the demand for fast and accurate analysis techniques is rising due to the increasing volume and diversity of information included in electronic disclosures, there are limitations for non-experts in understanding and utilizing these contents due to the need for specialized knowledge to analyze such information. Therefore, there is a demand for technology that utilizes the latest artificial intelligence technology, the 'Transformer-based NLP model,' to automatically summarize and visualize electronic disclosures released by companies, thereby enabling users to easily and accurately analyze and utilize corporate information. Korean Registered Patent No. 10-2442350 relates to an information analysis method and apparatus for performing automatic document creation based on artificial intelligence. The method comprises: a step of learning by applying a heuristic algorithm to a first criterion that performs analysis based on the location of information and a second criterion that performs analysis based on the form of information, in order to analyze information contained in a document subject to automatic document creation; a step of classifying the types of multiple documents registered on a platform that supports automatic document creation and analyzing the contents of the multiple documents based on a pre-matched analysis algorithm based on the classified types to extract multiple information items from the multiple documents; a step of performing automatic document creation based on artificial intelligence so that the extracted multiple information items are applied to the information items included in the standard document format provided by the platform, thereby creating the multiple documents as documents corresponding to the standard format; and a step of performing detailed corrections according to whether additional corrections are required for each information item of the multiple documents created as documents corresponding to the standard format, determined based on whether the multiple information items required for extraction through automatic document creation have been extracted. FIG. 1 is a diagram illustrating a transformer-based electronic disclosure document service platform system according to the present invention. Figure 2 is a diagram illustrating the system configuration of the transformer-based electronic disclosure document service platform device of Figure 1. Figure 3 is a diagram illustrating the functional configuration of the transformer-based electronic disclosure document service platform device of Figure 1. FIG. 4 is a flowchart illustrating an embodiment of a transformer-based electronic disclosure document service platform method according to the present invention. FIG. 5 is a diagram illustrating a transformer-based electronic disclosure document service platform process according to an embodiment of a transformer-based electronic disclosure document service platform device according to the present invention. The description of the present invention is merely an example for structural or functional explanation, and therefore the scope of the present invention should not be interpreted as being limited by the examples described in the text. That is, since the examples are subject to various modifications and may take various forms, the scope of the present invention should be understood to include equivalents capable of realizing the technical concept. Furthermore, the objectives or effects presented in the present invention do not imply that a specific example must include all of them or only such effects; therefore, the scope of the present invention should not be understood as being limited by them. Meanwhile, the meaning of the terms described in this application should be understood as follows. Terms such as "first," "second," etc., are intended to distinguish one component from another, and the scope