산업안전보건기준에 관한 규칙의 효과적 탐색과 이해를 위한 단어분포 지표와 Word2Vec 분석 방법

다국어 초록 (Multilingual Abstract)

The purpose of the Rule on Occupational Safety and Health Standards (hereinafter safety and health rules) is to stipulate the safety and health measures stipulated in the Occupational Safety and Health Act and specific instructions necessary for their implementation. However, the safety and health rules are broad and specialized, and the articles are intricately connected, making it difficult for users to navigate and approach key issues to be managed. Accordingly, in this study, the frequency, distribution, and relevance of terms included in the overall rules were analyzed in order for users to easily access the safety and health rules, and objects to be observed and methods for managing them were presented. To this end, statistical indicators were developed and applied to check the word distribution of safety and health rules, and items to be observed were summarized based on the connection between terms using the word embedding technique. First, the safety and health rules provided by the Korea Law Information Center were collected, and term distribution index were developed using the frequency and distribution of words extracted through text mining. The term distribution index derives whether it appears only in a specific chapter or in all rules through the ratio of the frequency of occurrence in all rule clauses to the frequency of occurrence in each chapter and clause. Through this, users can effectively explore for terms to be followed in a specific working environment and terms to be complied with in the overall working environment. Next, the Word2Vec algorithm was used to predict the surrounding words or the central word, and the words related to the previously derived key terms were visualized through t-SNE. This can help to prioritize the things that need to be managed first, focusing on key terms, without checking the overall rules. As such, it is believed that the results of this study can help the users to explore safety and health rules by understanding the distribution of words and visualizing key terms and related terms in the safety and health rules.

목차 (Table of Contents)

• 1. 서론 1
• 1.1 연구 필요성 1
• 1.2 연구 목적 및 내용 3
• 2. 배경 이론 4
• 2.1 TF-IDF 4
• 2.2 Word Embedding 8
• 3. 연구 설계 17
• 3.1 연구 절차 17
• 3.2 데이터 수집 19
• 3.3 데이터 전처리 21
• 3.4 단어의 빈도수 계산 25
• 3.5 단어분포 지표 개발 27
• 4. 연구 결과 34
• 4.1 Module 1 - 단어 간 지표 비교 34
• 4.2 Module 2 - 단어 간 연관성 분석 39
• 4.3 Module 2 - 분석 결과 시각화 42
• 5. 결론 50
• 참고문헌 52
• 부록 56
• A. Analysis code 56