Extracting ESG Evaluation Criteria Data and Building a Data Base with LLM

With the growing importance of sustainable development, ESG (Environmental, Social, and Governance) management has become a critical component of corporate strategy. Despite this, the domestic ESG evaluation system encounters significant challenges due to inconsistencies in evaluation criteria and the heterogeneous use of data among ESG rating agencies, which compromise the credibility of the evaluations. This study addresses these issues by proposing a methodology that utilizes Large Language Models (LLMs) and prompt engineering to automatically extract ESG evaluation factor data from corporate sustainability reports. The extraction process is based on the criteria established by the Ministry of Trade, Industry, and Energy (K-ESG Guidelines). The research involves collecting sustainability reports from companies across various industries and developing a systematic procedure to extract ESG factors classified into three domains: Environment (E), Social (S), and Governance (G). To ensure consistency and accuracy, synonym dictionaries and prompt engineering techniques are employed. The extracted data are organized in a structured database, and their accuracy and reliability are validated through comparisons with actual evaluation factors provided by ESG rating agencies. This study introduces an automated data extraction and management system that improves the standardization and efficiency of ESG evaluations, thereby contributing to the credibility and reliability of sustainability assessments.