This report presents findings from an AI-driven, large-scale textual analysis of Environmental, Social, and Governance (ESG) reporting practices among the 100 leading listed companies on the Stock Exchange of Thailand (SET100). Applying a state-of-the-art multilingual deep learning model—the Text Match Pre-Trained Transformer (TMPT)—we systematically analyzed 845 corporate disclosure documents comprising Annual Reports (AR) and Sustainability Reports (SR) in both English and Thai, covering fiscal years 2020 to 2024. This constitutes, to the best of our knowledge, the most comprehensive AI-based ESG textual analysis of the Thai capital market conducted to date.

The TMPT model scores the relevance of each text fragment in a corporate report against 13 predefined ESG topics spanning Environmental (4 topics), Social (6 topics), Governance (1 topic), and Economic (2 topics) dimensions. Scores are aggregated into company-level ESG Emphasis Scores and subsequently standardized to enable cross-company and temporal comparison.