Toward Real-Time ESG Evaluation: A Public Opinion-Based Approach Using Social Media Data

This study proposes a novel Social Index for evaluating ESG (Environmental, Social, and Governance) performance by leveraging deep learning-based text analysis of unstructured social media data. Traditional ESG assessment methods primarily rely on structured data sources, such as corporate disclosures or surveys, which fail to capture real-time public sentiment. To address this limitation, the proposed methodology utilizes the GloVe (Global Vectors for Word Representation) model to analyze social media texts in relation to the Sustainable Development Goals (SDGs), thereby quantifying public perception of corporate ESG activities in real time. The findings reveal three key contributions. First, the proposed Social Index significantly narrows the gap between reported ESG performance and public sentiment by incorporating dynamic, real-time data. Second, distinct patterns in public perception across ESG and SDG dimensions suggest the need for companies to adopt customized ESG strategies sensitive to regional and sectoral contexts. Third, this index offers a transparent and objective tool for external stakeholders to evaluate ESG performance, enhancing corporate accountability and alignment with societal expectations. Overall, the study demonstrates that integrating public opinion into ESG assessment provides a more responsive and socially attuned evaluation framework, supporting the development of sustainable and socially responsible corporate strategies.