Neurosymbolic AI for Mining Key Aspects of Socially Responsible Investing

Environmental, Social, and Governance (ESG) factors have become critical for assessing corporate sustainability and ethical responsibility. However, the vast volume of unstructured data available across corporate reports, social media, and news sources poses a challenge for systematic ESG analysis. This paper explores the application of neurosymbolic AI, which combines neural networks’ pattern recognition capabilities with the structured reasoning of symbolic AI, to mine key aspects of ESG from large-scale, diverse data sources. By leveraging SenticNet for concept parsing and deep learning for sentiment analysis, we extract relevant ESG metrics, classify corporate practices, and identify trends. This hybrid approach enhances both the interpretability and scalability of ESG analysis, providing more accurate insights into corporate behaviors and their impact on sustainability goals. Results demonstrate that neurosymbolic AI not only improves the extraction of meaningful ESG aspects but also enables real-time monitoring, supporting data-driven decision-making for investors, regulators, and stakeholders.