Reproduce “ESGSenticNet: A Neurosymbolic Knowledge Base for Corporate Sustainability Analysis” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Keane Ong, Rui Mao, Deeksha Varshney, Frank Xing, Ranjan Satapathy, Johan Sulaeman, Erik Cambria, Gianmarco Mengaldo
A neurosymbolic knowledge base, ESGSenticNet, leverages linguistic rules, GPT-4o, and graph-based learning to extract and categorize action-oriented ESG concepts from corporate sustainability reports. It offers substantially improved relevance, materiality, and structured organization for ESG topic analysis over current NLP methods.
Key insights
- Integrating symbolic linguistic rules with neural models (GPT-4o) and graph-based propagation enables the scalable and accurate construction of comprehensive, structured knowledge bases for complex domains like ESG.
- Task-specific evaluation metrics, such as ESG-relatedness and action-orientation, are crucial for effectively assessing the performance and practical utility of ESG topic analysis tools.
- Generative Large Language Models can be effectively leveraged within a neurosymbolic paradigm for high-quality, large-scale semantic annotation and knowledge acquisition, reducing reliance on extensive manual labeling.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.