Ontological Commitment

Ontological commitment refers to the underlying assumptions about the nature of reality embedded within a system, model, or text. Current research focuses on extracting and analyzing these implicit commitments from various sources, including large language models and word embeddings, often employing techniques like hierarchical clustering and feed-forward neural networks to map semantic relationships and build ontologies. This work is significant for improving the transparency and reliability of AI systems, facilitating knowledge integration across different models, and enabling more robust semantic interoperability in applications ranging from robotics to social service provision.

Papers