COMET: Commonsense Transformers for Automatic Knowledge Graph Construction
This work addresses the challenge of building commonsense knowledge bases for AI systems, offering a generative alternative to extractive methods, though it is incremental as it builds on existing knowledge graphs.
The paper tackled the problem of automatic commonsense knowledge graph construction by proposing COMET, a generative model that transfers implicit knowledge from pre-trained language models to generate explicit commonsense descriptions, achieving up to 77.5% precision on ATOMIC and 91.7% on ConceptNet, which approaches human performance.
We present the first comprehensive study on automatic knowledge base construction for two prevalent commonsense knowledge graphs: ATOMIC (Sap et al., 2019) and ConceptNet (Speer et al., 2017). Contrary to many conventional KBs that store knowledge with canonical templates, commonsense KBs only store loosely structured open-text descriptions of knowledge. We posit that an important step toward automatic commonsense completion is the development of generative models of commonsense knowledge, and propose COMmonsEnse Transformers (COMET) that learn to generate rich and diverse commonsense descriptions in natural language. Despite the challenges of commonsense modeling, our investigation reveals promising results when implicit knowledge from deep pre-trained language models is transferred to generate explicit knowledge in commonsense knowledge graphs. Empirical results demonstrate that COMET is able to generate novel knowledge that humans rate as high quality, with up to 77.5% (ATOMIC) and 91.7% (ConceptNet) precision at top 1, which approaches human performance for these resources. Our findings suggest that using generative commonsense models for automatic commonsense KB completion could soon be a plausible alternative to extractive methods.