Decker: Double Check with Heterogeneous Knowledge for Commonsense Fact Verification
It addresses a challenging problem in commonsense QA for AI systems, but appears incremental by combining existing knowledge types.
The paper tackles commonsense fact verification by proposing Decker, a model that bridges structured and unstructured knowledge to verify claims, achieving effectiveness on CSQA2.0 and CREAK benchmarks.
Commonsense fact verification, as a challenging branch of commonsense question-answering (QA), aims to verify through facts whether a given commonsense claim is correct or not. Answering commonsense questions necessitates a combination of knowledge from various levels. However, existing studies primarily rest on grasping either unstructured evidence or potential reasoning paths from structured knowledge bases, yet failing to exploit the benefits of heterogeneous knowledge simultaneously. In light of this, we propose Decker, a commonsense fact verification model that is capable of bridging heterogeneous knowledge by uncovering latent relationships between structured and unstructured knowledge. Experimental results on two commonsense fact verification benchmark datasets, CSQA2.0 and CREAK demonstrate the effectiveness of our Decker and further analysis verifies its capability to seize more precious information through reasoning.