Cognitive Graph for Multi-Hop Reading Comprehension at Scale
It addresses the problem of scalable and explainable multi-hop reasoning for large-scale QA systems, representing a novel method for a known bottleneck.
The paper tackles multi-hop question answering in web-scale documents by proposing the CogQA framework, which builds a cognitive graph using implicit extraction and explicit reasoning modules, achieving a joint F1 score of 34.9 on the HotpotQA dataset compared to 23.6 for the best competitor.
We propose a new CogQA framework for multi-hop question answering in web-scale documents. Inspired by the dual process theory in cognitive science, the framework gradually builds a \textit{cognitive graph} in an iterative process by coordinating an implicit extraction module (System 1) and an explicit reasoning module (System 2). While giving accurate answers, our framework further provides explainable reasoning paths. Specifically, our implementation based on BERT and graph neural network efficiently handles millions of documents for multi-hop reasoning questions in the HotpotQA fullwiki dataset, achieving a winning joint $F_1$ score of 34.9 on the leaderboard, compared to 23.6 of the best competitor.