I Know What You Asked: Graph Path Learning using AMR for Commonsense Reasoning

CommonsenseQA is a task in which a correct answer is predicted through\ncommonsense reasoning with pre-defined knowledge. Most previous works have\naimed to improve the performance with distributed representation without\nconsidering the process of predicting the answer from the semantic\nrepresentation of the question. To shed light upon the semantic interpretation\nof the question, we propose an AMR-ConceptNet-Pruned (ACP) graph. The ACP graph\nis pruned from a full integrated graph encompassing Abstract Meaning\nRepresentation (AMR) graph generated from input questions and an external\ncommonsense knowledge graph, ConceptNet (CN). Then the ACP graph is exploited\nto interpret the reasoning path as well as to predict the correct answer on the\nCommonsenseQA task. This paper presents the manner in which the commonsense\nreasoning process can be interpreted with the relations and concepts provided\nby the ACP graph. Moreover, ACP-based models are shown to outperform the\nbaselines.\n