Dynamic Relevance Graph Network for Knowledge-Aware Question Answering

Anonymous

Abstract: This work investigates the challenge of learning and reasoning for Commonsense Question Answering given an external source of knowledge in the form of a knowledge graph. We propose a novel graph neural network architecture, called dynamic relevance graph network (DRGN). DRGN operates on a given KG subgraph based on the question and answers entities and uses the relevance between the nodes to establish new edges dynamically for learning node representations in the graph network. Using the relevance between the graph nodes in learning representations helps the model to not only exploit the existing relationships in the KG subgraph but also recover the missing edges. Moreover, our model improves handling the negative questions due to considering the relevance between the global question node and the graph entities. Our proposed approach shows competitive performance on two QA datasets with commonsense knowledge, CommonsenseQA and OpenbookQA, and improves the state-of-the-art published results.

Paper Type: long