MKGN: A Multi-Dimensional Knowledge Enhanced Graph Network for Multi-Hop Question and Answering

• DOI: 10.1587/transinf.2021edp7154
• Publication Date: 2022-03-31T22:28:32Z
• Authors: Ying ZHANG; Fandong MENG; Jinchao ZHANG; Yufeng CHEN; Jinan XU; Jie ZHOU
• Keywords: Commonsense knowledge; Knowledge graph; Dependency graph

Machine reading comprehension with multi-hop reasoning always suffers from path breaking due to the lack of world knowledge, which results in wrong answer detection. In this paper, we analyze what knowledge previous work lacks, e.g., dependency relations and commonsense. Based on our analysis, propose a Multi-dimensional Knowledge enhanced Graph Network, named MKGN, exploits specific repair gap process. Specifically, approach incorporates not only entities through various graph neural networks, but also commonsense by bidirectional attention mechanism, aims enhance representations both question contexts. Besides, make most multi-dimensional investigate two kinds fusion architectures, i.e., sequential parallel manner. Experimental HotpotQA dataset demonstrate effectiveness verify that using especially commonsense, can indeed improve process contribute correct