Improving aspect-based sentiment analysis with Knowledge-aware Dependency Graph Network

Haiyan Wu, Chaogeng Huang, Shengchun Deng

Abstract: Highlights•Knowledge-aware Dependency Graph Network (KDGN) aspect-based sentiment analysis is proposed.•Dependency graph that integrates domain knowledge can improve sentiment analysis.•A relational graph attention network is built on a knowledge-aware dependency graph.•Experimental results demonstrate that KDGN significantly outperforms on benchmarking datasets.