Aspect Sentiment Triplet Extraction Incorporating Syntactic Constituency Parsing Tree and Commonsense Knowledge Graph

The aspect sentiment triplet extraction (ASTE) task aims to extract the target term and the opinion term, and simultaneously identify the sentiment polarity of target-opinion pairs from the given sentences. While syntactic constituency information and commonsense knowledge are both important and valuable for the ASTE task, only a few studies have explored how to integrate them via flexible graph convolutional networks (GCNs) for this task. To address this gap, this paper proposes a novel end-to-end model, namely GCN-EGTS, which is an enhanced Grid Tagging Scheme (GTS) for ASTE leveraging syntactic constituency parsing tree and a commonsense knowledge graph based on GCNs. Specifically, two types of GCNs are developed to model the information involved, namely span GCN for syntactic constituency parsing tree and relational GCN (R-GCN) for commonsense knowledge graph. In addition, a new loss function is designed by incorporating several constraints for GTS to enhance the original tagging scheme. The extensive experiments on several public datasets demonstrate that GCN-EGTS outperforms the state-of-the-art approaches significantly for the ASTE task based on the evaluation metrics. The outcomes of this research indicate that effectively incorporating syntactic constituency parsing information and commonsense knowledge is a promising direction for the ASTE task.