ABSTRACT Aspect‐based Sentiment Analysis (ABSA) is widely used in many applications. Recent research indicates that incorporating syntactic dependency trees from external parsers into graph neural networks can enhance the performance of ABSA. However, external parsers may produce inaccurate syntactic structures in syntax‐insensitive scenarios, which weakens the representation of word dependencies and makes long‐distance association modeling particularly difficult. To address these challenges, this paper constructs a novel integrated syntactic‐semantic partially directed graph (ISS‐PD graph) and proposes a syntax and semantics aware graph attention network with local context focus (SSGA‐LCF). Specifically, the ISS‐PD graph can combine syntactic and semantic features by adding labeled dependencies between aspect terms and contextual words. The proposed SSGA‐LCF effectively integrates multiple relational features from the ISS‐PD graph for refined attention computation and feature aggregation, thereby guiding the propagation of information. Experimental results on three public benchmark datasets demonstrate that our method outperforms other baseline methods overall.
Huang et al. (Mon,) studied this question.