A novel semantic dependency and aspect interaction graph convolutional network for aspect-level sentiment analysis

Aspect-based sentiment analysis (ABSA) is a task of fine-grained sentiment analysis that focuses on discerning the sentiment polarity of a sentence concerning specific aspects. This paper introduces a novel graph convolutional network, denoted as Semantic Dependency and Aspect Interaction Graph Convolutional Network(SDAIGCN), to overcome the limitations of existing graph-based sentiment analysis models in handling the semantic intricacies within aspectual phrases and the sentiment interactions among different aspects in a sentence. The SDAIGCN model achieves this by incorporating internal semantic correlations among aspect phrases through syntactic dependencies between context and aspect words in a sentence. Additionally, it associates affective interactions among different aspects to generate four types of adjacency matrix graphs. Experimental results on four publicly available datasets demonstrate that our proposed model yields improvements in both accuracy and F1 score.