S$^2$GSL: Incorporating Segment to Syntactic Enhanced Graph Structure Learning for Aspect-based Sentiment Analysis

TL;DR

This paper introduces S$^2$GSL, a novel graph structure learning framework for aspect-based sentiment analysis that effectively combines semantic and syntactic information while removing irrelevant dependencies.

Contribution

The paper proposes a segment-aware semantic graph learning and syntax-based latent graph learning, along with a self-adaptive aggregation network for improved ABSA performance.

Findings

Outperforms existing methods on four benchmark datasets.

Effectively removes irrelevant contexts and dependencies.

Demonstrates the benefit of combining semantic and syntactic graph learning.

Abstract

Previous graph-based approaches in Aspect based Sentiment Analysis(ABSA) have demonstrated impressive performance by utilizing graph neural networks and attention mechanisms to learn structures of static dependency trees and dynamic latent trees. However, incorporating both semantic and syntactic information simultaneously within complex global structures can introduce irrelevant contexts and syntactic dependencies during the process of graph structure learning, potentially resulting in inaccurate predictions. In order to address the issues above, we propose S$^2$GSL, incorporating Segment to Syntactic enhanced Graph Structure Learning for ABSA. Specifically,S$^2$GSL is featured with a segment-aware semantic graph learning and a syntax-based latent graph learning enabling the removal of irrelevant contexts and dependencies, respectively. We further propose a self-adaptive aggregation network that facilitates the fusion of two graph learning branches, thereby achieving complementarity across diverse structures. Experimental results on four benchmarks demonstrate the effectiveness of our framework.