文献类型：期刊文献

英文题名：Contextual Word Distance and Sentiment-Enhanced Dual Graph Convolutional Network for Aspect-Based Sentiment Analysis

作者：Qin, Shuai[1]; Guo, Weibin[1]

机构：[1] School of Information Science and Engineering, East China University of Science and Technology, Shanghai, 200237, China

年份：2025

起止页码：83

外文期刊名：2025 2nd International Conference on Smart Grid and Artificial Intelligence, SGAI 2025

收录：EI(收录号：20252818759904)

语种：英文

外文关键词：Convolution - Graph theory - Semantics - Sentiment analysis

摘要：Aspect-Based Sentiment Analysis (ABSA) is a fine-grained sentiment classification task that aims to determine the sentiment polarity of specific aspects within a sentence. Graph Convolutional Network (GCN) are widely used in ABSA for capturing dependencies between words. However, GCN relying on semantic and syntactic dependencies face challenges in processing informal or noisy texts due to syntactic parsing errors and the introduction of irrelevant contextual information by attention mechanisms. To address these issues, this paper proposes a Contextual Word Distance and Sentiment-Enhanced Dual Graph Convolutional Network (CWDS-DGCN). Specifically, we propose a contextual word distance matrix to explicitly capture the dependencies between aspect and sentiment words while reducing the impact of irrelevant information. We incorporate external sentiment polarity knowledge into the syntactic dependency tree to enhance its representation. In addition, we design a recurrent feature fusion module to combine information from both semantic and syntactic channels more effectively. Extensive experiments on three datasets show that the CWDS-DGCN model performs better than the most recent baseline approach. ? 2025 IEEE.