摘要
近年来,方面级情感分析利用图神经网络挖掘依赖句法信息逐渐成为趋势,但现有方法大多未考虑不同关系类型对内容词的影响,难以区分关键的关联词.此外,多视角信息的相互补充对捕捉情感特征起重要作用,但在过去的研究中融合机制常被忽视.为解决这些问题,提出一种多源信息融合图卷积网络(multi-source information graph convolutional network,MSI-GCN)有效捕获和集成三视角信息.首先,设计了一个双通道信息提取模块SSDGCN(syntax-semantics dual graph convolutional network),由类型嵌入的句法增强图卷积网络(TES-GCN)和语义图卷积网络(SEM-GCN)组成.TES-GCN通过引入类型嵌入层,使用句法模块学习不同类型的权重来增强句法信息.SEM-GCN对自注意矩阵进行编码,捕获语义信息,并引入正交正则化来增强语义关联.其次,嵌入外部知识图表示丰富词汇特征.最后,引入局部门控-全局卷积网络,充分利用视角之间的互补性,对其进行有效融合.本文在4个公开数据集上对提出的方法进行了评估,准确率和Macro-F1值相比于基线模型均有所提升.
In recent years,aspect-based sentiment analysis utilizing graph neural networks to exploit syntactic information has become increasingly popular.However,existing methods often overlook the impact of different types of relations on content words,making it challenging to distinguish key relational terms.In addition,the mutual supplementation of multiperspective information plays a crucial role in capturing sentiment features,but integration mechanisms have often been neglected in past research.To address these issues,a multi-source information graph convolutional network(MSI-GCN)is proposed to effectively capture and integrate three perspectives of information.First,a dual-channel information extraction module,syntax-semantics dual graph convolutional network(SSD-GCN),is designed,comprising a typeenhanced syntax graph convolutional network(TES-GCN)and a semantic graph convolutional network(SEM-GCN).TES-GCN enhances syntactic information by incorporating a type embedding layer and learning different weights through the syntactic module.SEM-GCN encodes self-attention matrices to capture semantic information and employs orthogonal regularization to strengthen semantic associations.Second,external knowledge graph embeddings enrich vocabulary features.Finally,a local-global convolutional network is introduced to leverage the complementarity between perspectives for effective integration.The proposed method is evaluated on four public datasets,showing improvements in accuracy and Macro-F1 scores compared to baseline models.
作者
高玮军
张玉莹
焦成寅
GAO Wei-Jun;ZHANG Yu-Ying;JIAO Cheng-Yin(School of Computer and Communication,Lanzhou University of Technology,Lanzhou 730050,China)
出处
《计算机系统应用》
2025年第8期14-24,共11页
Computer Systems & Applications
基金
国家自然科学基金(6172059)。
关键词
方面级情感分析
多源信息融合
图卷积网络
外部知识
句法类型嵌入
aspect-based sentiment analysis(ABSA)
multi-source information fusion
graph convolutional network(GCN)
external knowledge
syntactic type embedding
