Multi-task unified model for Chinese aspect-based sentiment analysis

IF 3.4 3区计算机科学 Q2 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Computer Speech and Language Pub Date : 2025-05-08 DOI:10.1016/j.csl.2025.101822

Yuewei Wu , Jialu Wang , Xiaoli Feng , Zhaoliang Wu , Jiakai Peng , Fulian Yin

{"title":"Multi-task unified model for Chinese aspect-based sentiment analysis","authors":"Yuewei Wu , Jialu Wang , Xiaoli Feng , Zhaoliang Wu , Jiakai Peng , Fulian Yin","doi":"10.1016/j.csl.2025.101822","DOIUrl":null,"url":null,"abstract":"<div><div>Aspect-Based Sentiment Analysis (ABSA) is crucial for in-depth mining and analysis of opinion expressions and sentiment tendencies in massive user review texts. Most of the existing researches on ABSA in Chinese only consider a single contextual semantic feature, while ignoring syntactic dependency feature, and rarely addresses the realization of multiple tasks in the same model. From the perspective of multi-task unification, this paper proposes a multi-task unified model for Chinese aspect-based sentiment analysis (MTUC-ABSA), which integrates Bi-directional Long Short-Term Memory (Bi-LSTM) and Graph Convolutional Network (GCN) to learn multiple features between context and sentiment elements, and uses the unified Machine Reading Comprehension (MRC) paradigm to build a multi-task model, which mainly focuses on the aspect sentiment triplet extraction (ASTE) task. Experimental results on real data sets show that our method can effectively improve the accuracy of aspect-based sentiment analysis compared with other existing methods.</div></div>","PeriodicalId":50638,"journal":{"name":"Computer Speech and Language","volume":"95 ","pages":"Article 101822"},"PeriodicalIF":3.4000,"publicationDate":"2025-05-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Computer Speech and Language","FirstCategoryId":"94","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S0885230825000476","RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE","Score":null,"Total":0}

引用次数: 0

Abstract

Aspect-Based Sentiment Analysis (ABSA) is crucial for in-depth mining and analysis of opinion expressions and sentiment tendencies in massive user review texts. Most of the existing researches on ABSA in Chinese only consider a single contextual semantic feature, while ignoring syntactic dependency feature, and rarely addresses the realization of multiple tasks in the same model. From the perspective of multi-task unification, this paper proposes a multi-task unified model for Chinese aspect-based sentiment analysis (MTUC-ABSA), which integrates Bi-directional Long Short-Term Memory (Bi-LSTM) and Graph Convolutional Network (GCN) to learn multiple features between context and sentiment elements, and uses the unified Machine Reading Comprehension (MRC) paradigm to build a multi-task model, which mainly focuses on the aspect sentiment triplet extraction (ASTE) task. Experimental results on real data sets show that our method can effectively improve the accuracy of aspect-based sentiment analysis compared with other existing methods.

查看原文本刊更多论文

中文面向方面的情感分析多任务统一模型

基于方面的情感分析（ABSA）对于深度挖掘和分析海量用户评论文本中的观点表达和情感倾向至关重要。现有的中文ABSA研究大多只考虑了单一的语境语义特征，而忽略了句法依赖特征，很少涉及同一模型中多个任务的实现。从多任务统一的角度出发，提出了面向中文面向情感分析的多任务统一模型（MTUC-ABSA），该模型集成了双向长短期记忆（Bi-LSTM）和图卷积网络（GCN）来学习上下文和情感元素之间的多个特征，并采用统一机器阅读理解（MRC）范式构建了面向面向情感三元提取（ASTE）任务的多任务统一模型。在真实数据集上的实验结果表明，与其他现有方法相比，我们的方法可以有效提高基于方面的情感分析的准确性。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文求助全文

来源期刊

Computer Speech and Language 工程技术-计算机：人工智能

CiteScore

11.30

自引率

4.70%

发文量

审稿时长

22.9 weeks

期刊介绍： Computer Speech & Language publishes reports of original research related to the recognition, understanding, production, coding and mining of speech and language. The speech and language sciences have a long history, but it is only relatively recently that large-scale implementation of and experimentation with complex models of speech and language processing has become feasible. Such research is often carried out somewhat separately by practitioners of artificial intelligence, computer science, electronic engineering, information retrieval, linguistics, phonetics, or psychology.