《计算机应用》唯一官方网站 ›› 2025, Vol. 45 ›› Issue (6): 1724-1731.DOI: 10.11772/j.issn.1001-9081.2024060903

• 第十二届CCF大数据学术会议 • 上一篇    

集成句法与情感知识的方面级情感分析模型

李自亮1,2, 朱广丽1,2(), 张玉雷1,2, 刘佳佳1,2, 焦熠璇1,2, 张顺香1,2   

  1. 1.安徽理工大学 计算机科学与工程学院,安徽 淮南 232001
    2.合肥综合性国家科学中心 人工智能研究院,合肥 230088
  • 收稿日期:2024-07-01 修回日期:2024-07-28 接受日期:2024-08-01 发布日期:2024-08-22 出版日期:2025-06-10
  • 通讯作者: 朱广丽
  • 作者简介:李自亮(1999—),男,安徽六安人,硕士研究生,CCF会员,主要研究方向:方面级情感分析、关系抽取
    朱广丽(1971—),女,安徽淮南人,副教授,硕士,主要研究方向:智能信息处理、情感计算 glzhu@aust.edu.cn
    张玉雷(1999—),男,河南安阳人,硕士研究生,主要研究方向:多模态情感分析、数据挖掘
    刘佳佳(2001—),男,安徽阜阳人,硕士研究生,主要研究方向:自然语言处理、多模态情感分析
    焦熠璇(2000—),女,河北衡水人,硕士研究生,主要研究方向:自然语言处理、文本分类
    张顺香(1970—),男,安徽无为人,教授,博士,主要研究方向:Web挖掘、语义搜索、关系抽取、复杂网络分析。
  • 基金资助:
    国家自然科学基金面上项目(62076006);安徽高校协同创新项目(GXXT-2021-008)

Aspect-based sentiment analysis model integrating syntax and sentiment knowledge

Ziliang LI1,2, Guangli ZHU1,2(), Yulei ZHANG1,2, Jiajia LIU1,2, Yixuan JIAO1,2, Shunxiang ZHANG1,2   

  1. 1.School of Computer Science and Engineering,Anhui University of Science and Technology,Huainan Anhui 232001,China
    2.Institute of Artificial Intelligence,Hefei Comprehensive National Science Center,Hefei Anhui 230088,China
  • Received:2024-07-01 Revised:2024-07-28 Accepted:2024-08-01 Online:2024-08-22 Published:2025-06-10
  • Contact: Guangli ZHU
  • About author:LI Ziliang, born in 1999, M. S. candidate. His research interests include aspect-based sentiment analysis, relation extraction.
    ZHU Guangli, born in 1971, M. S., associate professor. Her research interests include intelligent information processing, affective computing.
    ZHANG Yulei, born in 1999, M. S. candidate. His research interests include multi-modal sentiment analysis, data mining.
    LIU Jiajia, born in 2001, M. S. candidate. His research interests include natural language processing, multi-modal sentiment analysis.
    JIAO Yixuan, born in 2000, M. S. candidate. Her research interests include natural language processing, text classification.
    ZHANG Shunxiang, born in 1970, Ph. D., professor. His research interests include Web mining, semantic search, relation extraction, complex network analysis.
  • Supported by:
    National Natural Science Foundation of China(62076006);University Synergy Innovation Program of Anhui Province(GXXT-2021-008)

摘要:

方面级情感分析(ABSA)是一项细粒度的情感分析任务,旨在分析给定文本中特定方面词的情感极性。现有的ABSA方法采用图卷积网络(GCN)处理句法和语义信息,然而这些方法将方面词的所有句法依赖等同看待,忽略了远距离不相关词对目标方面词的影响,造成目标方面词和观点词权重分配的不适宜,且对语义信息提取不充分。针对这些问题,提出一种集成句法与情感知识的ABSA模型。首先,根据句法信息构建可达矩阵,以此为基础,利用方面词进行中心位置赋权构建句法增强图;其次,通过外部情感知识和方面增强构建语义增强图,利用图卷积分别对句法增强图和语义增强图进行充分建模形成不同的特征通道;再次,通过双仿射注意力更有效地交互融合句法信息和语义信息;最后,运用平均池化和拼接操作获取方面词对应的最终特征向量。实验结果表明,相较于深度依赖感知图卷积网络模型DA-GCN-BERT (deep Dependency Aware GCN+BERT(Bidirectional Encoder Representations from Transformers)),所提模型在5个公开数据集上的准确率分别提高了1.71、1.41、1.27、0.17和0.43个百分点。可见,所提模型在ABSA领域具有很强的适用性。

关键词: 自然语言处理, 方面级情感分析, 图卷积网络, 双仿射注意力, 平均池化

Abstract:

Aspect-Based Sentiment Analysis (ABSA) is a fine-grained sentiment analysis task aiming to analyze sentiment polarity of specific aspect words in a given text. Existing ABSA methods use Graph Convolutional Network (GCN) to process syntactic and semantic information, but they treat all syntactic dependencies of aspect words equally, ignoring the impact of distant unrelated words on target aspect words, resulting in inappropriate weight allocation of target aspect words and viewpoint words, and insufficient extraction of semantic information. Aiming at these issues, an ABSA model integrating syntax and sentiment knowledge was proposed. Firstly, a reachability matrix was constructed according to syntactic information. Based on this, a syntactic enhancement graph was constructed by weighting the central position through the aspect words. Secondly, a semantic enhancement graph was constructed by external emotional knowledge and aspect enhancement, and graph convolution was used to fully model the syntactic enhancement graph and semantic enhancement graph, respectively, so as to form different feature channels. Thirdly, biaffine attention was used to integrate syntactic and semantic information effectively. Finally, average-pooling and concatenation operations were used to obtain the final feature vectors corresponding to the aspect words. Experimental results indicate that compared with the deep dependency aware graph convolutional network model — DA-GCN-BERT (deep Dependency Aware GCN+BERT(Bidirectional Encoder Representations from Transformers)), the proposed model achieves the accuracy improvements of 1.71, 1.41, 1.27, 0.17, and 0.43 percentage points on five publicly available datasets, respectively. It can be seen that the proposed model has strong applicability in the ABSA field.

Key words: Natural Language Processing (NLP), Aspect-Based Sentiment Analysis (ABSA), Graph Convolutional Network (GCN), biaffine attention, average-pooling

中图分类号: