计算机科学

计算机科学 ›› 2026, Vol. 53 ›› Issue (5): 319-327.doi: 10.11896/jsjkx.250200126

• 人工智能 • 上一篇    下一篇

基于多视角图神经网络的跨度级方面情感三元组抽取

申奥, 周青凯, 夏天, 高瑞玲   

  1. 上海第二工业大学计算机与信息工程学院人工智能研究院 上海 201209
  • 收稿日期:2025-02-28 修回日期:2025-06-20 发布日期:2026-05-08
  • 通讯作者: 夏天(xiatian@sspu.edu.cn)
  • 作者简介:(shenao@sspu.edu.cn)
  • 基金资助:
    国家社会科学基金重大项目(20&ZD140);2024年上海高校青年教师培养资助计划(ZZEGD202412)

Span-based Aspect Sentiment Triplet Extraction Based on Multi-view Graph Neural Networks

SHEN Ao, ZHOU Qingkai, XIA Tian, GAO Ruiling   

  1. Institute of Artificial Intelligence, School of Computer and Information Engineering, Shanghai Polytechnic University, Shanghai 201209, China
  • Received:2025-02-28 Revised:2025-06-20 Online:2026-05-08
  • About author:SHEN Ao,born in 1993,Ph.D,lecturer,is a member of CCF(No.R7108M).Her main research interests include na-tural language processing and artificial intelligence.
    XIA Tian,born in 1979,Ph.D,professor,is a member of CCF(No.R7108M).His main research interests include natural language processing and text classification.
  • Supported by:
    Major Project of the National Social Science Fundation of China(20&ZD140) and 2024 Young Teachers Funding Project of Shanghai(ZZEGD202412).

PDF (PC)

217

摘要： 方面情感分析三元组抽取(Aspect Sentiment Triplet Extraction,ASTE)是细粒度情感分析中的一个新兴任务。传统跨度级别情感三元组抽取方法近期在ASTE任务上取得了显著的效果,但这些方法并未充分挖掘句法和语义依赖关系的潜力。为此,提出一种基于多视角边缘增强编码器的方法。该方法深度挖掘并全面整合词际间的句法与语义关联,以此实现方面与观点的精准甄别,并高效获取多元且深邃的深层信息。具体而言,首先基于RoBERTa提取句子基础语义信息,同时运用集双向长短期记忆网络和边增强的图神经网络为一体的边缘增强型编码器,全面捕获高阶语义及句法信息,通过多层次信息融合,有效增强跨度表征。此外,鉴于传统跨度级模型对跨度边界不敏感,引入多层图卷积网络以捕捉跨度之间的交叉关系,从而有效识别跨度边界,并采用互异消除策略来消除冲突三元组。在多个公共基准数据集上进行了大量实验,结果表明,该方法优于其他基线模型,验证了该方法在情感三元组抽取任务中的有效性。

关键词: 方面情感分析, 三元组抽取, 多视角图神经网络, 边缘增强, 自然语言处理

Abstract: Aspect sentiment analysis triplet extraction(ASTE) is an emerging task in fine-grained sentiment analysis.Traditional span-level sentiment triple extraction methods have recently achieved remarkable results on the ASTE task,but these methods have not fully exploited the potential of syntactic and semantic dependencies.This study proposes a multi-view edge-enhanced encoder that makes full use of the syntactic and semantic relationships between words to accurately distinguish aspect words and viewpoint words and obtain rich deep information.Specifically,a dual-channel encoder with RoBERTa is first used to obtain basic semantic information,and at the same time,a bi-directional long short-term memory network channel and the edge-enhanced graph neural network are utilized to comprehensively capture semantic and syntactic information.Considering the insufficient sensitivity of traditional span-level models to span boundaries,it introduces a multi-layer graph convolutional network to capture the cross-relationships between spans to effectively identify span boundaries.In addition,this study also uses the mutual difference elimination strategy to eliminate conflicting triples.Through extensive experiments on multiple public benchmark data sets,this method outperforms other baseline models and verifies its effectiveness in the emotional triple extraction task.

Key words: Aspect sentiment analysis, Triplet extraction, Multi-view graph neural network, Edge enhancement, Natural language processing

中图分类号: 

  • TP391
[1]LIU X,HOU R,GAN Y,et al.Aspect-oriented Opinion Alignment Network for Aspect-Based Sentiment Classification[M]//ECAI 2023.IOS Press,2023:1552-1559.
[2]WANG G,WANG B,XU F,et al.Domain-consistent syntacticrepresentation for cross-domain aspect sentiment triplet extraction[J].Expert Systems with Applications,2024,256:124854.
[3]MA F,ZHANG C,ZHANG B,et al.Aspect-specific contextmodeling for aspect-based sentiment analysis[C]//CCF International Conference on Natural Language Processing and Chinese Computing.Cham:Springer,2022:513-526.
[4]LI Y L,SUN C S,LUO L,et al.Aspect-level Sentiment Classification Based on Sentence Information to Enhance Word Information[J].Computer Science,2024,51(6):299-308.
[5]PENG H,XU L,BING L,et al.Knowing what,how and why:A near complete solution for aspect-based sentiment analysis[C]//Proceedings of the AAAI Conference on Artificial Intelligence.2020:8600-8607.
[6]WU Z,YING C,ZHAO F,et al.Grid Tagging Scheme for Aspect-oriented Fine-grained Opinion Extraction[C]//Findings of the Association for Computational Linguistics:EMNLP 2020.2020:2576-2585.
[7]CHEN W,DU J,ZHANG Z,et al.A Hierarchical Interactive Network for Joint Span-based Aspect-Sentiment Analysis[C]//Proceedings of the 29th International Conference on Computational Linguistics.2022:7013-7019.
[8]CHEN Y,KEMING C,SUN X,et al.A Span-level Bidirectional Network for Aspect Sentiment Triplet Extraction[C]//Procee-dings of the 2022 Conference on Empirical Methods in Natural Language Processing.2022:4300-4309.
[9]LIANG S,WEI W,MAO X L,et al.STAGE:span tagging and greedy inference scheme for aspect sentiment triplet extraction[C]//Proceedings of the AAAI Conference on Artificial Intelligence.2023:13174-13182.
[10]XU L,CHIA Y K,BING L.Learning Span-Level Interactions for Aspect Sentiment Triplet Extraction[C]//Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing.2021:4755-4766.
[11]PENG K,JIANG L,PENG H,et al.Prompt Based Tri-Channel Graph Convolution Neural Network for Aspect Sentiment Triplet Extraction[C]//Proceedings of the 2024 SIAM International Conference on Data Mining(SDM).Society for Industrial and Applied Mathematics,2024:145-153.
[12]LIU Z,LIN W,SHI Y,et al.A robustly optimized BERT pre-training approach with post-training[C]//China National Conference on Chinese Computational Linguistics.Cham:Springer,2021:471-484.
[13]CHEN H,ZHAI Z,FENG F,et al.Enhanced multi-channelgraph convolutional network for aspect sentiment triplet extraction[C]//Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics.2022:2974-2985.
[14]OUYANG J,XUAN C,WANG B,et al.Aspect-based sentiment classification with aspect-specific hypergraph attention networks[J].Expert Systems with Applications,2024,248:123412.
[15]CHEN S,WANG Y,LIU J,et al.Bidirectional machine reading comprehension for aspect sentiment triplet extraction[C]//Proceedings of the AAAI Conference on Artificial Intelligence.2021:12666-12674.
[16]MAO Y,SHEN Y,YU C,et al.A joint training dual-mrc framework for aspect based sentiment analysis[C]//Proceedings of the AAAI Conference on Artificial Intelligence.2021:13543-13551.
[17]YAN H,DAI J,JI T,et al.A Unified Generative Framework for Aspect-based Sentiment Analysis[C]//Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing.2021:2416-2429.
[18]ZHANG W,LI X,DENG Y,et al.Towards generative aspect-based sentiment analysis[C]//Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing.2021:504-510.
[19]JING H,LI Z,ZHAO H,et al.Seeking Common but Distinguishing Difference,A Joint Aspect-based Sentiment Analysis Model[C]//Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing.2021:3910-3922.
[20]LI D,YANG Z,LAN Y,et al.Simple Approach for Aspect Sentiment Triplet Extraction Using Span-Based Segment Tagging and Dual Extractors[C]//Proceedings of the 46th International ACM SIGIR Conference on Research and Development in Information Retrieval.2023:2374-2378.
[21]LI Y,ZENG X,ZENG Y,et al.Enhanced Packed Marker with Entity Information for Aspect Sentiment Triplet Extraction[C]//Proceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval.2024:619-629.
[22]ZHANG Y,YANG Y,LI Y,et al.Boundary-driven table-filling for aspect sentiment triplet extraction[C]//Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing.2022:6485-6498.
[23]SU G,WU M,HUANG Z,et al.Refine,align,and aggregate:multi-view linguistic features enhancement for aspect sentiment triplet extraction[C]//Findings of the Association for Computational Linguistics ACL 2024.2024:3212-3228.
[24]PENNINGTON J,SOCHER R,MANNING C D.Glove:Global vectors for word representation[C]//Proceedings of the 2014 Conference on Empirical Methods in Natural Language Process-ing(EMNLP).2014:1532-1543.
[25]SCHULDER M,WIEGAND M,RUPPENHOFER J,et al.Towards bootstrap** a polarity shifter lexicon using linguistic features[C]//Proceedings of the Eighth International Joint Conference on Natural Language Processing.2017:624-633.
[26]LI C,ZHANG J,TANG H,et al.Enhancing the Performance of Aspect-Based Sentiment Analysis Systems[J].arXiv:2404.03259,2024.
[27]CHEN S,LIU J,WANG Y,et al.Synchronous double-channel recurrent network for aspect-opinion pair extraction[C]//Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics.2020:6515-6524.
[28]XU L,LI H,LU W,et al.Position-Aware Tagging for Aspect Sentiment Triplet Extraction[C]//Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing(EMNLP).2020:2339-2349.
[29]FAN Z,WU Z,DAI X,et al.Target-oriented opinion words extraction with target-fused neural sequence labeling[C]//Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics:Human Language Technologies.2019:2509-2518.
[30]LOSHCHILOV I,HUTTER F.Fixing Weight Decay Regularization in Adam[J].arXiv:1171.05101,2018.
[31]SRIVASTAVA N,HINTON G,KRIZHEVSKY A,et al.Dropout:a simple way to prevent neural networks from overfitting[J].The Journal of Machine Learning Research,2014,15(1):1929-1958.
[32]YU BAI JIAN S,NAYAK T,MAJUMDER N,et al.Aspectsentiment triplet extraction using reinforcement learning[C]//Proceedings of the 30th ACM International Conference on Information & Knowledge Management.2021:3603-3607.
[33]MUKHERJEE R,KANNEN N,PANDEY S,et al.CONT-RASTE:Supervised Contrastive Pre-training With Aspect-based Prompts For Aspect Sentiment Triplet Extraction[C]//Fin-dings of the Association for Computational Linguistics:EMNLP 2023.2023:12065-12080.
[34]ZHAI Z,CHEN H,FENG F,et al.COM-MRC:A COntext-masked machine reading comprehension framework for aspect sentiment triplet extraction[C]//Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing.2022:3230-3241.
[35]LI Y,HE Q,ZHANG D.Dual graph convolutional networks integrating affective knowledge and position information for aspect sentiment triplet extraction[J].Frontiers in Neurorobotics,2023,17:1193011.
[36]NAGLIK I,LANGO M.Exploiting Phrase Interrelations inSpan-level Neural Approaches for Aspect Sentiment Triplet Extraction[C]//Pacific-Asia Conference on Knowledge Discovery and Data Mining.Cham:Springer,2023:222-233.
[37]ZOU W,ZHANG W,WU W,et al.A multi-task shared cascade learning for aspect sentiment triplet extraction using bert-mrc[J].Cognitive Computation,2024,16(4):1554-1571.
[38]SUN Q,YANG L,MA M,et al.Rethinking aste:A minimalist tagging scheme alongside contrastive learning[J].arXiv:2403.07342,2024.
[39]SHAO D G,HU Y J.Aspect Sentiment Triplet ExtractionBased on Multi-layer Attention and GCN[J].Journal of Chinese Computer Systems.2024,45(5):1062-1068.
Viewed
Full text


Abstract

Cited
  Shared   
  Discussed   
No Suggested Reading articles found!
渝ICP备16013121号-4
地址：重庆市两江新区洪湖西路18号    邮编：401121  
电话：023-63500828（编辑部） 023-67039612（会议/专题） 023-67039623（财务/发票）
E-mail：jsjkx12@163.com
本系统由北京玛格泰克科技发展有限公司设计开发