计算机科学

计算机科学 ›› 2025, Vol. 52 ›› Issue (6A): 240500051-7.doi: 10.11896/jsjkx.240500051

• 人工智能 • 上一篇    下一篇

融合情感词典和图对比学习的中文零样本立场检测

付书凡1, 王中卿2, 姜晓彤2   

  1. 苏州大学计算机科学与技术学院 江苏 苏州 215006
  • 出版日期:2025-06-16 发布日期:2025-06-12
  • 通讯作者: 王中卿(wangzq@suda.edu.cn)
  • 作者简介:(20234227002@stu.suda.edu.cn)
  • 基金资助:
    国家自然科学基金(62076175,61976146);江苏省双创博士计划

Zero-shot Stance Detection in Chinese by Fusion of Emotion Lexicon and Graph ContrastiveLearning

FU Shufan1, WANG Zhongqing2, JIANG Xiaotong2   

  1. School of Computer Science and Technology,Soochow University,Suzhou,Jiangsu 215006,China
  • Online:2025-06-16 Published:2025-06-12
  • About author:FU Shufan,born in 2001,postgraduate,is a member of CCF(No.V9926G).Her main research interests include natural language processing and so on.
    WANG Zhongqing,born in 1987,Ph.D,associate professor,is a member of CCF(No.94994M).His main research interests include natural language processing and so on.
  • Supported by:
    National Natural Science Foundation of China(62076175,61976146) and Jiangsu Innovation Doctor Plan.

PDF (PC)

47

摘要： 零样本立场检测的主要目的是在标注数据有限或没有标注数据的情况下识别作者对特定目标或主题的态度,目前已有的零样本立场检测主要是基于注意力机制或引入外部情感信息,该类方法忽略了原始文本中隐藏的情感信息和实体之间的语义关系。针对这一问题,提出了一个融合情感词典和图对比学习的零样本立场检测模型(EL-CL),利用思维链诱导的方法来挖掘原始文本中的情感信息,用于辅助构建新的输入文本,在对输入文本聚类生成原型图的训练过程中引入情感词典来增强原型图的文本向量中的情感信息。同时,采用自监督的图对比学习方法,对含有情感特征的向量进行数据增强,以提高模型对未知样本的推理能力。在NLPCC2016中文微博立场检测数据集上基于5个目标进行实验,结果显示,所提模型在多分类评价指标macro-F1上比基线模型提升了10%,证明了所提模型在零样本环境下具有良好的立场检测能力。

关键词: 零样本立场检测, 情感词典, 思维链诱导, 原型图, 图对比学习

Abstract: Zero-shot stance detection aims to identify the author’s attitude towards a specific target or topic when labeled data is limited or nonexistent.Currently,zero-shot stance detection methods are mainly based on attention mechanisms or the incorporation of external sentiment information.However,these methods often neglect the latent sentiment information within the original text and the semantic relationships between entities.To address this issue,a zero-shot stance detection model integrating a sentiment lexicon and graph contrastive learning(EL-CL) is proposed.It employs a chain of thought prompting method to uncover sentiment information within the original text,aiding in the construction of new input texts.During the clustering of input texts to generate prototype graphs,a sentiment lexicon is introduced to enhance the sentiment information within the text vectors of the prototype graph.Additionally,a self-supervised graph contrastive learning method is employed to augment the vectors containing sentiment features,the model’s ability to infer on unseen samples is improved.Experimental results on the public dataset NLPCC2016 Chinese Weibo stance detection demonstrate that,based on five targets,the proposed model improves the macro-F1 score by 10% over baseline models,which proves its effectiveness in zero-shot stance detection scenarios.

Key words: Zero-shot stance detection, Emotion dictionary, Chain-of-thought, Prototype, Contrastive learning

中图分类号: 

  • TP520
[1]ALLAWAY E,SRIKANTH M,MCKEOWNK,et al.Adversarial Learning for Zero-Shot Stance Detection on Social Media[C]//Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics.Human Language Technologies:2021:4756-4767.
[2]ZHANG H,LI Y Z,ZHU T F,et al.Commonsense-based adversarial learning framework for zero-shot stance detection[J].Neurocomputing,2024,563(2024):12693.
[3]UPADHYAYA A,FISICHELLA M,NEJDL W.Intensity-Va-lued Emotions Help Stance Detection of Climate Change Twitter Data[C]//Proceedings of the Thirty-Second International Joint Conference on Artificial Intelligence.2023.
[4]XU R F,ZHOU Y,WU D Y,et al.Overview of NLPCC Shared Task 4:Stance Detection in Chinese Microblogs[C]//NLPCC-ICCPOL.2016:907-916.
[5]SUN Q Y,WANG Z Q,ZHU Q M,et al.Exploring VariousLinguistic Features for Stance Detection[C]//NLPCC-ICCPOL.2016:840-847.
[6]KÜÇÜK D,CAN F L.Supplementary Material for:Stance Detection:A Survey[J].Association for Computing Machinery,2020,53(12).
[7]ALLAWAY E,MCKEOWN K.Zero-Shot Stance Detection:A Dataset and Model using Generalized Topic Representationsd[C]//EMNLP.2020:8913-8931.
[8]HE Z,MOKHBERIAN N,LERMAN K,et al.Infusing know-ledge from wikipedia to enhance stance detection[C]//Procee-dings of the 12th Workshop on Computational Approaches to Subjectivity.2022:71-77.
[9]LUO Y,LIU Z,SHI Y,et al.Exploiting sentiment and common sense for zero-shot stance detection[C]//Proceedings of the 29th International Conference on Computational Linguistics.2022:7112-7123.
[10]REVEILHAC M,SCHNEIDER G.Replicable semi-supervisedapproaches to state-of-the-art stance detection of tweets[J].Information Processing & Management,2023,60(2):103199.
[11]JI H,LIN Z,FU P,et al.Cross-target stance detection via refined meta-learning[C]//2022 IEEE International Conference on Acoustics(ICASSP 2022).2022:7822-7826.
[12]LIU R,LIN Z,FU P,et al.Connecting targets via latent topics and contrastive learning:A unified framework for robust zero-shot and few-shot stance detection[C]//IEEE International Conference on Acoustics.2022:7812-7816.
[13]CHEN T,KORNBLITH S,NOROUZI M,et al.A SimpleFramework for Contrastive Learning of Visual Representations[C]//International Conference on Machine Learning.2020.
[14]WEI P,MAO W.Modeling transferable topics for cross-targetstance detection[C]//Proceedings of the 42nd International ACM SIGIR Conference on Research and Development in Information Retrieval.2019:1173-1176.
[15]ALLAWAY E,SRIKANTH M,MCKEOWN K.Adversariallearning for zero-shot stance detection on social media[C]//Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics.2021:4756-4767.
[16]BALY R,DA SAN MARTINO G,GLASS J,et al.We can detect your bias:Predicting the political ideology of news articles[C]//EMNLP.2020:4982-4991.
[17]LIANG B,CHEN Z,GUI L,et al.Zero-shot stance detection via contrastive learning[C]//Proceedings of the ACM Web Conference.2022:2738-2747.
[18]ZUBIAGA A,KOCHKINA E,LIAKATA M,et al.Discourse-aware rumour stance classification in social media using sequential classifiers[J].Information Processing & Management,2018,54(2):273-290.
[19]SUN Q Y,WANG Z Q,LI S S,et al.detection via sentiment information and neural network model[J].Frontiers of Computer Science.2019,13(1):127-138.
[20]WANG C,ZHANG Y,YU X,et al.Adversarial network with external knowledge for zero-shot stance detection[C]//Procee-dings of the 22nd Chinese National Conference on Computational Linguistics.2023:824-835.
[21]ZHANG X L F,BEAUCHAMP N,LU W.Generative Entity-to-Entity Stance Detection with Knowledge Graph Augmentation[C]//Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing.2022:9950-9969.
[22]AUGENSTEIN I,ROCKTÄSCHEL T,VLA-CHOS A,et al.Stance detection with bidirectional conditional encoding[C]//Proceedings ofthe 2016 Conference on Empirical Methods in Natural Language Processing.2016:876-885.
[23]XU C,PARIS C,NEPAL S,et al.Cross-target stance classification with self-attention networks[C]//Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics.2018:778-783.
[24]DEVLIN J,CHANG M W,LEEK,et al.BERT:Pretraining of deep bidirectional transformers for language understandingd[C]//Proceedings of NAACL HCT.2019:4171-4186.
[25]YU N,PAN D,ZHANG M,et al.Stance detection in ChineseMicroblogs with Neural Networks[M]//Natural Language Understanding and Intelliegent Applications.Springer International Publishing,2016:893-900.
[26]ZUBIAGA A,KOCHKINA E,LIAKATA M,et al.Discourse-aware rumour stance classification in social media using sequential classifiers[J].Information Processing & Management,2018,54(2):273-290.
Viewed
Full text


Abstract

Cited
  Shared   
  Discussed   
No Suggested Reading articles found!
渝ICP备16013121号-4
地址：重庆市渝北区洪湖西路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