基于知识指导的零样本立场检测

doi:10.3778/j.issn.1002-8331.2405-0095

摘要/Abstract

摘要： 立场检测是从给定的文本中推测其对特定话题的看法和态度，如支持、反对、中立等。零样本立场检测则是针对未知话题的数据进行立场极性的检测。为解决零样本立场检测中由于测试数据和训练数据话题不匹配导致的检测准确率低的问题，提出基于知识指导的立场检测模型，通过引入方面级情感信息和维基百科知识来增强模型的通用检测性能。具体而言，训练一个支持跨多领域的方面级情感分析模型，将其输出作为立场检测模型中的方面级情感信息，再将待检测话题在维基百科中的解释一并输入立场检测模型，进行立场检测。实验结果表明，该模型在公开的立场检测数据集VAST上的检测效果优于已有的方法，F1分数相比其他主流方法至少提升了1.7个百分点。同时，消融实验也证明了模型中各知识模块的必要性。

关键词: 零样本立场检测（ZSSD）, 方面级情感分析, 维基百科

Abstract: Stance detection is to infer views and attitudes on a particular topic from a given text, such as support, opposition, neutrality, etc. Zero-shot stance detection is a stance polarity test based on data about unknown topics. In order to solve the problem of low detection accuracy caused by the mismatch between test data and training data in zero-shot stance detection, a knowledge-guided stance detection model is proposed, and the general detection performance of the model is enhanced by introducing aspect-based sentiment information and WikiPedia knowledge. Specifically, the model first trains an aspect-based sentiment analysis model that supports cross-domain support, and uses its output as the aspect-level sentiment information in the stance detection model, and then inputs the explanation of the topic to be detected in Wikipedia into the stance detection model for stance detection. Experimental results show that the detection effect of the model on the public stand detection dataset VAST is better than that of the existing methods, and the F1 score is at least 1.7 percentage points higher than that of other mainstream methods. At the same time, the ablation experiment also confirms the necessity of each knowledge module in the model.

Key words: zero-shot stance detection (ZSSD), aspect-based sentiment analysis, WikiPedia

韩冬晖, 李邵梅, 高超. 基于知识指导的零样本立场检测[J]. 计算机工程与应用, 2025, 61(14): 223-229.

HAN Donghui, LI Shaomei, GAO Chao. Knowledge-Guided Zero-Shot Stance Detection[J]. Computer Engineering and Applications, 2025, 61(14): 223-229.

参考文献

[1] ZHANG S, QIU L, CHEN F, et al. “We make choices we think are going to save us”: debate and stance identification for online breast cancer CAM discussions[C]//Proceedings of the 26th International Conference on World Wide Web Companion, 2017: 1073-1081.
[2] QIU M H, SIM Y, SMITH N A, et al. Modeling user arguments, interactions, and attributes for stance prediction in online debate forums[C]//Proceedings of the SIAM International Conference on Data Mining, 2015: 855-863.
[3] MOHAMMAD S M, SOBHANI P, KIRITCHENKO S. Stance and sentiment in tweets[J]. ACM Transactions on Internet Technology, 2017, 17(3): 1-23.
[4] Kü?üK D, CAN F. Stance detection[J]. ACM Computing Surveys, 2021, 53(1): 1-37.
[5] ALLAWAY E, MCKEOWN K. Zero-shot stance detection: a dataset and model using generalized topic representations[C]//Proceedings of the Conference on Empirical Methods in Natural Language Processing. Stroudsburg: ACL, 2020: 8913-8931.
[6] ALLAWAY E, SRIKANTH M, MCKEOWN K. Adversarial learning for zero-shot stance detection on social media[C]//Proceedings of the Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. Stroudsburg: ACL, 2021: 4756-4767.
[7] NGUYEN D Q, VU T, TUAN NGUYEN A. BERTweet: a pre?trained language model for English tweets[C]//Proceedings of the Conference on Empirical Methods in Natural Language Processing: System Demonstrations. Stroudsburg: ACL, 2020: 9-14.
[8] XU H Z, VUCETIC S, YIN W P. OpenStance: real-world zero-shot stance detection[C]//Proceedings of the 26th Conference on Computational Natural Language Learning. Stroudsburg: ACL, 2022: 314-324.
[9] HE Z H, MOKHBERIAN N, LERMAN K. Infusing knowledge from Wikipedia to enhance stance detection[C]//Proceedings of the 12th Workshop on Computational Approaches to Subjectivity, Sentiment & Social Media Analysis. Stroudsburg: ACL, 2022: 71-77.
[10] LUO Y, LIU Z H, SHI Y F, 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.
[11] WALKER M A, ANAND P, ABBOTT R, et al. Stance classification using dialogic properties of persuasion[C]//Proceedings of the Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, 2012: 592-596.
[12] HASAN K, NG V. Stance classification of ideological debates: data, models, features, and constraints[C]//Proceedings of the 6th International Joint Conference on Natural Language Processing, 2013: 1348-1356.
[13] LIANG B, FU Y H, GUI L, et al. Target-adaptive graph for cross-target stance detection[C]//Proceedings of the Web Conference. New York: ACM, 2021: 3453-3464.
[14] LIANG B, CHEN Z X, GUI L, et al. Zero-shot stance detection via contrastive learning[C]//Proceedings of the ACM Web Conference. New York: ACM, 2022: 2738-2747.
[15] WANG Z Q, SUN Q Y, LI S S, et al. Neural stance detection with hierarchical linguistic representations[J]. ACM Transactions on Audio, Speech, and Language Processing, 2020, 28: 635-645.
[16] SEN I, FL?CK F, WAGNER C. On the reliability and validity of detecting approval of political actors in tweets[C]//Proceedings of the Conference on Empirical Methods in Natural Language Processing. Stroudsburg: ACL, 2020: 1413-1426.
[17] BARACHI E M, ALKHATIB M, MATHEW S, et al. A novel sentiment analysis framework for monitoring the evolving public opinion in real-time: case study on climate change[J]. Journal of Cleaner Production, 2021, 312: 127820.
[18] ZHU L X, FANG Z, PERGOLA G, et al. Disentangled learning of stance and aspect topics for vaccine attitude detection in social media[C]//Proceedings of the Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. Stroudsburg: ACL, 2022: 1566-1580.
[19] BALY R, KARADZHOV G, AN J S, et al. What was written vs. who read it: news media profiling using text analysis and social media context[C]//Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics. Stroudsburg: ACL, 2020: 3364-3374.
[20] LIU R, LIN Z, TAN Y T, et al. Enhancing zero-shot and few-shot stance detection with commonsense knowledge graph[C]//Proceedings of the Association for Computational Linguistics. Stroudsburg: ACL, 2021: 3152-3157.
[21] DENG Y, ZHANG W X, PAN S J, et al. Bidirectional generative framework for cross-domain aspect-based sentiment analysis[C]//Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics. Stroudsburg: ACL, 2023: 12272-12285.
[22] SCHLICHTKRULL M, KIPF T N, BLOEM P, et al. Modeling relational data with graph convolutional networks[C]//Proceedings of the 15th International Semantic Web Conference. Cham: Springer International Publishing, 2018: 593-607.
[23] YANG B S, YIH W T, HE X D, et al. Embedding entities and relations for learning and inference in knowledge bases[J]. arXiv:1412.6575, 2014.
[24] AUGENSTEIN I, ROCKT?SCHEL T, VLACHOS A, et al. Stance detection with bidirectional conditional encoding[C]//Proceedings of the Conference on Empirical Methods in Natural Language Processing. Stroudsburg: ACL, 2016: 876-885.
[25] 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. Stroudsburg: ACL, 2018: 778-783.
[26] KIPF T N, WELLING M. Semi-supervised classification with graph convolutional networks[J]. arXiv:1609.02907, 2016.
[27] VASHISHTH S, SANYAL S, NITIN V, et al. Composition-based multi-relational graph convolutional networks[J]. arXiv:1911.03082, 2019.