基于对比学习的零样本对象谣言检测

doi:10.3785/j.issn.1008-973X.2024.09.004

浙江大学学报(工学版)

2024, Vol. 58

Issue (9): 1790-1800 DOI: 10.3785/j.issn.1008-973X.2024.09.004

计算机与控制工程

基于对比学习的零样本对象谣言检测

陈珂1(

),张文浩2

1. 广东石油化工学院计算机学院，广东茂名 525000
2. 广东石油化工学院电子信息工程学院，广东茂名 525000

Zero-shot object rumor detection based on contrastive learning

Ke CHEN1(

),Wenhao ZHANG2

1. School of Computer, Guangdong University of Petrochemical Technology, Maoming 525000, China
2. School of Electronic and Information Engineering, Guangdong University of Petrochemical Technology, Maoming 525000, China

全文: PDF(931 KB) HTML

摘要：

现有的谣言检测模型通常依赖大规模人工标注的谣言数据集，标注成本高且谣言特征来源于已被辟谣的谣言. 为了提高模型对未知谣言的检测能力，提出面向不同对象的谣言检测方法. 基于零样本学习，将谣言数据集按照不同的对象划分为样本与内容互不重叠的多个数据集，从而实现零样本对象谣言检测任务；为了表征对象之间的关系构建通义掩码特征，从而设计区分通义掩码特征的代理任务；为了减少数据增强带来的噪声，引入面向对象的信息辅助文本作为特征，并将其与原语义向量进行线性变换. 在此基础上，提出面向零样本对象谣言检测的基于代理任务的分层对比学习模型(ZPTHCL)，可以通过迁移学习进行谣言检测. 在一个基于对象的零样本谣言数据集和Ma-Weibo、Weibo20、Twitter15、Twitter16这4个公开数据集上进行实验，结果表明所提出的对比学习零样本对象谣言检测模型性能更优.

关键词： 谣言检测; 零样本学习; 迁移学习; 代理任务; 对比学习

Abstract:

Existing rumor detection models often rely on large-scale manually annotated rumor datasets, which are costly and limited in their ability to detect unknown rumors due to the reliance on features derived from debunked rumors. To address this limitation, an approach for rumor detection targeted at different objects was proposed. Leveraging the zero-shot learning, the rumor dataset was divided into multiple datasets with non-overlapping samples and contents based on different objects, enabling the zero-shot object-oriented rumor detection task. Correspondingly, a universal mask feature was constructed to represent the relationship between objects, and a proxy task was designed to differentiate the universal mask feature. Additionally, object-oriented information-assisted text was introduced to reduce noise caused by data augmentation and was linearly transformed with the original vector semantics. Then, a proxy task-based hierarchical contrastive learning model (ZPTHCL) was presented for zero-shot object-oriented rumor detection, which leveraged transfer learning for rumor detection. Finally, experiments were conducted on a zero-shot rumor dataset based on objects and four publicly available datasets, Ma-Weibo, Weibo20, Twitter15 and Twitter16, demonstrating superior performance of the proposed contrastive learning zero-shot object-oriented rumor detection model.

Key words: rumor detection zero-shot learning transfer learning proxy task contrastive learning

收稿日期: 2023-05-20 出版日期: 2024-08-30

CLC:

TP 18

基金资助: 国家自然科学基金资助项目(61172145)；广东省自然科学基金资助项目(2018A030307032)；广东省普通高校重点科研平台和项目(2020ZDZX3038).

作者简介: 陈珂（1964—2024），男，教授，从事机器学习与数据挖掘、自然语言处理研究. orcid.org/0000-0002-9341-9526

	服务
	把本文推荐给朋友
	加入引用管理器
	E-mail Alert
	作者相关文章
	陈珂
	张文浩

引用本文:

陈珂,张文浩. 基于对比学习的零样本对象谣言检测[J]. 浙江大学学报(工学版), 2024, 58(9): 1790-1800.

Ke CHEN,Wenhao ZHANG. Zero-shot object rumor detection based on contrastive learning. Journal of ZheJiang University (Engineering Science), 2024, 58(9): 1790-1800.

链接本文:

https://www.zjujournals.com/eng/CN/10.3785/j.issn.1008-973X.2024.09.004 或 https://www.zjujournals.com/eng/CN/Y2024/V58/I9/1790

图 1 ZPTHCL的模型结构

图 2 数据增强过程

图 3 主题相关词示例

图 4 掩码样本示例

图 5 各个对象所对应的信息辅助文本

表 1 4个谣言检测数据集的数据统计

表 2 零样本对象谣言数据集Zeo-Weibo统计

表 3 不同方法在4个谣言检测数据集上的准确度

表 4 不同方法在Zeo-Weibo数据集上的实验结果

表 5 由中文训练集至英文测试集上的谣言检测结果

表 6 由英文训练集至中文测试集上的谣言检测结果

表 7 ZPTHCL模型在4个谣言检测数据集上缺乏标签的情况下的准确度

表 8 在7个对象数据集上的消融实验结果

1	KANTAR M. Social Media Trends [R]. London: Kantar Media, 2019.
2	KAPFERER J. Rumeurs-Le plus vieux média du monde [M]// Pari: Editions du Seuil, 1987: 31−33.
3	LAROCHELLE H, ERHAN D, BENGIO Y. Zero-data learning of new tasks [C]// Proceedings of the 23rd AAAI Conference on Artificial Intelligence . Chicago: AAAI Press, 2008: 646−651 .
4	CHANG M W, RATINOV L, ROTH D, et al. Importance of semantic representation: dataless classification [C]// Proceedings of the 23rd AAAI Conference on Artificial Intelligence. Chicago: AAAI Press, 2008: 830−835.
5	LIN H, YI P, MA J, et al. Zero-shot rumor detection with propagation structure via prompt learning [C]// Proceedings of the AAAI Conference on Artificial Intelligence . Washington: AAAI Press, 2023: 5213−5221.
6	SONG Y, UPADHYAY S, PENG H, et al Toward any-language zero-shot topic classification of textual documents[J]. Artificial Intelligence, 2019, 274 (C): 133- 150
7	SONG Y, UPADHYAY S, PENG H, et al. Cross-lingual dataless classification for many languages [C]// Proceedings of the 25th International Joint Conference on Artificial Intelligence . New York: AAAI Press, 2016: 2901−2907.
8	GOODFELLOW I, POUGET-ABADIE J, MIRZA M, et al Generative adversarial networks[J]. Communications of the ACM, 2020, 63 (11): 139- 44 doi: 10.1145/3422622
9	KINGMA D P, WELLING M. Auto-encoding variational bayes [C]// Proceedings of the International Conference on Learning Representations . Ithaca: ArXiv, 2014: 14−16.
10	CHEN T, KORNBLITH S, NOROUZI M, et al. A simple framework for contrastive learning of visual representations [C]// Proceedings of the International Conference on Machine Learning . [s. l. ]: PMLR, 2020: 1597−1607.
11	HE K, FAN H, WU Y, et al. Momentum contrast for unsupervised visual representation learning [C]// Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition . Seattle: IEEE, 2020: 9726−9735.
12	LIANG B, CHEN Z X, GUI L, et al. Zero-shot stance detection via contrastive learning [C]// Proceedings of the ACM Web Conference. Lyon: ACM, 2022: 2738−2747.
13	VICARIO M D, QUATTROCIOCCHI W, SCALA A, et al Polarization and fake news: early warning of potential misinformation targets[J]. ACM Transactions on the Web, 2019, 13 (2): 1- 22
14	MEEL P, VISHWAKARMA D K Fake news, rumor, information pollution in social media and web: a contemporary survey of state-of-the-arts, challenges and opportunities[J]. Expert Systems with Applications, 2020, 153 (1): 112986
15	WANG Z, GUO Y Rumor events detection enhanced by encoding sentimental information into time series division and word representations[J]. Neurocomputing, 2020, 397 (2): 224- 243
16	KUMAR S, CARLEY K M. Tree LSTMs with convolution units to predict stance and rumor veracity in social media conversations [C]// Proceedings of the 57th annual meeting of the association for computational linguistics . Florence: ACL, 2019: 5047−5058.
17	BIAN T, XIAO X, XU T, et al. Rumor detection on social media with bi-directional graph convolutional networks [C]// Proceedings of the AAAI Conference on Artificial Intelligence . New York: AAAI Press, 2020: 546−556.
18	ZHANG Q, LIPANI A, LIANG S, et al. Reply-aided detection of misinformation via bayesian deep learning [C]// Proceedings of the World Wide Web Conference . San Francisco: ACM, 2019: 2333−2343.
19	RIEDEL B, AUGENSTEIN I, SPITHOURAKIS G P, et al. A simple but tough-to-beat baseline for the fake news challenge stance detection task [EB/OL]. (2018−05−21). https://doi.org/10.48550/arXiv.1707.03264.
20	LU Y J, LI C T. GCAN: graph-aware co-attention networks for explainable fake news detection on social media [C]// Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics . [s. l. ]: ACL, 2020: 505−514.
21	RAO D, MIAO X, JIANG Z, et al. STANKER: stacking network based on level-grained attention-masked BERT for rumor detection on social media [C]// Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing . Online and Punta Cana: ACL, 2021: 3347−3363.
22	CHEN X, ZHOU F, TRAJCEVSKI G, et al Multi-view learning with distinguishable feature fusion for rumor detection[J]. Knowledge-Based Systems, 2022, 240 (8): 108085
23	XU Y, GUO J, QIU W, et al. "Comments matter and the more the better!": improving rumor detection with user comments [C]// International Conference on Trust, Security and Privacy in Computing and Communications . Wuhan: IEEE, 2022: 383−390.
24	PUSHP P K, SRIVASTAVA M M. Train once, test anywhere: zero-shot learning for text classification [EB/OL]. (2017−12−23). https://doi.org/10.48550/arXiv.1 712.05972.
25	陆恒杨, 范晨悠, 吴小俊. 面向网络社交媒体的少样本新冠谣言检测 [J]. 中文信息学报, 2022, 36(1): 135−144. LU Hengyang, FAN Chenyou, WU Xiaojun. Few-shot COVID-19 rumor detection for online social media [J]. Journal of Chinese Information Processing . 2022, 36(1): 135−144.
26	ZHOU H, MA T, RONG H, et al MDMN: multi-task and domain adaptation based multi-modal network for early rumor detection[J]. Expert Systems with Applications, 2022, 195 (3): 116517
27	RAN H, JIA C. Unsupervised cross-domain rumor detection with contrastive learning and cross-attention [C]// Proceedings of the AAAI Conference on Artificial Intelligence . Washington: AAAI Press, 2023: 13510−13518.
28	MA J, GAO W, MITRA P, et al. Detecting rumors from microblogs with recurrent neural networks [C]// Proceedings of the 25th International Joint Conference on Artificial Intelligence . New York: AAAI Press, 2016: 3818−3824.
29	DEVLIN J, CHANG M, LEE K, et al. BERT: pre-training of deep bidirectional transformers for language understanding [C]// Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics . Minneapolis: ACL, 2019: 4171−4186.
30	BLEI D M, NG A Y, JORDAN M I Latent dirichlet allocation[J]. Journal of Machine Learning Research, 2003, 3 (1): 993- 1022
31	MA J, GAO W, WEI Z, et al. Detect rumors using time series of social context information on microblogging websites [C]// Proceedings of the 24th ACM International on Conference on Information and Knowledge Management . Melbourne : ACM , 2015: 1751−1754.
32	MA J, GAO W, WONG K F. Detect rumors in microblog posts using propagation structure via kernel learning [C]// Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics . Vancouver: ACL, 2017: 708−717.
33	LIU Z, WEI Z, ZHANG R Rumor detection based on convolutional neural network[J]. Journal of Computer Applications, 2017, 37 (11): 3053
34	SUJANA Y, LI J, KAO H Y. Rumor detection on twitter using multiloss hierarchical bilstm with an attenuation factor [C]// Asian Chapter of the Association for Computational Linguistics . [s. l. ]: ACL, 2020: 18−26.
35	RANI N, DAS P, BHARDWAJ A K. A hybrid deep learning model based on CNN-BiLSTM for rumor detection [C]// Proceedings of the 2021 6th International Conference on Communication and Electronics Systems . Coimbatre: IEEE, 2021: 1423−1427.
36	MA J, GAO W, JOTY S, et al An attention-based rumor detection model with tree-structured recursive neural networks[J]. ACM Transactions on Intelligent Systems and Technology, 2020, 11 (4): 1- 28
37	TU K, CHEN C, HOU C, et al Rumor2vec: a rumor detection framework with joint text and propagation structure representation learning[J]. Information Sciences, 2021, 560 (1): 137- 151
38	LIU Y, OTT M, GOYAL N, et al. Roberta: a robustly optimized Bert pretraining approach [C]// Proceedings of the 20th Chinese National Conference on Computational Linguistics . Huhhot: Chinese Information Processing Society of China, 2021: 1218−1227.
39	BELTAGY I, PETERS M E, COHAN A. Longformer: the long-document transformer [EB/OL]. [2020-12-02]. https://doi.org/10.48550/arXiv.2004.05150.
40	KHOO L M S, CHIEU H L, QIAN Z, et al. Interpretable rumor detection in microblogs by attending to user interactions [C]// Proceedings of the AAAI Conference on Artificial Intelligence . California: AAAI Press, 2020: 8783-8790.
41	WU Y, ZENG Y, YANG J, et al Weibo rumor recognition based on communication and stacking ensemble learning[J]. Discrete Dynamics in Nature and Society, 2020, 2020: 1- 12
42	RISCH J, KRESTEL R. Bagging bert models for robust aggression identification [C]// Proceedings of the Second Workshop on Trolling, Aggression and Cyberbullying . Marseille: ELRA, 2020: 55−61.
43	GENG Y, LIN Z, FU P, et al. Rumor detection on social media: a multi-view model using self-attention mechanism [C]// Proceedings of the Computational Science-ICCS 2019: 19th International Conference . Faro: Springer-Verlag, 2019: 339−352.

[1]	周姿含,王叙萌,陈为. 基于迁移学习的交互时序数据可视化生成方法[J]. 浙江大学学报(工学版), 2024, 58(2): 239-246.
[2]	周天琪,杨艳,张继杰,殷少伟,郭增强. 基于无负样本损失和自适应增强的图对比学习[J]. 浙江大学学报(工学版), 2023, 57(2): 259-266.
[3]	顾桂梅,贾耀华,赵岩浩,张文辉,闫炳旭. 基于组合零样本学习的接触网吊弦线缺陷识别[J]. 浙江大学学报(工学版), 2023, 57(11): 2285-2293.
[4]	华夏,王新晴,芮挺,邵发明,王东. 视觉感知的无人机端到端目标跟踪控制技术[J]. 浙江大学学报(工学版), 2022, 56(7): 1464-1472.
[5]	王友卫,童爽,凤丽洲,朱建明,李洋,陈福. 基于图卷积网络的归纳式微博谣言检测新方法[J]. 浙江大学学报(工学版), 2022, 56(5): 956-966.
[6]	高一聪,王彦坤,费少梅,林琼. 基于迁移学习的机械制图智能评阅方法[J]. 浙江大学学报(工学版), 2022, 56(5): 856-863, 889.
[7]	曾素佳,庞善民,郝问裕. 深度监督对齐的零样本图像分类方法[J]. 浙江大学学报(工学版), 2022, 56(11): 2204-2214.
[8]	付晓峰,牛力. 基于深度卷积和自编码器增强的微表情判别[J]. 浙江大学学报(工学版), 2022, 56(10): 1948-1957.
[9]	陈智超,焦海宁,杨杰,曾华福. 基于改进MobileNet v2的垃圾图像分类算法[J]. 浙江大学学报(工学版), 2021, 55(8): 1490-1499.
[10]	康庄,杨杰,郭濠奇. 基于机器视觉的垃圾自动分类系统设计[J]. 浙江大学学报(工学版), 2020, 54(7): 1272-1280.
[11]	沈宗礼,余建波. 基于迁移学习与深度森林的晶圆图缺陷识别[J]. 浙江大学学报(工学版), 2020, 54(6): 1228-1239.
[12]	付晓峰,牛力,胡卓群,李建军,吴卿. 基于过渡帧概念训练的微表情检测深度网络[J]. 浙江大学学报(工学版), 2020, 54(11): 2128-2137.

Viewed

Full text

Abstract

Cited

Shared

Discussed