基于多视图表征的虚假新闻检测

doi:10.3969/j.issn.1671-1122.2024.03.009

摘要/Abstract

摘要：

社交网络已经成为人们日常生活中获取和分享信息的主要渠道，同时也为虚假新闻的传播提供了捷径。如今，针对网络虚假新闻的检测问题受到学术界的广泛关注，但目前的检测方法缺乏基于新闻多个视角的深度探索或忽视了新闻中不同信息传播方向不同的问题，有待改进。文章提出一种基于新闻内容、用户信息和新闻传播3种视角的多视图表征和检测的模型MVRFD（Multi-View Representations for Fake News Detection），为虚假新闻检测任务提供更全面的视角。首先，利用协同注意力机制表征新闻内容中的多模态信息，使用具有不同方向的图神经网络聚合新闻传播过程中的用户信息和观点信息；然后，利用双协同注意力机制实现多个视角间的信息交互；最后，将新闻内容特征和新闻上下文特征进行融合。在公开数据集上的实验结果表明，文章所提出的模型实现了96.7%的准确率和96.8%的F1值，优于主流的文本处理模型以及基于单视角的检测模型。

关键词: 虚假新闻检测, 图神经网络, 多模态表征, 注意力机制, 多视图表征

Abstract:

Social networks have become a major channel for people to access and share information in their daily lives, while also providing shortcuts for the spread of fake news. Nowadays, the detection of online fake news has been widely concerned and studied by the academic community, but the current methods lack in-depth exploration based on multiple perspectives of news or ignore the different directions of different information in news. In order to provide a more comprehensive perspective for fake news detection task, this paper proposed a multi-view representations for fake news detection (MVRFD) model based on three perspectives: news content, user information and news propagation. Firstly, the co-attention mechanism was used to represent the multimodal information in news content, and the graph neural network with different directions was used to aggregate user information and views in the process of news transmission. Then the dual-co-attention mechanism was used to realize the information interaction between multiple perspectives. Finally, the features of news content and the news context were integrated. Experiments on the publicly available dataset show that the proposed model achieves 96.7% accuracy and 96.8% F1 score, which are better than the mainstream text processing models and single-view-based detection models.

Key words: fake news detection, graph neural network, multimodal representation, attention mechanism, multi-view representation

中图分类号:

TP309

张新有, 孙峰, 冯力, 邢焕来. 基于多视图表征的虚假新闻检测[J]. 信息网络安全, 2024, 24(3): 438-448.

ZHANG Xinyou, SUN Feng, FENG Li, XING Huanlai. Multi-View Representations for Fake News Detection[J]. Netinfo Security, 2024, 24(3): 438-448.

图/表 8

图1

图2

图3

表1

表2

图4

图5

图6

参考文献 30

[1]	YANG Fan, YU Xiaohui, LIU Yang, et al. Automatic Detection of Rumor on Sina Weibo[C]// ACM. Proceedings of the ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. New York: ACM, 2012: 1-7.
[2]	ZHAO Zhe, RESNICK P, MEI Qiaozhu. Enquiring Minds: Early Detection of Rumors in Social Media from Enquiry Posts[C]// ACM. The 24th International Conference on World Wide Web. New York: ACM, 2015: 1395-1405.
[3]	HOOI B, SHAH N, BEUTEL A, et al. BIRDNEST: Bayesian Inference for Ratings-Fraud Detection[C]// SIAM. The 2016 SlAM Internationa Conference on Data Mining. Philadelphia: SIAM, 2016: 495-503.
[4]	GIRGIS S, AMER E, GADALLAH M. Deep Learning Algorithms for Detecting Fake News in Online Text[C]// IEEE. The 13th International Conference on Computer Engineering and Systems. New York: IEEE, 2018: 93-97.
[5]	DEVLIN J, CHANG Mingwei, LEE K, et al. BERT: Pretraining of Deep Bidirectional Transformers for Language Understanding[C]// ACL. Proceedings of NAACL-HLT. Stroudslourg: ACL 2019, 1: 4171-4186.
[6]	DU Pengfei, LI Xiaoyong, GAO Yali. Survey on Multimodal Visual Language Representation Learning[J]. Journal of Software, 2021, 32(2): 327-348.
	杜鹏飞, 李小勇, 高雅丽. 多模态视觉语言表征学习研究综述[J]. 软件学报, 2021, 32(2): 327-348.
[7]	WU Ke, SONG Yang, ZHU K Q. False Rumors Detection on Sina Weibo by Propagation Structures[C]// IEEE. The 31st IEEE International Conference on Data Engineering. New York: IEEE, 2015: 651-662.
[8]	BIAN Tian, XIAO Xi, XU Tingyang, et al. Rumor Detection on Social Media with Bi-Directional Graph Convolutional Networks[C]// AAAI. Proceedings of the AAAI Conference on Artificial Intelligence. Menlo Park: AAAI, 2020, 34(1): 549-556.
[9]	SONG Chenguang, TENG Yiyang, ZHU Yangfu, et al. Dynamic Graph Neural Network for Fake News Detection[J]. Neurocomputing, 2021(505): 27-31.
[10]	SONG Chenguang, SHU Kai, WU Bin, et al. Temporally Evolving Graph Neural Network for Fake News Detection[J]. Information Processing & Management, 2021, 58(6): 102712. doi: 10.1016/j.ipm.2021.102712 URL
[11]	SONG Chenguang, NING Nianwen, ZHANG Yunlei, et al. A Multimodal Fake News Detection Model Based on Crossmodal Attention Residual and Multichannel Convolutional Neural Networks[J]. Information Processing & Management, 2021, 58(1): 102437. doi: 10.1016/j.ipm.2020.102437 URL
[12]	XIONG Shufeng, ZHANG Guipeng, BATRA V, et al. TRIMOON: Two-Round Inconsistency-Based Multi-Modal Fusion Network for Fake News Detection[J]. Information Fusion, 2023, 93: 150-158. doi: 10.1016/j.inffus.2022.12.016 URL
[13]	QI Peng, CAO Juan, LI Xirong, et al. Improving Fake News Detection by Using an Entity-Enhanced Framework to Fuse Diverse Multimodal Clues[C]// ACM. Proceedings of the 29th ACM International Conference on Multimedia. New York: ACM, 2021: 1212-1220.
[14]	CHEN Yiyuan, LI Dongsheng, ZHANG Peng, et al. Cross-Modal Ambiguity Learning for Multimodal Fake News Detection[C]// ACM. The ACM Web Conference. New York: ACM, 2022: 2897-2905.
[15]	MA Jing, GAO Wei, MITRA P, et al. Detecting Rumors from Microblogs with Recurrent Neural Networks[C]// ACM. The 25th International Joint Conference on Artificial Intelligence. New York: ACM, 2016: 3818-3824.
[16]	WEI Lingwei, HU Dou, ZHOU Wei, et al. Uncertainty-Aware Propagation Structure Reconstruction for Fake News Detection[C]// ACL. The 29th International Conference on Computational Linguistics. Stroudsburg: ACL, 2022, 29(1): 2759-2768.
[17]	SUN Ling, RAO Yuan, LAN Yuqian, et al. HG-SL: Jointly Learning of Global and Local User Spreading Behavior for Fake News Early Detection[C]// AAAI. Proceedings of the 37th AAAI Conference on Artificial Intelligence. Menlo Park: AAAI, 2023, 37(4): 5248-5256.
[18]	DOU Yingtong, SHU Kai, XIA Congying, et al. User Preference-Aware Fake News Detection[C]// ACM. The 44th International ACM SIGIR Conference on Research and Development in Information Retrieval. New York: ACM, 2021: 2051-2055.
[19]	RAN Hongyan, JIA Caiyan, ZHANG Pengfei, et al. MGAT-ESM: Multi-Channel Graph Attention Neural Network with Event-Sharing Module for Rumor Detection[J]. Information Sciences, 2022, 592: 402-416. doi: 10.1016/j.ins.2022.01.036 URL
[20]	HAN Xiaohong, ZHAO Mengfan, ZHANG Yutao. Joint Heterogeneous Graph Convolutional Network and Attention Mechanism for Fake News[EB/OL]. (2022-11-28)[2023-12-01]. https://doi.org/10.20009/j.cnki.21-1106/TP.2022-0412.
	韩晓鸿, 赵梦凡, 张钰涛. 联合异质图卷积网络和注意力机制的假新闻检测[EB/OL]. (2022-11-28)[2023-12-01]. https://doi.org/10.20009/j.cnki.21-1106/TP.2022-0412.
[21]	BAZMI P, ASADPOUR M, SHAKERY A. Multi-View Co-Attention Network for Fake News Detection by Modeling Topic-Specific User and News Source Credibility[J]. Information Processing and Management, 2023, 60(1): 103146. doi: 10.1016/j.ipm.2022.103146 URL
[22]	YING Qichao, HU Xiaoxiao, ZHOU Yangming, et al. Bootstrapping Multi-View Representations for Fake News Detection[C]// AAAI. The 37th AAAI Conference on Artificial Intelligence. Menlo Park: AAAI, 2023, 37(4): 5384-5392.
[23]	SIMONYAN K, ZISSERMAN A. Very Deep Convolutional Networks for Large-Scale Image Recognition[C]// IEEE. The 3th International Conference on Learning Representations. New York: IEEE, 2015: 730-734.
[24]	LU Jiasen, BATRA D, PARIKH D, et al. ViLBERT: Pretraining Task-Agnostic Visiolinguistic Representations for Vision-and-Language Tasks[C]// ACM. The 33nd International Conference on Neural Information Processing Systems. New York: ACM, 2019: 13-23.
[25]	KIPF T N, WELLING M. Semi-Supervised Classification with Graph Convolutional Networks[C]// IEEE. The 5th International Conference on Learning Representations. New York: IEEE, 2017: 11305-11312.
[26]	VELIČKOVIĆ P, CUCURULL G, CASANOVA A, et al. Graph Attention Networks[EB/OL]. (2017-10-30)[2023-12-01]. https://arxiv.org/abs/1710.10903.
[27]	SINGH J P, KUMAR A, RANA N P, et al. Attention-Based LSTM Network for Rumor Veracity Estimation of Tweets[J]. Information Systems Frontiers, 2022, 24(2): 459-474. doi: 10.1007/s10796-020-10040-5
[28]	KIM Y. Convolutional Neural Networks for Sentence Classification[C]// ACL. 2014 Conference on Empirical Methods in Natural Language Processing. Stroudsburg: ACL, 2014: 1746 -1751.
[29]	LIU Xiaoyang, ZHAO Zhengyang, ZHANG Yihao, et al. Social Network Rumor Detection Method Combining Dual-Attention Mechanism with Graph Convolutional Network[J]. IEEE Transactions on Computational Social Systems, 2023, 10(5): 2350-2361. doi: 10.1109/TCSS.2022.3184745 URL
[30]	BAI Lin, HAN Xueming, JIA Caiyan. A Rumor Detection Model Incorporating Propagation Path Contextual Semantics and User Information[J]. Neural Processing Letters, 2023, 55(7): 9831-9850. doi: 10.1007/s11063-023-11229-w

统计数据	数量
帖子	3805656个
用户	2746818个
图片	3845张
新闻事件	4664个
真实新闻	2351个
虚假新闻	2313个
事件的平均传播时间	2460.7 h
每个事件平均帖子数	816个
单个事件最大帖子数	59318个
单个事件最小帖子数	10个
最小转发深度	2
最大转发深度	73
平均转发深度	6.19
转发深度超过5的事件数	2198个

Model	准确率	虚假新闻			真实新闻
Model	准确率	精确度	召回率	F1	精确度	召回率	F1
LSTM	87.8 %	88.5 %	87.7 %	87.9 %	86.5 %	88.1 %	87.0 %
BERT	89.7 %	88.7 %	91.4 %	89.9 %	90.5 %	87.9 %	89.0 %
BERT+Attention	91.1 %	91.0 %	91.8 %	91.3 %	91.3 %	90.1 %	90.6 %
TextCNN	90.4 %	90.3 %	91.5 %	90.5 %	91.0 %	89.9 %	90.0 %
CARMN	92.5 %	92.2 %	92.7 %	92.4 %	92.7 %	92.3 %	92.5 %
Bi-GCN	94.4 %	92.6 %	96.8 %	94.6 %	96.6 %	92.0 %	94.1 %
DA-GCN	94.4 %	94.1 %	94.6 %	94.4 %	94.7 %	94.1 %	94.4 %
DAN-Tree	95.8 %	94.6 %	97.2 %	95.8 %	97.2 %	94.5 %	95.8 %
MVRFD	96.7 %	96.0 %	97.7 %	96.8 %	97.5 %	95.5 %	96.4 %