Please wait a minute...
Journal of ZheJiang University (Engineering Science)  2022, Vol. 56 Issue (5): 956-966    DOI: 10.3785/j.issn.1008-973X.2022.05.013
    
New inductive microblog rumor detection method based on graph convolutional network
You-wei WANG1(),Shuang TONG1,Li-zhou FENG2,Jian-ming ZHU1,Yang LI1,Fu CHEN1
1. School of Information, Central University of Finance and Economics, Beijing 100081, China
2. School of Statistics, Tianjin University of Finance and Economics, Tianjin 300222, China
Download: HTML     PDF(1246KB) HTML
Export: BibTeX | EndNote (RIS)      

Abstract  

A new inductive microblog rumor detection method based on graph convolutional networks (GCN) was proposed to solve the problems faced by traditional GCN in rumor detection, such as the insufficient consideration of word semantic information and the difficulty of selecting pooling methods. Firstly, the semantic relationship between words was considered. A microblog event graph construction method based on word semantic correlation was proposed by combining the traditional word co-occurrence based graph construction method, and the node information aggregation was realized by combining GCN and gate recurrent unit (GRU). Then, in order to effectively fuse the feature information of different nodes, a multiple pooling methods fusion strategy based on attention mechanism, which fused max-pooling, average-pooling and global-pooling, was proposed to obtain the final graph level vector. Finally, in order to improve the efficiency of microblog rumor detection, the influence of microblog comment time on detection results was explored, and the best comment utilization time threshold for model training was obtained. Experimental results show that the performance of the proposed method is generally better than that of Text-CNN, Bi-GCN, TextING and other typical methods on the given datasets, verifying its effectiveness in the field of microblog rumor detection.



Key wordsrumor detection      graph convolutional network      microblog event      gate recurrent unit      attention mechanism     
Received: 14 November 2021      Published: 31 May 2022
CLC:  TP 391  
Fund:  国家自然科学基金资助项目(61906220); 教育部人文社科资助项目(19YJCZH178); 国家社科基金资助项目(18CTJ008); 天津市自然科学基金资助项目(18JCQNJC69600); 内蒙古纪检监察大数据实验室2020-2021年度开放课题资助项目(IMDBD202002, IMDBD202004); 中央财经大学新兴交叉学科建设项目;中国高校产学研创新基金项目(2021FNA01002)
Cite this article:

You-wei WANG,Shuang TONG,Li-zhou FENG,Jian-ming ZHU,Yang LI,Fu CHEN. New inductive microblog rumor detection method based on graph convolutional network. Journal of ZheJiang University (Engineering Science), 2022, 56(5): 956-966.

URL:

https://www.zjujournals.com/eng/10.3785/j.issn.1008-973X.2022.05.013     OR     https://www.zjujournals.com/eng/Y2022/V56/I5/956


基于图卷积网络的归纳式微博谣言检测新方法

为了解决传统图卷积神经网络在进行谣言检测时面临的未充分考虑单词语义信息以及池化方法选择困难的问题,提出基于图卷积网络(GCN)的归纳式微博谣言检测新方法. 考虑单词之间的语义关系,结合传统词共现建图方法提出基于词语义相关性的微博事件建图方法,并结合图卷积网络和门循环单元(GRU)实现节点信息聚合;为了有效融合不同节点状态的特征信息,提出基于注意力机制的多池化方法融合策略融合最大池、平均池和全局池以获取最终的图级向量;为了提高微博谣言检测效率,探究微博评论时间对检测结果的影响,获得用于模型训练的最佳评论利用时间阈值. 实验结果表明,本研究方法在给定数据集上的表现普遍优于Text-CNN、Bi-GCN、TextING等典型方法,验证了其在微博谣言检测领域的有效性.


关键词: 谣言检测,  图卷积网络,  微博事件,  门循环单元,  注意力机制 
Fig.1 Example of microblog event
Fig.2 Flowchart of inductive graph convolutional network
Fig.3 Example of problems in graph construction of reference [20]
数据集 nu ne nr nt nc
Ma_Dataset 2 746 818 4 664 2 351 2 313 3 805 656
Song_Dataset 1 067 410 3 387 1 838 1 849 1 275 180
Tab.1 Details of datasets
tm/h 典型评论内容
1 这是真的么?
真的?
2 真噶?![吃惊][哈哈]
真的还是假的现在醒着还是醉了[围观]
3 真的还是假的,咋没新闻?
假的吧
4 假的吧P的吧
[汗] 真的?
5 真的吗,求真相
真葛..真葛..?
6 真的吗?~小综!
真的吗
7 真的假的[思考]
求真相···
8 真的的还是假的?震惊~
这是骗我的吧?
9 不是吧[抓狂]
是假的是吗
10 啥?真的假的?
这么假也有人信
Tab.2 One source microblog and its related comments in ten hours
Fig.4 Rumor detection accuracy at different time points within ten hours after source microblog being sent out
本研究建图方法分类 th 本研究建图方法分类 th
WR-1 0.95 WR-4 0.80
WR-2 0.90 WR-5 0.75
WR-3 0.85 WR-6 0.70
Tab.3 Classification of proposed graph construction methods of different threshold values
Fig.5 Accuracy comparison of different graph construction methods
Fig.6 Accuracy Comparison of different pooling methods
对比方法 实验设定
DT-Rank[4] 所选特征包括来源可信度、来源身份、来源多样性、来源地址、语言态度、事件传播特征,特征选择方法为信息增益.
SVM-TS[3] 所选特征为内容特征、用户特征和传播特征,核函数为RBF.
Text-CNN[23] 卷积核尺寸分别等于3、4、5,卷积核数量为256.
GRU-2[8] GRU层数为2,词典大小为5 000.
dEFEND[24] 注意力层维度为100,共注意力层潜在维度为200.
Text-GCN[18, 22] GCN层数为2.
Bi-GCN[25] 模型早停忍耐批次为10.
GLAN[26] 卷积核尺寸分别等于3、4、5,卷积核数量为100.
TextING[20] 滑动窗口大小为3.
Tab.4 Parameter settings of different methods for comparisons
方法 Ma_Dataset Song_Dataset
Acc Pre Rec F1 Acc Pre Rec F1
DT-Rank 0.727 0.736 0.731 0.733 0.653 0.637 0.665 0.651
SVM-TS 0.829 0.814 0.823 0.818 0.746 0.751 0.761 0.756
Text-CNN 0.848 0.839 0.854 0.846 0.801 0.807 0.812 0.809
GRU-2 0.902 0.895 0.891 0.893 0.842 0.837 0.846 0.841
dEFEND 0.917 0.912 0.929 0.920 0.881 0.873 0.898 0.885
Text-GCN 0.924 0.915 0.919 0.917 0.889 0.892 0.885 0.888
Bi-GCN 0.929 0.931 0.924 0.927 0.901 0.897 0.906 0.901
GLAN 0.930 0.935 0.932 0.933 0.903 0.908 0.912 0.910
TextING 0.938 0.937 0.943 0.940 0.912 0.906 0.915 0.910
本研究方法 0.946 0.939 0.943 0.941 0.923 0.925 0.922 0.923
Tab.5 Comparison of microblog rumor detection results of proposed method and existing typical methods
方法 Ma_Dataset Song_Dataset
Acc Pre Rec F1 Acc Pre Rec F1
DT-Rank 0.723 0.728 0.725 0.726 0.647 0.635 0.669 0.652
SVM-TS 0.824 0.810 0.817 0.813 0.743 0.753 0.764 0.758
Text-CNN 0.839 0.833 0.849 0.841 0.800 0.813 0.809 0.811
GRU-2 0.899 0.896 0.885 0.890 0.839 0.835 0.847 0.841
dEFEND 0.915 0.913 0.931 0.922 0.877 0.869 0.899 0.883
Text-GCN 0.925 0.916 0.913 0.914 0.892 0.887 0.880 0.883
Bi-GCN 0.928 0.933 0.921 0.927 0.902 0.895 0.911 0.903
GLAN 0.929 0.936 0.930 0.933 0.902 0.907 0.916 0.911
TextING 0.937 0.936 0.939 0.937 0.909 0.908 0.911 0.909
本研究方法 0.945 0.938 0.941 0.940 0.921 0.920 0.925 0.923
Tab.6 Validation of effectiveness of optimal comment utilization time threshold
[1]   ZUBIAGA A, AKER A, BONTCHEVA K, et al Detection and resolution of rumours in social media: a survey[J]. ACM Computing Surveys (CSUR), 2018, 51 (2): 1- 36
[2]   新浪微博虚假消息辟谣官方账号. 2020年度微博辟谣数据报告[EB/OL]. (2020-02-07) [2021-11-05]. https://weibo.com/1866405545/K0QaImwsK.
[3]   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: CIKM, 2015.
[4]   ZHAO Z, RESNICK P, MEI Q. Enquiring minds: early detection of rumors in social media from enquiry posts [C]// Proceedings of the 24th International Conference on World Wide Web. New York: WWW, 2015.
[5]   张仰森, 彭媛媛, 段宇翔, 等 基于评论异常度的新浪微博谣言识别方法[J]. 自动化学报, 2020, 46 (8): 1689- 1702
ZHANG Yang-sen, PENG Yuan-yuan, DUAN Yu-xiang, et al The method of Sina Weibo rumor detecting based on comment abnormality[J]. Acta Automatica Sinica, 2020, 46 (8): 1689- 1702
[6]   曾子明, 王婧 基于LDA和随机森林的微博谣言识别研究: 以2016年雾霾谣言为例[J]. 情报学报, 2019, 38 (1): 89- 96
ZENG Zi-ming, WANG Jing Research on Microblog rumor identification based on LDA and random forest[J]. Journal of the China Society for Scientific and Technical Information, 2019, 38 (1): 89- 96
doi: 10.3772/j.issn.1000-0135.2019.01.010
[7]   CAI G, BI M, LIU J. A novel rumor detection method based on labeled cascade propagation tree [C]// Proceedings of the 13th International Conference on Natural Computation, Fuzzy Systems and Knowledge Discovery. Guilin: ICNC-FSKD, 2017.
[8]   MA J, GAO W, MITRA P, et al. Detecting rumors from microblogs with recurrent neural networks [C]// International Joint Conference on Artificial Intelligence. New York: IJCAI, 2016.
[9]   WANG Z, GUO Y, WANG J, et al Rumor events detection from chinese microblogs via sentiments enhancement[J]. IEEE Access, 2019, 7: 103000- 103018
doi: 10.1109/ACCESS.2019.2928044
[10]   尹鹏博, 潘伟民, 彭成, 等 基于用户特征分析的微博谣言早期检测研究[J]. 情报杂志, 2020, 39 (7): 81- 86
YIN Peng-bo, PAN Wei-min, PENG Cheng, et al Research on early detection of Weibo rumors based on user characteristics analysis[J]. Journal of Intelligence, 2020, 39 (7): 81- 86
doi: 10.3969/j.issn.1002-1965.2020.07.014
[11]   SONG C, YANG C, CHEN H, et al CED: credible early detection of social media rumors[J]. IEEE Transactions on Knowledge and Data Engineering, 2019, 33 (8): 3035- 3047
[12]   刘政, 卫志华, 张韧弦 基于卷积神经网络的谣言检测[J]. 计算机应用, 2017, 37 (11): 3053- 3056
LIU Zheng, WEI Zhi-hua, ZHANG Ren-xian Rumor detection based on convolutional neural network[J]. Journal of Computer Applications, 2017, 37 (11): 3053- 3056
[13]   胡斗, 卫玲蔚, 周薇, 等 一种基于多关系传播树的谣言检测方法[J]. 计算机研究与发展, 2021, 58 (7): 1395- 1411
HU Dou, WEI Ling-wei, ZHOU Wei, et al A rumor detection approach based on multi-relational propagation tree[J]. Journal of Computer Research and Development, 2021, 58 (7): 1395- 1411
doi: 10.7544/issn1000-1239.2021.20200810
[14]   WU Z, PI D, CHEN J, et al Rumor detection based on propagation graph neural network with attention mechanism[J]. Expert Systems with Applications, 2020, 158: 113595
doi: 10.1016/j.eswa.2020.113595
[15]   杨延杰, 王莉, 王宇航 融合源信息和门控图神经网络的谣言检测研究[J]. 计算机研究与发展, 2021, 58 (7): 1412- 1424
YANG Yan-jie, WANG Li, WANG Yu-hang Rumor detection based on source information and gating graph neural network[J]. Journal of Computer Research and Development, 2021, 58 (7): 1412- 1424
doi: 10.7544/issn1000-1239.2021.20200801
[16]   YANG X, LYU Y, TIAN T, et al. Rumor detection on social media with graph structured adversarial learning [C]// Proceedings of the 29th International Conference on International Joint Conferences on Artificial Intelligence. Montreal: IJCAI, 2021.
[17]   HU L, YANG T, SHI C, et al. Heterogeneous graph attention networks for semi-supervised short text classification [C]// Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing. Hong Kong: EMNLP-IJCNLP, 2019.
[18]   YAO L, MAO C, LUO Y. Graph convolutional networks for text classification [C]// Proceedings of the AAAI Conference on Artificial Intelligence. New Orleans: AAAI, 2019.
[19]   LIU X, YOU X, ZHANG X, et al. Tensor graph convolutional networks for text classification [C]// Proceedings of the AAAI Conference on Artificial Intelligence. New York: AAAI, 2020.
[20]   ZHANG Y, YU X, CUI Z, et al. Every document owns its structure: inductive text classification via graph neural networks [C]// Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics.[s.l.]:ACL, 2020.
[21]   LI Y, TARLOW D, BROCKSCHMIDT M, et al. Gated graph sequence neural networks [C]// Proceedings of the 4th International Conference on Learning Representations. Puerto Rico: ICLR, 2016.
[22]   米源, 唐恒亮 基于图卷积网络的谣言鉴别研究[J]. 计算机工程与应用, 2021, 57 (13): 161- 167
MI Yuan, TANG Heng-liang Rumor identification research based on graph convolutional network[J]. Computer Engineering and Applications, 2021, 57 (13): 161- 167
doi: 10.3778/j.issn.1002-8331.2003-0357
[23]   KIM Y. Convolutional neural networks for sentence classification [C]// Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing. Doha: EMNLP, 2014.
[24]   SHU K, CUI L, WANG S, et al. dEFEND: explainable fake news detection [C]// Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. Anchorage: KDD, 2019.
[25]   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, 2020.
[1] Xiao-chen JU,Xin-xin ZHAO,Sheng-sheng QIAN. Self-attention mechanism based bridge bolt detection algorithm[J]. Journal of ZheJiang University (Engineering Science), 2022, 56(5): 901-908.
[2] Xue-qin ZHANG,Tian-ren LI. Breast cancer pathological image classification based on Cycle-GAN and improved DPN network[J]. Journal of ZheJiang University (Engineering Science), 2022, 56(4): 727-735.
[3] Meng XU,Dan WANG,Zhi-yuan LI,Yuan-fang CHEN. IncepA-EEGNet: P300 signal detection method based on fusion of Inception network and attention mechanism[J]. Journal of ZheJiang University (Engineering Science), 2022, 56(4): 745-753, 782.
[4] Chang-yuan LIU,Xian-ping HE,Xiao-jun BI. Efficient network vehicle recognition combined with attention mechanism[J]. Journal of ZheJiang University (Engineering Science), 2022, 56(4): 775-782.
[5] Qiao-hong CHEN,Hao-lei PEI,Qi SUN. Image caption based on relational reasoning and context gate mechanism[J]. Journal of ZheJiang University (Engineering Science), 2022, 56(3): 542-549.
[6] Ting WANG,Xiao-fei ZHU,Gu TANG. Knowledge-enhanced graph convolutional neural networks for text classification[J]. Journal of ZheJiang University (Engineering Science), 2022, 56(2): 322-328.
[7] Yuan-jun NONG,Jun-jie WANG,Hong CHEN,Wen-han SUN,Hui GENG,Shu-yue LI. A image caption method of construction scene based on attention mechanism and encoding-decoding architecture[J]. Journal of ZheJiang University (Engineering Science), 2022, 56(2): 236-244.
[8] Ying-li LIU,Rui-gang WU,Chang-hui YAO,Tao SHEN. Construction method of extraction dataset of Al-Si alloy entity relationship[J]. Journal of ZheJiang University (Engineering Science), 2022, 56(2): 245-253.
[9] Xin WANG,Qiao-hong CHEN,Qi SUN,Yu-bo JIA. Visual question answering method based on relational reasoning and gating mechanism[J]. Journal of ZheJiang University (Engineering Science), 2022, 56(1): 36-46.
[10] Zhi-chao CHEN,Hai-ning JIAO,Jie YANG,Hua-fu ZENG. Garbage image classification algorithm based on improved MobileNet v2[J]. Journal of ZheJiang University (Engineering Science), 2021, 55(8): 1490-1499.
[11] Zi-ye YONG,Ji-chang GUO,Chong-yi LI. weakly supervised underwater image enhancement algorithm incorporating attention mechanism[J]. Journal of ZheJiang University (Engineering Science), 2021, 55(3): 555-562.
[12] Han-juan CHEN,Fei-peng DA,Shao-yan GAI. Deep 3D point cloud classification network based on competitive attention fusion[J]. Journal of ZheJiang University (Engineering Science), 2021, 55(12): 2342-2351.
[13] Yue-lin CHEN,Wen-jing TIAN,Xiao-dong CAI,Shu-ting ZHENG. Text matching model based on dense connection networkand multi-dimensional feature fusion[J]. Journal of ZheJiang University (Engineering Science), 2021, 55(12): 2352-2358.
[14] Wen-bin XIN,Hui-min HAO,Ming-long BU,Yuan LAN,Jia-hai HUANG,Xiao-yan XIONG. Static gesture real-time recognition method based on ShuffleNetv2-YOLOv3 model[J]. Journal of ZheJiang University (Engineering Science), 2021, 55(10): 1815-1824.
[15] Chuang LIU,Jun LIANG. Vehicle motion trajectory prediction based on attention mechanism[J]. Journal of ZheJiang University (Engineering Science), 2020, 54(6): 1156-1163.