关注社交异配性的社交机器人检测框架

doi:10.3969/j.issn.1671-1122.2024.02.015

信息网络安全 ›› 2024, Vol. 24 ›› Issue (2): 319-327.doi: 10.3969/j.issn.1671-1122.2024.02.015

关注社交异配性的社交机器人检测框架

余尚戎¹^,²^,³, 肖景博¹^,²^,³, 殷琪林¹^,²^,³, 卢伟¹^,²^,³()

1.中山大学计算机学院，广州 510006
2.中山大学信息技术教育部重点实验室，广州 510006
3.广东省信息安全技术重点实验室，广州 510006

收稿日期:2023-10-31 出版日期:2024-02-10 发布日期:2024-03-06
通讯作者: 卢伟 E-mail:luwei3@mail.sysu.edu.cn
作者简介:余尚戎（1999—），男，湖北，硕士研究生，主要研究方向为多媒体内容安全|肖景博（2001—），男，河南，硕士研究生，主要研究方向为多媒体内容安全|殷琪林（1995—），男，江苏，博士研究生，主要研究方向为数字多媒体取证|卢伟（1979—），男，河南，教授，博士，CCF会员，主要研究方向为人工智能安全与对抗、信息取证与安全
基金资助:
国家自然科学基金(U2001202);国家自然科学基金(62072480)

A Social Heterophily Focused Framework for Social Bot Detection

YU Shangrong¹^,²^,³, XIAO Jingbo¹^,²^,³, YIN Qilin¹^,²^,³, LU Wei¹^,²^,³()

1. School of Computer Science and Engineering, Sun Yat-sen University, Guangzhou 510006, China
2. Ministry of Education Key Laboratory of Information Technology, Sun Yat-sen University, Guangzhou 510006, China
3. Guangdong Province Key Laboratory of Information Security Technology, Guangzhou 510006, China

Received:2023-10-31 Online:2024-02-10 Published:2024-03-06
Contact: LU Wei E-mail:luwei3@mail.sysu.edu.cn

摘要/Abstract

摘要：

随着社交机器人的迭代，其倾向于与正常用户进行更多交互，对其检测变得更具挑战性。现有检测方法大多基于同配性假设，由于忽视了不同类用户间存在的联系，难以保持良好的检测性能。针对这一问题文章提出一种关注社交异配性的社交机器人检测框架，以社交网络用户间的联系为依据，通过充分挖掘用户社交信息来应对异配影响，并实现更精准的检测。文章分别在同配视角和异配视角下看待用户之间的联系，将社交网络构建为图，通过消息传递机制实现同配边和异配边聚合，以提取节点的频率特征，同时利用图中各节点特征聚合得到社交环境特征，将以上特征混合后用于检测。实验结果表明，文章所提方法在开源数据集上的检测效果优于基线方法，证明了该方法的有效性。

关键词: 社交机器人检测, 同配性与异配性, 图神经网络

Abstract:

As social bot technology advances, these bots increasingly interact with human users, making their detection a more challenging problem. Existing detection methods primarily rely on the homophily assumption, often overlooking the connections between different classes of users, particularly the impact of heterophily. This oversight impairs their detection performance. To address this issue, this paper presented an innovative social bot detection framework that emphasizes social heterophily. It leveraged user connections within social networks and extensively explored various types of social information to mitigate the effects of heterophily and achieved more accurate detection. This paper examined user relationships from both homophily and heterophily perspectives. It constructed the social network as a graph and employed a message-passing mechanism to aggregate information from both homophilic and heterophilic edges, allowing for the extraction of frequency-based node features. Furthermore, it aggregated features from various nodes within the graph to generate social context features. These features are then blended and utilized for the detection task. The experimental results validate the method’s superiority over comparative approaches on publicly available datasets, confirming its effectiveness.

Key words: social bot detection, homophily and heterophily, graph neural network

中图分类号:

TP309

余尚戎, 肖景博, 殷琪林, 卢伟. 关注社交异配性的社交机器人检测框架[J]. 信息网络安全, 2024, 24(2): 319-327.

YU Shangrong, XIAO Jingbo, YIN Qilin, LU Wei. A Social Heterophily Focused Framework for Social Bot Detection[J]. Netinfo Security, 2024, 24(2): 319-327.

图/表 8

图1

图2

表1

表2

表3

图3

表4

表5

参考文献 26

[1]	BESSI A, FERRARA E. Social Bots Distort the 2016 Us Presidential Election Online Discussion[J]. First Monday, 2016, 16(21): 7-11.
[2]	WENG Zixuan, LIN Aijun. Public Opinion Manipulation on Social Media: Social Network Analysis of Twitter Bots During the Covid-19 Pandemic[EB/OL]. (2022-12-07)[2023-10-20]. https://www.mdpi.com/1660-4601/19/24/16376.
[3]	WANG Gang, MOHANLAL M, WILSON C, et al. Social Turing Tests: Crowdsourcing Sybil Detection[EB/OL]. (2012-12-07)[2023-10-20]. https://arxiv.org/pdf/1205.3856.pdf.
[4]	DAVIS C A, VAROL O, FERRARA E, et al. Botornot: A System to Evaluate Social Bots[C]// ACM. 25th International Conference Companion on World Wide Web. New York: ACM, 2016: 273-274.
[5]	DEWANGAN M, KAUSHAL R. Socialbot: Behavioral Analysis and Detection[C]// Springer. 4th International Symposium on Security in Computing and Communications (SSCC). Heidelberg:Springer, 2016: 450-460.
[6]	KUDUGUNTA S, FERRARA E. Deep Neural Networks for Bot Detection[J]. Information Sciences, 2018, 467: 312-322. doi: 10.1016/j.ins.2018.08.019 URL
[7]	PING Heng, QIN Sujuan. A Social Bots Detection Model Based on Deep Learning Algorithm[C]// IEEE. 18th IEEE International Conference on Communication Technology (ICCT). New York:IEEE, 2018: 1435-1439.
[8]	ALI A S, BIN T R, NAJAFI P, et al. Detect Me If You Can: SPAM Bot Detection Using Inductive Representation Learning[C]// ACM. Companion Proceedings of the 2019 World Wide Web Conference. New York: ACM, 2019: 148-153.
[9]	FENG Shangbin, WAN Herun, WANG Ningnan, et al. BotRGCN: Twitter Bot Detection with Relational Graph Convolutional Networks[C]// ACM. 2021 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining. New York: ACM, 2021: 236-239.
[10]	FENG Shangbin, WAN Herun, WANG Ningnan, et al. Satar: A Self-Supervised Approach to Twitter Account Representation Learning and Its Application in Bot Detection[C]// ACM. 30th ACM International Conference on Information & Knowledge Managemen. New York: ACM, 2021: 3808-3817.
[11]	FENG Shangbin, TAN Zhaoxuan, LI Rui, et al. Heterogeneity-Aware Twitter Bot Detection with Relational Graph Transformers[C]// AAAI. AAAI Conference on Artificial Intelligence. Menlo Park: AAAI, 2022, 36(4): 3977-3985.
[12]	LE T, TRAN-THANH L, LEE D. Socialbots on Fire: Modeling Adversarial Behaviors of Socialbots via Multi-Agent Hierarchical Reinforcement Learning[C]// ACM. 31st ACM Web Conference (WWW). New York:ACM, 2022: 545-554.
[13]	DES M N G, HUNTER D S, EL H Z, et al. Detecting Bots and Assessing Their Impact in Social Networks[J]. Operations Research, 2022, 70(1): 1-22. doi: 10.1287/opre.2021.2118 URL
[14]	MCPHERSON M, SMITH-LOVIN L, COOK J M. Birds of a Feather: Homophily in Social Networks[J]. Annual Review of Sociology, 2001, 27(1): 415-444. doi: 10.1146/soc.2001.27.issue-1 URL
[15]	SHEN Xiao, SUN Dewang, PAN Shirui, et al. Neighbor Contrastive Learning on Learnable Graph Augmentation[C]// AAAI. AAAI Conference on Artificial Intelligence. Menlo Park: AAAI, 2023, 37(8): 9782-9791.
[16]	YANG Peipei, ZHENG Zhuoyuan. Fake Account Detection with Attention-Based Graph Convolution Networks[C]// IEEE. 2020 IEEE 3rd International Conference on Automation, Electronics and Electrical Engineering (AUTEEE). New York:IEEE, 2020: 106-110.
[17]	FENG Shangbin, WAN Herun, WANG Ningnan, et al. Twibot-20: A Comprehensive Twitter Bot Detection Benchmark[C]// ACM. 30th ACM International Conference on Information & Knowledge Management. New York: ACM, 2021: 4485-4494.
[18]	FENG Shangbin, TAN Zhaoxuan, WAN Herun, et al. TwiBot-22: Towards Graph-Based Twitter Bot Detection[J]. Advances in Neural Information Processing Systems, 2022(35): 35254-35269.
[19]	VASWANI A, SHAZEER N, PARMAR N, et al. Attention is All You Need[J]. Advances in Neural Information Processing Systems, 2017, (30): 6000-6010.
[20]	WU Qi, YANG Yingguan, HE Buyun, et al. Heterophily-Aware Social Bot Detection with Supervised Contrastive Learning[EB/OL]. (2023-06-15)[2023-10-20]. https://doi.org/10.48550/arXiv.2306.07478.
[21]	GILANI Z, FARAHBAKHSH R, TYSON G, et al. Of Bots and Humans (on Twitter)[C]// IEEE. The 2017 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining. New York: IEEE, 2017: 349-354.
[22]	LUAN Sitao, HUA Chenqing, LU Qincheng, et al. Is Heterophily a Real Nightmare for Graph Neural Networks to Do Node Classification?[EB/OL]. (2021-09-12)[2023-10-20]. https://doi.org/10.48550/arXiv.2109.05641.
[23]	BO Deyu, WANG Xiao, SHI Chuan, et al. Beyond Low-Frequency Information in Graph Convolutional Networks[C]// AAAI. AAAI Conference on Artificial Intelligence. Menlo Park: AAAI, 2021, 35(5): 3950-3957.
[24]	LIU Nian, WANG Xiao, BO Deyu, et al. Revisiting Graph Contrastive Learning from The Perspective of Graph Spectrum[J]. Advances in Neural Information Processing Systems, 2022, (35): 2972-2983.
[25]	EKAMBARAM V N. Graph-Structured Data Viewed through a Fourier Lens[M]. Berkeley: University of California, 2014.
[26]	VAN D M L, HINTON G. Visualizing Data Using T-SNE[J]. Journal of Machine Learning Research, 2008, 9(11): 2579-2605.

超参数	值
优化器	AdamW
批大小	64
训练轮数	200
学习率	8e-4
权重衰减因子	8.9e-4
聚合特征丢弃率	1.38e-2
Transformer编码器输入维度	(2, 32)
Transformer编码器自注意力头	2
Transformer编码器层数	1

数据集	模型	$Acc$	$F1\text{-Score}$
TwiBot-20	BotRGCN	83.41 %±0.41 %	85.94 %±0.36 %
	RGT	83.92 %±0.27 %	86.57 %±0.17 %
	BotSCL	83.81 %±0.42 %	86.47 %±0.26 %
	本文方法	84.76 %±0.27 %	87.15 %±0.21 %
TwiBot-22	BotRGCN	85.02 %±0.29 %	64.46 %±1.7 %
	RGT	85.11 %±0.06 %	64.61 %±0.21 %
	BotSCL	84.82 %±0.15 %	62.58 %±2.78 %
	本文方法	85.02 %±0.08 %	65.97 %±1.57 %

低频特征	高频特征	环境特征	$Acc$	$F1\text{-Score}$
?	?	√	$83.44$%	$86.08$%
√	?	√	$83.77$%	$86.37$%
?	√	√	$84.15$%	$86.43$%
√	√	√	$84.76$%	$87.15$%

MLP	Transformer编码器	$Acc$	$F1\text{-Score}$
√	?	$83.90$%	$86.52$%
√	√	$84.76$%	$87.15$%

关注社交异配性的社交机器人检测框架

A Social Heterophily Focused Framework for Social Bot Detection

RichHTML

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

图/表 8

参考文献 26

相关文章 4

编辑推荐

Metrics

本文评价

[1]	秦中元, 马楠, 余亚聪, 陈立全. 基于双重图神经网络和自编码器的网络异常检测[J]. 信息网络安全, 2023, 23(9): 1-11.
[2]	仝鑫, 金波, 王靖亚, 杨莹. 一种面向Android恶意软件的多视角多任务学习检测方法[J]. 信息网络安全, 2022, 22(10): 1-7.
[3]	朱丽娜, 马铭芮, 朱东昭. 基于图神经网络和通用漏洞分析框架的C类语言漏洞检测方法[J]. 信息网络安全, 2022, 22(10): 59-68.
[4]	秦中元, 胡宁, 方兰婷. 基于免疫仿生机理和图神经网络的网络异常检测方法[J]. 信息网络安全, 2021, 21(8): 10-16.