目前基于网络的垃圾用户检测方法只考虑了简单社会关系,缺乏对更多复杂社会语义关系的利用,难以达到最优性能.针对这一挑战,提出一种基于层次注意力机制的垃圾用户检测模型(HAM-SD).模型首先使用异质信息网络对社交媒体进行建模,挖掘丰富的语义与结构信息,接着利用节点级注意力层聚合元路径邻居增强节点表示,同时利用自适应层级聚合模块选择不同层级特征提升表征能力,然后通过语义级注意力层融合不同元路径下的节点表示,最后带入分类检测模块实现垃圾用户检测.在公开数据集上的实验结果表明该模型能够有效检测垃圾用户,并在不平衡数据分布时保持较强的稳定性.
The current network-based spammer detection methods only consider simple social relations and lack the utilization of more complex social semantic relations, so it is difficult to achieve optimal performance. Aiming at this challenge, this paper proposes a spammer detection model based on hierarchical attention mechanism(HAM-SD). The model first uses a heterogeneous information network to model social media to mine rich semantic and structural information, then uses a node-level attention layer to aggregate meta-path neighbors to enhance node representation, and uses an adaptive hierarchical aggregation module to select features at different levels to improve representation Then, the node representations under different meta-paths are fused through the semantic-level attention layer, and finally brought into the classification detection module to realize the detection of spammer. The results on public datasets show that the model can effectively detect spammer and maintain strong stability when the data distribution is unbalanced.
[1] ANAS S M, KUMARI S. Opinion mining based fake product review monitoring and removal system[C] //2021 6th International Conference on Inventive Computation Technologies, Coimbatore, India: IEEE, 2021: 985-988. DOI:10.1109/ICICT50816.2021.9358716.
[2] 李忠,靳小龙,庄传志,等.面向图的异常检测研究综述[J].软件学报, 2021, 32(1): 167-193.DOI: 10.13328/j.cnki.jos.006100.
[3] 张宇翔,孙菀,杨家海,等.新浪微博反垃圾中特征选择的重要性分析[J].通信学报, 2016, 37(8): 24-33.DOI: 10.11959/j.issn.1000-436x.2016152.
[4] ZHENG X H, ZENG Z P, CHEN Z Y, et al. Detecting spammers on social networks[J]. Neurocomputing, 2015, 159: 27-34. DOI:10.1016/j.neucom.2015.02.047.
[5] SONG J, LEE S, KIM J. Spam filtering in Twitter using sender receiver relationship[C] //International workshop on recent advances in intrusion detection, Berlin, German: Springer, 2011: 301-317. DOI:10.1007/978-3-642-23644-0_16.
[6] CHEN H, LIU J, LUY Y Z, et al. Semi-supervised clue fusion for spammer detection in Sina Weibo[J]. Information Fusion, 2018, 44: 22-32. DOI:10.1016/j.inffus.2017.11.002.
[7] 杨晓晖,梁笑.基于多视图融合的微博垃圾用户检测方法[J].华南理工大学学报(自然科学版), 2020, 48(12): 125-134.
[8] DOU Y T, MA G X, YU P S, et al. Robust spammer detection by Nash reinforcement learning[C] //Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, New York, USA: ACM, 2020: 924-933. DOI:10.1145/3394486.3403135.
[9] WU F Z, WU C H, LIU J X. Semi-supervised collaborative learning for social spammer and spam message detection in microblogging[C] //Proceedings of the 27th ACM International Conference on Information and Knowledge Management, New York, USA: ACM, 2018: 1791-1794. DOI:10.1145/3269206.3269324.
[10] YU D G, CHEN N, JIANG F, et al. Constrained NMF-based semi-supervised learning for social media spammer detection[J]. Knowl Based Syst, 2017, 125: 64-73. DOI:10.1016/j.knosys.2017.03.025.
[11] LI C Z, WANG S Z, HE L F, et al. SSDMV: semi-supervised deep social spammer detection by multi-view data fusion[C] //Proceedings of 2018 IEEE International Conference on Data Mining, Singapore: IEEE, 2018: 247-256. DOI:10.1109/ICDM.2018.00040.
[12] 宋畅,禹可,吴晓非.基于改进边权重的成对马尔可夫随机场模型的社交异常账号检测方法[J].计算机科学, 2020, 47(2): 251-255.DOI: 10.11896/jsjkx.190600172.
[13] KIPF T N, WELLING M. Semi-supervised classification with graph convolutional networks[C] //International Conference on Learning Representations, 2017. DOI:10.48550/arXiv.1609.02907.
[14] VELICKOVIC P, CUCURULL G, CASANOVA A, et al. Graph attention networks[C] //Proceedings of the International Conference on Learning Representations, 2018. DOI:10.48550/arXiv.1710.10903.
[15] WU Y J, LIAN D F, XU Y H, et al. Graph convolutional networks with Markov random field reasoning for social spammer detection[J]. Proceedings of the AAAI Conference on Artificial Intelligence, 2020, 34(1): 1054-1061. DOI:10.1609/aaai.v34i01.5455.
[16] 石川,王睿嘉,王啸.异质信息网络分析与应用综述[J].软件学报, 2022, 33(2): 598-621. DOI: 10.13328/j.cnki.jos.006357.
[17] 龚卫华,沈松,裴小兵,等.基于位置的社交网络中双重异质社区的聚类与关联方法[J].计算机学报, 2020, 43(10): 1909-1923. DOI: 10.11897/SP.J.1016.2020.01909
[18] WANG S, CHEN Z Z, YU X et al. Heterogeneous graph matching networks for unknown malware detection[C] //Proceedings of the 28th International Joint Conference on Artificial Intelligence, California, USA: IJCAI, 2019: 3762-3770. DOI:10.24963/ijcai.2019/522. (
