Community Detection in Heterogeneous Networks Based on Graph MaskedAutoencoder and Attention Mechanism
[1]ZHANG Zhen,ZHANG Xinfang,GAO Sihan.Community Detection in Heterogeneous Networks Based on Graph MaskedAutoencoder and Attention Mechanism[J].Journal of Zhengzhou University (Engineering Science),2027,48(XX):1-8.[doi:10. 13705 / j. issn. 1671-6833. 2025. 06. 018]
Copy
Journal of Zhengzhou University (Engineering Science)[ISSN
1671-6833/CN
41-1339/T] Volume:
48
Number of periods:
2027 XX
Page number:
1-8
Column:
Public date:
2027-12-10
- Title:
- Community Detection in Heterogeneous Networks Based on Graph MaskedAutoencoder and Attention Mechanism
- Keywords:
- heterogeneous network; community detection; graph masked autoencoder; graph attention; dynamic mask
- CLC:
- TP389.1 文献标码：0 引言
- Abstract:
- Graph representation learning has attracted extensive attention in the field of community detection. However, existing methods neglect the effective fusion of heterogeneity and network feature structure information, are highly dependent on the distinguishability of features, and fail to fully combine with the community detection task. Therefore, this paper proposes a heterogeneous network community detection model based on a graph masked autoencoder and an attention mechanism. Firstly, the mask preprocessing module is optimized: nodes are pre-clustered, followed by dynamic masking, and noise is introduced to enhance the robustness of the graph masked autoencoder and the performance of feature reconstruction. Secondly, a heterogeneous network hierarchical encoder integrating spatial attention is designed to encode the node features of the heterogeneous network and the structure information based on meta-paths. Finally, self-training clustering loss, feature reconstruction loss, and meta-path reconstruction loss are jointly trained to obtain graph vectors suitable for the community detection task, which are then used for clustering processing. Experimental results on four datasets (DBLP, ACM, AMiner, and Freebase) show that the model’s NMI and ARI metrics have increased by an average of 3.16% and 3.2% compared with the current state-of-the-art methods. The maximum improvement in the Purity metric reaches 3.71%, and the visualization effect is prominent, which proves the effectiveness of the model
- References:
-
[1].Gasparetti F, SansoneRi G, Micarelli A. Community detection in social recommender systems: A survey[J]. Applied Intelligence,2021:3975-3995.
[2].冯拓宇,刘佳宁,曹子奇,等.社区发现方法研究综述[J]. 中国电子科学研究院学报,2024,19(6): 487-498.
[3].FENG T Y,LIU J N,CAO Z Q, et al. Comprehensive review of community detection methods[J]. Journal of China Academy of Electronics and Information Technology,2024,19(6): 487-498.
[4].Kojaku S, Radicchi F, Ahn Y, et al. Network community detection via neural embeddings[J]. Nat Commun,2024, 15,9446.
[5].Tang, F, Li, J, Liu, X, et al. GATFELPA integrates graph attention networks and enhanced label propagation for robust community detection[J]. Sci Rep，2025，15, 3952 .
[6].He C, Cheng J, Chen G, et al. Detecting communities with multiple topics in attributed networks via self - supervised adaptive graph convolutional network [J]. Information Fusion, 2024, 105: 102254.
[7].李文举,姬倩倩,沙利业,等.基于图游走和图注意力的点云分类与分割[J].郑州大学学报(工学版),2024,45(02):33-41.
[8].LI W J,JI Q Q , SHA L Y, et al. Point cloud classification and segmentation based on graph walk and graph attention[J].Journal of Zhengzhou University (Engineering Science Edition),2024,45(02):33-41.
[9].Dong Y X, Chawla N V, Swami A. Metapath2vec: scalable representation learning for heterogeneous networks [C].// Proceedings of the 23rd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. Halifax, Nova Scotia, Canada: SIGKDD. 2017: 135-144.
[10].Shi C, Hu B B, Zhao W X, et al. Heterogeneous information network embedding for recommendation[J]. IEEE Transactions on Knowledge and Data Engineering, 2019:357 - 370.
[11].Luo L H, Fang Y, Cao X , et al. Detecting communities from heterogeneous graphs: a context path-based graph neural network model [C]. //Proceedings of the 30th ACM International Conference on Information & KnowledgeManagement.Queensland,Australia:CIKM,2021:1170-1180.
[12].Sun Y D, Zhu D J, WANG Y S, et al. GTC: GNN-Transformer co-contrastive learning for self-super-vised heterogeneous graph representation[J].Neural Networks,2025, 181: 106645.
[13].Duan H R，Xie C, Li L Y. Reserving-Masking-Reconstruction model for self-supervised[C].//Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining. Barcelona, Spain: ACM SIGKDD,2024:25-29.
[14].Tian Y, Dong K, Zhang C, et al. Heterogeneous graph masked autoencoders[C].//Proceedings of the AAAI Conference on Artificial Intelligence. Washington DC, USA: AAAI Press,2023: 9997-10005.
[15].Hou Z, Liu X, Dong Y, et al. Graph MAE: self-supervised masked graph autoencoders[C].// Proceedings of the 28th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining. New York: ACM SIGKDD, 2022:594-604．
[16].Zhao Z W, Li Y H ,Zou Y X ,et al. Masked graph autoencoder with non-discrete bandwidths[C].// Proceedings of the 33rd ACM Web Conference. Singapore: ACM, 2024:377-388.
[17].Wang X, Ji H, Shi C, et al. Heterogeneous graph attention network[C]// The World Wide Web Conference. New York:ACM,2019:2022-2032.
[18].Liu M, Liu Y, Liang K, et al. Deep temporal graph clustering [C]//Proceedings of the twelfth international conference on learning representations. Vienna Austria:ICLR,2024:2001-2012.
[19].Ji S Y, Zhang Z Z, Ying S H et al. Kullback–leibler divergence metric learning[J]. IEEE Transactions on Cybernetics,2022, 52(04): 2047-2058.
[20].Fu X Y, Zhang J N, Meng Z Q, et al. MAGNN: metapath aggregated graph neural network for heterogeneous graph embedding[C]//Proceedings of The Web Conference 2020. Taipei, Taiwan :In WWW,2020:2331–2341.
[21].Zhao J N, Wang X, and Shi C, et al. Network sc-hema preserving heterogeneous information networkembedding[C]//Proceedings of the twenty-ninth international joint conference on artificial intelligence. Yokohama, Japa:IJCAI,2021:1366–1372.
[22].Hu B B, Fang Y, and Shi C. Adversarial learning on heterogeneous information networks[C]// Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining (2019). New York: SIGKDD,2019:120–129.
[23].Li X, Ding D, Kao B, et al. Leveraging meta-pathcontexts for classification in heterogeneous information networks[C]//IEEE 37th International Conferen-ce on Data Engineering. Chania:ICDE,2021:912-923.
[24].周万珍,宋健,许云峰．异质网络社区发现方法研究综述[J]．河北科技大学学报,2021,42(03)：231-240．
[25].Zhou W Z, Song J, Xu Y F. Survey of community discovery methods of heterogeneous networks[J]. Journal of Hebei University of Science and Technology,2021, 42(03)231-240．
[26].Hamilton W L, Ying R, Leskovec J. Inductive representation learning on large graphs[C]// Advances in Neural Information Processing Systems 30. Long Beach, CA, USA: NeurIPS, 2017:1024–1034.
[27].Kifp T N, Welling M. Variational graph auto-encoders[C]// Advances in Neural Information Processing Systems 29. Barcelona, Spain: NeurIPS, 2016:1251-1258
[28].Park C, Kim D, Han J W, et al. Unsupervised attributed multiplex network embedding[C]//Proceedings of the AAAI Conference on Artificial Intelligence 34.New York :In AAAI,2020:5371–5378.
[29].Wang X, Liu N, Han H, et al. Self-supervised heterogeneous graph neural network with co-contrastive learning[C]//Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining. New York: SIGKDD,2021:1726-1736.
[30].Zhang C X, Song D J, Huang C ,et al. Heterogeneous graph neural network[C]. // Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining. New York, USA: SIGKDD, 2019:793–803
[31].ZHAO Y, LI W, LIU F, et al. Integrating heterogeneous structures and community semantics for unsupervised community detection in heterogeneous networks[J]. Expert Systems with Applications, 2024, 238: 121821.
[32].武永亮,窦世卯,李景辉等.融合异质性和动态性的社区发现研究综述[J].计算机工程与应用,2024,60(21):55-72.
[33].WU Y L,DOU S M,LI J H, et al. Survey of community detection from perspectives of dynamics and heterogeneity[J].Computer Engineering and Applications,2024,60(21):55-72.Community Detection in Heterogeneous Networks Based on Graph Masked Autoencoders and Graph Attention
[34].ZHANG Zhen,ZHANG Xinfang, GAO Siihan
[35].（School of Electrical and Information Engineering, Zhengzhou University, Zhengzhou 450001,China；
[36].School of Computer Science and Artificial Intelligence, Zhengzhou University, Zhengzhou 450001,China；.）
[37].Abstract: Graph representation learning has attracted extensive attention in the field of community detection. However, existing methods neglect the effective fusion of heterogeneity and network feature structure information, are highly dependent on the distinguishability of features, and fail to fully combine with the community detection task. Therefore, this paper proposes a heterogeneous network community detection model based on a graph masked autoencoder and an attention mechanism. Firstly, the mask preprocessing module is optimized: nodes are pre-clustered, followed by dynamic masking, and noise is introduced to enhance the robustness of the graph masked autoencoder and the performance of feature reconstruction. Secondly, a heterogeneous network hierarchical encoder integrating spatial attention is designed to encode the node features of the heterogeneous network and the structure information based on meta-paths. Finally, self-training clustering loss, feature reconstruction loss, and meta-path reconstruction loss are jointly trained to obtain graph vectors suitable for the community detection task, which are then used for clustering processing. Experimental results on four datasets (DBLP, ACM, AMiner, and Freebase) show that the model’s NMI and ARI metrics have increased by an average of 3.16% and 3.2% compared with the current state-of-the-art methods. The maximum improvement in the Purity metric reaches 3.71%, and the visualization effect is prominent, which proves the effectiveness of the model.
- Similar References:
Memo
-
Last Update:
2026-06-12
