基于关系强度理论与反馈机制的信息传播动态网络表示

doi:10.19678/j.issn.1000-3428.0063656

摘要/Abstract

摘要： 现有多数网络表示学习方法不能很好地贴合真实世界的信息传播网络，且无法对信息传播动态网络的时间特性与动力学演化特征进行有效建模。提出一种新的信息传播动态网络表示模型，基于关系强度将信息传播动态网络划分为关系网络与传播网络，并分别计算变化节点对的概率密度和邻接矩阵。通过更新节点注意力强度矩阵，聚合节点邻域变化信息，并融合节点邻域变化信息、自身历史信息以及外部影响因素，对信息传播动态网络进行归纳式表示学习。引入反馈机制，将最新的节点表示反馈到邻居节点，解决网络表示不及时的问题，提升网络表示性能。实验结果表明，与Know-Evolve、DyRep、LDG等模型相比，该模型的命中率和平均排名提升显著，与LDG模型相比，其时间效率在Social Evolution数据集和Github数据集上分别提升了91.8%、87.2%。

关键词: 信息传播, 动态网络表示, 关系强度, 反馈机制, 链接预测

Abstract: Most of the existing Network Representation Learning (NRL) methods can not fit well with the real world information dissemination network, and can not effectively model the time characteristics and dynamic evolution characteristics of the information dissemination dynamic network.A new Dynamic Network Representation(DNRep) model of information dissemination is proposed.First, the information transmission dynamic network is divided into relationship and communication networks based on the relationship strength.Second, the probability density and adjacency matrix of variable node pairs are calculated.Then, the attention intensity matrix of the nodes was updated to aggregate the node neighborhood change information.Finally, the inductive representation learning of the information transmission dynamic network is performed by integrating the change information, historical information of the nodes, and external influencing factors.Meanwhile, a feedback mechanism is introduced to feed back the latest node representation information to the neighbor node to solve the delay in network presentation and improve the network representation effect.The experimental results indicate that the hit rate (HITS@10) and Mean Average Rank (MAR) of the proposed model are higher than that of the Know-Evolve, DyRep, LDG, and other models.Compared with the LDG model, the time efficiency of the proposed model based on the social evolution and Github datasets increased by 91.8% and 87.2%, respectively.

Key words: information diffusion, Dynamic Network Representation(DNRep), relationship strength, feedback mechanism, link prediction

中图分类号:

TP391

潘乐, 李弼程, 万旺, 曾荣燊. 基于关系强度理论与反馈机制的信息传播动态网络表示[J]. 计算机工程, 2023, 49(2): 246-253.

PAN Le, LI Bicheng, WAN Wang, ZENG Rongshen. Dynamic Network Representation of Information Diffusion Based on Relationship Strength Theory and Feedback Mechanism[J]. Computer Engineering, 2023, 49(2): 246-253.

https://www.ecice06.com/CN/Y2023/V49/I2/246

图/表 8

20230216182019

20230216182023

20230216182026

20230216182029

20230216182032

20230216182035

20230216182038

20230216182042

参考文献

[1] PEROZZI B, AL-RFOU R, SKIENA S.DeepWalk:online learning of social representations[C]//Proceedings of the 20th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining.New York, USA:ACM Press, 2014:701-710.
[2] BOURIGAULT S, LAMPRIER S, GALLINARI P.Representation learning for information diffusion through social networks:an embedded cascade model[C]//Proceedings of the 9th ACM International Conference on Web Search and Data Mining.New York, USA:ACM Press, 2016:573-582.
[3] LI C, MA J Q, GUO X X, et al.DeepCas:an end-to-end predictor of information cascades[C]//Proceedings of the 26th International Conference on World Wide Web.New York, USA:ACM Press, 2017:577-586.
[4] JOLLIFFE I T.Pincipal component analysis[J].Journal of Marketing Research, 2002, 25(4):513-521.
[5] BELKIN M, NIYOGI P.Laplacian eigenmaps and spectral techniques for embedding and clustering[C]//Proceedings of the 14th International Conference on Neural Information Systems.New York, USA:ACM Press, 2001:585-591.
[6] GROVER A, LESKOVEC J.Node2vec:scalable feature learning for networks[C]//Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining.New York, USA:ACM Press, 2016:855-864.
[7] YANG C, LIU Z Y, ZHAO D L, et al.Network representation learning with rich text information[EB/OL].[2021-11-10].https://www.semanticscholar.org/paper/Network-Representation-Learning-with-Rich-Text-Yang-Liu/fce14c6aa64e888456256ac6796744683165a0ff.
[8] YAO W, KUMAR A, ROLIA J, et al.A context-aware framework for patient navigation and engagement[C]//Proceedings of the 8th IEEE International Conference on Collaborative Computing:Networking, Applications and Worksharing.Washington D.C., USA:IEEE Press, 2012:316-325.
[9] TU C C, ZHANG W C, LIU Z Y, et al.Max-margin deepwalk:discriminative learning of network representation[C]//Proceedings of the 25th International Joint Conference on Artificial Intelligence.New York, USA:ACM Press, 2016.3889-3895.
[10] GOYAL P, KAMRA N, HE X R, et al.DynGEM:deep embedding method for dynamic graphs[EB/OL].[2021-11-10].https://arxiv.org/abs/1805.11273.
[11] LIU W Q, LI H X, XIE B.Real-time graph partition and embedding of large network[C]//Proceedings of the18th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing.Washington, USA:IEEE Press, 2018:432-441.
[12] WANG Q, YU Y H, GAO H Y, et al.Network representation learning enhanced recommendation algorithm[J].IEEE Access, 2019(7):61388-61399.
[13] XIONG Y, ZHANG Y, FU H J, et al.DynGraphGAN:dynamic graph embedding via generative adversarial networks[C]//Proceedings of International Conference on Database Systems for Advanced Applications.Berlin, Germany:Springer, 2019:536-552.
[14] GOYAL P.Dyngraph2vec:capturing network dynamics using dynamic graph representation learning[J].Knowledge-Based Systems, 2020, 187(1):104816-104822.
[15] TRIVEDI R, FARAJTABAR M, BISWAL P, et al.DyRep:learning representations over dynamic graphs[EB/OL].[2021-11-10].https://arxiv.org/abs/1803.04051.
[16] FRIEDKIN N E.Information flow through strong and weak ties in intraorganizational social networks[J].Social Networks, 1982, 3(4):273-285.
[17] KEUCHENIUS A, TÖRNBERG P, UITERMARK J.Adoption and adaptation:a computational case study of the spread of Granovetter's weak ties hypothesis[J].Social Networks, 2021, 66(7):10-25.
[18] 宾宁, 冼文峰, 胡凤.基于强弱关系的社交网络信息传播博弈模型[J].现代情报, 2016, 36(12):30-35. BIN N, XIAN W F, HU F.Game model of information transmission in social networks:based on strong and weak relationship theory[J].Journal of Modern Information, 2016, 36(12):30-35.(in Chinese)
[19] Social evolution dataset.[EB/OL].[2021-11-10].http://realitycommons.media.mit.edu/socialevolution.html.
[20] GH Archive.[EB/OL].[2021-11-10].https://www.gharchive.org.
[21] KNYAZEV B, AUGUSTA C, TAYLOR G W.Learning temporal attention in dynamic graphs with bilinear interactions[J].PLoS One, 2021, 16(3):36-45.
[22] KONG C, CHEN B, LI S, et al.D2NE:deep dynamic network embedding[C]//Proceedings of International Conference on Advanced Data Mining and Applications. Berlin, Germany:Springer, 2020:168-175.
[23] KINGMA D P, BA J.Adam:a method for stochastic optimization[EB/OL].[2021-11-10].https://arxiv.org/abs/1412.6980.
[24] HAMILTON W L, YING R, LESKOVEC J.Inductive representation learning on large graphs[EB/OL].[2021-11-10].https://arxiv.org/abs/1706.02216.
[25] VELIČKOVIĆ P, CUCURULL G, CASANOVA A, et al.Graph attention networks[EB/OL].[2021-11-10].https://arxiv.org/abs/1710.10903.
[26] TRIVEDI R, DAI H J, WANG Y C, et al.Know-Evolve:deep temporal reasoning for dynamic knowledge graphs[EB/OL].[2021-11-10].https://arxiv.org/abs/1705.05742.

选择文件类型/文献管理软件名称

选择包含的内容