电子学报 ›› 2022, Vol. 50 ›› Issue (8): 1830-1839.DOI: 10.12263/DZXB.20211288

• 学术论文 • 上一篇    下一篇

基于元路径的动态异质网络表示学习

刘群, 谭洪胜, 张优敏, 王国胤   

  1. 重庆邮电大学计算机科学与技术学院,重庆 400065
  • 收稿日期:2021-09-17 修回日期:2022-04-21 出版日期:2022-08-25
    • 作者简介:
    • 刘 群 女,1969年出生于江西省南昌市,博士,重庆邮电大学教授,博士生导师.主要研究方向为复杂网络、可解释人工智能和数据挖掘.E-mail: liuqun@cqupt.edu.cn
      谭洪胜 男,1996年出生于重庆市.硕士生,主要研究方向为网络表示学习.E-mail: redtan2000@163.com
      张优敏 女,1987年出生于陕西省兴平市.博士生,主要研究方向为图神经网络的可解释.E-mail: ymzhang0103@hotmail.com
      王国胤 男,1970年出生于重庆市,博士,重庆邮电大学教授,博士生导师.教育部“长江学者”特聘教授(2015‒2019)、中组部“万人计划”科技创新领军人才(2014)、人社部“新世纪百千万人才工程”国家级人选、国务院特殊津贴专家、中科院“百人计划”专家、教育部“新世纪优秀人才”.主要研究方向为粗糙集、粒计算、数据挖掘、认知计算、大数据、人工智能等.E-mail: wanggy@cqupt.edu.cn
    • 基金资助:
    • 国家自然科学基金重点项目 (61936001); 重庆市教委重点合作项目 (HZ2021008); 重庆市自然科学基金 (cstc2021ycjh-bgzxm0013)

Dynamic Heterogeneous Network Representation Method Based on Meta-Path

LIU Qun, TAN Hong-sheng, ZHANG You-min, WANG Guo-yin   

  1. College of Computer Science and Technology,Chongqing University of Posts and Telecommunications,Chongqing 400065,China
  • Received:2021-09-17 Revised:2022-04-21 Online:2022-08-25 Published:2022-09-08

摘要:

对网络表示学习的研究已经取得了很多成果,但是大部分网络表示学习模型忽略了网络动态性和异质性,无法区分网络中耦合的时间和空间(结构)特征,也不能捕获网络的丰富语义信息.本文提出了基于元路径的动态异质网络表示学习方法.首先将节点的邻域结构按照时间划分出不同的子空间结构,并为每个节点采样出所有时间加权元路径的序列.其次通过门控循环单元将节点的全部时间加权元路径序列上的邻域信息进行集成,最后利用带注意力机制的双向门控循环单元对融合后的节点序列进行时空上下文信息学习,获得每个节点的最终表示向量.通过在真实数据集上的实验表明,在节点分类、聚类和可视化的下游任务测试中,本文提出的算法较基线方法在性能上均有较大提升.节点分类任务中的Micro-F1平均提高了1.09%~3.72%,节点聚类任务中的ARI值提高了3.23%~14.49%.

关键词: 网络表示学习, 动态异质网络, 元路径, 注意力机制, 门控循环单元

Abstract:

The researches of network representation learning have made many achievements. Since most of the researches ignore the dynamics and heterogeneity of the networks, coupled temporal and spatial structure features can not be distinguished, and rich semantic information of the network cannot be captured well. In this paper, meta-path based dynamic heterogeneous network representation learning method is proposed. Firstly, the neighborhood structures of nodes are divided into different sub-spaces according to their time, then the sequences of all time-weighted meta-paths for each node are sampled. Secondly, the neighborhood information on all time-weighted meta-paths of each node is integrated by a gated recurrent unit network(GRU). Furthermore, a bi-directional gated recurrent unit network(Bi-GRU) with an attention mechanism is used to learn the spatio-temporal contextual information from the merged sequences, and the final node representation will be received. Experiments on real data sets show that our algorithm has greatly improved performance on the downstream network tasks, such as node classification, clustering and visualization. Compared with state-of-the-art baseline methods,the Micro-F1 value has been raised by 1.09%~3.72% averagely on classification tasks, and the ARI value has been increased by 3.23%~14.49% on clustering tasks.

Key words: network representation learning, dynamic heterogeneous network, metapath, attention mechanism, gated recurrent unit

中图分类号: