作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程 ›› 2021, Vol. 47 ›› Issue (6): 188-196. doi: 10.19678/j.issn.1000-3428.0058719

• 移动互联与通信技术 • 上一篇    下一篇

基于信息传播节点集的CTDN节点分类算法

黄鑫1,2, 李赟2, 熊瑾煜2   

  1. 1. 中国人民解放军战略支援部队信息工程大学 信息系统工程学院, 郑州 450001;
    2. 盲信号处理国家级重点实验室, 成都 610041
  • 收稿日期:2020-06-22 修回日期:2020-08-11 发布日期:2020-08-17
  • 作者简介:黄鑫(1988-),男,工程师、硕士研究生,主研方向为智能信息处理;李赟,助理研究员、博士;熊瑾煜,副研究员、博士。

Node Classification Algorithm Based on Information Propagation Node Set for CTDN

HUANG Xin1,2, LI Yun2, XIONG Jinyu2   

  1. 1. College of Information System Engineering, PLA Strategic Support Force Information Engineering University, Zhengzhou 450001, China;
    2. National Key Laboratory of Science and Technology on Blind Signal Processing, Chengdu 610041, China
  • Received:2020-06-22 Revised:2020-08-11 Published:2020-08-17
  • Contact: 国防科技重点实验室基金。 E-mail:Huangx_0735@163.com

摘要: 针对连续时间动态网络的节点分类问题,根据实际网络信息传播特点定义信息传播节点集,改进网络表示学习的节点序列采样策略,并设计基于信息传播节点集的连续时间动态网络节点分类算法,通过网络表示学习方法生成的节点低维向量以及OpenNE框架内的LogicRegression分类器,获得连续时间动态网络的节点分类结果。实验结果表明,与CTDNE和STWalk算法相比,该算法在实验条件相同的情况下,网络表示学习结果的二维可视化效果更优且最终的网络节点分类精度更高。

关键词: 信息传播节点集, 连续时间动态网络, 网络表示学习, 节点分类, 随机游走, Skip-Gram模型

Abstract: The study described in this paper addresses the problem of node classification in Continuous-Time Dynamic Network(CTDN).In this work, an information propagation node set is defined according to the features of the actual network information propagation, and the node sequence sampling strategy in network representation learning is improved.Based on the defined information propagation node set, a node classification algorithm for CTDN is designed.The algorithm employs the network representation method to generate the low-dimensional node vector, and uses the LogicRegression classifier to obtain the node classification results of CTDN.Experimental results show that the proposed algorithm outperforms the existing classic algorithms such as CTDNE and STWalk under the same experimental conditions, providing better 2D visualized network representation learning results and higher network node classification accuracy.

Key words: information propagation node set, Continuous-Time Dynamic Network(CTDN), network representation learning, node classification, random walk, Skip-Gram model

中图分类号: