《计算机应用研究》|Application Research of Computers

有向动态网络中基于模体演化的链路预测方法

Link prediction method based on motif evolution in directed dynamic networks

免费全文下载 (已被下载 次)  
获取PDF全文
作者 杜凡,刘群
机构 重庆邮电大学 计算智能重庆市重点实验室,重庆 400065
统计 摘要被查看 次,已被下载
文章编号 1001-3695(2019)05-034-1441-05
DOI 10.19734/j.issn.1001-3695.2017.11.0738
摘要 以往传统的链路预测方法大多数针对无向网络,而实际上大多数社交网络是有向的,并且没有考虑网络中同一节点对之间的重复边以及微观演化信息,因此不能较好地解决有向动态网络中的链路预测问题。针对有向网络,将节点对之间的重复边信息转换为该节点对之间连边的权值;接着采用了基于三元组模体的演化模型,对滑动窗口中相邻时间片的模体转换概率进行统计后,采用指数加权滑动平均法对其进行时序分析得到不同模体转换概率的预测矩阵,进而使用该矩阵对网络中的链边进行预测。这不仅充分利用了网络微观演化信息,而且解决了动态网络中重复边的问题。最后对实验结果进行分析发现,在高全局聚类系数高平均度的网络中AUC相比Triad Transition Matrix方法提高了近0.01,而相比CN方法提高更多。因此,所提方法能够较好地应用网络微观演化信息进行链路预测。
关键词 时序链路预测; 有向网络; 模体演化; 时序分析
基金项目 国家自然科学基金资助项目(61572091,61075019)
重庆市自然科学基金资助项目(CSTC2014jcyjA40047)
重庆市教委研究项目(KJ1400403)
重庆邮电大学博士启动资助项目(A2014-20)
本文URL http://www.arocmag.com/article/01-2019-05-034.html
英文标题 Link prediction method based on motif evolution in directed dynamic networks
作者英文名 Du Fan, Liu Qun
机构英文名 Chongqing Key Laboratory of Computational Intelligence,Chongqing University of Posts & Telecommunication,Chongqing 400065,China
英文摘要 In the past, most of the traditional link prediction methods are oriented to the undirected network, in fact, most social networks are directional, and do not consider the duplication between the same node pair and the microscopic evolution information in the network, therefore they can not solve link prediction in directed dynamic networks better. This paper focused on the directional network, and transformed the repeated edge information between the pair of nodes into the weight of the edge between the pair of nodes, then it used the evolution model based on the triad motif, calculated the motif transformation probability matrix between the adjacent time slice in the move window, analyzed the probability matrix by exponentially weighted moving average, and then it used the matrix to predict the chain edge in the network. This method not only makes full use of the network micro evolution information, but also solves the problem of overlapping edges in dynamic network. Experiments show that this method can get better results than CN, Triad Transition Matrix and other methods in network with high global clustering coefficient and high average degree.Therefore, this method can apply the network microscopic information to the link prediction better.
英文关键词 time series link prediction; directed network; motif evolution; time series analysis
参考文献 查看稿件参考文献
 
收稿日期 2017/11/9
修回日期 2018/1/5
页码 1441-1445,1453
中图分类号 TP181
文献标志码 A