动态有向网络中的时序链路预测问题研究

Research on the Problem of Time-series Link Prediction in Dynamic Directed Network

由于动态向网络具有独特的特征,在对其时序链路预测过程中存在许多问题,为此提出动态有向网络中的时序链路预测问题研究。首先针对动态有向网络特征,采用拓扑结构解决其网络定义问题;利用指数加权滑动平均法对网络时序进行分析,得到T+1时刻预测值,解决时序分析问题;利用聚类算法计算出欧式距离最短的链路路径,解决动态有效网络聚类倾向问题,以此完成动态有向网络中的时序链路预测问题研究。针对上述3个问题提出相对应的解决方法之后,该文中通过仿真实验在真实的动态有向网络中进行了验证,证明通过上述方法解决了相关问题后链路预测效果更佳。

Due to the unique characteristics of dynamic directed networks,there are many problems in the process of predicting its time-series links.To this end,research on the problem of time-series links in dynamic directed networks is proposed.Firstly,according to the characteristics of the dynamic directed network,the topology is used to solve the network definition problem;the exponentially weighted moving average method is used to analyze the network time series to obtain the time prediction value to solve the time series analysis problem;the clustering algorithm is used to calculate the shortest Euclidean distance chain Road path,to solve the problem of clustering tendency of dynamic and effective networks,in order to complete the research on the problem of time-series link prediction in dynamic directed networks.After the corresponding solutions to the above three problems are proposed,the paper verifies the real dynamic directed network through simulation experiments,and proves that the link prediction effect is better after the above methods are solved.