摘要
链路预测是一种挖掘数据之间潜在关系的重要方法。传统的链路预测算法主要基于静态网络,而现实生活中绝大多数网络是动态的,因此原有的算法性能受到了限制。文章首先在网络局部结构信息的基础上,引入节点共同邻居之间的连边信息,提出了一种混合结构相似性指标。通过建立网络的时间序列,文章将该指标与线性回归预测模型相结合,得到了一种适用于动态网络的混合结构线性回归算法。该算法充分利用了网络的时间信息与结构信息,真实的实验结果表明,混合结构线性回归算法性能优于传统的链路预测算法,具有更高的预测精度。
Link prediction is an important method to mine potential relationships of data. Most networks in real life are the dynamic networks, but the traditional link prediction algorithms assume that the networks are static. Thus, their performance are restricted. In this paper, we firstly consider the edge of common neighborhoods of node pairs and propose a similarity index using hybrid structural information. And then we establish the time series and propose the hybrid structural linear regression algorithm（HS_LR） based on the proposed similarity index and linear regression model. The HS_LR takes advantage of both network temporal information and structural information, and the results in real dynamic networks show that HS_LR algorithm outperforms the traditional methods and achieves great improvement in precision.
作者
刘继嘉
王童
何兴盛
傅忠谦
Lin Jijia;Wang Tong;He Xingsheng;Fu Zhongqian(Department of Electronic Science and Technology,University of Science and Technology of China,Hefei 230027,Chin)
出处
《电子技术（上海）》
2018年第7期53-59,共7页
Electronic Technology
关键词
链路预测
动态网络
混合结构
时间序列
线性回归
linkprediction
dynamic n＇etworks
hybrid structure
time series
linear regression
