关键词: 链路预测, 有向网络, 时序分析

Abstract: Most real networks are directed, and the network structure changes dynamically with time. Traditional link prediction methods are mostly applicable to undirected networks, and their analysis methods cannot effectively mine information in real networks. Aiming at the above problems, this paper proposes a timing-directed link prediction algorithm based on normalized AA and LAS. Based on the common neighbor, node degree attribute and local community similarity, the algorithm assigns a time impact factor to each link and substitutes it into the NALAS indicator for calculation. It considers the impact of network directionality and network history structure. The algorithm is simulated on real social network datasets and compared with Salton and Jaccard algorithms. The results show that the proposed algorithm is improved compared with other algorithms, which indicates that the algorithm can effectively predict the link in a time-oriented social network.

Key words: link prediction, directed network, time series analysis