基于节点相似性的时序网络节点重要性识别算法
Node Importance Identification Algorithm for Temporal Networks Based on Node Similarity

时序网络可以更加准确地描述网络节点之间的交互顺序。本文提出基于节点相似性的时序路径聚集方法识别时序网络节点重要性，具体思想是在聚集时序路径的邻居信息评估目标节点重要性时，与目标节点相似性越高的邻居节点具有更高的影响力。同时，通过引入衰减因子区分不同时序路径长度邻居的信息权重，融合节点相似性系数和衰减因子构建基于节点相似性的时间信息聚集模型度量目标节点的重要性。在两个实证网络数据集上的实验结果显示相比于经典的方法，本文方法的肯德尔相关系数最高提高13.85%。该结果表明节点相似性系数的引入能够有效邻居信息来评估时序网络节点重要性。
Temporal networks can more accurately describe the interaction order between network nodes. In this paper, we propose a node similarity-based temporal path aggregation method to identify the importance of temporal network nodes, the specific idea is that when aggregating the neighbor information of temporal paths to assess the importance of the target node, the neighbor nodes with higher similarity to the target node have higher influence. At the same time, by introducing attenuation factors to distinguish the information weights of neighbors with different temporal path lengths, the node similarity coefficient and attenuation factors are fused to construct a node similarity-based temporal information aggregation model to measure the importance of target nodes. The experimental results on two empirical network datasets show that the Kendall correlation coefficient of this paper’s method is improved by up to 13.85% compared with the classical method. This result indicates that the introduction of node similarity factor can effectively neighbor information to assess the importance of temporal network nodes.