关键词: 复杂网络, 影响力节点, 集体影响, 节点识别

Abstract: To study the identification of a group of influential nodes in complex network, a key and challenging problem is how to get a good balance between accuracy and time complexity. Firstly, considering the interaction between the nodes, this paper proposes a heuristic central index, namely local collective influence index (LCII) that evaluates the influence of nodes, and then designs a local collective influence rank algorithm (LCIR) for influential nodes identification. Although the time complexity of this algorithm is low, in some special networks, the number of selected influential nodes is related to the heterogeneity of networks, it may have the limitation of insufficient selection of influential nodes. Thus, by introducing the idea of influential nodes candidate set and adaptive recalculation method, this paper puts forward a local collective influence rank-adaptive recalculation algorithm (LCIR-AR), which is an improved algorithm of LCIR, and verifies its good performance through the independent cascade model (IC) and network destructive experiments on four real data sets, meanwhile, reveals that low-degree nodes play a major broker role in the network can be powerful influencers.

Key words: complex network, influential nodes, collective influence, identification of nodes