Identifying Influential Nodes in Complex Networks Based on Weighted Formal Concept Analysis

Abstract

The identification of influential nodes is essential to research regarding network attacks, information dissemination, and epidemic spreading. Thus, techniques for identifying influential nodes in complex networks have been the subject of increasing attention. During recent decades, many methods have been proposed from various viewpoints, each with its own advantages and disadvantages. In this paper, an efficient algorithm is proposed for identifying influential nodes, using weighted formal concept analysis (WFCA), which is a typical computational intelligence technique. We call this a WFCA-based influential nodes identification algorithm. The basic idea is to quantify the importance of nodes via WFCA. Specifically, this model converts the binary relationships between nodes in a given network into a knowledge hierarchy, and employs WFCA to aggregate the nodes in terms of their attributes. The more nodes aggregated, the more important each attribute becomes. WFCA not only works on undirected or directed networks, but is also applicable to attributed networks. To evaluate the performance of WFCA, we employ the SIR model to examine the spreading efficiency of each node, and compare the WFCA algorithm with PageRank, HITS, K-shell, H-index, eigenvector centrality, closeness centrality, and betweenness centrality on several real-world networks. Extensive experiments demonstrate that the WFCA algorithm ranks nodes effectively, and outperforms several state-of-the-art algorithms.