Influence of multiple spreaders through farthest first traversal

Abstract

Abstract Identifying influential spreaders in complex networks is crucial for improving the efficiency of spreading processes. In this paper, we use a farthest first traversal to partition the network into communities and identify influential spreaders. The spreaders selected by this method satisfy the two criteria of being dispersed as well as influential in their neighborhood. Using an SIR-based epidemic spread on network datasets we examine the spreading ability of the influential spreaders. We compare the epidemic size when initial spreaders are selected from each community with ranked initial spreaders but no assurance of representation from each community. A larger epidemic size is observed when the initial influential spreaders are dispersed and selected from each community. In addition, the spread ability of influential spreaders selected by the proposed methods is similar to those obtained by creating non-overlapped communities using the Louvain algorithm, provided they are selected by using the same criteria.

Keywords

• Multiple influential spreaders
• Farthest first traversal
• Non-overlapping communities
• SIR