Cost effective approach to identify multiple influential spreaders based on the cycle structure in networks

Identifying influential spreaders has theoretical and practical significance in complex networks. Traditional centrality methods can efficiently find a single spreader, but it could lead to influence redundancy and high initializing costs when used to identify a set of multiple spreaders. A cycle structure is one of the most crucial reasons for the complexity of a network and the cornerstone of the feedback effect. From this novel perspective, we propose a new method based on basic cycles in networks to identify multiple influential spreaders with superior spreading performance and low initializing costs. Experiments on six empirical networks show that the spreaders selected by the proposed method are more scattered in the network and yield the best spreading performance compared with those on seven well-known methods. Importantly, the proposed method is the most cost effective under the same spreading performance. The cycle-based method has the advantage of generating multiple solutions. Our work provides new insights into identifying multiple spreaders and hence can benefit wide applications in practical scenarios.