Identifying Influential Spreaders in Complex Networks by an Improved Spectralrank Algorithm

Abstract

Spreading process is a common natural phenomenon in complex networks, such as the spreading of rumors, diseases, information and so on. The small-world and scale-free characteristics of complex networks facilitate the fast transmission of infectious diseases and various behaviors. In order to control the rapid spread of disease or false information, it is important to design reasonable algorithms to predict the influential spreaders. This paper develops an improved SpectralRank algorithm to identify the influential spreaders in complex networks. The proposed algorithm improved the SpectralRank algorithm by adding a weight on the edge from each node to the ground node, which is called the SpectralRank algorithm with weighted out-degree (SRWO). Numerical simulations reveal that the SRWO can well identify actually influential spreaders in several real-world complex networks. Comparing with existing methods, the SRWO has several merits, and importantly, it is robust to random attacks on edges and nodes. The related investigations have potential applications in the prevention and control of infectious diseases and rumors, as well as the formulation of proper propaganda strategies.