Identifying influential spreaders in complex networks

Abstract

In this project, three algorithms are implemented to find the influential spreaders in the complex network and compared to help people choosing the proper algorithm for finding the suitable spreaders in social network to spread the idea or information effectively. The implemented algorithms are K-shell decomposition, heuristic group discovery algorithm (HGD algorithm) and genetic algorithm (GA). The SIR model is used to simulate the influence of spreaders in the social network. Three sizes of graph network are used to test the ability to find influential spreaders and contain around 4000 nodes, 2000 nodes and 600 nodes respectively. The sum of the infected node and the recovered node is recorded to determine the effectiveness and the running time of the algorithms are recorded to determine the time consumption. The k-shell decomposition is suitable for the close network (average shortest path < 2.5) which HGD algorithm is suitable for the wide network (average shortest path ≥ 2.5) when a quick result is needed. GA can generate the good result in different network type, but it is very time consuming and require more runs to try more combination.