Tip Over Community: Backbone Nodes Detection Based on Community Structure

Key nodes detection is conducive to the information promotion and public opinion monitoring. Traditional works mainly focus on the nodes affecting the breadth of information diffusion. However, there are several works concentrate on the depth of information diffusion. Breadth dictates the number of nodes receiving information in one-hop propagation. Depth implies the length of the retweeting cascade of information and the number of social circles getting information. Depth is as important as breadth, for the nodes which contribute to the depth of information diffusion can promote society propagation (SP). If we have a good control of these nodes, information will ignite the community and minority will tip over the cognitive of the group. The purpose of this paper is to identify the influential nodes which can influence the SP in breadth and depth. With the analysis of SP, we proposed backbone node detection based on community structure (BNDCS) to detect these nodes. In this paper, we defined backbone nodes (BNs) by the location feature and the process of information diffusion. Besides, we presented a cohesion subgroup detection method based on hybrid seed expansion. It can adapt the network with no significant community structure and fit the star topology well. We proposed an information tension factor to measure the influence of bridge nodes. Through the experiment contrasts, the BNDCS is superior to compared methods. It showed that the BNs play a critical role in SP. Therefore, monitoring the BNs can prevent the mass incidents occurring and stop the spread of panic.