Fractal Property Of Mobile Social Networks And Its Application

Mobile social networks are a new social network where individuals access social networking sites to connect with one another through their mobile phone and/or tablet.Network topology properties and applications for MSNs have attracted much attention in recent years. In this work, it is focused on the fractal features and link prediction for MSNs. The main work of this paper is shown as follows:Firstly, a box-covering method based on the ratio of excluded mass to closeness centrality(REMCC) is presented to investigate the fractal feature of MSNs. It would need the fewer boxes of a given size to cover the network than Song's MEMB method.Our simulation results indicate that the REMCC method would get fewer boxes compared with the Song's method and Zhang's method in some real unweighted networks, the results show that the REMCC method is more efficient.Secondly, an improved REMCC method is proposed to analyze the fractal features of MSNs. Each edge is assigned to have a weight and the weight of each edge relies on the average contact duration of the two linked nodes. Then MSNs can be transformed into a weighted network. Expanding the size of the box from the integer domain to the real domain, which can be better applied for analyzing the fractal features of MSNs. Our simulation results indicate that some MSNs are fractal at some time intervals. At last, we analyze the weighted assortativity coefficient of network.We find the weighted assortativity coefficient is negative, which shows that the MSNs are disassortative.Finally, we study the link prediction of MSNs. A link prediction method based on common neighbor and closeness centrality is proposed. In order to consider the time information into the link prediction, we construct a time series model for all unconnected nodes, and take the proposed method into the time series model.Compared with some typical link prediction algorithm by AUC indicators for different datasets. Our simulation results indicate that the proposed method is more efficient than other methods.