The Structure Characteristics Empirical Analysis And Research In Temporal Network

Complex networks exist everywhere in real life, researching complex networks’ structure and characteristics can help us to understand, predict and optimize the network’s dynamic behavior. In the past, most scholars abstracted networks into static diagrams that ignored the networks’ time attribute. But in real life, networks especially human behavior networks are closely related humans schedule, events in these networks are temporal order, each side has an active time sequence. Complex networks study includes different areas. Considering these networks’ time attribute makes all kinds of their features change.This thesis’ main job is to abstract networks with time properties into temporal diagrams, and analyze the network structural characteristics, the node importance and the transmission dynamics, etc. This thesis introduces temporal networks’ definition firstly, and uses the method of getting the discrete signal cycle sequence by power spectrum in signal processing to acquire temporal networks’ period, and does the empirical analysis on actual data. Then it introduces the temporal network structure characteristics, proposes the temporal shortest path algorithm based on Dijkstra algorithm, uses mathematical methods to prove its correctness, analyzes its complexity and compares it with Holme’s algorithm. Furthermore, it defines the node centricity metrics in temopral networks such as betweenness, closeness, etc. And it does the analysis and comparison in constructing networks. After that, it studies mining the importance nodes in the temporal networks based on the shapley value method in static networks, considers the time propertie of networks and proposes events related contagion and get the temporal social networks’ node importance evaluation index and sorting algorithm. The results show that the proposed algorithm can effectively mining important nodes in the temporal network. Since temporal networks have their unique characteristics, in recent years, many scholars research the effects of temporal network’s features on the dissemination of information, such as bursty and cyclical. In the final, this thesis adopts the SIS and SIR propagation model and the history-dependent contagion to study the effect of bursty on information dissemination, and uses the data sets Sexual Escort and Infectious to do example analysis.This thesis proposes my own algorithm, and improves the others’ algorithm to adapt to the temporal network. The shortest distance algorithm proposed in this thesis has high time complexity, but it is accurate. The important nodes mining algorithm is proven to be effective to mining important nodes on temporal social networks.