Modeling And Searching For Time- Varying Network Based On Spatial Activity

A lot of real networks have time-varying characteristics. Modeling for time varying network is helpful for people to understand the structure and function of real networks. When the time scale of the evolution of the network topology is equivalent to that of the propagation dynamics, there are some deviations in the dynamic process based on some static models. As a basic dynamic process, the searching for complex networks has a wide range of applications, such as the searching for web pages in the Internet. Many classical searching strategies are usually based on static network, the effectiveness of which in time-varying network remains to be verified. The research of searching strategies for time-varying network provides reference for solving practical problems such as quick delivery of messages in social networks.Based on these problems above, this paper proposes a time-varying spatial network model driven by activity based on the characteristics of online social networks, which is called spatial activity network model. And the searching process on the time-varying network is discussed. The main contribution of this paper is as follows:1. An online social network with time-varying property is constructed by Twitter dataset and the spatial activity network model is proposed combined with the empirical analysis results. A series of topology in a certain interval of time for online social network are constructed by Twitter dataset. Some results are found that the activity distribution of users is virtually independent of time scale and degree distribution is heterogeneous as well as geographical distance distribution through analysis. Combined with the empirical analysis results, a spatial activity network model driven by both activity and preferential attachment is proposed. The statistical characteristics are consistent with empirical analysis results, which is proved that the building mechanisms of the model are accurate.2. Searching process is carried on spatial activity network. We realize strategies of random walk, maximum activity search and greedy search separately and compare efficiency after proposing three measure indexes for searchingstrategy. In order to study searching on spatial activity network, firstly, we introduce search time, search path length and waiting time as evaluation indexes for search strategy. Secondly, we propose the maximum activity searching strategy according to activity property of nodes and realize the strategies of random walk, maximum activity searching and greedy searching. Finally, we use these searching strategies to search on the spatial activity network and calculate measure indexes, finding the efficiency of greedy searching strategy is the highest and that of random walk is the lowest.3. Greedy searching strategy is improved and a maximum activity minimum distance searching strategy is proposed for searching optimization through analysis of spatial activity network. In order to ensure that the direction of the searching is not deviate from the destination, we improve greedy searching strategy by means of taking the distance from the current node to the target node into consideration. The simulation results show that the improved searching strategy is more efficient than the classical greedy searching strategy. In addition, we propose a maximum activity minimum distance searching strategy which is in combination with activity and geographical distance. It is found that using this strategy to search on the spatial activity network will get higher efficiency than any of other strategies discussed before so as to optimize the searching process.