Centralidade de tempo em grafos variantes no tempo

Abstract

Currently, there is a great interest in investigating dynamics in Time-Varying Graphs (TVGs). These graphs contemplate the temporal evolution, both nodes and edges. In this scenario, similar to static graphs, centrality usually refers to metrics that assess the relative importance of vertices. However, in TVGs it is possible to assess the importance of time instants (or states) of a graph throughout its existence. Determining important time instants in this context may have strong practical applications and is particularly useful for defining best times to spread, generate models and predict the behavior of TVGs. In this paper, we define time centrality in Time-Varying Graphs. Time centrality evaluates the relative importance of time instants. We present two time centrality metrics focused on information dissemination processes and another based on layout of network connections.. We evaluate metrics we define relying in a real dataset from an hospital environment. Our results show that the best classified time instants, according to created metrics, can make a faster and more efficient diffusion process. Compared to a random choice, the diffusion process starting at best rated time instants can up to 2.5 times faster, and it also can reach almost double the number of nodes in the network in some cases.