Connecting node age with centrality measurement in growing networks

Inferring the evolution history of complex networks is a crucial problem relating to a wide range of real problems. A representative application is that one can significantly improve the link prediction accuracy via network time information. In this context, we systematically study the performance of centrality metrics in identifying node ages in growing networks. Interestingly, we find that the accuracy is strongly related to the decay speed of nodes' attractiveness during networks' growth. We reveal the range of decay factor where the centrality metrics are suitable for detecting node ages, and identify several metrics that perform stably in this task. These findings are finally validated in real-world growing networks. (paper: interdisciplinary statistical mechanics)