Time-Dependent Influence Measurement in Citation Networks

Abstract

In every scientific discipline, researchers face two common dilemmas: where to find bleeding-edge papers and where to publish their own articles. We propose to answer these questions by looking at the influence between communities, e.g. conferences or journals. The influential conferences are those which papers are heavily cited by other conferences, i.e. they are visible, significant and inspiring. For the task of finding such influential places-to-publish, we introduce a Running Influence model that aims to discover pairwise influence between communities and evaluate the overall influence of each considered community. We have taken into consideration time aspects such as intensity of papers citations over time and difference of conferences starting years. The community influence analysis is tested on real-world data of Computer Science conferences.