Tracking the Evolution of Communities and Research Topics in a Dynamic Citation Network

Abstract

Understanding the evolution of research topics is essential to the development of any discipline. For funding agencies, academic institutions, individual researchers, and academic conference organizers, it helps them to understand the trends in their disciplines from a macro perspective and to make better decisions. In this paper, a method for understanding research topic evolution is proposed to answer the following questions: how can we use community detection approaches to locate research topics in citation networks? And how can we track the evolution of research topics in a dynamic citation network? This study used modularity-based algorithm for community detection, keyword word frequency for topic recognition using the tf-idf algorithm, and a clear definition of seven community events (Birth, Death, Growth, Contraction, Merging, Splitting and Continue). Based on this approach, research topics and disciplinary frontier developments can be better predicted and understood.