An attribute-based Node2Vec model for dynamic community detection on co-authorship network

Networks offer a wide range of applications in various domains of life and scientific research. Community detection, which aims at understanding the structure and function of complex networks, is a basic and essential task in network analysis. In this study, we propose an approach for community detection in a dynamic network based on network embedding, incorporating both network topology and node attributes. Furthermore, we analyze the evolution of statistician collaborative patterns and statistical research topics based on dynamic co-authorship networks through publications that are collected from 43 statistical journals from 2001 to 2021. Specifically, we explore the dynamic community detection results based on the newly proposed approach and conduct statistical analysis from the following perspectives. First, the evolution information of the community center is mined. Second, we explore the collaboration mode of community institutions. Finally, we track the evolution of community research content. This study provides a novel method for exploring network representation with node attributes and the analysis of dynamic community detection, as well as offers multiple perspectives for community detection analysis.