Conference: Quantitative Network Science


Clustering citation networks and some results on the power law for degree distributions

{{_Ltalk:R}} Dr. Clement Lee
Date: 26.11.20   Time: 11.00 - 11.45   Room:

Clustering methods for networks are different to those for other kinds of data as network data are relational in nature. While stochastic block models are becoming more popular among such methods, the degree heterogeneity among the nodes in real-life data needs to be accounted for. In this talk, I will introduce a method that addresses this issue for a particular kind of networks, namely directed acylic graphs (DAGs). I will then present some results on modelling the degree distribution directly by extending the power law, which is often concise but sometimes inadequate. Finally, we will look at a statistical application to information diffusion on Twitter, in which the degree is found to be a highly useful covariate for predicting influence on social networks.