B0455
Title: Modeling temporal networks of relational events data
Authors: Subhadeep Paul - The Ohio State University (United States) [presenting]
Abstract: Temporal networks observed through timestamped relational events data are commonly encountered in applications, including online social media, human mobility, financial transactions, and international relations. Temporal networks often exhibit community structure and strong dependence patterns among node pairs. High-dimensional, mutually-exciting Hawkes processes are combined with the stochastic block model to model community structure and node pair dependence. An upper bound is obtained on the misclustering error of spectral clustering of the event count matrix as a function of the number of nodes and communities, time duration, and a quantity measuring the amount of dependence in the model. The theoretical results provide insights into the effects of dependencies in the mutually-exciting Hawkes processes on the accuracy of spectral clustering.