Scaling Up Integrated Structural and Content-Based Network Analysis

Abstract

Structural analysis of social networks can provide important insights into the clusters and important nodes. However, it is silent on the content-based reasons for importance or commonality. This additional level of insight requires sampling content from nodes and processing it to distill new insights. That is done effectively by human analysts, but as networks grow into Big Data scale, human analysis is not possible. This raises the question of whether automated techniques can mimic the same results humans find. In this paper, we demonstrate how topic modeling can be applied, filtered, and adapted to produce easy-to-understand keywords that represent important clusters in a network. Those keywords reflect the insights achieved by human analysts doing a manual content-based analysis of the network features. While humans should never be removed from the analysis process, this work shows how automated techniques can be integrated to scale humans’ ability to gain insights in large networks.