Social media analytics for understanding inter-region interest change dynamics and their causes

Abstract

We propose a set of methods to enable analysis of the dynamics of a topic among different regions over time and their causes. The sub-topic distributions of a topic computed using the Tweets collected from different regions are used to build a graph structure and cluster regions for their common sub-topic interests. The clustering results are further used to reveal the level of consensus and dissensus among the regions through "bubble charts" that can show convergence and divergence patterns of sub-topic interests over time. Through the case analyses, we demonstrate that the proposed methods can progressively pin down how inter-region sub-topic interests changed and what influenced the changes in volume/versatility and consensus/dissensus.