Visual Exploration of Spatial and Temporal Variations of Tweet Topic Popularity

Abstract

We present a visual analytical approach to exploring variation of topic popularity in social media (such as Twitter) over space and time. Our approach includes an analytical pipeline and a multi-view visualization tool. As attempts of topic extraction from very short texts like tweets may not produce meaningful results, we aggregate the texts prior to applying topic modelling techniques. Interactive visualisations support detection of burst events in social media posting activities at different locations, show the spatial, temporal, quantitative, and semantic aspects of these events, and enable the user to explore how popularity of topics varies over cities and time. A case study has been conducted using a real-world tweet dataset.