Visual analytics of text streams through multiple dynamic frequency matrices

We propose a Visual Analytics tool that supports situation awareness and exploration tasks for text streams. To reach this goal, we design our own data model to encode streaming text in multiple dynamic frequency matrices, handling multiple aspects of data. Our visualizations are composed first of two dynamic Theme Rivers. They allow real time exploration of all the aspects extracted from texts stored in both, short-term and long-term buffers. In addition, we visualize the geographical location of messages on a map. We use these visualizations, enhanced by efficient user interaction mechanisms, to answer the questions of the third mini-challenge of 2014 VAST Challenge. As a result we identify several challenging issues that we will investigate in future work.