Three-level Visualization of Internet Discussion with Extruded Word Clouds

Abstract

Visualization of discussions is an intensively explored information visualization area. Many existing approaches explore the discussions from the point of view of individual participants, often omitting interesting information about the discussion itself. We present a visualization technique that centers around the discussion, its dynamics, intensity and topic changes. The flow of the text is divided into uniform time spans that aggregate the data and where the text analyzer runs. Then, the discussion is visualized using three different levels of details. The topmost shows the global view of the discussion development over time, the middle shows an inside view on discussions' main topics, and the bottom level displays individual word clouds. The technique is demonstrated on data from an Internet Relay Chat (IRC) but it can be used for visual analysis of any time-dependent text data such as email communication, search terms, keywords and similar.