WordBridge: Using Composite Tag Clouds in Node-Link Diagrams for Visualizing Content and Relations in Text Corpora

Abstract

We introduce WordBridge, a novel graph-based visualization technique for showing relationships between entities in text corpora. The technique is a node-link visualization where both nodes and links are tag clouds. Using these tag clouds, WordBridge can reveal relationships by representing not only entities and their connections, but also the nature of their relationship using representative keywords for nodes and edges. In this paper, we apply the technique to an interactive web-based visual analytics environment - Apropos-where a user can explore a text corpus using Word-Bridge. We validate the technique using several case studies based on document collections such as intelligence reports, co-authorship networks, and works of fiction.