Colored-sketch of Text Information

This paper presents an information visualization method, which transforms text into abstracted visual representations. The proposed color-coding algorithm converts text into a sequence of colored icons that inform users about the distributional patterns of given queries, as well as the structural overview of a document simultaneously. By presenting the compact, but instructive visual abstraction of texts concurrently, users can compare multiple documents intuitively while alleviating the need to reference the underlying text. The system provides interactive navigation tools to support users' decision-making processes - including multi-level viewing, a tree hierarchy recording previous search activities, and suggestive words for refinement of the search scope. An experimental study evaluating this visual approach for delivering search results has been conducted on text corpora in comparison with a traditional information retrieval system. By informing search results to clientele in a perceptive form, the users' performance in obtaining desired information has been improved, while maintaining the accuracy.