Using map representations to visualize, explore and understand large collections of dynamically categorized documents

Abstract

This paper presents VOROSOM, a novel visualization scheme that supports collection understanding and exploration of large, distributed collections. Using metadata harvested from diverse collections, VOROSOM produces a map representation in which regions are associated with categories of documents. The shape of each region in the map reflects the relationships among documents in each of the categories. Thus, the distance between two regions directly corresponds to their semantic affinity. Maps are produced in such a way that the number of categories is maintained within a manageable size, considering the user's cognitive capabilities. Maps are organized hierarchically, which supports the exploration and navigation within categories and subcategories of documents using map representations consistently. We report initial results of user studies with a prototypical implementation of our visualization scheme over an actual network of digital libraries. © 2013 Springer International Publishing.