Visualization and Clustering of Document Collections using a Flock-based Swarm Intelligence Technique

Electronic availability of documents continues to increase, yet identifying documents relevant to the user remains a primary constraint in electronic document use. Visual representations of document collections can facilitate search by representing large collections of documents in a manner that is complementary to linear, text based representations. Visual representations can provide a means to make the overall structure of a collection comprehensible, as well as a mechanism to identify groups of useful documents and access relevant individual documents. The current work employs flock-based clustering to both organize documents and provide visual representations of documents. Reynolds’ three rule flocking scheme is augmented with additional rules to provide document clustering. A unified visual representation supplies facilities for overview of the entire document collection, filtering documents, and retrieving individual documents. The system also utilizes visualization tools for individual cluster identification and exploration based on keyword search. Stereoscopic viewing is provided to enhance users’ perception of 3D organization.