Visualizing Knowledge Domain Citation and Semantic Structure

Researchers are faced with a wide range of tasks when interacting with the literature of a scientific field. These tasks range from determining the field’s seminal documents, for the individual beginning investigation in an area, to keeping abreast of current literature and emerging trends, for a scientist working in the field. Visualization can provide one mechanism for ordering documents and revealing structure and relations within a knowledge domain. The system described in this paper provides visual representations of document collections, using both citation information for individual documents and the semantic structure of the document collection, to form interactive visualizations that the user can explore. The system is currently in use with the Citeseer index and provides tools that display the citation structure of a user-defined domain. This bibliometric network is augmented by semantic information derived using a cosine term vector analysis of documents to provide a similarity metric among documents. Supplementary network information from this semantic analysis is used to augment the citation network and provide domain information that reflects documents’ content relations.