Relevance Ranking of Documents Using Relations and Text

This project will develop technology to provide researcher-friendly metasearch results of textual knowledge-bases. Results will be improved by creating precision searches through better relevance ranking and by clustering results. The approach will augment standard text retrieval methods with network-based inference, an area that has received little attention with respect to text retrieval of research articles. The technology will take advantage of the fact that research articles have relational information such as citations and co-authorships. This relational information will provide both better relevance scores and clustering, and also will customize the results to the user. Commercial Applications and other Benefits as described by the awardee: The technology should improve text retrieval in domains where documents are networked in meaningful ways.