Multi-Faceted Clustering of Meta-Search Results

It is not unusual to retrieve thousands or millions of results for a search query. Even though search engines try to rank search results by importance, the order of the results often does not reflect the interest of the person doing the search, and it is still necessary to browse through many pages of sparsely relevant information. When the results of several search engines are combined, as in meta-searching, the problem of ranking and organization becomes very crucial. This project will design and develop an advanced search-result clustering technology for organizing meta-search retrieval sets from heterogeneous sources. Clustering is a natural way of ordering the search results by textual, linguistic, or conceptual similarity because documents with common features are brought together. Clustering enables the selection, reordering, or exclusion of a subset of the retrievals with a single click of the mouse. Three complementary methodologies will be investigated: (1) the identification of meta-data and/or information patterns that can be used for multi-dimensional and multi-faceted clustering; (2) the display and navigation of multiple information dimensions and multiple information facets simultaneously, in order to provide a better grasp of the relationships between textual, structured, and multi-media data and information; and (3) dynamic augmentation of cluster sets with in-context semantic information from diverse sources. Commercial Applications and other Benefits as described by the awardee: The technology should be useful for science, education, commerce, and intelligence.