Concept-based Information Retrieval and Fusion Engine

Existing search technology is inadequate for addressing the information retrieval requirements of today's complex tasks that require the integration of information from multiple and heterogeneous sources, the discovery of arbitrary and non-obvious relations between the documents, and the uncovering of information that is intentionally trying to hide. The goal of this project is to advance the state-of-the-art in text-based search technology by developing an experimental prototype that combines best-of-bread technologies to address these problems. Our concept-based information retrieval system combines background knowledge with knowledge discovered dynamically by globally analyzing the target collection(s) and locally analyzing the retrieved documents to identify the appropriate concept space(s) for each query and progressively follow all the clues that are present in them. Our information fusion system takes advantage of alternate representations of the underlying collection and information retrieval approaches to identify the documents that are the most relevant to the desired concepts. Our pattern discovery system finds relationships and interactions among entities present in the documents.