Real-Time Full Text Analysis

Relevance ranking of documents retrieved from federated search sources is typically performed by analyzing the words in the document's summary information, which is provided by the underlying source. This type of analysis is highly dependent on the quality of the summary provided, which varies greatly by source, and often leads to a poor match that ranks high or to a good match that ranks poorly. This project will design and develop a system that performs full-text document analysis to provide improved ranking. The cost of retrieving and analyzing full-text documents will be assessed, and the benefits of performing much of the analysis near the source of the content will be characterized. Commercial Applications and other Benefits as described by the awardee: Researchers with sufficient resources (time, storage, compute and network power) and access to the full text of a large number of potentially relevant documents should achieve significantly improved relevance ranking. An improved ranking can translate to time saved skipping less relevant documents, and can mitigate the risk associated with missing key documents. In turn, the time savings will result in cost savings.