Computational approaches that mine large amounts of free text to extract relational information and support applications such as information retrieval and literature based discovery, have gained popularity recently. Most of these tend to focus on specific areas in biomedicine, for example gene-gene interactions or disease-treatment relations. The only effort that extracts a broad set of relations adhering to a large standardized vocabulary is the rule-based SemRep program being developed by researchers at the National Library of Medicine (NLM).
Dr. Kavuluru's project, selected for funding through NLM's R21 grant mechanism, aims to improve the recall of SemRep without major sacrifices in precision through an alternative statistical learning approach that adapts the open information extraction paradigm to biomedicine. This effort will measure how well his methods complement those used by SemRep in terms of performance improvements in both application agnostic and task specific evaluations.
This will be the first R-series grant from the NLM awarded to the University of Kentucky.