Exploiting semantic relations for literature-based discovery.

We propose using semantic predications to enhance literature-based discovery (LBD) systems, which currently depend exclusively on co-occurrence of words or concepts in target documents. In this paper, the predications, which are produced by the combined application of two natural language processing systems, BioMedLEE and SemRep, are coupled with an LBD system BITOLA. Initial experiments suggest this approach can uncover new associations that were not possible using previous methods.