Improving an automatically extracted corpus for UMLS Metathesaurus word sense disambiguation

Manually annotated data is expensive, so manually covering a large terminological resource like the UMLS Metathesaurus is infeasible. In this paper, we evaluate two approaches used to improve the quality of an automatically extracted corpus to train statistical learners to performWSD. The first one contributes to more specific terms while the second filters out false positives. Using both approaches, we have obtained an improvement on the original automatic extracted corpus of approximately 6% in F-measure and 8% in recall.