BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Disambiguation of Biomedical Text - Mark Stevenson - Sheffield Uni
 versity
DTSTART:20090306T120000Z
DTEND:20090306T130000Z
UID:TALK16445@talks.cam.ac.uk
CONTACT:Johanna Geiss
DESCRIPTION:Like text in other domains\, biomedical documents contain a ra
 nge of\nterms with more than one possible meaning. These ambiguities form 
 a\nsignificant obstacle to the automatic processing of these\ntexts. Previ
 ous approaches to resolving this problem have made use of\na variety of kn
 owledge sources including the context in which the\nambiguous term is used
  and domain-specific resources (such as\nUMLS). We compare a range of know
 ledge sources which have been\npreviously used and introduce a novel one: 
 MeSH terms. The best\nperformance is obtained using linguistic features in
  combination with\nMeSH terms. Performance exceeds previously reported res
 ults on a\nstandard test set.\n\nOur approach is supervised and therefore 
 relies on annotated training\nexamples. A novel approach to automatically 
 acquiring additional\ntraining data\, based on the relevance feedback tech
 nique from\nInformation Retrieval\, is presented. Applying this method to 
 generate\nadditional training examples is shown to lead to a further incre
 ase in performance.
LOCATION:SW01\, Computer Laboratory
END:VEVENT
END:VCALENDAR