BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Learning to Classify Noun-Noun Semantic Relations - Diarmuid O'Sea
 ghdha\, Computer Laboratory\, University of Cambridge
DTSTART:20080201T120000Z
DTEND:20080201T130000Z
UID:TALK10368@talks.cam.ac.uk
CONTACT:Johanna Geiss
DESCRIPTION:Identifying the semantic relations between entities mentioned 
 in a\nsentence is an important NLP task. Variants of the task crop up in m
 any\nguises. One of these is the problem of classifying the semantic relat
 ion\nin a noun-noun compound (e.g. "kitchen table" is a locational compoun
 d\,\nwhereas "plastic table" describes the composition of the table). This
 \nproblem has received a lot of attention in recent years but remains\ndif
 ficult to solve\, in part because standard relation classifcation\nmethods
  fall down when the context of the entity mentions do not make the\nsemant
 ic relation explicit.\n\nI'll be talking about the kind of corpus-driven/m
 achine-learning methods\nwe can use for classifying semantic relations in 
 compounds and whether\nthese methods extend to more standard relation task
 s such as the Semeval\n2007 task on "Classification of Semantic Relations 
 between Nominals". In\nparticular I'll describe kernels on vectors\, strin
 gs\, trees and sets and\nhow multiple kernels can be combined to integrate
  different\nrepresentations of the data at hand.
LOCATION:SW01 Computer Laboratory
END:VEVENT
END:VCALENDAR