Learning to Classify Noun-Noun Semantic Relations

Add to your list(s) Download to your calendar using vCal

• Diarmuid O'Seaghdha, Computer Laboratory, University of Cambridge
• Friday 01 February 2008, 12:00-13:00
• SW01 Computer Laboratory.

If you have a question about this talk, please contact Johanna Geiss.

Identifying the semantic relations between entities mentioned in a sentence is an important NLP task. Variants of the task crop up in many guises. One of these is the problem of classifying the semantic relation in a noun-noun compound (e.g. “kitchen table” is a locational compound, whereas “plastic table” describes the composition of the table). This problem has received a lot of attention in recent years but remains difficult to solve, in part because standard relation classifcation methods fall down when the context of the entity mentions do not make the semantic relation explicit.

I’ll be talking about the kind of corpus-driven/machine-learning methods we can use for classifying semantic relations in compounds and whether these methods extend to more standard relation tasks such as the Semeval 2007 task on “Classification of Semantic Relations between Nominals”. In particular I’ll describe kernels on vectors, strings, trees and sets and how multiple kernels can be combined to integrate different representations of the data at hand.

This talk is part of the NLIP Seminar Series series.

This talk is included in these lists:

• All Talks (aka the CURE list)
• Cambridge Centre for Data-Driven Discovery (C2D3)
• Cambridge Forum of Science and Humanities
• Cambridge Language Sciences
• Cambridge talks
• Chris Davis' list
• Department of Computer Science and Technology talks and seminars
• Graduate-Seminars
• Guy Emerson's list
• Interested Talks
• Language Sciences for Graduate Students
• NLIP Seminar Series
• PMRFPS's
• SW01 Computer Laboratory
• Simon Baker's List
• Trust & Technology Initiative - interesting events
• bld31
• ndk22's list
• ob366-ai4er
• rp587

Note that ex-directory lists are not shown.