BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Unsupervised Entailment Detection between Dependency Graph Fragmen
 ts - Marek Rei\, University of Cambridge
DTSTART:20110605T113000Z
DTEND:20110605T120000Z
UID:TALK31696@talks.cam.ac.uk
CONTACT:Thomas Lippincott
DESCRIPTION:Entailment detection systems are generally designed to work ei
 ther on single words\, relations or full sentences. We propose a new task 
 – detecting entailment between dependency graph fragments of any type 
 – which relaxes these restrictions and leads to much wider entailment di
 scovery. An unsupervised framework is described that uses intrinsic simila
 rity\, multi-level extrinsic similarity and the detection of negation and 
 hedged language to assign a confidence score to entailment relations betwe
 en two fragments. The final system achieves 84.1% average precision on a d
 ata set of entailment examples from the biomedical domain. 
LOCATION:FW26\, Computer Laboratory
END:VEVENT
END:VCALENDAR