BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Exploiting Large Corpora for Parsing - Dominick Ng\, University of
  Sydney
DTSTART:20140321T130000Z
DTEND:20140321T140000Z
UID:TALK51620@talks.cam.ac.uk
CONTACT:Tamara Polajnar
DESCRIPTION:Parsers have come a long way since the advent of research in N
 LP\, with much of the field settling on supervised techniques trained on l
 inguistically annotated corpora. However\, in recent years\, accuracy impr
 ovements have stalled at around 92% to 94%\, and parsing mistakes account 
 for a substantial proportion of errors made in downstream applications.\n\
 nThis talk will discuss some of my previous work in nbest CCG parsing and 
 reranking\, and then go on to potential methods of overcoming this parsing
  bottleneck using large external corpora. Much of the work is in-progress 
 and speculative as part of my PhD at the University of Sydney\, and I welc
 ome any feedback or suggestions.
LOCATION:FW26\, Computer Laboratory
END:VEVENT
END:VCALENDAR