BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Apposition Extraction and Named Entity Linking - Will Radford
DTSTART:20130723T110000Z
DTEND:20130723T120000Z
UID:TALK45484@talks.cam.ac.uk
CONTACT:Tamara Polajnar
DESCRIPTION:Named Entity Linking attempts to ground textual entity mention
 s to an external knowledge base (e.g.\, Wikipedia). Mentions are assigned 
 a KB entry or NIL if they are absent from the KB. The task requires resolv
 ing name polysemy and synonymy to disambiguate their references. I'll disc
 uss the NEL task\, datasets and evaluation as well as a comparison of some
  systems from the literature. \n\nMore recently\, we've investigated how t
 o better extract entity descriptions for disambiguation. While apposition 
 is used as a component in several tasks (e.g.\, Coreference Resolution\, T
 extual Entailment)\, apposition extraction performance is not often direct
 ly evaluated.  We propose systems exploiting syntactic and semantic  const
 raints to extract appositions from OntoNotes 4.  Our joint log-linear mode
 l outperforms the state-of-the-art model (Favre and Hakkani-Tür in Inters
 peech 2009)\, by around 10% on Broadcast News\, and achieves 54.3% F-score
  on multiple genres. I'll talk about our apposition system and some more g
 eneral work on sentence-local description and its application to NEL. \n\n
 Finally\, I'll demonstrate an application developed with our industry part
 ner. Fairfax Media operates some of the main metropolitan newspapers and n
 ews websites in Australia and has recently launched "zoom"\, showing how N
 EL can provide a compelling view of 25 years of stories.
LOCATION:FW11\, Computer Laboratory
END:VEVENT
END:VCALENDAR