BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Bayesian Word Sense Induction - Andreas Vlachos (University of Cam
 bridge)
DTSTART:20091026T123000Z
DTEND:20091026T133000Z
UID:TALK21121@talks.cam.ac.uk
CONTACT:Diarmuid Ó Séaghdha
DESCRIPTION:At this session of the NLIP Reading Group we'll be discussing 
 the following paper:\n\nSamuel Brody. 2009. "Bayesian word sense induction
 ":http://www.aclweb.org/anthology/E/E09/E09-1013.pdf. In Proceedings of EA
 CL-09.\n\n*Abstract:*\nSense induction seeks to automatically identify wor
 d senses directly from a corpus. A key assumption underlying previous work
  is that the context surrounding an ambiguous word is indicative of its me
 aning. Sense induction is thus typically viewed as an unsupervised cluster
 ing problem where the aim is to partition a word's contexts into different
  classes\, each representing a word sense. Our work places sense induction
  in a Bayesian context by modeling the contexts of the ambiguous word as s
 amples from a multinomial distribution over senses which are in turn chara
 cterized as distributions over words. The Bayesian framework \nprovides a 
 principled way to incorporate a wide range of features beyond lexical co-o
 ccurrences and to systematically assess their utility on the sense inducti
 on task. The proposed approach yields improvements over state-of-the-art s
 ystems on a benchmark dataset.\n\nLike some work presented at recent *ACLs
 \, it builds on the Latent Dirichlet Allocation model (a.k.a. the standard
  "topic model"). For a more thorough introduction to the latter\, the foll
 owing paper is recommended:\n\nThomas L. Griffiths and Mark Steyvers. 2004
 . "Finding scientific topics":http://www.pnas.org/content/101/suppl.1/5228
 .full.pdf. Proceedings of the National Academy of Sciences 101: 5228-5235.
 \n
LOCATION:GS15\, Computer Laboratory
END:VEVENT
END:VCALENDAR