BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Hierarchical Passage Retrieval - Phil Cowans\, Inference Group
DTSTART:20060419T140000Z
DTEND:20060419T150000Z
UID:TALK4851@talks.cam.ac.uk
CONTACT:Phil Cowans
DESCRIPTION:Much recent research in information retrieval has concentrated
  on approaches which make use of language models to score documents. Typic
 ally\, language models are constructed from the collection and relevance i
 s defined in terms of the probability of the query string under these mode
 ls. In particular\, use of hierarchical Dirichlet models of the whole coll
 ection can be shown to naturally provide many desirable features\, includi
 ng term weighting similar to the well known tf.idf scheme. \n\nIn this tal
 k I will describe the extension of this model to include the notion of a f
 urther subdivision of documents into passages. This model can be used to s
 core documents while at the same time provides an indication of parts of t
 he documents which are particularly relevant. A simple application of this
  model is the construction of summary information which can be displayed a
 longside the search results\, although there are many other cases in which
  this information is of use. I shall present a comparison of this model wi
 th alternate approaches using standard test sets.
LOCATION:Ryle Seminar Room\, Cavendish Laboratory
END:VEVENT
END:VCALENDAR