BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Acquiring syntactic and semantic transformations in question answe
 ring - Michael Kaisser (Microsoft)
DTSTART:20100701T160500Z
DTEND:20100701T162500Z
UID:TALK25236@talks.cam.ac.uk
CONTACT:Dr Fabien Petitcolas
DESCRIPTION:*Abstract*: One and the same fact in natural language can be e
 xpressed in many different ways by using different words and/or a differen
 t syntax. This phenomenon\, commonly called paraphrasing\, is the main rea
 son why Natural Language Processing (NLP) is such a challenging task. This
  becomes especially obvious in Question Answering (QA) where the task is t
 o automatically answer a question posed in natural language\, usually in a
  text collection also consisting of natural language texts. It cannot be a
 ssumed that an answer sentence to a question uses the same words as the qu
 estion and that these words are combined in the same way by using the same
  syntactic rules.\n\nIn my thesis I describe methods that can help to addr
 ess this problem. Firstly\nI explore how lexical resources\, i.e. FrameNet
 \, PropBank and VerbNet can be used\nto recognize a wide range of syntacti
 c realizations that an answer sentence to a given\nquestion can have. Furt
 hermore\, I use a corpus of question and answer sentence pairs (QASPs) to 
 develop an approach to QA based on matching dependency relations between a
 nswer candidates and question constituents in the answer sentences. In thi
 s talk\, I will describe these two approaches in more detail and present e
 valuation results.
LOCATION:Large public lecture room\, Microsoft Research\, Roger Needham Bu
 ilding\, 7 J J Thomson Avenue\, Cambridge CB3 0FB
END:VEVENT
END:VCALENDAR