BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:IC Monday Seminar: Inducing Meaning Representations from Text
DTSTART:20120507T161500
DTEND:20120507T173000
DTSTAMP:20260407T175717Z
UID:389c6fd819b4075cb637db8705cf5b5d3f69381cf1d13dbd05c4762b
CATEGORIES:Conferences - Seminars
DESCRIPTION:Dr. Ivan Titov\, Saarland University - IC Faculty candidate\nA
 bstract\nLanguage understanding by machines is one of the principal object
 ives of artificial intelligence research. Though full understanding of unr
 estricted texts is still a remote goal\, in recent years\, statistical app
 roaches have been developed to predict more shallow forms of semantics\, s
 uch as underlying predicate-argument structure of sentences. Most existing
  statistical techniques for tackling these problems rely on large human-an
 notated datasets\, which are expensive to create and exist only for a very
  limited number of languages. Even then\, they are not very robust\, cover
  only a small proportion of semantic constructions appearing in the labele
 d data\, and are domain-dependent. We investigate Bayesian models which do
  not use any labeled data but induce semantic representations from unannot
 ated texts. Unlike semantically-annotated data\, unannotated texts are ple
 ntiful and available for many languages and many domains which makes our a
 pproach particularly promising. We show that these models induce linguisti
 cally-plausible semantic representations\, significantly outperform curren
 t state-of-the-art approaches\, and yield competitive results on question 
 answering in the biomedical domain. We also look into several extensions o
 f the model\, and specifically consider multilingual induction of semantic
 s\, where we show that multilingual parallel texts provide a valuable sour
 ce of indirect supervision for induction of shallow semantic representatio
 ns. \n\nBiography\nIvan Titov joined the Saarland University as a junior 
 faculty and head of a research group in November 2009\, following a postdo
 c at the University of Illinois at Urbana-Champaign. He received his Ph.D.
  in Computer Science from the University of Geneva in 2008 and his master'
 s degree in Applied Mathematics and Informatics from the St. Petersburg St
 ate Polytechnic University (Russia) in 2003. His current research interest
 s are in statistical natural language processing (models of syntax\, seman
 tics and sentiment) and machine learning (structured prediction methods\, 
 latent variable models\, Bayesian methods).
LOCATION:INM 202 http://plan.epfl.ch/?lang=fr&room=INM202
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR