BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Advanced Graph-based and Transition-based Dependency Parsing Appro
 aches and Recent Trends towards Joint Syntactic and Morphologic Disambigua
 tion  - Bernd Bohnet\, University of Birmingham
DTSTART:20140523T110000Z
DTEND:20140523T120000Z
UID:TALK51328@talks.cam.ac.uk
CONTACT:Tamara Polajnar
DESCRIPTION:In this talk\, we compare higher order graph-based and transit
 ion-based parsing approaches in terms of their properties and give some ba
 ckground on the development in the last decade. The differences in those t
 wo approaches have lead to number of combination approaches to gain the be
 st of both worlds. We briefly compare these advanced parsers with constitu
 ent parsers that perform often competitive in terms of accuracy despite mo
 re than a decade of intensive research into dependency parsing.\n \nRecent
 ly\, dependency-parsing systems aim to join morphological and syntactic an
 alysis. Starting from a transition-based model for joint part-of-speech ta
 gging and dependency parsing\, we explore different ways of integrating mo
 rphological features into the model. We also investigate the use of rule-b
 ased morphological analysers and the use of word clusters to tackle the sp
 arsity of lexical features. We will present an evaluation on five morpholo
 gically rich languages that shows consistent improvements in both morpholo
 gical and syntactic accuracy for joint prediction over a pipeline model. T
 he final results improve the state of the art in dependency parsing for th
 ese languages.
LOCATION:FW26\, Computer Laboratory
END:VEVENT
END:VCALENDAR