Home

About

Advanced Search

Browse by Discipline

Scientific Societies

E-print Alerts

Add E-prints

E-print Network
FAQHELPSITE MAPCONTACT US


  Advanced Search  

 
PROBABILISTIC MODELLING OF ISLAND-DRIVEN PARSING Alicia Ageno and Horacio Rodrguez
 

Summary: PROBABILISTIC MODELLING OF ISLAND-DRIVEN PARSING
Alicia Ageno and Horacio Rodríguez
TALP Research Center
Universitat Politècnica de Catalunya (UPC)
Jordi Girona, 1-3. E-08034 Barcelona, Spain
{ageno,horacio}@lsi.upc.es
Abstract
Two methods for stochastically modelling bidirectionality in chart parsing are presented. A probabilistic island-
driven parser which uses such models (either isolated or in combination) has been built and tested on wide-coverage
corpora. The best results are accomplished by the hybrid approaches that combine both methods.
1 Introduction
Although most methods for context-free grammar parsing are based on a uniform way of guiding the parsing
process (e.g. top-down, bottom-up, left-corner), there have recently been several attempts to introduce more
flexibility, allowing bidirectionality, in order to make parsers more sensitive to linguistic phenomena
([1],[2],[3]).
We can roughly classify such approaches into head-driven and island-driven parsing. They respectively
assume the existence of a distinguished symbol in each rule, the head, and certain distinguished words in the
sentence to be parsed, the islands, playing a central role on the respective parsing approach. While assigning
heads to rules is a heavy knowledge intensive task, selecting islands can be carried out straightforwardly:
unambiguous words, base NPs (in the case of textual input), accurately recognised fragments (in the case of

  

Source: Ageno, Alicia - Departament of Llenguatges i Sistemes Informátics, Universitat Politècnica de Catalunya

 

Collections: Computer Technologies and Information Sciences