An AGFL grammar for the Russian
language is under development at the department of Applied
Linguistics, Philological faculty of St.Petersburg State
University. The main goal of the project is to create the
Rus4IR system (Russian parser for Information Retrieval) - a
powerful natural language processing tool aimed to generate
parses from texts written in Russian. That means to create an
efficient AGFL grammar for the Russian language and to provide
it with an appropriate lexicon.
In our project we use the AGFL parser generation system
which was developed by the group of Kees Koster at the
Computer Science Department of the University of Nijmegen. As
a result, Rus4IR has the same options as other current
AGFL-based parsers (i.e. various output modes: tree format or
labeled brackets format; a possibility to use transduction
etc.).
This system is the first parser system for the Russian
language based on AGFL grammars, which have already proved (on
other European languages) to be a good solution for
representing a language in NLP technologies. Rus4IR, being an
information retrieval tool, can deal not only with
gramatically well-formed sentences, but also with "segments"
extracted left-to-right, therefore it can cope with the
ill-formed sentences modern Internet is full of. As a result,
the whole system should be both flexible and robust.
(c) First version of the AGFL parser for Russian,
Irina Azarova, SPbSU, 1995. (c) Rus4IR, Irina Azarova,
SPbSU, 2004.
Irina
Azarova, Assistant Professor of the Department of Applied
Linguistics, Philological Faculty, St.Petersburg State
University, Russia.
|