| | |
Summary: FGT - A Framework for Generating
Rule-based Taggers
Zoltan Alexin 1 , Tamas Varadi 2 , Csaba Oravecz 2 ,
Gabor Proszeky 3 , Janos Csirik 4 , and Tibor Gyimothy 4
1 Dept. of Applied Informatics, Jozsef Attila University,
Arpad ter 2, H-6720 Szeged Hungary, alexin@inf.u-szeged.hu
2 Research Institute for Linguistics, Hungarian Academy of Sciences,
varadi, oravecz@inf.u-szeged.hu
3 Morphologic Ltd. Hungary, proszeky@morphologic.hu
4 Research Group on Arti cial Intelligence, Hungarian Academy of Sciences,
Aradi vertanuk tere 1, H-6720 Szeged, csirik, gyimi@inf.u-szeged.hu
Abstract. In this paper we give an overview of a tool called FGT
(Framework for Tagger generation) which is able to assist in the de-
velopment of rule-based POS taggers. In the tagging generation process
we need programs to determine ambiguity tag classes, produce lexicons,
generate training and test data for di erent learners, and transform the
learnt disambiguation rules into working taggers. C and Prolog programs
have been written for such tasks and we are going to integrate them into
the FGT. The modules of the FGT system have already been tested for
|
