http://www.coli.uni-sb.de/~thorsten/tnt/
TnT is a very efficient statistical part-of-speech tagger that is trainable
on different languages and tagsets. The system incorporates several methods
of smoothing and of handling unknown words.
The tagger comes with two pre-compiled language models, one for German, and
one for English. It can immediately be used with these models. Additionally,
TnT is trainable with only small effort on a large variety of corpora and
tagsets.
TnT is available free of charge for non-commercial research purposes.
The current version runs under Solaris and Linux. For details, please visit
http://www.coli.uni-sb.de/~thorsten/tnt/
Thorsten Brants
Saarland University
Computational Linguistics
P.O.Box 151150
D-66041 Saarbruecken
Germany