ptnplanet / nltk-contributions Goto Github PK

NLTK http://nltk.org/ related code snippets and contributions.

The ClassifierBasedGermanTagger is a part-of-speech tagger extending the nltk's ClassifierBasedTagger with a modified feature_detector. It is designed for tagging German text and achieves an accuracy of 96.09% after being trained on 90% of the German TIGER corpus. More information on the creation of this classifier based tagger can be found at http://experimentallabor.de/?p=162. A full evaluation of the tagger can be found at http://experimentallabor.de/?p=207.

The NegraCorpusReader is a corpus reader for Negra format corpora like the TIGER corpus. The class extends the ConllCorpusReader from the NLTK.

My experiments with the NLTK can be found here: http://www.experimentallabor.de

Unfortunately, the blog articles and the experiment's setup are lost until I can recover the server's full backup.