Part-of-speech tagging for Swedish
Klas Prütz · 2002
This paper describes the work with a part-of-speech tagger for Swedish. The tagger used in the work was originally designed by Brill (1992) and may be adapted to different languages using annotated training corpora. The training corpus in this case is very small and may be the reason why the tagger is not very accurate in its original form. Extending the lexicon using different methods has enhanced the performance of the tagger. This work also indicates that a more fine-grained tagset may contribute to the performance when a much smaller tagset is needed.