This paper deals with spelling normalization of historical texts with regard to further processing with modern part-of-speech taggers. Different methods for this task are presented and evaluated on a set of historical German texts from the 15th-18th century, and specific problems inherent to the processing of historical data are discussed. A chain combination using word-based and character-based techniques is shown to be best for normalization, while POS tagging of normalized data is shown to benefit from ignoring punctuation marks. Using these techniques, when 500 manually normalized tokens are used as training data for the normalization, the tagging accuracy of a manuscript from the 15th century can be raised from 28.65% to 76.27%.
| ||||||||||||||||||||||||||||
