Automatic classification of multi-word expressions in print dictionaries

Summary This work demonstrates the assignment of multi-word expressions in print dictionaries to POS classes with minimal linguistic resources. In this application, 32,000 entries from the Wörterbuch der deutschen Idiomatik (H. Schemann 1993) were classified using an inductive description of POS sequences in conjunction with a Brill Tagger trained on manually tagged idiomatic entries. This process assigned categories to 86% of entries with 88% accuracy. This classification supplies a meaningful preprocessing step for further applications: the resulting POS-sequences for all idiomatic entries might be used for the automatic recognition of multi-word lexemes in unrestricted text.