Assigning verbs to semantic classes via WordNet
Anna Korhonen · 2002
We propose a method for semi-automatic classification of verbs to Levin classes via the semantic network of WordNet. The method involves first classifying entire WordNet senses to semantic classes and then classifying individual verbs on the basis of their WordNet senses. We report evaluation which shows that the method can be used to build a verb classification accurate enough for practical NLP use. The WordNet-Levin mapping produced as a by-product, may, in turn, be used to supplement WordNet with novel information.