Hungarian WordNet and representation of verbal event structure

Abstract

During the development of the Hungarian WordNet we found that the primarily hierarchical structure of the Princeton WordNet (PWN) had to be expanded in order that it could be used for a more accurate representation of relations among verbal meanings in Hungarian, as well. Treating verbs as eventualities we made some aspectual information explicitly available in the representation of verbs. Using the notion of nucleus introduced by Moens&Steedman we added new relations to the WordNet and classified verbal synsets according to aspectual characteristics. This enabled the representation of some psycholinguistically relevant pieces of information and a wider possible usage of the thus extended Hungarian verbal WordNet in the field of computational linguistics.