Constructing lexicon with morpho-syntactic features from untagged corpora
Résumé
This article presents a computational method of morpho-syntactic rules which automatically creates a lexicon with morphological features after disambiguation and PoS tagging in large non annotated corpora. The method is tested and implemented in two different languages: French and Greek which are very diverse to the complexity of their morphology. Although semantic features are missing from the lexicon (they could be supplied manually) we can see this method as a promising one for the creation of large scale lexicons used as data base morpho-syntactic features for automatic annotation of untagged corpora