Abstract
Previous work in corpus annotation focused on the automatic annotation of syntactic information (PoS-Tagging and Shallow Parsing). We go a step beyond by distinguishing between different types of proper names (person names, geographical names, company names) as well as between temporal and local phrases. This kind of semantic annotation can be used, for instance, to learn verb subcategorization requirements. In the talk we will present recall and precision values for the automatic annotation of a computer magazine corpus.