Annotating the argument structure of deverbal nominalizations in Spanish

摘要: Over recent years, there has been a growing interest in the computational treatment of nominalized Noun Phrases due to rich semantic information they contain. These can be understood as verbal paraphrases and, just like them, also denote argument and thematic-role relations. This paper presents methodology followed annotate structure deverbal nominalizations Spanish AnCora-Es corpus. We focus on automated annotation process that is mostly based specified lexicon but syntactic annotated The heuristic rules make use this rely linguistic assumptions are evaluated we evaluate reliability process. was manually checked order ensure accuracy final resource. demonstrate its feasibility (77% F-measure) show it facilitates corpus annotation, which always time-consuming costly result enrichment with thematic roles nominalizations. It first kind freely available.