Description:
This is the LMF version of the SenSem database created by the Spanish Inter-University Research Group GRIAL. As part of SenSem project, a corpus of sentences annotated at the semantic and syntactic levels was created. The source corpus is made up of around 13 million words extracted from the online versions of a Spanish newspaper. From this corpus, 25.000 sentences have been randomly selected, 100 for each of the 250 more frequent verbs in current Spanish. Each sentence has been labeled according to the verb sense it exemplifies, the type of complements it takes (arguments or adjunts), their syntactic category and function, and finally each argument has been labelled with a semantic role. The sentence has also been annotated as to its semantics both in relation with aspectual information and the type of construction being expressed. From this annotated corpus a lexical data base of verbs was created in which all the previous information will be recollected. The unit of description of the verbs is the sense. In the description of the verbs, argument structure is included, incorporating subcategorization patterns, with the information of frequency of them, semantic roles and information regarding sentence semantics. The lexicon and the corpus are associated at sense level and together shape up what we call the data bank of the sentential semantic of the Spanish verbs. Both resources are available via web and will form a very important source of linguistic information which we hope will be of utility in different areas of the natural language processing and linguistic research in general. The LMF conversion has been done by the Universitat Pompeu Fabra.