Methods and tools of computational linguistics for the classification of natural non-referential ellipsis in Spanish

This article represents a brief survey of the few works, dedicated to the modern approaches of natural language processing (NLP) to the analysis of impersonal sentences in Spanish. Such an analysis consists in classification of non-referential ellipsis that can be used in machine translation systems. The NLP approaches related with Spanish are mainly based on the work of Rello published in 2010. These approaches do not make use of a proper classification of impersonal models, but of a relative descriptive distribution without strict criteria. The structured classification presented in this article, based on historical and semantic data of interlingual nature, can be also applied for creation of linguistically-motivated classes for machine learning methods. The automatic classification method, employed in the work of Rello, is based on the use of the well-known WEKA package instance-based learner.