A machine learning approach for identifying zero pronouns in Bulgarian
Diana Grigorova · 2014
This paper presents a machine learning approach for identifying clauses with zero pronouns in Bulgarian. A parser, based on dependency grammar is used as a pre-processing tool. The feature vector consists of 16 attributes and the training set counts 7009 instances. The experiments were carried out with Weka package. The algorithm JRip produces results with accuracy of 92%.