A machine learning approach for correcting the errors of a Treebank

The Treebank is one of the most useful resources for supervised or semi-supervised learning in many NLP tasks such as speech recognition, spoken language systems, parsing and machine translation. Treebank can be developded in different ways that could be, generally, categorized in manually and statistical approaches. While the resulted Treebank in each of these methods has the annotation error, one which accomplished by statistical method has much more errors than the other. Error in Treenabanks causes that they are not useful anymore. In this paper an statistical method is proposed which aims to correct the errors in a specific English LTAG-Treebank. The proposed method was applied to a automatically generated Treebank and an improvement from 68% to 79% respect to F-measure is retrieved.