Dependency Parsing Dataset

The Dependency Parsing Dataset contains texts taken from the ISDT (Italian Stanford Dependency Treebank) for a total of 178553 tokens, split into a training set (165,975 tokens) and a validation set (12,578 tokens). The resource is annotated according to the Stanford dependencies scheme and complies with the standard CoNLL-X format. The data set has been used in the context of the Dependency Parsing Task (http://medialab.di.unipi.it/Evalita2014/), organised as part of the EVALITA 2014 evaluation campaign (​​http://www.evalita.it/2014).