Dependency Parsing Task Dataset

The Dependency Parsing Task Dataset contains the dataset used for the Main Dependency Parsing Task (http://www.evalita.it/2009/tasks/parsing) organised as part of the EVALITA 2009 evaluation campaign (http://www.evalita.it/2009). It comprehends data from TUT, the treebank for Italian developed by the Natural Language Processing group of the Department of Computer Science of the University of Turin.<p>The training set includes 2,400 sentences that correspond to 72,149 annotated tokens in TUT native format, and 66,055 tokens in CoNLL format. The corpus can be separated in three subcorpora, i.e. one from Italian newspapers (1,100 sentences and 30,561 tokens in CoNLL format), one from the Italian Civil Law Code (1,100 sentences and 28,048 tokens), and one from the Italian section of the JRC-Acquis Multilingual Parallel Corpus, a collection of declarations of the European Community (200 sentences and 7,446 tokens).</p><p>The test set of the MDS includes 240 sentences (5,287 tokens) and features a balancement alike to that of the training set: 100 sentences (1,782 tokens) from newspapers, 100 sentences (2,293 tokens) from Civil Law Code and 40 sentences (1,212 tokens) from the Passage/JRC-Acquis corpus.</p>