The PASCAL Challenge on Grammar Induction
Douwe Gelling, Trevor Cohn, Phil Blunsom, Joäo Graça · 2012
This paper presents the results of the PASCAL Challenge on Grammar Induction, a competition in which competitors sought to predict part-of-speech and dependency syntax from text. Although many previous competitions have featured dependency grammars or partsof-speech, these were invariably framed as supervised learning and/or domain adaption. This is the first challenge to evaluate unsupervised induction systems, a sub-field of syntax which is rapidly becoming very popular. Our challenge made use of a 10 different treebanks annotated in a range of different linguistic formalisms and covering 9 languages. We provide an overview of the approaches taken by the participants, and evaluate their results on each dataset using a range of different evaluation metrics. 1