Simple semi-supervised training of part-of-speech taggers

TY - GEN

T1 - Simple semi-supervised training of part-of-speech taggers

AU - Søgaard, Anders Østerskov

PY - 2010

Y1 - 2010

N2 - Most attempts to train part-of-speech taggers on a mixture of labeled and unlabeled data have failed. In this work stacked learning is used to reduce tagging to a classification task. This simplifies semisupervised training considerably. Our prefered semi-supervised method combines tri-training (Li and Zhou, 2005) and disagreement-based co-training. On the Wall Street Journal, we obtain an error reduction of 4.2% with SVMTool (Gimenez and Marquez, 2004).

AB - Most attempts to train part-of-speech taggers on a mixture of labeled and unlabeled data have failed. In this work stacked learning is used to reduce tagging to a classification task. This simplifies semisupervised training considerably. Our prefered semi-supervised method combines tri-training (Li and Zhou, 2005) and disagreement-based co-training. On the Wall Street Journal, we obtain an error reduction of 4.2% with SVMTool (Gimenez and Marquez, 2004).

M3 - Article in proceedings

BT - Proceedings of the 48th Annual Meeting of the Association for Computational Linguistics

PB - Association for Computational Linguistics

ER -