Browse by topic
Type of publication
Statistical Confidence Measures for Probabilistic Parsing. Proceedings of the International Conference on Recent Advances in Natural Language Processing (RANLP'09), 2009. pp. 388-392.We introduce a formal framework that allows the calculation of new purely statistical confidence measures for parsing, which are estimated from posterior proba- bility of constituents. These measures allow us to mark each constituent of a parse tree as correct or incorrect. Experimental assessment using the Penn Treebank shows favorable results for the classical confidence evaluation metrics: the CER and the ROC curve. We also present preliminar experiments on application of confidence measures to improve parse trees by automatic constituent relabeling.