Syntax-based Semi-Supervised Named Entity Tagging

We report an empirical study on the role of syntactic features in building a semi-supervised named entity (NE) tagger. Our study addresses two questions: What types of syntactic features are suitable for extracting potential NEs to train a classifier in a semi-supervised setting? How good is the resulting NE classifier on testing instances dissimilar from its training data? Our study shows that constituency and dependency parsing constraints are both suitable features to extract NEs and train the classifier. Moreover, the classifier showed significant accuracy improvement when constituency features are combined with new dependency feature. Furthermore, the degradation in accuracy on unfamiliar test cases is low, suggesting that the trained classifier generalizes well.