posted on 2011-06-01, 00:00authored byVamshi Ambati, Sanjika Hewavitharana, Stephan Vogel, Jaime G. Carbonell
Supervised learning algorithms for identifying comparable sentence pairs from a dominantly non-parallel corpora require resources for computing feature functions as well as training the classifier. In this paper we propose active learning techniques for addressing the problem of building comparable data for low-resource languages. In particular we propose strategies to elicit two kinds of annotations from comparable sentence pairs: class label assignment and parallel segment extraction. We also propose an active learning strategy for these two annotations that performs significantly better than when sampling for either of the annotations independently
History
Publisher Statement
Copyright 2011 The Association for Computational Linguistics