Unsupervised Detection of Downward-Entailing Operators By Maximizing Classification Certainty
Jackie Chi Kit Cheung, Gerald M. Penn · 2012
We propose an unsupervised, iterative method for detecting downward-entailing operators (DEOs), which are important for deducing entailment relations between sentences. Like the distillation algorithm of Danescu-Niculescu-Mizil et al. (2009), the initialization of our method depends on the correlation between DEOs and negative polarity items (NPIs). However, our method trusts the initialization more and aggressively separates likely DEOs from spurious distractors and other words, unlike distillation, which we show to be equivalent to one iteration of EM prior re-estimation. Our method is also amenable to a bootstrapping method that co-learns DEOs and NPIs, and achieves the best results in identifying DEOs in two corpora. 1