Corrected Co-training

This paper investigates bootstrapping for statis- tical parsers to reduce their reliance on manu- ally annotated training data. We consider both a mostly-unsupervised approach, co-training, in which two parsers are …...