We describe how simple, commonly understood statistical models, such as statistical dependency parsers, probabilistic context-free grammars, and word-to-word translation models, c...
This paper shows how the best data-driven dependency parsers available today [1] can be improved by learning from unlabeled data. We focus on German and Swedish and show that label...