CCG Supertagging Using Morphological and Dependency Syntax Information

After presenting a new CCG supertagging algorithm based on morphological and dependency syntax information, we use this to create French Tree Bank corpus (20,261 sentences) the FTB by Abeillé et al. We then corpus, as well Groningen for English language, train BiLSTM+CRF neural architecture that uses (a) morphosyntactic input features (b) feature correlations features. show experimentally an inflected language like French, information allows significant improvement of accuracy task, when using deep learning techniques.