We present a novel neural network model that learns POS tagging andgraph-based dependency parsing jointly. Our model uses bidirectional LSTMs tolearn feature representations shared for both POS tagging and dependencyparsing tasks, thus handling the feature-engineering problem. Our extensiveexperiments, on 19 languages from the Universal Dependencies project, show thatour model outperforms the state-of-the-art neural network-basedStack-propagation model for joint POS tagging and transition-based dependencyparsing, resulting in a new state of the art. Our code is open-source andavailable together with pre-trained models at:https://github.com/datquocnguyen/jPTDP
展开▼
