Dependency Parsing as Head Selection
This approach simplifies dependency parsing by removing structural constraints during training, potentially benefiting NLP researchers and practitioners with efficient and accurate parsing tools.
The paper tackles dependency parsing by reformulating it as independent head selection for each word, using a bidirectional RNN model called DeNSe, and achieves state-of-the-art performance on four languages with varying non-projectivity.
Conventional graph-based dependency parsers guarantee a tree structure both during training and inference. Instead, we formalize dependency parsing as the problem of independently selecting the head of each word in a sentence. Our model which we call \textsc{DeNSe} (as shorthand for {\bf De}pendency {\bf N}eural {\bf Se}lection) produces a distribution over possible heads for each word using features obtained from a bidirectional recurrent neural network. Without enforcing structural constraints during training, \textsc{DeNSe} generates (at inference time) trees for the overwhelming majority of sentences, while non-tree outputs can be adjusted with a maximum spanning tree algorithm. We evaluate \textsc{DeNSe} on four languages (English, Chinese, Czech, and German) with varying degrees of non-projectivity. Despite the simplicity of the approach, our parsers are on par with the state of the art.