A Dynamic Oracle for Linear-Time 2-Planar Dependency Parsing
This improves dependency parsing efficiency and accuracy for natural language processing applications, though it is incremental over existing transition-based methods.
The paper tackles the problem of training a linear-time 2-Planar dependency parser by proposing an efficient dynamic oracle, which outperforms static training in most languages and beats an enhanced arc-hybrid parser on most datasets.
We propose an efficient dynamic oracle for training the 2-Planar transition-based parser, a linear-time parser with over 99% coverage on non-projective syntactic corpora. This novel approach outperforms the static training strategy in the vast majority of languages tested and scored better on most datasets than the arc-hybrid parser enhanced with the SWAP transition, which can handle unrestricted non-projectivity.