Toleu A.   Tolegen G.   Mussabayev R.  

Comparison of Various Approaches for Dependency Parsing

Reporter: Toleu A.

This paper presents results for dependency parsing using various discrete feature-based and deep learning-based approaches for two distinct languages: Kazakh and English.
We apply graph-based, transition-based approaches for the task to report the typed and untyped accuracy.
The comparisons were made with different settings: a cross-linguistic and mono-linguistic comparison, for those approaches with discrete features or dense features.
Experimental results show that discrete feature-based approaches(graph-based) seems to performs well than other when the size of data-set is relatively small.
For a little bit larger data set, the results of those approaches are very competitive with each other, no significant improvement can be observed.
In terms of training speed, the results show that discrete feature-based parsers take much less training time than the neural network-based parser, but with comparable performances.


To reports list