On the Frailty of Universal POS Tags for Neural UD Parsers
This work addresses a specific problem for NLP researchers and practitioners in dependency parsing, showing incremental insights into feature limitations.
The study analyzed how Universal POS (UPOS) tag accuracy affects neural dependency parsing performance, finding that using UPOS tags as features requires very high tagging accuracy and that gold tags provide a non-linear performance boost, indicating exceptionality.
We present an analysis on the effect UPOS accuracy has on parsing performance. Results suggest that leveraging UPOS tags as features for neural parsers requires a prohibitively high tagging accuracy and that the use of gold tags offers a non-linear increase in performance, suggesting some sort of exceptionality. We also investigate what aspects of predicted UPOS tags impact parsing accuracy the most, highlighting some potentially meaningful linguistic facets of the problem.