annotation inconsistencies between data sets can cause problems for
low-resource NLP, where noisy or inconsistent data cannot be as easily replaced
compared with resource-rich languages. In this paper, we propose a method for
automatically detecting annotation mismatches between depend