Dataset

We investigate the impact of word-level quality estimation on MT post-editing with 42 professional post-editors.

DivEMT is a publicly available post-editing study of Neural Machine Translation over a typologically diverse set of target languages.