Readability Metrics for Machine Translation in Dutch: Google vs. Azure & IBM

This paper introduces a novel method to predict when Google translation is better than other machine translations (MT) in Dutch. Instead of considering fidelity, this approach considers fluency and readability indicators for ranked best. research explores an alternative the field quality estimation. The contributes by publishing dataset with sentences from English Dutch, human-made classifications on best-worst scale. Logistic regression shows correlation between T-Scan output, such as measurements like lemma frequencies, was Azure IBM. last part results section prediction possibilities. First logistic second generated automated learning model. Respectively, they have accuracy 0.59 0.61.