Paper: Estimating Grammar Correctness for a Priori Estimation of Machine Translation Post-Editing Effort

ACL ID W14-0303
Title Estimating Grammar Correctness for a Priori Estimation of Machine Translation Post-Editing Effort
Venue EACL Workshop on Humans and Computer-assisted Translation
Session
Year 2014
Authors

We present a supervised learning pilot ap- plication for estimating Machine Transla- tion (MT) output reusability, in view of supporting a human post-editor of MT content. We train our model on typed dependencies (labeled grammar relation- ships) extracted from human reference and raw MT data, to then predict gram- mar relationship correctness values that we aggregate to provide a binary segment- level evaluation. In view of scaling up to larger data, we provide implemented Na??ve Bayes and Stochastic Gradient De- scent with Support Vector Machine loss function approaches and their evaluation, and verify the correlation of predicted val- ues with human judgement.