Management of Large Annotation Projects Involving Multiple Human Judges: a Case Study of GALE Machine Translation Post-editing