An empirical study of segment prioritization for incrementally retrained post-editing-based SMT

18619

Knowledge Triples