... Association for Computational Linguistics, pages 930–939,Jeju, Republic of Korea, 8-14 July 2012.c2012 Association for Computational LinguisticsPORT: a Precision-Order-Recall MT Evaluation Metric for ... not optimize its performance as an evaluation metric, but rather, to optimize system tuning performance. There are some other possible reasons why PORT did not outperform METEOR v1.0 at system ... with BLEU, PORT, and Qmean. PORT outperforms Qmean on seven of the eight automatic scores shown for small and large Chinese-English. 936However, for the European language pairs, PORT and...