Login
Computer Science @ Rochester
Friday, March 23, 2007
12:30 PM
CSB 703
Ding Liu
University of Rochester
Maximum Correlation Training for Machine Translation Evaluation
We propose three new features for MT evaluation: source-sentence constrained n-gram precision, source-sentence re-ordering metrics, and discriminative unigram precision, as well as a method of learning linear feature weights to directly maximize correlation with human judgments. Our source-sentence constrained n-gram precision achieves, among all the testing metrics including BLEU, NIST, ROUGE, METEOR, SIA, the best adequacy and overall evaluation results and the second best result in fluency evaluation. We further improve the evaluation performance by combining the individual metrics using maximum correlation training, which is shown to be better than the classification-based framework.