The Best Lexical Metric for Phrase-Based Statistical MT System Optimization

It is shown that people tend to prefer BLEU and NIST trained models to those trained on edit distance based metrics like TER or WER, and that using BLEu or NIST produces models that are more robust to evaluation by other metrics and perform well in human judgments.

Abstract

Translation systems are generally trained to optimize BLEU, but many alternative metrics are available. We explore how optimizing toward various automatic evaluation metrics (BLEU, METEOR, NIST, TER) affects the re-sulting model. We train a state-of-the-art MT system using MERT on many parameteriza-tions of each metric and evaluate the result-ing models on the other metrics and also us-ing human judges. In accordance with popular wisdom, we find that it’s important to train on the same metric used in testing. However, we also find that training to a newer metric is only useful to the extent that the MT model’s struc-ture and features allow it to take advantage of the metric. Contrasting with TER’s good cor-relation with human judgments, we show that people tend to prefer BLEU and NIST trained models to those trained on edit distance based metrics like TER or WER. Human prefer-ences for METEOR trained models varies de-pending on the source language. Since using BLEU or NIST produces models that are more robust to evaluation by other metrics and per-form well in human judgments, we conclude they are still the best choice for training. 1