AMBER : A modified BLEU, enhanced ranking metric

This paper proposes a new automatic machine translation evaluation metric: AMBER, which is based on the metric BLEU but incorporates recall, extra penalties, and some text processing variants. There is very little linguistic information in AMBER. We evaluate its system-level correlation and sentence-level consistency scores with human rankings from the WMT shared evaluation task; AMBER achieves state-of-the-art performance.Peer reviewed: NoNRC publication: Ye