Christophe Servan

Optimising Multiple Metrics with MERT

Optimising Multiple Metrics with MERT Optimisation in statistical machine translation is usually made toward the BLEU score, but this metric is questioned about its relevance to an human evaluation. Many other metrics exist but none of them are in perfect harmony with human evaluation. On the other hand, most evaluation campaigns use multiple metrics (BLEU, TER, METEOR, etc.). Statistical machine translation systems can be optimised for other metrics than BLEU, but usually the optimisation with other metrics tends to decrease the BLEU score, the main metric used in MT evaluation campaigns. In this paper we extend the minimum error training tool of the popular Moses SMT toolkit with a scorer for the TER score, and any linear combination of the existing metrics. The TER scorer was reimplemented in C++ which results in a ten times faster execution than the reference java code. We have performed experiments with two large-scale phrase-base SMT systems to show the benefit of the new options of the minimum error training in Moses. The first one translates from French into English (WMT 2011 evaluation). The second one was developed in the frame work of the DARPA Gale project to translate from Arabic to English in three different genres (news, web and transcribed broadcast news and conversations).

Translation project adaptation for MT-enhanced computer assisted translation

The effective integration of MT technology into computer-assisted translation tools is a challenging topic both for academic research and the translation industry. In particular, professional translators consider the ability of MT systems to adapt to the feedback provided by them to be crucial. In this paper, we propose an adaptation scheme to tune a statistical MT system to a translation project using small amounts of post-edited texts, like those generated by a single user in even just one day of work. The same scheme can be applied on a larger scale in order to focus general purpose models towards the specific domain of interest. We assess our method on two domains, namely information technology and legal, and four translation directions, from English to French, Italian, Spanish and German. The main outcome is that our adaptation strategy can be very effective provided that the seed data used for adaptation is ‘close enough’ to the remaining text to be translated; otherwise, MT quality neither improves nor worsens, thus showing the robustness of our method.