Amplexor MTExpert - machine translation adapted to the translation workflow
暂无分享,去创建一个
MTExpert is AMPLEXOR’s proprietary machine translation (MT) system based on state-of-the-art statistical and linguistic algorithms, easily integrated with existing linguistic assets, delivering quality results tailored to different communication objectives. Description AMPLEXOR MTExpert is a fully automated MT service based on the Moses open-source platform with language-specific linguistic optimizations (Ceausu and Hunsicker, 2014), as well as terminology integration and format handling. The available MT engines include specialised MT domains, as well as customisable MT engines adapted to a particular content type. MTExpert provides translation for specialised and generic MT language domains for most European languages, Chinese, Japanese and Arabic. It includes domains ranging from the EU official publication domain to the technical domain or life sciences. Existing translation engines can be customised to a particular content type, based on existing language assets like translation memories or terminology databases. MTExpert confidence score can automatically estimate the reliability of the machine translated content, for a better quality and cost-benefit assessment. The confidence score can be fine-tuned for each translation workflow using the feedback from translators (Hunsicker and Ceausu, 2015). System integration into standard translation / post-editing processes is achieved using flexible interfaces such as CAT system plugins that allow a flexible threshold definition for the application of MT, e.g. for words in segments below a defined match quality. The output of MT is used as additional linguistic resource for translators and post-editors who can work in their usual translation environment.
[1] Sabine Hunsicker,et al. Pre-ordering of phrase-based machine translation input in translation workflow , 2014, LREC.
[2] Sabine Hunsicker,et al. Machine translation quality estimation adapted to the translation workflow , 2014, TC.