SMT - CAT Integration in a Technical Domain. Handling XML mark-up using pre and post-editing processing methods
Arda Tezcan, Vincent Vandeghinste · Lirias · 2011
The increasing use of eXtensible Markup Language (XML) is bringing additional challenges to statistical machine translation (SMT) and computer assisted translation (CAT) workflow integration in the translation industry. This paper analyzes the need to handle XML markup as a part of the translation material in a technical domain. It explores different ways of handling such markup by applying transducers in pre and post-processing steps. A series of experiments indicates that XML markup needs a specific treatment in certain scenarios. One of the proposed methods not only satisfies the SMT-CAT integration need, but also provides slightly improved translation results on English-to-Spanish and English-to-French translations, compared to having no additional pre or post-processing steps.