Project-specific Machine Translation
John J. Morgan · 2011
Abstract : This report details a project-specific method for translating a book-length text. The method used a human-in-the-loop strategy to iteratively and incrementally produce a progressively more accurate statistical machine translation system. The system generated rough-draft translations of chunks of text that were post-edited by an expert translator. The method was used to produce a high-quality translation of the Army Field Manual (FM) 7-8 Infantry Rifle Platoon and Squad from English into the Afghan language Pashto. The results show that through a process of over-fitting, this method leads to systems that are optimized for aiding a single translator.