file.pdf (525.57 kB)

A Framework for Interactive and Automatic Refinement of Transfer-based Machine Translation

Download (525.57 kB)
journal contribution
posted on 01.01.2005 by Ariadna Font Llitjós, Jaime G. Carbonell, Alon Lavie

Most current Machine Translation (MT) systems do not improve with feedback from post-editors beyond the addition of corrected translations to parallel training data (for statistical and example-base MT) or to a memory database. Rule based systems to date improve only via manual debugging. In contrast, we propose a largely automated method for capturing more information from human post-editors, so that corrections may be performed automatically to translation grammar rules and lexical entries. This paper introduces a general framework for incorporating a refinement module into rule-based transfer MT systems. This framework allows for generalizing post-editing efforts in an effective way, by identifying and correcting rules semi-automatically on order to improve coverage and overall translation quality.