A New Search Approach for Interactive-Predictive Computer-Assisted Translation

Although significant improvements have been achieved in statistical machine translation (SMT), even the best machine translation technology is far from competing with human translators. An alternative approach to obtain high quality translation is to use a human translator who is assisted by an SMT. In interactive-predictive computer-assisted translation (IPCAT) paradigm, the human translator begins to type the translation of a given source text; by typing each character the MT system interactively offers the choices to complete the translation. Human translator may continue typing or accept the whole completion or part of it. In this paper, we propose a new search approach for increasing the performance of the IPCAT. This new search approach consists of a new search method and a hybrid back-off model. We achieve 2.3% and 1.16% absolute improvements by using the proposed search approach for two different corpora.