Black-box integration of heterogeneous bilingual resources into an interactive translation system

Abstract

The objective of interactive translation prediction (ITP) is to assist human translators in the translation of texts by making context-based computer-generated suggestions as they type. Most of the ITP systems in literature are strongly coupled with a statistical machine translation system that is conveniently adapted to provide the suggestions. In this paper, however, we propose a resource-agnostic approach in which the suggestions are obtained from any bilingual resource (a machine translation system, a translation memory, a bilingual dictionary, etc.) that provides targetlanguage equivalents for source-language segments. These bilingual resources are considered to be black boxes and do not need to be adapted to the peculiarities of the ITP system. Our evaluation shows that savings of up to 85% can be theoretically achieved in the number of keystrokes when using our novel approach. Preliminary user trials indicate that these benefits can be partly transferred to real-world computer-assisted translation interfaces.