Final report of the 2005 language engineering workshop on statistical machine translation by parsing


Designers of SMT system have begun to experiment with tree-structured translation models. Unfortunately, SMT systems driven by such models are even more difficult to build than the already complicated WFST-based systems. The purpose of our workshop was to lower the barriers to entry into research involving such SMT systems. Our goals were inspired by the successful 1999 MT workshop, which had a similar purpose. Specifically, we wanted to follow that precedent to 1. build a publicly available toolkit for experimenting with tree-structured translation models; 2. build a tool for visualizing the predictions of such models; 3. demonstrate the feasibility of SMT with tree-structured models by running baseline experiments on large datasets; and 4. demonstrate that it is easy to retarget the toolkit to new language pairs.

Johns Hopkins University