New study into Automated Language Translation highlights need for shared, clean and normalized data. In cooperation with the Translation Automation User Society (TAUS), Asia Online conducted an experiment to determine the optimum approaches for building statistical machine translation engines. The findings indicate that significant improvements in translation machine quality can be achieved with smaller pools of shared, clean data.

Asia Online performed extensive analysis and created a total of 29 separate SMT engines by combining the data from the three companies in various configurations, as well as performing comparisons of output quality (using the BLEU and the F-Measure metric) of all the systems that were built.

Great care was taken to ensure that the key input into the measurement systems was consistent and clean, to ensure that the final BLEU scores are meaningful and indicative of likely real-life experience results of data consolidation.

Asia Online encourages comparison of its output with that of its industry peers. We provide a selection of pre-translated examples and comparisons against other third-party translation engines. We also provide the ability to translated a small amount of your own material for a direct evaluation of key language pairs.

Examples are also provided which demonstrate the Asia Online itterative improvement cycle over a 3 month period.