Smartling, an enterprise translation solutions company, announced a major product expansion for its Neural Machine Translation (NMT) Hub, available to all Smartling customers. Smartling uses a combination of technologies such as Neural Machine translation, which is capable of providing translation quality comparable to that of human translation, their proprietary Machine Learning models and “human-in-the-loop” delivery platform.
Smartling spent years collecting vast volumes of Machine Translation Quality data across multiple translation directions and content types, based on a widely accepted Translation Error Rate (TER) automated MT evaluation metric. From there, the data was used to train the AI models to not just make accurate decisions for the best MT engine selection, but consider fallback engine suggestions as well. This solves for the main deficiency associated with single-engine approach. The AI algorithms in NMT Hub are designed to dynamically learn from billions of data points available through Smartling’s closed-loop enterprise content delivery platform to ensuring the best quality MT output.
Designed to tackle translation issues that traditional single-engine approaches are not able to solve, Smartling’s NMT Hub is built to solve many of the challenges web developers and content managers.