The act or process of translating something into a different language, connects people around the world. Facebook, who pride themselves on being renowned connection-makers, have progressed us past the age of direct translations - unleashing their days-old AI network.
The site now uses translators powered by neural networks, which look at full sentences and the context they are used in to produce translations with increased accuracy. Before, was a more limiting phrase-based machine which would use direct translations of words or phrases to form a sentence translation on Facebook. The first model was riddled with shortcomings that were most obvious to the user when translating between languages that require different sentence structures.
The most fundamental advantage (for those of us who don’t have a strong grasp on programming) is that the sentence structure in the new model is both grammatically and functionally correct. All the data necessary for a translation is processed at once. It is pointed out that this takes a greater toll on a processor - but aren’t computers getting stronger everyday?
and Facebook Researchers are not stopping there...
We have just started being able to use more context for translations. Neural networks open up many future development paths related to adding further context, such as a photo accompanying the text of a post, to create better translations.