Why is Machine Translation Better than Ever...and Why it Matters to Your Business (Part 2)

Continuing on the first part of this Article, where we discussed the early technologies used in Machine Translation - statistical-based and rule-based. And now we are getting to the really good stuff because after rule-based and statistical-based machine translation methodology came Neural Machine Translation (NMT), the type of MT we are using now.

NMT consists of neural networks that, much like human neural pathways, can be trained and optimized for translation in virtually any human activity area. NMT can analyze translations completed by human translators and learn from them. Perhaps most importantly, NMT can understand the context. Because of all the above, NMT is more "fluent" and "human-like" than either of its predecessors.

It sounds somewhat futuristic, doesn't it? And what if I told you that NMT could also learn from its mistakes and adjust the next time, accordingly? You see, the future is already here.

But don't jump to the conclusion that you can replace your human translation service with MT just yet!

Even though NMT produces phenomenal results, its output is still far from perfect. In most cases, for any content that will be used in the course of business, and especially for any public use, post-editing, completed by human translators, will be needed.

Come back for Part 3 of this post next week and find out just what you can expect from NMT and what human involvement should look like for the best outcomes.  

Can't wait for the next installment? Contact me and let's talk!