Machine Translation

It is Maven International policy that only human translators are engaged in our projects due to their superiority over available machine translation tools. However, the exponential growth of short-lived content on social media has created demand for machine translation. Unlike Google Translate and other online translation engines, our technology enables us to greatly improve the final outcome of the translation by customizing engine settings for each client, and by using a predefined term base.

It doesn’t stop there; at the final stage, we engage specially trained post-machine translation human editors (PMTE) to review and correct any possible mistranslations. Machine translation results in an even faster translation process and greater savings. 

Post Machine Translation Editors know how machine translation behaves in each language pair and direction and know where to look for potential shortcomings. 

Applicability: 

Large volume of standardized data, chunks of content, (phrases or titles only), short lived content (running lines, sometimes machine translation is applicable in translation of subtitles). 

If you have translation memories of 100 k – 500 k translated words in your language pair and domain, we may use this as feed to one of our machine translation engines powered by neuron learning to train our engines to produce very similar output for similar content in same language pairs and domain.

Quality of machine translation output varies greatly, it depends on your domain, language pair & direction, complexity of the content, level of coherence and cohesion, availability of translation memories and previously translated content.

The higher quality the less post machine translation editing is required, the lower its cost. 

Please note that PMTE service is outside of scope ISO 17100