Continual improvement is vital to making the most of your Machine Translation (MT) investment, but natural language processing engineers won’t spot linguistic and translation issues. So what if you could have an expert linguist check your output? Someone who has specific in-domain knowledge and also understands machine learning?

That’s what this linguistic feedback service does for Asian Absolute’s MT clients in the UK and around the world. It’s the easiest way to have your MT engine output reviewed and assessed by an expert native-speaking translator. Without the cost of post-editing each segment along the way.

Of course, if you need Machine Translation post-editing, you’ve come to the right place too. Here you’ll find specialist engineers and highly skilled linguists. With expertise in your domain. And native-speaking mastery of your target language.

You’ll find it easy to get the specialist translation expertise you need to assess and analyse your data. This includes specific knowledge and qualifications in your domain. And long experience of what clean data looks like from an MT perspective.

• Get your data assessed by linguists with extensive in-domain qualifications and experience
• Receive feedback from native-speaking specialists in more than 120 languages
• Outsource any aspects of your MT projectwhich make commercial sense – linguistic feedback, post-editing, cleaning and fixing bilingual data…
• Rely on ISO:9001-certified and award-winningproject management skills

Client testimonials

You will receive a report which collates and analyses:

• Issues in your MT output and the category and sub-category they fall under
• Any patterns and rules underlying those issues

It’s an often vital precursor to a more technical assessment by an engineer. Because few engineers will have the language mastery required to give in-depth linguistic feedback.

Armed with this report, you can then have your own engineers – or our experts – draw up a list of actions. Everything required to fix the root causes of the identified problems.

Linguistic feedback is often carried out alongside our Machine Translation post-editing service. But it can be provided as a standalone service, without the linguist making any of the edits which a post-editor would.

Instead, our linguist will gather and categorise all of their feedback. So your engineers will know exactly what you need to do to train your engine and improve its performance.

And by not spending time editing, the linguist will assess a much higher volume of output for every hour worked. So if your objective by post-editing the MT output is to train your MT engine, you get a better return on your spend by requesting a feedback service only.

For Statistical Machine Translation (SMT) engines, your feedback can be used without you needing to retrain your entire engine from scratch. Fix the identified issues and you’ll prevent them from occurring ever again.

If you’re using an NMT engine you may need to repeat the training from scratch. But this time you will be armed with the understanding needed to build a better engine.

Either way, you will achieve higher quality engines faster than by relying on data processing engineers to identify issues, and conventional post-editing to provide clean bilingual data.

Frequently Asked Questions