So, you’re thinking about introducing Machine Translation into your company?
Are you a Language Service Provider (LSP)? If you are then you’ve probably got a client request for Machine Translation, or you are a progressive company who wants to list the latest technologies on your website and you’ve heard about the productivity gains others are getting from MT and want part of it.
Are you an Enterprise? Then you’re probably under pressure to reduce translation costs and/or introduce new languages online and offline.
Either way, you will travel down a similar path when introducing Machine Translation into your company and many people and departments will be touched– translators, engineers, project managers, managers, solutions engineers, finance, legal etc. That’s why it’s important to be clear about what to expect from the process and to put down measures to ensure that it all runs smoothly.
During the investigating period you will research in both print and online publications to get an idea of which types of Machine Translation solutions are available and which might suit your business needs. You’ll read about rule-based MT, statistical MT, and possibly hybrid too, learning that each option has its pros and its cons.
After you narrow down your preferences to a small group of providers, you should look for peer based reviews of the Machine Translation products – this could be in the form of case studies, webinars, video testimonials, forums, or face to face at conferences. Doing this will give you a better understanding of how the product works in a real life scenario and might bring to light issues you should be aware of, that only users will tell you.
At this point you will want to contact some Machine Translation sales companies to get a better overview of the product and service they offer and to see if the pricing and support meets your needs. Ask about automatic post-editing tools and analytics as these are key!
It’s important to see the Machine Translation system in action so you know if it’s manageable for your team. Most providers will offer you a demo after or during your first meeting, but don’t be afraid to ask if they don’t.
So, after you’ve seen a few systems in action you may be in a situation where you are choosing between a custom build MT system and a platform which allows you to develop, manage and deploy your own MT systems. Have you tried a pilot? This is a great way to see whether your data works better with one system or another.
Things to think about when you start using Machine Translation (statistical):
– Do I have sufficient data to get started?
Realistically you need some good data to start off with. While MT providers (like us) offer to manufacture data – it is never as good as your own. You’ll need a TM (of roughly 2-5 million words), some terminology and some monolingual data (which can be easily generated).
– Does the MT provider have a developer friendly API?
Whether you’re intending to use Machine Translation as a pre-translation tool or to translate content on-demand you’ll want to ask the MT provider about its API capabilities. The speed and flexibility of an API will vary greatly so remember to ask about this!
– Have you discussed post-editing with your translators?
Some translators are less open to Machine Translation than others, and rightly so – there are some questionable free MT systems available online today! While many have had bad experiences, some just don’t like the thought of a computer doing their work and others would prefer to translate from scratch. It’s important to talk to your translators before, during and after deciding to introduce Machine Translation into your company so they are part of the process and more in tune with the technology. Translators and linguists play an integral part in the development of Machine Translation engines and will help be a great help when deciding which MT solution produces to choose.
This is not an exhaustive list of items to think about and discuss with your team but it should give you some idea of the process you will likely go through when evaluating Machine Translation systems. If you would like more information about the buying process, or would like to book a demo of KantanMT.com please contact Niamh (niamhl@kantanmt.com)