Over the last few years, there’s been a buzz about machine learning and how automated systems can improve the efficiencies of business and everyday life. From the algorithms used to power search engines to machine learning used to optimise retail pricing, it’s safe to say that these automated systems are now crucial technologies for navigating life in the 21st century.
When it comes to language automation, machine translation (MT) is fast becoming a popular tool used by businesses to streamline localisation workflow and reduce costs. Even Google, Microsoft and Amazon are all trying to outdo each other when it comes to their research and machine translation offering.
But it wasn’t so complex as this in the early days of MT. Although machine translation is a relatively new technology, its origins are based on a more rudimentary rule-based and example-based systems.
Today you’ll find statistical machine translation (SMT) as a common form of MT used to adapt the content. These systems are trained by feeding them large amounts of bilingual translations which are typically phrase-based. This means that they focus on analysing sequences of words using statistical techniques.
Neural machine translation (NMT), on the other hand, is fast becoming a popular technology used by businesses such as retailers. Some have even increased translation efficiencies when using NMT for product descriptions.
Unlike the traditional phrase-based translation systems, NMT uses a single, large neural network that is trained with existing content and a feedback loop to produce more accurate translations over time. The more you use the NMT system, the better the quality of translation. In some cases, neural machine translation systems have been used by retail brands to translate large batches of product descriptions.
It’s safe to say that language automation services like machine translation has a very long way to go in understanding the complexities of human communication and culture; so you won’t be seeing MT taking over the entire localisation process any time soon.
But machine translation has proven to be an impressive tool to be accompanied by native human translators to increase efficiencies and help brands go to market faster.
Want more information about how some of the world’s top retail and travel brands use automotive technology to reach consumers in new markets? We partnered with Econsultancy on a research report identifying localisation opportunities and challenges with a panel of eight leading brand marketers.
You can download the report for free below.
