Amazon Translate tops machine translation ranking
Tech giant Amazon’s Amazon Translate service uses neural deep learning models to automatically localise content.
The benefits of such an approach include both more accurate and more natural sounding translations, as compared to traditional, more literal approaches.
As a non-static model, Amazon says the solution is continually improving its capabilities across areas such as language localisation, analytics and real-time translation for use in areas such as customer service.
The service has now been ranked #1 in a report by Intento, an organisation that assists enterprises with AI procurement, which looked at the top 15 machine translation providers. While it didn’t come out top for all the language pairs, industry sectors and content types the solutions were assessed for, Amazon’s offering nevertheless had the highest number of wins, followed by Google in second place.
In a blog post, Greg Rushing, US Air Force Fellow in Amazon’s BRIDGE program, said: “The data used included examples from across 16 industry sectors, with 8 content types, including topics such as financial documentation, patents, sales and marketing material. These inputs were translated between 14 common language pairs to determine the best engine for a given translation scenario. It ranked the results of each MT engine based on how they compared to a reference human translation.”
Amazon also uses Translate in a consumer-focused setting, with its own Alexa smart assistant, which is likely the area where most will be familiar with the system. With Alexa speaking eight languages, Amazon uses Translate to understand questions asked by users without needing eight duplicate, hard-coded rules for each language.
In another blog post, Heike Schirme, Senior Manager at Amazon Alexa, said: “Instead of manually translating thousands of utterance lists offline, Alexa uses [Amazon Translate] to translate utterances instantly and on demand. If an utterance in one language doesn’t map to a given intent, Alexa translates the message to English and sends it again for a second attempt.”