“These are certainly bold assertions, but there`s not a lot of information about how they did their evaluations so that someone could support them at this point” – John Tinsley, CEO, Iconic Tanslation Machines In this context, enter DeepL, an online machine translator developed by the founder of Linguee (the database of dictionaries and online translations , which is usually one of Google`s top 3 organic results when they do an English translation). The CEO of Linguee is former Google researcher Gereon Frahling. NMT systems are hungry for data and electricity. At the end of August, the Chinese internet company Sogou invested in the translation data provider UTH to safeguard a large body of quality data. “The verification is preliminary, and Chinese antitrust authorities generally verify this information before deciding whether a formal investigation is necessary” (translation in the image below). Barely two years after the burglary of the translation technology scene, neural machine translation (NMT) is all the MT community is talking about. Microsoft, Google, Facebook and other major technology companies have all switched to NMT, as have the European Patent Office and the World Intellectual Property Organization. Even end buyers are starting to build their own systems from open source models. Slator approached two machine translation experts to use DeepL. We also contacted DeepL, but we did not receive a response at the press time.

Vashee continued: “It will depend on the level of adjustment they will allow for the business market, but they already have enough Linguee traffic (advertising revenue) to heavily subsidize these efforts and I assume they will put pressure on the (small) marginal MT players. Vashee concluded that it was too early to say more at this point, until they published more details. Co-founder of Slator. Linguist, economic developer and mountain runner. Based in the beautiful seaside city of Zurich, Switzerland. Not that other MT suppliers tend to be overly modest when marketing their products, but DeepL, headquartered in Germany, is releasing door-to-door pistols. Yet Kirti Vanshee rightly concludes that DeepL “is definitely someone you can observe.” Kirti Vashee, a freelance technology and marketing consultant and editor-in-chief of eMpTy Pages, commented: “I think they use well-understood and public test kits and say they have better results. In my opinion, they are a little better, but far from revolutionary. Anyway, to be better is to be worse. The graphics are a little too dramatic, based on actual facts, but they have a deep knowledge of the data, and Linguee is one of the best voice data implementations of big data, so this will certainly be a challenge for other MT players in the market. John Tinsley, CEO of Iconic Translation Machines, said: “These are certainly bold assertions, but there is not much information about how they did their assessments so that someone could support them at this point. I look forward to them publishing their API in the coming months.

I am sure that some of the mt research groups will be interested in applying some academic rigors to evaluations. It`s also quite impressive to have the 23rd most powerful supercomputer in the world…┬áVisit us on October 12, 2017 at SlatorCon New York and receive an update on the state of the art in neural machine translation by one of the leading researchers in this field, Kyunghyun Cho, assistant professor of computer science and data at New York University. “You have a deep knowledge of data, and Linguee is one of the best big data implementations of voice data, so it will certainly be a challenge for other MT players on the market” – Kirti Vashee, independent technology consultant and publisher of eMpTy DeepL Pages translation, was just and actually much more fluid for that particular phrase than Google Translate.