As the worlds get smaller and smaller, the need for language translation is becoming essential. Now, Flitto is helping machines get better at translation.
As the worlds get smaller and smaller, the need for language translation is becoming essential. Big techs know about this and that’s why Facebook, Amazon, Microsoft, and Google are working on their form of machine-generated translation service. However, when compared to a professional translation service, machines are still miles away. Now, Flitto is helping machines get better at translation.
Flitto is a translation company that started back in 2012. Their main base of operation is in Seol. Flitto provides translation services, which are powered by real-human translators through a crowdsourcing platform.
Flitto has over 7.5 million users. And yet, these days 80% of its revenue comes from selling language data to tech companies such as NTT DoCoMo, Tencent, Microsoft, and Baidu.
When Flitto initially started to open their services, the main at the time was Google Translate. We all know that Google Translate is a hit-and-miss business. With human-powered translation services of Flitto, the company offered way better translation results. Flitto crowdsources with more than 1.2 million translators and backed by Colopl, a Japanese game developer giant.
Last 2016, the game changed a bit when Google Translate incorporated their neural machine for translation services. The addition drastically improved Google Translate’s accuracy. Now, more and more tech companies are competing in the same arena. Apple, Facebook, Amazon, and Microsoft are only a few of the tech giants that are getting in the translation industry ring.
While it’s true that machine-generated translation systems are getting more accurate, it’s still far from being perfect. Translation machines need huge amounts of data, and this is where Flitto comes into the picture.
Flitto can help machine learning translation systems by introducing sentences, words or phrases that come from its crowdsourcing platform. Local dialects, pop culture references and slang are often inputs that need to be done by a human translator as these kinds of elements are very difficult for machines to figure out on their own. Since Flitto started their services, they have already compiled no less than 100 million sets of language data.
There are some computer learning programs, like Deepminds AlphaGo, can learn to train themselves without any need for human inputs. However, human language is something that may seem simple to us, but it’s very complex for machines to learn. Computers are good at calculation, but creativity and connecting of dots are something that computers are not really good at. According to Flitto’s CEO, machines generated translations have a place in the future, but you can’t completely pull out human translators in the picture.