Jul 22, 2020
The History of Google Translate
What if we could understand each other more easily?
Automatic translation has to some extent become part and parcel of our online experience. On Facebook, if you have international friends, you can ask directly to see a translation of the conversation. Not all too long ago automatically translating texts were not a given. In this article I want to focus, in short, on one of the most common ways of translating in the world.
In 2016 it was claimed that over 500 million people used the service daily. I must admit that I am one of the users that have gotten a great deal of joy out of trying to understand international friends or getting help to translate documents in a foreign language. The understanding and comprehension is of course not perfect, however the simple fact that I can understand part of a text is already incredible. It foes not only translate simple words typed into a translator. It translates multiple forms of texts and media such as words, phrases and webpages!
How did Google Translate start?
Allegedly the idea for Google Translate originated from 2004, when:
…co-founder Sergey Brin became frustrated with a translation program the company was licensing after it translated a Korean email into “The sliced raw fish shoes it wishes. Google green onion thing!” (TechCrunch, 2016)
Google Translate was developed by Google in April 2006.
Originally Google Translate was released as a statistical machine translation service.
Before you had to take the mandatory step of translating the text into English, however that is no longer necessary.
At that point it used predictive algorithms to translate text.
The service had, as you would expect for these earlier times, poor grammatical accuracy.
It was a considerable cost as well for Google to resolve the limitation of ever-evolving language.
In 2010 Google introduced an Android app to server as personal interpreter. The same year it was also integrated into browsers such as Chrome.
Google acquired Word Lens in 2014 to improve the quality of visual and voice translation. They were able to scan text or picture to have it translated instantly with a device.
Beyond this the system automatically identifies foreign languages and translates speech without requiring individuals to tap the microphone button whenever speech translation is needed.
→ With the advent of deep learning and these methods increasing in usage Google transitioned in 2016 to neural machine translation.
This translated whole sentences at a time and was measured to be more accurate between English and French, German, Spanish, and Chinese.
Machine translation is a huge application for NLP that allows us to overcome barriers to communicating with individuals from around the world as well as understand tech manuals and catalogs written in a foreign language.
Google Translate uses a combination of machine learning and human volunteers to make translations better.
They have a community for translators.
For Google it is now like having a personal interpreter in your pocket.
It now has 103 languages. In addition they can translate in a lot of different ways.
Although the features vary according to language.
They have both a translation API and AutoML
In fact the AutoML Translation is in BETA, but is a product offered by Google.
You can use Google to help with this task.
As you can see this is a product that has benefited a lot of users, however it is additionally a commercial product being offered by Google on the market.
It is certainly interesting to see how Google Translate has gone from a small project in a large company to one of the leading services out there to help us understand each other better used by people across most of the world.
In a world increasingly polarised, perhaps it helps us slightly to attempt understanding each other.
A good step forward is more accurate translation.
This is #500daysofAI and you are reading article 414. I am writing one new article about or related to artificial intelligence every day for 500 days.
