Facebook has launched a multilingual composer that lets the user to compose and translate their posts into 44 languages. It is said that there are almost 1.5 billion users of Facebook. But, only half of the users speak the English language. The rest of the users speak other languages.
Reports say that the new tool lets the user translate the posts in a semi-effective way. The translated posts will automatically show up in the users’ News Feed.
Ayan (who oversees the company’s translation efforts) and team designed this multilingual composer especially for the multilingual audience. Further, the composer lets the user to edit the automatic translation. One can even provide their own translation.
But the eventual goal is to automate the process for each Facebook user. Although, machine translation is not so perfect, yet it is steadily improving.
Facebook will handle the translation task largely with traditional algorithmic models. These models rely on the language statistics. But when it comes to translating English to German, the company is majorly leaning over the deep neural networks. According to Ayan, this provides a noticeable improvement.
In recent years, the deep neural networks have been known to be adept at learning varied tasks, such as identifying spoken words, recognizing faces in photos etc. But, now, they are also improving their understanding about the natural language and machine translation.
It is said that Skype translation service also leans over the neural networks.
The users will get an option of clicking to create additional versions of the post or update in different languages. The user can either write his/her post from scratch, or edit the text generated by the Facebook’s translation software.
Comments are closed.