"Translate Coursera": how to make a platform for translating the best courses in the world into Russian

On March 24, ABBYY Language Services and the Digital October center announced the official launch of the “ Translate Coursera ” crowdsourcing project. His goal is to volunteer to translate the best Coursera courses into Russian and make them available to all Runet users. You can register and start translating lectures now on the project website.

While we were developing and preparing this project for launch, a lot of interesting things happened. We solved a lot of organizational and technical tasks, met a lot of interesting people, including the bright team and the founders of Coursera, as well as the team of our local partner for this project Digital October. We think that it will be more interesting for you to learn about technologies, so now we will tell you what the platform we have developed consists of and what it can do. Almost certainly this is not the last post about the project, so write in the comments what else you would like to know.

The development began last year, and until yesterday the platform was in the closed beta testing stage, attended by employees of ABBYY, Digital October, as well as students and young specialists of Moscow State University, Russian State Humanitarian University, Moscow City Psychological and Pedagogical University and other universities. Thanks to the comments of the first translators, our specialists were able to finalize the platform, make it more convenient and fully prepare for work.
')
The resource is based on the cloud platform developed by ABBYY Language Services for the automation of SmartCAT translation. As you remember , this system allows the use of modern technologies in the translation industry: translation memory, automated support for the integrity of terminology, machine translation. All this is collected in a convenient editor interface that is designed to maximize user productivity. In "We will translate Coursera" these professional technologies are available to everyone - special training is not required for "inclusion" in the work. To make the project interesting for volunteers, our developers have added a few additional functions to the system, which we will discuss below.

Anyone can register on the project website , choose any course you like and start working with it. The texts of lectures are divided into sentences, the time frame for translation is not: only you decide how much time to spend on participation - translate 2-3 sentences or 20 pages. In addition, for each sentence, you can add several translation options, if you cannot choose the best one.

After registration, each participant will have his own profile, in which, besides his personal data, information is displayed on the number of proposals translated, the rating, the courses being translated and the voting results of other participants. Volunteers will be able to discuss translations with each other right in the process, and the most diligent will receive virtual rewards for their achievements.

Not less important stage - voting for the best translation option. You can vote directly in the translation editor; Also on the site there is a special widget.

Then the translation options will go to the expert for approval and then go to the Coursera subtitles. In the role of experts are teachers of relevant specialties of Russian universities.

User activity is expressed in the form of a personal rating. On the main page there is a list of project leaders in which participants compete for the title of the most active translator of Coursera.

The rating is made up of the following parameters:
1. The number of translations added by the user;
2. The number of votes of other participants cast for translations of this user;
3. The number of votes given by the user for other people's transfers;
4. The number of user translations that the expert chose as the final.
Parameters 2 and 4 make the greatest contribution to the rating, so it’s profitable to translate less, but better.

As mentioned above, the platform has automated support for terminology integrity. For each of the courses being translated, a glossary is pre-compiled: a special program analyzes lectures and selects common words. The resulting list of terms is checked by experienced editors of ABBYY Language Services, “weed out” unnecessary and translate - this is how a pair of terms is obtained in the source language and the target language. In the working interface of the translator, the system highlights the terms, shows their translation and allows you to substitute it in the edit field. This ensures the uniformity and integrity of the translation of the entire course.

In the translation editor there is a special panel in which users can view the video course that they translate. Firstly, it helps to better understand the meaning of the text; secondly, each project participant sees at once how his translation version looks on the lecture video fragment: at the time of editing, the user watches in real time for changes to the subtitles in the video and can manually specify the positions for the subtitle line breaks. Thus, participants have the opportunity to select phrases for translation that will look harmonious on the screen. In addition, the system allows you to quickly rewind the video to the fragment that the user is currently translating, and vice versa - go to the sentence that sounds at the moment. To display the video, we took an open source web player. It can play subtitles that it accepts as .srt files through the API: SmartCAT dynamically breaks the translation text into subtitle lines, generates a .srt file from them and sends it to the player.


We would very much like to tell you about the details of the development of the platform, but it’s impossible to distract our specialists yet, so we’ll introduce you to other technical details a little later. For now, join the Coursera translation ! All questions, comments and suggestions are sent to our address: coursera@abbyy-ls.com or simply post here in the comments. You will learn more about the project, its mission and Coursera from our blog .