Crowdsourcing Better Translations of Indian Roads Congress Standards

About This Project

Greetings! Jai Hind! Jai Gyan!

On this page we present an experiment in machine-language translation into Indian languages. We were inspired by the National Translation Mission, “a Government of India scheme to establish translation as an industry in general and, to facilitate higher education by making knowledge texts accessible to students and academics in Indian languages in particular.”

Public Resource has put considerable effort into making available rules and regulations and other important knowledge of use to students, teachers, government officials, and citizens of India. Among those resources are the codes and standards of the esteemed Indian Roads Congress. All of those road codes have been made available on our servers and on the Internet Archive on a strictly non-commercial basis to assist the students and the lifelong learners of India in their pursuit of an education so that they may better their status and their opportunities and to secure for themselves and for others justice, social, economic and political.

Recently, we have embarked on an effort to transform these codes and standards into ever more useful fashions. Specifically, we have rekeyed the standards below into HTML, then used open source code we have developed to translate the HTML files into 9 Indian languages: Bangla (Ben), Gujarati (Guj), Hindi (hin), Kannada (Kan), Malayalam (Mal), Marathi (Mar), Punjabi (Pan), Tamil (Tam), and Telugu (Tel).

The technology we are using is Google Vision to do neural network based machine translation. However, these translations are only the first step! We need your help! With neural networks, one can upload a glossary file, which is used by the neural network to adjust how it does translations. A glossary file is a comma separated list of words or short phrases. On the left is the English word or phrase. On the right is the proper translation.

What we need from you is this. If you are fluent in English and also one of the 9 languages, we ask that you start reading the English and then the Indian language translation. When a translated word or phrase is off, write those down into your glossary file. For example, on a previous project, we noted that the word "jurors" (as in "jury instructions" translated into Hindi as "gamblers." In this case, you'd note that discrepancy, and add the following line to your glossary file:

juror,जूरी-सदस्य

You only need one glossary file, we are trying to build up a vocabulary that spans multiple standards. Please name your file as follows with the appropriate language tag:

glossary.hin.txt

When you are done, email your file to Carl Malamud at carl at media dot org. Feel free to ask me questions by email or on twitter at @carlmalamud. If you are embarking on this, please drop me a note and let me know which language you have picked.

After we receive the glossary files, we will rerun the translation a second time, and we will continue to crowdsource better glossaries. Needless to say, we are very happy to make all this information available to the esteemed Indian Roads Congress and Public Resource practices a strict policy of non-commercial activity and asserts “No Rights Reserved” on any of the material we produce. This is all in the public domain, we are all public workers. Please join us!

Jai Gyan!

Carl Malamud for Public Resource

The Road Codes