The saara Machine Translation Systems

'saara' is a comprehensive system of computational tools for Indian languages designed and developed by Prof. Kavi Narayana Murthy and his team at the School of Computer and Information Sciences, University of Hyderabad, India. The term 'saara' includes the theoretical foundations, the architectures, data, tools and technologies for various language processing tasks such as Spell Checking, Stemming, Tagging, Morphological Analysis and Generation, Syntactic Parsing, Glossing, Paraphrasing, and Translation. A Machine Translation system developed in the saara approach is termed "The saara Translator".

The saara Translators are NOT intended to be a fully automatic systems, the aim is to produce HIGH QUALITY TRANSLATIONS with minimal human effort. The saara Translartors are NOT directly available to end users, instead, translation services are offered to interested users.

The saara Translators are superfast: WE CAN NOW TRANSLATE 1,00,000 SENTENCES PER SECOND ON ANY ORDINARY DESKTOP PC or Laptop. A typical 150 page book has about 3000 sentences. This means we can translate some 33 books per second, about 1,00,000 books per hour! The goal, however, is to get high quality and this will need a bit of manual effort in addition.

Currently, saara Translator for Kannada to Telugu is ready. To translate any book or article, there will be a pre-check phase where the system databases are tuned. Then comes the automatic translation phase - We can now translate 1,00,000 sentences per second on any ordinary desktop PC, and produce quality translations with 85%-95% comprehensibility. In the last phase, manual post-editing is done to produce publishable quality translations. Overall, this saves a good deal of time, effort and money compared to completely manual translations.

A highly popular 150 page book ("Kshana-Hottu Ani-Muttu" by "S Shadakshari") has been translated and released on 15th December 2018 at a function in Hyderabad.

A Live Demo of the saara Translator was given at an Interactive Session at Hyderabad on 17th January 2019. Hon'ble Vice-President of India, Sri M Venkaiah Naidu garu was the Chief Guest. Chancellor of the University, Justice L Narasimha Reddy garu presided. Prof. Appa Rao Podile, Vice-Chancellor of the University, gave the welcome speech. Dr Pammi Pavan Kumar, Department of Telugu, University of Hyderabad, gave the words of thanks. The saara MT systems were highly appreciated by the Hon'ble Vice-President of India. It has been suggested that Machine Translation should be taken as a National Level Mission so as to make all relevant and useful information available to people of our country in their own languages and scripts.

Telugu - Kannada saara Translator is also available.

Also, a generic MT Development Framework, known as'The saara MTDF' has been developed. Using this framework, it should be possible to develop high quality MT systems for any two languages with minimal time and effort, following our theory and guidelines. You will not need a large parallel corpus, nor will you need in-depth lingustic experitise. Only people who know the two languages will be required.

Please Note that The saara Translator and The saara MT Development Framework are NOT PRODUCTS. They are NOT available for download. THEY ARE AVAILABLE AS SERVICES FROM OUR WEBSITE. Interested users who wish to get some books or articles translated may contact us at knmuh@yahoo.com Researchers who wish to collaborate with us in MT System Development for languages of their choice may also contact us at knmuh@yahoo.com Also, please note that Machine Translation works best for Modern Prose only, not for Poetry or Archaic Texts.