Monday, May 18, 2009

Panini in the Computer

The digital computer is almost completely a 20th century invention, save for the invention of electricity and the development of Boolean Logic. One rarely, if ever, hears that something developed hundreds, leave alone thousands of years ago will suddenly be found useful for the computer programs. Let's thank Panini - the legendary 5th century BCE Sanskrit grammarian - for this new development. I'll reproduce the entire Hindu news report because it makes interesting reading:

Panini to the rescue
Research team turns to the “world’s first computational grammarian!”.

K.V. Kurmanath

Panini, the legendary Sanskrit grammarian of 5th century BC, is the world’s first computational grammarian! Panini’s work, Ashtadhyayi (the Eight-Chaptered book), is considered to be the most comprehensive scientific grammar ever written for any language.

According to Prof Rajeev Sangal, Director of IIIT (Hyderabad) and an expert on language computation, Panini’s epic treatise on grammar came to the rescue of language experts in making English unambiguous. English is more difficult (as far as machine translations are concerned) with a high degree of ambiguity.

Some words have different meanings, making the analysis (to facilitate translations) a difficult process. Making it disambiguous is quite a task, where Panini’s principles might be of use.

Ashtadhyayi, the earlier work on descriptive linguistics, consists of 3,959 sutras (or principles). These highly systemised and technical principles, some say, marked the rise of classical Sanskrit.

Sampark, the multi-institute effort launched to produce a translation engine, enabling users to translate tests from English to various languages, will use some of the technical aspects enunciated by Panini. “We looked at alternatives before choosing Panini,” Prof Sangal says.

Incidentally, Prof Sangal co-authored a book, Natural Language Processing – A Panini Perspective, a few years ago.

Besides the technical side, Panini would be of great help to researchers on the translation engine on the language side too.

A good number of words in almost all the Indian languages originate from Sanskrit. “That is great because Indian languages are related to each other,” Prof Sangal points out.

No comments: