Sunday, November 12, 2006

Sanskrit and NLP


Although I wasn't intending to, one of my friend's friend claimed that Sanskrit was object oriented while I said that its regular grammar is what makes it more interesting for computer scientists. I ended up digging out quite some information on the topic.

It turns out, the whole interest in Sanskrit first started when a NASA researcher, Rick Briggs, published a paper "Knowledge Representation in Sanskrit and Artificial Intelligence" in the AI Magazine in 1985. The paper proposes that natural languages could also be used for knowledge representation. It details how Sanskrit is ideally suited for the job due to its regular, unambiguous grammar. You can read the entire contents of the paper here: http://www.gosai.com/science/sanskrit-nasa.html

The part about Sanskrit being OO is also probably true. Prof M A Lakshmi Thathachar (then Director of Academy of Sanskrit Research) says so in a '03 article in the GoodNewsIndia magazine. He says that:
"Words in Sanskrit are instances of pre-defined classes, a concept that drives object oriented programming [OOP] today. For example, in English 'cow' is a just a sound assigned to mean a particular animal. But if you drill down the word 'gau' --Sanskrit for 'cow'-- you will arrive at a broad class 'gam' which means 'to move. From these derive 'gamanam', 'gatih' etc which are variations of 'movement'. All words have this OOP approach, except that defined classes in Sanskrit are so exhaustive that they cover the material and abstract --indeed cosmic-- experiences known to man. So in Sanskrit the connection is more than etymological."
Although not entirely convincing, it is quite interesting. The complete article can be read here:
http://www.goodnewsindia.com/index.php/Magazine/story/melkote-sanskrit-academy/

Another interesting aspect about Sanskrit seems to be that Panini formalized the entire Sanskrit's grammar in his own BNF-like form using meta-rules, transformations, and recursion with just 4000 sutras (rules). All this was way back in 400 - 200 B.C. while the well-known BNF was developed sometime in 1958-60. Details here: http://www.infinityfoundation.com/mandala/t_es/t_es_rao-t_syntax.htm

Looks like Sanskrit has the ideal grammar structures suited for machine representation of knowledge. However, it hasn't been shown yet that Sanskrit is object oriented.

Whatever be the case, I am just happy and proud that I got to learn not only Sanskrit (in my school days) but also about BNF and Machine Learning so that I can appreciate it all !!

2 comments:

Karthik said...

Daii ide pathi seshu sir evloooo vaati classla sollirupar .. bava oda sanda potunde kekkama irunthuttu friend's friend sonnanama.. dog da nee

Aravind said...

Dei class la enakku thoongarathu thaan velai - nee thaan kavanippae - sonna apporam peela vidaraennu solluvae....:P

Post a Comment

Back to Top