Science is Fun Fridays!
Sometime between the 6th and 4th century BCE, Indian scholar Daksiputra Panini wrote down some rules describing the workings of Sanskrit. Over 2,000 years later these rules helped establish the science of linguistics.
He studied how words are formed and how they relate to others in the same language, now known as morphology. He described syntax and semantics, as well as an algorithm to construct grammatically correct Sanskrit words, but it's been hardly understood.
Now, a PhD student from the University of Cambridge, claims to have decoded this "language machine." Rishi Rajpopat believes this will have implications in teaching computers to understand human speech.
Panini provided a set of rules for word building, as well as a metarule - In the event of a conflict between two rules of equal strength, the rule that comes later in the grammar's serial order wins. Rajpopat believes this has been misinterpreted, and that is why scholars have struggled so long.
He applied the metarule to the left and right side of a word, with rules that applied to the right side given precedence - later in the word, not the order - and this almost always produced grammatically correct results.
"Computer scientists working on Natural Language Processing gave up on rule-based approaches 50 years ago," Rajpopat said in a statement. That may change now....
Comments
Post a Comment