Printable Version of this PageHome PageRecent ChangesSearchSign In

:28 May 2016

Recently, Google open sourced their dependency parsing library called SyntaxNet. This library is built on top of Tensorflow, a hugely popular library for numerical computation using data flow graphs. This has generated significant buzz among NLP researchers. I wonder how much of it is because of the Google brand name. Okay, I think the problem of syntactic parsing is an important one. But how big a step forward is SyntaxNet? Conceptually, the contribution of SyntaxNet is pretty subtle. But the devil is in the details and the bulk of the contribution is about careful experimentation, tuning, scale up, and refinement - the hallmarks of Google engineering. Can wait to try it out!

Last modified 12 July 2016 at 4:40 pm by svattam