Add a function to support Google's Word2Vec

[yuming wang]

Hi:



I have a function to load Google’s Word2Vec generated binary file and spark
can use this model. If it is convenient, I'm going to open a JIRA and Pull
Request.



My code is:

https://github.com/wangyum/spark/commit/7c80d311722d67ed4b9746537e0a21c2dc1a9670



Thanks