Spark - Find most common words in dataset

Question

I have a Dataset below like

+----------+-----------------------------------------------------------------------------+
|prediction|text