Identifying collocation in Stanford POS Tagger?

Question

Hi guys,

Is the Stanford POS tagger able to detect collocation? If so, how do I use it?

If I want to provide my own training file for the Stanford POS Tagger, do I have to tag the words according to the one like the WSJ

This means that I have to 'bracket" the words into Entities and collocation right?

If so, how do I find collocations from the tagger?

I am avoiding the need of using a parser.

Answer 1

No, the Stanford tagger neither needs nor provides collocations. It just puts part of speech labels on individual words. (If you are training a tagger, you don't have to use WSJ tags, but you do have to provide training data with a tag for each word.)