Natural language processing in Python

I want to be able to take a whole bunch of text, say, individual blog posts, and pick out the few keywords that describe that post. Think something like automatic tagging.

I am quite familiar with Python. Should I use scikit learn or the Natural Language Processing toolkit (NLTK)?

Also, any suggestions on what to read? The only thing I can think of at this point is just trying to follow some of the tutorials/reference documentation and see how I go.