Any NLP or machine-learning people here?
I'll be working on some statistical NLP stuff over the next month, and it would be nice to be able to discuss it some. The field's being relatively small and new makes finding resources challenging.
At the moment, I am attempting to do feature induction for a CRF for NP-chunking on the standard WSJ sections. The assignment is actually to just play around with features and see what kind of performance we can get. But I figured feature induction might save me time. Is this a really stupid idea?