Does anyone know if these are/can be good for named entity recognition?
I am stuck implementing second order CRFs right now for the lack of a good implementation, and this seems a lot easier.
I'm not aware of any strong RNN results for NER, no.
You'd probably find the paper here: http://aclweb.org/anthology/ (everything in CL is open access). You want the proceedings of CL, TACL, ACL, EMNLP, EACL, and NAACL. Don't bother with the workshops.
