Which one is better for coreference resolution in terms of speed and accuracy: Stanford CoreNLP or Apache OpenNLP?

Which one is better for coreference resolution in terms of speed and accuracy: Stanford CoreNLP or Apache OpenNLP?

I can’t speak to speed and accuracy; however, I can say that for the year 2014, CoreNLP is cited over twice as many times as OpenNLP within the context of coreference resolution in Google Scholar.

See also my quick and dirty webpages:

• Apache OpenNLP & Coreference Resolution 2014 | Meta-Guide.com
• Stanford CoreNLP & Coreference Resolution 2014 | Meta-Guide.com