Closed zhonghao0077 closed 6 years ago
By default ner
runs a lot of rule based stuff now. You can deactivate that by setting ner.applyFineGrained
to false and should speed up performance (though you'll lose the fine-grained tags such as CITY, STATE, and COUNTRY, etc...)
Hi,
Recently I am trying to use the server mode of CoreNLP 3.9.1, The annotation contains ssplit, tokenize, pos, lemma and ner. The speed seems much slower than 3.8.0. Do you know why is that thanks,
Peter