Does removing stop words from text affect stanford core nlp NER performance?

问题

we are trying to implement name entity recognition on millions of comments/feedback and the process appears to be slow. We are thinking of removing stop words/frequent words from the texts and apply ner on them. Does removing stop words affect the accuracy of ner?

回答1:

I think it's plausible you will get respectable F1 scores if you run on a sentence with the stop words removed. Ultimately you will have to experiment with it and see if the quality is acceptable for your needs.

来源：https://stackoverflow.com/questions/42393503/does-removing-stop-words-from-text-affect-stanford-core-nlp-ner-performance