Rethinking Evaluation Metrics for Grammatical Error Correction: Why Use a Different Evaluation Process than Human?

Open in new window