Detecting corrections in a sentence, with an NLP model


I shared a correction Detector on github. We use this to auto-complete language tutors' feedbacks to learner’s writings. This helps to high-light changes, and provide rationales behind fixes. For a complete disccussion, please refer to our PAPER.

Demo: Try correct the sentence below (click to try another).

Now, what errors have been corrected? And how?

{{chunk[0]}} {{chunk[0]}} {{chunk[1]}} ({{chunk[2]}})