noise in User Generated Annotations #1
Description
User annotations cannot be trusted, if we use whatever annotations the users provide, eventually the model will start giving poor classification results.
There has to be a way to verify if the user has annotated the data correctly before deciding to use that as training data.