Evaluation of the preprocessing to make a dataset anonymous

I have a very huge dataset from the NLP area and I want to make it anonymous. Is there any way to check if my pre-processing is correct? Generaly, is there any way to evaluate how good is the pre-processing for the anonyminity?

I want to mention that the dataset is really huge, therefore it can be cheched manually.