harmful slurs present in dataset about charmanteau-camready HOT 2 CLOSED

Thanks a lot for the feedback! We have now included a disclaimer about the potential harmful slurs/ offensive language in the repository's Data Readme . [Wherever possible, we will also try to update the copies of our paper to state this clearly]

If there are any particular words you would recommend us to remove, we could release a filtered version of the dataset which has those words removed and explicitly mention this dataset in the README as well. Ideally, we would have liked to remove all such words but lot of the portmanteaus come from slang/new headlines etc and we did not have objective labels or a clear understanding ourselves of which specific words/terms would be not offensive and which ones would.

We apologize for not being proactive about doing this ourselves, and will make sure to be more careful on such issues in the future.

from charmanteau-camready.

Thank you so much for including the warning! A useful starting point for filtering the list might be the words in Darius Kazemi's wordfilter list, though it is by no means comprehensive (and may also lead to false positives). In the end, I think manually curating the list item-by-item is the only way to achieve satisfactory results, and/or publishing the precise methodology that you used to arrive at the list.

from charmanteau-camready.