Identifying personally identifiable information in datasets upon upload/publish

Some datasets uploaded to the Harvard Dataverse contain personally identifiable information (PII), particularly IP address and geolocation data (longitude and latitude). PII is included (as default) in survey data when researchers use internet survey software, such as Qualtrics. Researchers should delete this PII when they upload datasets to Harvard Dataverse. I did a quick search for "LocationLatitude" (the standard variable name for latitude data in Qualtrics) and it turned up 130 such datasets. Many of these datasets do not require geolocation data as part of their analysis. I suspect that researchers had uploaded the PII because they were not careful. @adam3smith suggested that Dataverse have an automated check and alert users when they are uploading datasets with PII.

The issue names no files, tests, or entry points. Begin by tracing the dataset upload and publish workflows, then determine where PII detection and user alerts would belong; done would require an agreed scope and a verified check covering the cited IP address and geolocation data.