I'm working on a research project, which is focused on demonstrate ineffectiveness of document redaction methods. I'm looking for a dataset of pdf documents which contains generalities about real people (or generate pdfs out of a dataset containing personal information).
To better explain, an ideal scenario would be having a dataset of publicly available and real resume in pdf; the pipeline would be:
- Redact personal information from the pdfs;
- Unredact with my "magic tool" (give it for granted);
- Test the outcome (unredacted pdfs) against the ground truth (initial clean dataset).
What I found so far:
- https://www.kaggle.com/datasets/snehaanbhawal/resume-dataset (easy to generate pdf from the dataset, but no "ground truth, no real person behind the id);
- https://www.kaggle.com/datasets/palaksood97/resume-dataset/data (no license specified and restricted sample).
I'm also thinking about public open data that governments are forced to publish. It really depends on the country, but I though about public law sue reports or death/birth certificates (I already saw it's not easy to collect).
Do you have in mind some reference which might work for my needs?