A Dataset of over 500.000 commercial email newsletters, as collected by PrivacyMail.info

This dataset contains the data from roughly two years of operating PrivacyMail.info, an Open Source Email privacy measurement platform. It contains slightly over 500.000 commercial newsletters, as crowdsourced by users of PrivacyMail.info. You can find the methodology discussed in our paper: Max Maass, Stephan Schwär, and Matthias Hollick. "Towards transparency in email tracking." Annual Privacy Forum, 2019. The source code can be found on github.com/privacymail/privacymail

Please note that, due to its crowdsourced nature, this dataset is a sample of opportunity - it is not representative for all newsletters on the Internet, and likely contains biases based on how it was collected. Notably, German-language newsletters will likely be heavily over-represented.

Dataset Structure
The dataset is structured as follows: On the top level are folders describing the website the newsletter belongs to. Inside that folder are subfolders for each identity that was registered for that website. Inside each of these folders are a series of .eml files that represent the received email messages.

Copyright and Licensing
This dataset is set to non-public due to copyright concerns: The contents of the email messages are (presumably) protected by copyright in most jurisdictions. Most copyright doctrines contain exceptions for non-commercial research use - thus, we feel it is appropriate and acceptable to share the data on a case-by-case basis, the same way we did before shutting down PrivacyMail.info. When requesting access to the data, please briefly describe what research you want to conduct with it, and we will grant you access.

We thus do not put any explicit license on this dataset. Please do not share the raw data publicly. We request that you cite the above-mentioned paper and this dataset in any publications that result from it.