DOI (v2.0): 10.5281/zenodo.11229402.
This repository provides the experimental results of the paper The Privacy-Utility Trade-off in the Topics API (DOI: 10.1145/3658644.3670368; arXiv: 2406.15309).
The notebooks were run using:
Pythonv3.11.9bvmlibv1.0.0matplotlib3.8.0numpy1.24.3pandas2.0.1qif1.2.3requests2.31.0scipy1.11.3tldextract5.1.2tqdm4.66.1urllib31.26.16
Please note that qif, as of version 1.2.3, is not compatible with Apple Silicon processors. This is a known issue already reported on the module's GitHub repository: Support for M1 Mac.
The datasets produced for the experimental analyses can be found on Zenodo: AOL Dataset for Browsing History and Topics of Interest (DOI: 10.5281/zenodo.11229615).
For convenience, create a Python virtual environment before installing the required packages:
python3 -m venv venv
source venv/bin/activate
pip install -r requirements.txt
The data treatment notebooks are primarily for reference and were used to generate the AOL Dataset for Browsing History and Topics of Interest (DOI: 10.5281/zenodo.11229615) files.
To execute the analyses notebooks, please make sure to add to the folder data-treatment/ the respective files from the AOL Dataset for Browsing History and Topics of Interest (DOI: 10.5281/zenodo.11229615).
To execute the analyses notebooks, please make sure to add to the folder analyses/ the respective files from the AOL Dataset for Browsing History and Topics of Interest (DOI: 10.5281/zenodo.11229615).
To understand how the various GNU licenses are compatible with each other, please refer to the GNU licenses FAQ.