topics_api_analysis

This is the code artifact of the paper A Public and Reproducible Assessment of the Topics API on Real Data

@inproceedings{topics_secweb24_beugin,
      title={A Public and Reproducible Assessment of the Topics API on Real Data},
      author={Yohan Beugin and Patrick McDaniel},
      booktitle={2024 IEEE Security and Privacy Workshops (SPW)},
      year={2024},
      month={may},
      doi={10.48550/arXiv.2403.19577}
}

Check out also our other topics_analysis repository and PETS'24 paper:

@inproceedings{topicsweb24_beugin,
      title={Interest-disclosing Mechanisms for Advertising are Privacy-Exposing (not Preserving)},
      author={Yohan Beugin and Patrick McDaniel},
      booktitle={Proceedings on {Privacy} {Enhancing} {Technologies} {Symposium} ({PETS})},
      year={2024},
      month={july},
      doi={10.56553/popets-2024-0004}
}

---

Getting Started

• Clone this topics_api_analysis repository and the topics_classifier submodule at once with:

  • git clone --recurse-submodules git@github.com:yohhaan/topics_api_analysis.git (SSH)
  • git clone --recurse-submodules https://github.com/yohhaan/topics_api_analysis.git (HTTPS)

• Note: the .devcontainer/ directory contains the config for integration with VS Code (see guide here).

• Then, follow either set of instructions (or install dependencies manually):

Using the Docker image from the Container Registry

This GitHub workflow automatically builds and pushes the Docker image to GitHub's Container Registry when the Dockerfile or the requirements.txt files are modified.

1. Pull the Docker image:

docker pull ghcr.io/yohhaan/topics_api_analysis:main

2. Launch the Docker container, attach the current working directory (i.e., run from the root of the cloned git repository) as a volume, set the context to be that volume, and provide an interactive bash terminal:

docker run --rm -it -v ${PWD}:/workspaces/topics_api_analysis \
    -w /workspaces/topics_api_analysis \
    --entrypoint bash ghcr.io/yohhaan/topics_api_analysis:main

Using a locally built Docker image

1. Build the Docker image:

docker build -t topics_api_analysis:main .

2. Launch the Docker container, attach the current working directory (i.e., run from the root of the cloned git repository) as a volume, set the context to be that volume, and provide an interactive bash terminal:

docker run --rm -it -v ${PWD}:/workspaces/topics_api_analysis \
    -w /workspaces/topics_api_analysis \
    --entrypoint bash topics_api_analysis:main

Reproduction Steps

Topics classification: refer to and execute the bash scripts in the corresponding folder under ./data to classify the different datasets with the Topics API:

• CrUX: cd data/crux && ./crux.sh
• Tranco: cd data/tranco && ./tranco.sh
• Real Browsing Histories: cd data/web_data && ./web_data.sh

Topics evaluation: refer to the topics_simulator.py script to evaluate the Topics API (simulation of the API for users, denoising, and re-identification across epochs)

usage: python3 topics_simulator.py [-h]
                                   users_topics_tsv nb_epochs config_model_json top_list_tsv
                                   unobserved_topics_threshold repeat_each_user_n_times output_prefix

Simulate the Topics API and evaluate its privacy guarantees

positional arguments:
  users_topics_tsv
  nb_epochs
  config_model_json
  top_list_tsv
  unobserved_topics_threshold
  repeat_each_user_n_times
  output_prefix

Examples:

• python3 topics_simulator.py data/web_data/users_topics_5_weeks.tsv 5 topics_classifier/chrome5/config.json data/crux/crux_202406_chrome5_topics-api.tsv 10 1 data/reidentification_exp/5_weeks_10_unobserved
• python3 topics_simulator.py data/web_data/users_topics_5_weeks.tsv 5 topics_classifier/chrome5/config.json data/crux/crux_202406_chrome5_topics-api.tsv 10 100 data/denoise_exp/5_weeks_100_repetitions_10_unobserved

Analysis: to extract statistics and plot the figures, refer to the analysis.py script.