On the Vulnerability of Text Sanitization

This repo is the official python implementation for the paper On the Vulnerability of Text Sanitization (NAACL' 25, Oral).

Setup Environment

Step 1 Install required packages

conda create -n RT python=3.10
conda activate RT
git clone https://github.com/mengtong0110/On-the-Vulnerability-of-Text-Sanitization
cd On-the-Vulnerability-of-Text-Sanitization
pip install -r requirements.txt

Step 2 Install CUSTEXT+ and SANTEXT+

Please refer to the repositories for the files of text sanitization in experiments: CUSTEXT+ and SANTEXT+.

Reconstruction Attacks

You can start the attcks against the CUSTEXT+ using the commands:

conda create -n RT python=3.10
conda activate RT
cd CUSTEXT+
./ pipline.ps1

Also, you can start the attcks against the SANTEXT+ using the commands:

conda create -n RT python=3.10
conda activate RT
cd SANTEXT+
./ pipline.ps1

Citation

If you find this repository useful for your work, please consider citing it as follows:

@inproceedings{tong2025vulnerability,
    title = "On the Vulnerability of Text Sanitization",
    author={Tong, Meng and Chen, Kejiang and Yuan, Xiaojian and Liu, Jiayang and Zhang, Weiming and Yu, Nenghai and Zhang, Jie},
    booktitle = "Proceedings of the 2025 Conference of the North {A}merican Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers)",
    year = "2025",
    publisher = "Association for Computational Linguistics",
}