This repository contains code for the Authorship Obfuscation tool called "Mutant-X" presented in PoPETs 2019 (https://petsymposium.org/2019/files/papers/issue4/popets-2019-0058.pdf).
In order to run the code, you need to fulfill following requirements:
- Install Python3
- Install Java
- Download pycocoevalcap folder from (https://www.dropbox.com/sh/vwyjxmnbr9ytglx/AAAfkvJr8GlDv2ut-f1lmnGsa?dl=0) and put it inside the main folder. This is used for calculation of METEOR score. Original code can be found at (https://github.com/tylin/coco-caption/tree/master/pycocoevalcap)
- Download Word Embeddings folder from (https://www.dropbox.com/sh/y3srrf82n9jbx8x/AAAlHlICEftupAJ3WZnS8W3Aa?dl=0) and also place it inside the main project folder.
- Install libraries required for this project using the following command:
pip3 install -r requirements.txt
In order to create the obfuscated document using writeprintsRFC Mutant-X, run Obfuscator.py in the following way.
python3 Obfuscator.py
Following is a list of parameters along-with their explanation.
| Parameter | Description |
|---|---|
| generation | Number of documents to be generated per document |
| topK | Top K highest fitness selection |
| crossover | Crossover probability |
| iterations | Maximum number of iterations |
| alpha | weight assigned to probability in fitness function |
| beta | weight assigned to METEOR in fitness function |
| replacements | percentage of document to change |
| documentName | Name of document for obfuscation |
Running this script generates a document named Obfuscated_text.txt which contains the obfuscated text for the input document.
Please use the following citation when using the code.
@article{mahmood2019mutantx,
title={A Girl Has No Name: Automated Authorship Obfuscation using Mutant-X},
author={Mahmood, Asad and Ahmad, Faizan and Shafiq, Zubair and Srinivasan, Padmini and Zaffar, Fareed},
journal={Proceedings on Privacy Enhancing Technologies},
volume={2019},
number={4},
pages={54--71},
year={2019},
publisher={Sciendo}
}