Summary of Alison: Fast and Effective Stylometric Authorship Obfuscation, by Eric Xing et al.
ALISON: Fast and Effective Stylometric Authorship Obfuscation
by Eric Xing, Saranya Venkatraman, Thai Le, Dongwon Lee
First submitted to arxiv on: 1 Feb 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary This paper proposes a new method for Authorship Obfuscation (AO), called ALISON, which significantly reduces the training and obfuscation time compared to state-of-the-art (SOTA) AO methods. The proposed approach achieves better obfuscation success by attacking transformer-based AA models on two benchmark datasets, typically performing 15% better than competing methods. Additionally, ALISON does not require direct signals from a target AA classifier during obfuscation and utilizes unique stylometric features for explainable obfuscation. The paper also demonstrates the effectiveness of ALISON in preventing SOTA AA methods from accurately determining authorship of ChatGPT-generated texts while minimizing changes to original text semantics. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary ALISON is a new method that makes it harder for AI models to figure out who wrote something. It does this by changing the writing style, but not so much that the meaning of what was written gets lost. This method is special because it’s fast and good at hiding the author’s identity. In fact, it’s better than other methods like this, which often take a long time to work. The authors also tested ALISON on texts generated by ChatGPT, and it was able to keep the AI models from guessing who wrote them. |
Keywords
* Artificial intelligence * Semantics * Transformer
