Adversarial stylometry: Circumventing authorship recognition to preserve privacy and anonymity

By Michael Brennan, Sadia Afroz, Rachel Greenstadt. 2012

228 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Computer Vision and Pattern Recognition

Citations in the corpus

Practical Evasion of a Learning-Based Classifier: A Case Study (2014, methodology): “The essential components comprising every learning-based classifier system are: • the set of features used by the classifier • the training dataset used for classifier training • the classification algorithm with its parameters.”
Auditing Data Provenance in Text-Generation Models (2018, methodology): “nces (which would not be preserved by obfuscation) rather than higher-level linguistic features (which might be). We use an obfuscation technique, previously considered for evading author attribution [4], that machine-translates the text to a different language and back. We obfuscate the training an”
Towards Making Systems Forget with Machine Unlearning (2015, background): “They are widely used in many areas of science [25, 35, 37, 46, 55, 63–65].”
All You Need is "Love": Evading Hate Speech Detection (2018, background): “fooling sentiment analysis, author anonymization [2, 3], or avoiding content-based text classification to escape censorship.”
Small Language Models are the Future of Agentic AI (2025, unknown): “Application specific inputs (e.g. legal or internal documents) can be often be automatically paraphrased to obfuscate named entities and numerical details without compromising the general information content [9, 76, 73].”
The Limitations of Stylometry for Detecting Machine-Generated Fake News (2019, background): “While these classiﬁers could be fooled by an aware writer that intentionally imitates other’s style (Brennan, Afroz, and Greenstadt 2012), this approach was found useful for de-anonymizing cybercriminals in forums (Afroz et al. 2014), identifying programmers (Caliskan-Islam et al. 2015) and more…”
The Natural Auditor: How To Tell If Someone Used Your Words To Train Their Model (2018, methodology): “We use a speciﬁc obfuscation technique, previously considered in the context of author attribution and evasion thereof [6], that machine-translates the text to a different language and back.”
Obfuscating Gender in Social Media Writing (2016, background): “The most relevant existing work is that of Brennan et al. (2012) who explore the related problem of modifying text to defeat authorship detectors.”
Topic or Style? Exploring the Most Useful Features for Authorship Attribution (2018, background): “Previous work on feature exploration for authorship at-tribution, focused on the overall effectiveness of features without considering the characteristics of the datasets to which they were applied, e.g. (Grieve, 2007; Guthrie, 2008; Stamatatos, 2009; Brennan et al., 2012; Sapkota et al., 2015).”
Mobile Game Evaluation Method Based on Data Mining of Affective Time Series (2025, unknown): “Below, we describe the founding concepts of the method, which were jointly developed by the research team with the experts from the gaming company we worked with.”