The Case for Being Average: A Mediocrity Approach to Style Masking and Author Obfuscation

Abstract

Users posting online expect to remain anonymous unless they have logged in, which is often needed for them to be able to discuss freely on various topics. Preserving the anonymity of a text's writer can be also important in some other contexts, e.g., in the case of witness protection or anonymity programs. However, each person has his/her own style of writing, which can be analyzed using stylometry, and as a result, the true identity of the author of a piece of text can be revealed even if s/he has tried to hide it. Thus, it could be helpful to design automatic tools that can help a person obfuscate his/her identity when writing text. In particular, here we propose an approach that changes the text, so that it is pushed towards average values for some general stylometric characteristics, thus making the use of these characteristics less discriminative. The approach consists of three main steps: first, we calculate the values for some popular stylometric metrics that can indicate authorship; then we apply various transformations to the text, so that these metrics are adjusted towards the average level, while preserving the semantics and the soundness of the text; and finally, we add random noise. This approach turned out to be very efficient, and yielded the best performance on the Author Obfuscation task at the PAN-2016 competition.

A Girl Has No Name: Automated Authorship Obfuscation using Mutant-X

“…McDonald et al [29] Manual incorporation of suggested changes * Karadzhov et al [26] Target specific features Mansoorizadeh et al [28] Replacement with synonyms Keswani et al [27] Round trip translation Shetty et al [43] Generative Adverserial Networks (GANs)…”

“…More specifically, they replaced a word with one of its synonyms obtained from WordNet [33]. Karadzhov et al [26], from PAN 2016, transformed stylometric features to their average values using different rules such as sentence splitting, stop words removal, and spelling correction to obfuscate text. The authors reported that splitting sentences and removal of certain words helps with obfuscation but significantly hurts the sensibility of the obfuscated text to a human reviewer.…”

“…Other automated approaches (e.g. Karadzhov et al [26], Mansoorizadeh et al [28]) use predefined rules. While these have the advantage that they do not require knowledge of the adversary's attribution tool, they are unable to effectively evade authorship attribution classifiers and/or cannot preserve the semantics.…”

Generalised Differential Privacy for Text Document Processing

“…This approach was used in Anonymouth [35], a semi-automated tool that provides feedback to authors on which features to modify to effectively anonymise their texts. A similar approach was also followed by Karadhov et al [23] as part of the PAN@Clef 2017 task.…”

Recursive Style Breach Detection with Multifaceted Ensemble Learning

“…Authorship attribution Previous work on authorship attribution and related problems (e.g., author obfuscation [3,4,12,15]) used primarily term frequencies [5,9] and features from stylometry [8,9,18]. We borrowed ideas for traditional features from [2,13], but we also designed some new ones, related to tautology, grammar contractions, quote use discrepancies, beginning and ending author statement words, and named entity spellings (see Section 3.4).…”