Unfair Mistakes on Social Media: How Demographic Characteristics influence Authorship Attribution

2025-10-22

AI Summary

Author attribution techniques—widely deployed in sock-puppet detection and cross-platform identity linkage—lack rigorous fairness evaluation across demographic groups, risking discriminatory misattribution. This work presents the first systematic audit of demographic bias in author attribution across gender, native language, and age dimensions, under both closed-world and open-world settings. Through controlled-variable experiments, we quantify misattribution probabilities across population subgroups. Results reveal no significant performance disparity in closed-world tasks; however, under open-world conditions—where the true author is absent from the candidate set—misattributions exhibit strong demographic homophily, disproportionately assigning authorship to individuals sharing demographic traits with the true author. Moreover, errors concentrate on stylistically similar texts, exposing a critical limitation of closed-world fairness assessments. Our findings uncover a previously unrecognized implicit bias mechanism in author attribution models, providing empirical evidence essential for developing socially robust text provenance systems.

Technology Category

• Natural Language Processing: Ethics — Bias, Fairness, Transparency & Privacy
• Computer Vision: Bias, Fairness & Privacy
• Machine Learning: Ethics, Bias, and Fairness

Application Category

• Social Networks and Social Media: Fairness and bias in social network and social media analysis
• Web Mining and Content Analysis: Web data provenance, reliability, and authenticity
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Authorship attribution techniques are increasingly being used in online contexts such as sock puppet detection, malicious account linking, and cross-platform account linking. Yet, it is unknown whether these models perform equitably across different demographic groups. Bias in such techniques could lead to false accusations, account banning, and privacy violations disproportionately impacting users from certain demographics. In this paper, we systematically audit authorship attribution for bias with respect to gender, native language, and age. We evaluate fairness in 3 ways. First, we evaluate how the proportion of users with a certain demographic characteristic impacts the overall classifier performance. Second, we evaluate if a user's demographic characteristics influence the probability that their texts are misclassified. Our analysis indicates that authorship attribution does not demonstrate bias across demographic groups in the closed-world setting. Third, we evaluate the types of errors that occur when the true author is removed from the suspect set, thereby forcing the classifier to choose an incorrect author. Unlike the first two settings, this analysis demonstrates a tendency to attribute authorship to users who share the same demographic characteristic as the true author. Crucially, these errors do not only include texts that deviate from a user's usual style, but also those that are very close to the author's average. Our results highlight that though a model may appear fair in the closed-world setting for a performant classifier, this does not guarantee fairness when errors are inevitable.

Problem

• Auditing authorship attribution bias across gender, language, and age demographics
• Evaluating fairness when classifiers misattribute texts to incorrect authors
• Identifying demographic-based errors in open-world authorship attribution scenarios

Innovation

• Auditing authorship attribution for demographic bias
• Evaluating fairness through misclassification probability analysis
• Identifying demographic-based errors in open-world scenarios