Explainable Disentangled Representation Learning for Generalizable Authorship Attribution in the Era of Generative AI

2026-04-23

AI Summary

Existing authorship attribution methods suffer from limited generalization in cross-domain scenarios due to the entanglement of content and stylistic features. To address this, this work proposes EAVAE, a novel framework that explicitly disentangles textual style from content—a first for this task—by integrating dedicated encoders, a variational autoencoder, and supervised contrastive learning. Furthermore, an interpretable discriminator is introduced to jointly enhance model performance and explainability. The proposed method achieves state-of-the-art results on multiple benchmark datasets, including Amazon Reviews, PAN21, and HRS, and demonstrates strong performance in few-shot detection of AI-generated text on the M4 dataset.

Technology Category

Natural Language Processing: Sentiment Analysis, Stylistic Analysis, and Argument MiningMachine Learning: Deep Generative Models & AutoencodersComputer Vision: Generative Adversarial Networks (GANs) for Vision

Application Category

Web Mining and Content Analysis: Large pretrained models with web dataSearch and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for rankingEconomics, Online Markets and Human Computation: Trust and reliance of crowd workers and data experts on GenAI

Abstract

Learning robust representations of authorial style is crucial for authorship attribution and AI-generated text detection. However, existing methods often struggle with content-style entanglement, where models learn spurious correlations between authors' writing styles and topics, leading to poor generalization across domains. To address this challenge, we propose Explainable Authorship Variational Autoencoder (EAVAE), a novel framework that explicitly disentangles style from content through architectural separation-by-design. EAVAE first pretrains style encoders using supervised contrastive learning on diverse authorship data, then finetunes with a Variational Autoencoder (VEA) architecture using separate encoders for style and content representations. Disentanglement is enforced through a novel discriminator that not only distinguishes whether pairs of style/content representations belong to the same or different authors/content sources, but also generates natural language explanation for their decision, simultaneously mitigating confounding information and enhancing interpretability. Extensive experiments demonstrate the effectiveness of EAVAE. On authorship attribution, we achieve state-of-the-art performance on various datasets, including Amazon Reviews, PAN21, and HRS. For AI-generated text detection, EAVAE excels in few-shot learning over the M4 dataset. Code and data repositories are available online\footnote{https://github.com/hieum98/avae} \footnote{https://huggingface.co/collections/Hieuman/document-level-authorship-datasets}.

Problem

• authorship attribution
• style-content entanglement
• generalization
• disentangled representation
• generative AI

Innovation

• disentangled representation learning
• explainable AI
• authorship attribution
• generative AI detection
• contrastive learning