XAM: Interactive Explainability for Authorship Attribution Models

2025-12-07

AI Summary

To address the weak interpretability and reliance on predefined stylistic labels in embedded author attribution models, this paper proposes IXAM, an interactive explanation framework. IXAM enables multi-granularity exploration—lexical, sentential, and document-level—within the embedding space, allowing users to dynamically construct prediction explanations grounded in actual writing features, thereby overcoming limitations of static rules or fixed stylistic dimensions. By integrating explainable AI (XAI) techniques with interactive visualization, IXAM supports user-driven explanation generation and validation. A user study demonstrates that IXAM significantly improves explanation credibility, depth of understanding, and interaction satisfaction (p < 0.01). This work establishes a novel explanation paradigm for black-box text classification models, characterized by flexibility, transparency, and practical usability.

Technology Category

• Humans and AI: Explainable AI (XAI) for Human Understanding
• Machine Learning: Transparent, Interpretable, Explainable ML
• Natural Language Processing: Interpretability, Analysis, and Evaluation of NLP Models

Application Category

• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

We present IXAM, an Interactive eXplainability framework for Authorship Attribution Models. Given an authorship attribution (AA) task and an embedding-based AA model, our tool enables users to interactively explore the model's embedding space and construct an explanation of the model's prediction as a set of writing style features at different levels of granularity. Through a user evaluation, we demonstrate the value of our framework compared to predefined stylistic explanations.

Problem

• Develops an interactive framework for explaining authorship attribution models
• Enables exploration of embedding spaces and style feature explanations
• Compares interactive explanations with predefined stylistic analysis methods

Innovation

• Interactive exploration of embedding space for authorship attribution
• Constructs explanations as multi-granularity writing style features
• User evaluation shows value over predefined stylistic explanations