InstructAV: Instruction Fine-tuning Large Language Models for Authorship Verification

TL;DR

This work tackles the gap in authorship verification (AV) by introducing InstructAV, which integrates instruction-tuning with parameter-efficient fine-tuning (PEFT) to simultaneously achieve high AV accuracy and produce transparent linguistic explanations. The framework collects explanation data, enforces consistency between labels and explanations, and applies LoRA-based fine-tuning to align predictions with robust rationales. Across IMDB, Twitter, and Yelp AV datasets, InstructAV demonstrates state-of-the-art performance and superior explanation quality, with substantial gains when explanation data is used for training. The approach enhances both the reliability and interpretability of AV systems, offering practical benefits for forensics, digital security, and textual analysis while highlighting avenues for improving efficiency and evaluation methodologies.

Abstract

Large Language Models (LLMs) have demonstrated remarkable proficiency in a wide range of NLP tasks. However, when it comes to authorship verification (AV) tasks, which involve determining whether two given texts share the same authorship, even advanced models like ChatGPT exhibit notable limitations. This paper introduces a novel approach, termed InstructAV, for authorship verification. This approach utilizes LLMs in conjunction with a parameter-efficient fine-tuning (PEFT) method to simultaneously improve accuracy and explainability. The distinctiveness of InstructAV lies in its ability to align classification decisions with transparent and understandable explanations, representing a significant progression in the field of authorship verification. Through comprehensive experiments conducted across various datasets, InstructAV demonstrates its state-of-the-art performance on the AV task, offering high classification accuracy coupled with enhanced explanation reliability.

Figures (1)

• Figure 1: An illustration of InstructAV's architectures.