Beyond the Final Actor: Modeling the Dual Roles of Creator and Editor for Fine-Grained LLM-Generated Text Detection

TL;DR

This paper introduces RACE, a novel method for fine-grained detection of LLM-generated text, distinguishing creator and editor roles to better support nuanced policy enforcement.

Contribution

It proposes a new four-class detection framework and RACE method that leverages Rhetorical Structure Theory to differentiate creator and editor signatures.

Findings

RACE outperforms 12 baselines in fine-grained detection accuracy.

RACE achieves low false alarm rates in identifying nuanced text types.

The approach aligns with policy needs for regulating LLM-generated content.

Abstract

The misuse of large language models (LLMs) requires precise detection of synthetic text. Existing works mainly follow binary or ternary classification settings, which can only distinguish pure human/LLM text or collaborative text at best. This remains insufficient for the nuanced regulation, as the LLM-polished human text and humanized LLM text often trigger different policy consequences. In this paper, we explore fine-grained LLM-generated text detection under a rigorous four-class setting. To handle such complexities, we propose RACE (Rhetorical Analysis for Creator-Editor Modeling), a fine-grained detection method that characterizes the distinct signatures of creator and editor. Specifically, RACE utilizes Rhetorical Structure Theory (RST) to construct a logic graph for the creator's foundation while extracting Elementary Discourse Unit (EDU)-level features for the editor's style. Experiments show that RACE outperforms 12 baselines in identifying fine-grained types with low false alarms, offering a policy-aligned solution for LLM regulation.