Mitigating Hallucination in Fictional Character Role-Play

TL;DR

This paper addresses hallucination in fictional character role-play by introducing a new dataset and a method called RoleFact, which reduces hallucinations and improves factual accuracy in generated responses.

Contribution

The paper presents a novel dataset for evaluating hallucination in role-play and introduces RoleFact, a method that reduces hallucination by adjusting the influence of parametric knowledge.

Findings

18% improvement in factual precision for adversarial questions

44% reduction in temporal hallucination

Effective mitigation of hallucination in role-playing scenarios

Abstract

Role-playing has wide-ranging applications in customer support, embodied agents, and computational social science. The influence of parametric world knowledge of large language models (LLMs) often causes role-playing characters to act out of character and to hallucinate about things outside the scope of their knowledge. In this work, we focus on the evaluation and mitigation of hallucination in fictional character role-play. We introduce a dataset with over 2,000 characters and 72,000 interviews, including 18,000 adversarial questions. We propose RoleFact, a role-playing method that mitigates hallucination by modulating the influence of parametric knowledge using a pre-calibrated confidence threshold. Experiments show that the proposed method improves the factual precision of generated responses by 18% for adversarial questions with a 44% reduction in temporal hallucination for time-sensitive interviews. The code and the dataset are available at https://github.com/NafisSadeq/rolefact.git.