VicSim: Enhancing Victim Simulation with Emotional and Linguistic Fidelity

TL;DR

VicSim is a novel victim simulation model that enhances realism in scenario-based training by integrating emotional, linguistic, and informational fidelity through adversarial training, outperforming GPT-4 in human-likeness.

Contribution

Introduces VicSim, a victim simulation model combining GAN-based training and key-information prompting to improve realism in scenario-based training.

Findings

VicSim outperforms GPT-4 in human-likeness according to human raters.

The model effectively integrates emotional and linguistic cues for more realistic victim simulation.

Adversarial training enhances the detection of grammar and emotional cues in synthetic content.

Abstract

Scenario-based training has been widely adopted in many public service sectors. Recent advancements in Large Language Models (LLMs) have shown promise in simulating diverse personas to create these training scenarios. However, little is known about how LLMs can be developed to simulate victims for scenario-based training purposes. In this paper, we introduce VicSim (victim simulator), a novel model that addresses three key dimensions of user simulation: informational faithfulness, emotional dynamics, and language style (e.g., grammar usage). We pioneer the integration of scenario-based victim modeling with GAN-based training workflow and key-information-based prompting, aiming to enhance the realism of simulated victims. Our adversarial training approach teaches the discriminator to recognize grammar and emotional cues as reliable indicators of synthetic content. According to evaluations by human raters, the VicSim model outperforms GPT-4 in terms of human-likeness.