Persona-Based Synthetic Data Generation Using Multi-Stage Conditioning with Large Language Models for Emotion Recognition

TL;DR

This paper presents PersonaGen, a multi-stage persona-based framework using large language models to generate diverse, realistic emotional text for improving emotion recognition models, addressing data scarcity issues.

Contribution

Introducing PersonaGen, a novel multi-stage conditioning approach with large language models for synthesizing emotionally rich, diverse text data based on layered virtual personas.

Findings

PersonaGen outperforms baselines in diversity and realism.

Synthetic data improves emotion classification accuracy.

Framework effectively captures subjective emotional expressions.

Abstract

In the field of emotion recognition, the development of high-performance models remains a challenge due to the scarcity of high-quality, diverse emotional datasets. Emotional expressions are inherently subjective, shaped by individual personality traits, socio-cultural backgrounds, and contextual factors, making large-scale, generalizable data collection both ethically and practically difficult. To address this issue, we introduce PersonaGen, a novel framework for generating emotionally rich text using a Large Language Model (LLM) through multi-stage persona-based conditioning. PersonaGen constructs layered virtual personas by combining demographic attributes, socio-cultural backgrounds, and detailed situational contexts, which are then used to guide emotion expression generation. We conduct comprehensive evaluations of the generated synthetic data, assessing semantic diversity through clustering and distributional metrics, human-likeness via LLM-based quality scoring, realism through comparison with real-world emotion corpora, and practical utility in downstream emotion classification tasks. Experimental results show that PersonaGen significantly outperforms baseline methods in generating diverse, coherent, and discriminative emotion expressions, demonstrating its potential as a robust alternative for augmenting or replacing real-world emotional datasets.