Text-based Personality Prediction Using Large Language Models

This paper addresses the task of text-based personality detection using Large Language Models (LLMs). The study focuses primarily on exploring the text-based detection of extraversion, a trait represented in both the Big Five personality traits and MBTI models. The study utilized three datasets that included annotations for different personality models and encompassed data from various sources. We examined both the performance metrics evaluating the LLMs and the models’ decision-making processes while applying various prompt engineering methods.