Published in Springer Lecture Notes in Computer Science, vol. 14763, pp. 291–301.
The paper examines whether LLMs can infer personality traits from text. It uses the Big Five framework, with a more detailed experimental focus on extraversion.
The research evaluates prompt-engineering strategies, ways of describing personality traits in instructions, and the effect of prompt wording on classification quality. Results are compared against traditional approaches to text-based personality prediction.