Fine-Tuned Multi-Agent Framework for Detecting OCEAN in Life Narratives

2026-07-13

AI Summary

This work addresses the challenge of inferring the Big Five personality traits (OCEAN) from lengthy life narratives, where trait expressions are often implicit, context-dependent, and susceptible to pretraining biases in large language models, leading to unstable predictions. To overcome these limitations, the authors propose a fine-tuned multi-agent framework that trains multiple sub-agents—each specialized to model a specific personality dimension from high, low, or neutral perspectives—using masked language modeling augmented with psychometric supervision. A referee large language model then aggregates these multi-perspective outputs to produce a final, coherent prediction. This approach not only enhances interpretability but also effectively mitigates model bias, significantly outperforming baseline and ablation models on a life narrative dataset in both accuracy and prediction stability.

Abstract

Accurately assessing personality from text is challenging because traits are latent, context-dependent, and often subtly expressed across long narratives. Large language models (LLMs) offer new opportunities by processing extensive textual contexts, but pretraining of these models can induce latent "personality-like" biases, making single-model inferences inconsistent. We propose a fine-tuned multi-agent framework for detecting OCEAN personality traits, in which sub-agents are conditioned to adopt high, low, or neutral perspectives for each trait through masked language modeling (MLM) and psychometric supervision. A judge LLM aggregates and compares sub-agent outputs to generate final trait predictions, capturing multiple complementary perspectives while mitigating individual model biases. We evaluate the framework on life narrative dataset through quantitative and qualitative experiments, including baselines, ablations, and inference quality analyses. Our approach offers a scalable and interpretable method for text-based personality inference, highlighting the benefits of multi-agent reasoning grounded in psychometric supervision.

Problem

• personality inference
• OCEAN traits
• life narratives
• latent traits
• LLM bias

Innovation

• multi-agent framework
• OCEAN personality traits
• psychometric supervision
• masked language modeling
• personality bias mitigation