Integrating Multi-view Analysis: Multi-view Mixture-of-Expert for Textual Personality Detection

TL;DR

MvP addresses the challenge of detecting personality traits from text by integrating multiple perspectives through a Multi-view Mixture-of-Experts (MoE) framework and a novel User Consistency Regularization. Post representations are enhanced via learnable Parameter Whitening, and a gating mechanism fuses multiple views into a cohesive multi-view user representation for MBTI trait prediction. A multi-task objective combines supervised detection with self-supervised consistency constraints, validated on Kaggle and Pandora MBTI datasets, where MvP outperforms strong baselines and ablations highlight the importance of both the MoE and regularization. The approach demonstrates that multi-view modeling of user posts captures richer signals than single-view methods and can inform downstream tasks requiring granular personality profiling. Practical implications include more accurate user modeling for personalization and targeted content generation across social platforms.

Abstract

Textual personality detection aims to identify personality traits by analyzing user-generated content. To achieve this effectively, it is essential to thoroughly examine user-generated content from various perspectives. However, previous studies have struggled with automatically extracting and effectively integrating information from multiple perspectives, thereby limiting their performance on personality detection. To address these challenges, we propose the Multi-view Mixture-of-Experts Model for Textual Personality Detection (MvP). MvP introduces a Multi-view Mixture-of-Experts (MoE) network to automatically analyze user posts from various perspectives. Additionally, it employs User Consistency Regularization to mitigate conflicts among different perspectives and learn a multi-view generic user representation. The model's training is optimized via a multi-task joint learning strategy that balances supervised personality detection with self-supervised user consistency constraints. Experimental results on two widely-used personality detection datasets demonstrate the effectiveness of the MvP model and the benefits of automatically analyzing user posts from diverse perspectives for textual personality detection.

Figures (3)

• Figure 1: The architecture of the MvP model. The left side illustrates the overall framework of MvP, while the right side depicts the structure of Multi-view MoE.
• Figure 2: Performance curves for different number of views.
• Figure 3: Performance curves for different trade-off parameters.