WPG-MoE: Weak-Prior-Guided Dense Mixture-of-Experts for User-Level Social Media Depression Detection

2026-07-05

AI Summary

This work addresses the limitations of existing social media–based depression detection methods, which rely on single classifiers and struggle with user expression heterogeneity, often misclassifying individuals who do not explicitly self-disclose symptoms. To overcome this, the authors propose the WPG-MoE framework, which constructs a dense mixture-of-experts architecture built upon a shared large language model. It incorporates a user-level weak semantic prior to guide the routing mechanism, directing distinct evidence patterns to specialized experts for personalized risk assessment. Furthermore, the framework adopts the Learning Using Privileged Information (LUPI) paradigm to decouple training and inference processes. Experimental results demonstrate that WPG-MoE significantly outperforms strong baselines on both English and Chinese datasets, exhibits interpretable routing behavior, and notably improves detection accuracy for non–self-disclosing users.

Abstract

Online social media posts provide scalable signals for early depression screening, and recent studies mainly improve pre-classification evidence through risk-post selection, symptom grounding, and clinically informed feature construction. However, these screening-stage designs often leave final decisions to a single detector, overlooking how users heterogeneously express depressive risk after screening. A monolithic classifier must average across heterogeneous users, which may dilute localized evidence and cause misclassification, especially for non-self-disclosing users. To address this issue, we propose WPG-MoE, a weak-prior-guided dense mixture-of-experts framework built on a shared large language model (LLM) backbone. WPG-MoE derives user-level weak semantic priors to softly route users to experts matched to different evidence layouts. We formulate this process as learning using privileged information (LUPI): rich LLM-extracted structured evidence guides training-time routing, while inference retains only Patient Health Questionnaire-9 (PHQ-9) template screening and the deployable backbone. Experiments on Chinese and English datasets show that WPG-MoE outperforms strong baselines with interpretable routing behavior.

Problem

• depression detection
• user heterogeneity
• social media
• misclassification
• weak priors

Innovation

• Mixture-of-Experts
• Weak-Prior Guidance
• Privileged Information
• User-Level Heterogeneity
• Depression Detection