PersonaAct: Simulating Short-Video Users with Personalized Agents for Counterfactual Filter Bubble Auditing

2026-01-30

AI Summary

This study addresses the challenge of auditing personalized recommendation systems on short-video platforms, which often trap users in information cocoons yet resist large-scale, privacy-preserving evaluation. To overcome this, the authors propose a multimodal agent framework that integrates real user behavioral traces with automated structured interviews, enabling the first interpretable and high-fidelity synthetic user profiling. They also introduce the first open-source multimodal short-video dataset to support reproducible recommender system audits. Through supervised fine-tuning and reinforcement learning, the agent significantly outperforms general-purpose large language model baselines in behavioral simulation fidelity. Empirical analysis reveals substantial content narrowing during user interactions, while Bilibili demonstrates the strongest potential for breaking users out of information cocoons.

Technology Category

• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval
• Machine Learning: Large Multimodal Models (LMMs)
• Multiagent Systems: Agent-Based Simulation and Emergent Behavior

Application Category

• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Economics, Online Markets and Human Computation: Economics and fairness of platforms and recommendation systems

Abstract

Short-video platforms rely on personalized recommendation, raising concerns about filter bubbles that narrow content exposure. Auditing such phenomena at scale is challenging because real user studies are costly and privacy-sensitive, and existing simulators fail to reproduce realistic behaviors due to their reliance on textual signals and weak personalization. We propose PersonaAct, a framework for simulating short-video users with persona-conditioned multimodal agents trained on real behavioral traces for auditing filter bubbles in breadth and depth. PersonaAct synthesizes interpretable personas through automated interviews combining behavioral analysis with structured questioning, then trains agents on multimodal observations using supervised fine-tuning and reinforcement learning. We deploy trained agents for filter bubble auditing and evaluate bubble breadth via content diversity and bubble depth via escape potential. The evaluation demonstrates substantial improvements in fidelity over generic LLM baselines, enabling realistic behavior reproduction. Results reveal significant content narrowing over interaction. However, we find that Bilibili demonstrates the strongest escape potential. We release the first open multimodal short-video dataset and code to support reproducible auditing of recommender systems.

Problem

• filter bubbles
• short-video platforms
• personalized recommendation
• user simulation
• content diversity

Innovation

• personalized agents
• multimodal simulation
• filter bubble auditing
• persona synthesis
• reinforcement learning