MAEPS: Multi-agent Event Prediction System Based on Human Expert Team Collaboration Simulation

Event prediction (EP), the accurate forecasting of future events, is vital for strategic planning and risk management in both governmental and business contexts. The rapid advancement of large language models (LLMs) has positioned AI-based automated prediction methods at the forefront of academic and industrial research. However, current LLM prediction systems exhibit several shortcomings. Firstly, their information retrieval mostly searches based on the question itself, failing to gather relevant data from multiple perspectives as human expert teams do. Secondly, their temporal analysis is inadequate, as the collected information often includes subjective opinions or speculations and lacks the ability to reconcile contradictory information across different time points during real-time prediction. To address these issues this paper introduces MAEPS (Multi-Agent Event Prediction System), which emulates the collaborative efforts of human expert teams through 12 specialized agents. Each agent collects data from a specific professional dimension. The system automatically identifies and resolves conflicting information, ensuring that predictions prioritize recent and consistent facts. Experiments on EP datasets from real prediction platforms demonstrate that MAEPS significantly outperforms existing LLM prediction systems by 7% in accuracy, thereby validating the efficacy of simulating expert team collaboration for prediction purposes.