Abstract: Massive social simulation plays a vital role in predicting real-world trends. Previous studies use Large Language Models (LLMs) to replace traditional methods to enrich the scenarios and improve the simulation accuracy. However, they are faced with limitations such as rigid frameworks, small-scale simulations, and narrow evaluation criteria. To this end, we introduce Simulate Anything: a generalized social simulation framework driven by LLM agents, which is composed of a 10-million-user real-world pool, a demographic distribution sampling strategy, and a unified simulation evaluation method. We evaluate the framework by conducting massive simulations under political, journalistic, and economic scenarios. The results prove that our framework can support diverse and trustworthy massive social simulations with a standard pipeline and minimal changes. Upon acceptance, we will release all three simulations with the corresponding user pool.
Paper Type: Long
Research Area: Computational Social Science and Cultural Analytics
Research Area Keywords: Social Simulation, LLM agents, Large Language Models
Contribution Types: NLP engineering experiment, Data resources
Languages Studied: English, Chinese
Submission Number: 7561