AgentSociety

AgentSociety

by Tsinghua University
AgentSociety is a social simulator developed by Tsinghua University, using large language models (LLM) to simulate complex social behaviors and phenomena through human-like intelligent agents.

What is AgentSociety?

AgentSociety is a social simulator developed by Tsinghua University, leveraging large language models (LLM) to create intelligent agents with human-like minds. These agents are endowed with emotions, needs, and cognitive abilities, enabling them to simulate complex social behaviors in urban environments.

Main Features of AgentSociety

  • LLM-Driven Social Agents: Constructs intelligent agents with human-like minds, endowing them with emotions, needs, motivations, and cognitive abilities.
  • Realistic Urban Social Environment Simulation: Accurately simulates urban spaces, including transportation, infrastructure, and public resources.
  • Large-Scale Social Simulation Engine: Employs an asynchronous simulation architecture and the Ray distributed computing framework for efficient, scalable agent interaction.
  • Intelligent Social Science Research Toolbox: Provides tools for sociological research methods such as experiments, interviews, and questionnaires.
  • Real-Time Interactive Visualization: Offers a real-time interface for monitoring and interacting with agents during experiments.

Technical Principles of AgentSociety

  • Mental Level: Agents have stable individual profiles and dynamic personal states, ensuring personalized behavior patterns.
  • Mind-Behavior Coupling: Agent behavior is driven by emotions, needs, and cognition, based on Maslow's hierarchy of needs and the theory of planned behavior.
  • Behavioral Level: Agents perform simple and complex social behaviors, dynamically adjusting based on environmental feedback.
  • Urban Space: Simulates urban road networks, areas of interest (AOI), and points of interest (POI).
  • Social Space: Supports online and offline social interactions, simulating dynamic changes in social networks.
  • Economic Space: Simulates macroeconomic activities, including employment, consumption, taxation, and interest mechanisms.
  • Asynchronous Simulation Architecture: Each agent acts as an independent simulation unit, exchanging information through a messaging system.
  • Distributed Computing: Based on the Ray framework and Python's asyncio mechanism, supporting distributed cluster expansion.
  • MQTT Communication Protocol: Supports high-concurrency, low-latency message transmission among large-scale agents.

Project Address of AgentSociety

Application Scenarios of AgentSociety

  • Social Opinion Propagation: Simulating the spread and impact of information in social networks.
  • Public Policy Response: Evaluating the impact of policies on individual and group behaviors.
  • Social Polarization: Studying the mechanisms of opinion divergence and the formation of opposing camps.
  • Natural Disaster Response: Simulating crowd behavior and social dynamics under extreme events.

Framework Features

Supported Tasks
Social Behavior Simulation Policy Testing Crisis Warning Social Science Research Urban Environment Modeling
Tags
Social Simulation LLM AI Agents Urban Environment Sociology Distributed Computing Behavioral Modeling Policy Testing Crisis Simulation Research Tool

Getting Started

Pricing
free

Screenshots & Images

Primary Screenshot
Additional Images
View Repository View Demo Documentation

Stats

0 Views
0 Favorites
249 GitHub Stars

Community & Support

GitHub Repository

Similar Frameworks

TPO
0
Phantom by ByteDance
0
DualPipe by DeepSeek
0