Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Casevo: A Cognitive Agents and Social Evolution Simulator

Zexun Jiang, Yafang Shi, Maoxu Li, Hongjiang Xiao, Yunxiao Qin, Qinglan Wei, Ye Wang, Yuan Zhang

TL;DR

Casevo presents a Mesa-based framework that uses LLM-driven cognitive agents to simulate complex social dynamics on customizable networks, with Chain of Thought reasoning, RAG memory, and memory mechanisms. It extends traditional agent-based approaches by enabling semantic reasoning and parallel request optimization for large-scale runs. The evaluation uses a 2020 US presidential debate to illustrate dynamic voter behavior, opinion diffusion, and decision-making, showing more realistic interactions than prior methods. The work contributes an open-source platform for social science research, with potential applications in public opinion dynamics, information diffusion, and election studies.

Abstract

In this paper, we introduce a multi-agent simulation framework Casevo (Cognitive Agents and Social Evolution Simulator), that integrates large language models (LLMs) to simulate complex social phenomena and decision-making processes. Casevo is designed as a discrete-event simulator driven by agents with features such as Chain of Thoughts (CoT), Retrieval-Augmented Generation (RAG), and Customizable Memory Mechanism. Casevo enables dynamic social modeling, which can support various scenarios such as social network analysis, public opinion dynamics, and behavior prediction in complex social systems. To demonstrate the effectiveness of Casevo, we utilize one of the U.S. 2020 midterm election TV debates as a simulation example. Our results show that Casevo facilitates more realistic and flexible agent interactions, improving the quality of dynamic social phenomena simulation. This work contributes to the field by providing a robust system for studying large-scale, high-fidelity social behaviors with advanced LLM-driven agents, expanding the capabilities of traditional agent-based modeling (ABM). The open-source code repository address of casevo is https://github.com/rgCASS/casevo.

Casevo: A Cognitive Agents and Social Evolution Simulator

TL;DR

Casevo presents a Mesa-based framework that uses LLM-driven cognitive agents to simulate complex social dynamics on customizable networks, with Chain of Thought reasoning, RAG memory, and memory mechanisms. It extends traditional agent-based approaches by enabling semantic reasoning and parallel request optimization for large-scale runs. The evaluation uses a 2020 US presidential debate to illustrate dynamic voter behavior, opinion diffusion, and decision-making, showing more realistic interactions than prior methods. The work contributes an open-source platform for social science research, with potential applications in public opinion dynamics, information diffusion, and election studies.

Abstract

In this paper, we introduce a multi-agent simulation framework Casevo (Cognitive Agents and Social Evolution Simulator), that integrates large language models (LLMs) to simulate complex social phenomena and decision-making processes. Casevo is designed as a discrete-event simulator driven by agents with features such as Chain of Thoughts (CoT), Retrieval-Augmented Generation (RAG), and Customizable Memory Mechanism. Casevo enables dynamic social modeling, which can support various scenarios such as social network analysis, public opinion dynamics, and behavior prediction in complex social systems. To demonstrate the effectiveness of Casevo, we utilize one of the U.S. 2020 midterm election TV debates as a simulation example. Our results show that Casevo facilitates more realistic and flexible agent interactions, improving the quality of dynamic social phenomena simulation. This work contributes to the field by providing a robust system for studying large-scale, high-fidelity social behaviors with advanced LLM-driven agents, expanding the capabilities of traditional agent-based modeling (ABM). The open-source code repository address of casevo is https://github.com/rgCASS/casevo.
Paper Structure (21 sections, 4 figures, 7 tables)

This paper contains 21 sections, 4 figures, 7 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Multi-Agent System (MAS)
  4. Social network analysis
  5. Overview
  6. Module Implementation
  7. The Model Module
  8. Agent Module
  9. Parallel Optimization Module
  10. Network Module
  11. Key Features
  12. Scenario Example
  13. Scenario Settings
  14. Agent Configuration
  15. Simulation Steps
  16. ...and 6 more sections

Figures (4)

  • Figure 1: Schematic implementation of Casevo rounds.
  • Figure 2: The architecture of the system.
  • Figure 3: The structure of the network between voters.
  • Figure 4: Simulation Result