A survey of social network simulation in the LLM era: From classical models to generative agents
Author Identifier (ORCID)
Abstract
Social network simulation enables controlled study of information dynamics and collective behavior on social graphs, yet the literature remains fragmented across disciplines. This survey provides a unified treatment of the field, organizing 97 papers across seven decades into five paradigms: network structure models, information diffusion, opinion dynamics, rule-based agent-based models, and large language model (LLM)-based agentic simulation. We bridge classical and generative approaches by mapping the progression from analytical update rules to language-grounded agents with semantic memory and platform-native actions, and present a systematic architectural comparison of LLM-based simulation systems. We analyze the literature across five application domains—misinformation, political discourse, platform design, information diffusion, and user behavior—tracing a shift from descriptive analysis toward prescriptive policy prototyping. We outline seven research frontiers spanning representational alignment, multi-level validation, scalability, and responsible use. An accompanying online repository at https://github.com/tamlhp/awesome-sns catalogs 48 implementations (42 open-source), 21 benchmarks and datasets, and 22 evaluation metrics.
Keywords
agent-based modeling, generative agents, large language models, opinion dynamics, social network simulation
Document Type
Journal Article
Date of Publication
10-25-2026
Article Number
116947
ISSN
09507051
Volume
352
Publication Title
Knowledge-Based Systems
Publisher
Elsevier
School
School of Science
Funding received from the Australian Research Council (ARC)
DP240101108, DE260100673
Copyright
subscription content
Recommended Citation
Nguyen, T. T., Pham, T., Huynh, V., Nguyen, M. H., Vo, B., Qu, L., Li, J., Yin, H., & Nguyen, Q. V. H. (2026). A survey of social network simulation in the LLM era: From classical models to generative agents. Knowledge-Based Systems, 352, Article 116947. https://doi.org/10.1016/j.knosys.2026.116947