![[2602.14043] Beyond Static Snapshots: Dynamic Modeling and Forecasting of Group-Level Value Evolution with Large Language Models](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.14043] Beyond Static Snapshots: Dynamic Modeling and Forecasting of Group-Level Value Evolution with Large Language Models
Summary
This article presents a novel framework for dynamic modeling and forecasting of group-level value evolution using large language models (LLMs), addressing the limitations of static snapshot approaches in social simulation.
Why It Matters
Understanding the dynamics of group-level values is crucial for accurate social evolution predictions, especially in the context of data-driven decision-making. This research enhances the capabilities of LLMs in social simulation, providing insights that can inform policy and social science.
Key Takeaways
- Proposes a framework integrating historical value trajectories into LLM-based modeling.
- Demonstrates significant improvements in prediction accuracy over traditional methods.
- Highlights cross-group heterogeneity in value volatility between U.S. and Chinese groups.
- Younger demographics show greater sensitivity to external changes.
- Advances the field of social simulation by bridging gaps in longitudinal data analysis.
Computer Science > Social and Information Networks arXiv:2602.14043 (cs) [Submitted on 15 Feb 2026] Title:Beyond Static Snapshots: Dynamic Modeling and Forecasting of Group-Level Value Evolution with Large Language Models Authors:Qiankun Pi, Guixin Su, Jinliang Li, Mayi Xu, Xin Miao, Jiawei Jiang, Ming Zhong, Tieyun Qian View a PDF of the paper titled Beyond Static Snapshots: Dynamic Modeling and Forecasting of Group-Level Value Evolution with Large Language Models, by Qiankun Pi and 7 other authors View PDF HTML (experimental) Abstract:Social simulation is critical for mining complex social dynamics and supporting data-driven decision making. LLM-based methods have emerged as powerful tools for this task by leveraging human-like social questionnaire responses to model group behaviors. Existing LLM-based approaches predominantly focus on group-level values at discrete time points, treating them as static snapshots rather than dynamic processes. However, group-level values are not fixed but shaped by long-term social changes. Modeling their dynamics is thus crucial for accurate social evolution prediction--a key challenge in both data mining and social science. This problem remains underexplored due to limited longitudinal data, group heterogeneity, and intricate historical event impacts. To bridge this gap, we propose a novel framework for group-level dynamic social simulation by integrating historical value trajectories into LLM-based human response modeling. We select Ch...
