UniGO: A Unified Graph Neural Network for Modeling Opinion Dynamics on Graphs

TL;DR

UniGO introduces a unified graph neural framework for opinion dynamics on graphs, addressing the challenge of combining multiple classical fusion rules with equilibrium behavior while mitigating over-smoothing. It leverages a coarsen-refine mechanism to model dynamics on graph skeletons, enabling equilibrium-like predictions without collapsing node distinguishability. A synthetic data synthesis module parametrizes diverse fusion rules and noise to train the model, and pretraining on synthetic data enhances generalization to real-world networks. Experiments on synthetic and real datasets demonstrate strong predictive performance and robust transfer from synthetic to real domains, with ablations confirming the importance of the coarsening-refinement design and synthetic pretraining.

Abstract

Polarization and fragmentation in social media amplify user biases, making it increasingly important to understand the evolution of opinions. Opinion dynamics provide interpretability for studying opinion evolution, yet incorporating these insights into predictive models remains challenging. This challenge arises due to the inherent complexity of the diversity of opinion fusion rules and the difficulty in capturing equilibrium states while avoiding over-smoothing. This paper constructs a unified opinion dynamics model to integrate different opinion fusion rules and generates corresponding synthetic datasets. To fully leverage the advantages of unified opinion dynamics, we introduces UniGO, a framework for modeling opinion evolution on graphs. Using a coarsen-refine mechanism, UniGO efficiently models opinion dynamics through a graph neural network, mitigating over-smoothing while preserving equilibrium phenomena. UniGO leverages pretraining on synthetic datasets, which enhances its ability to generalize to real-world scenarios, providing a viable paradigm for applications of opinion dynamics. Experimental results on both synthetic and real-world datasets demonstrate UniGO's effectiveness in capturing complex opinion formation processes and predicting future evolution. The pretrained model also shows strong generalization capability, validating the benefits of using synthetic data to boost real-world performance.

Figures (2)

• Figure 1: (a) Opinion Interactions in Different Scenarios. (b) A simple example illustrating the similarity between opinion update in opinion dynamics (Degroot model) and node embedding update in graph neural networks.
• Figure 2: The framework of UniGO.