Agent-based modelling of polarized news and opinion dynamics in social networks: a guidance-oriented approach.

This article focuses on addressing opinion polarization in social networks through a novel news opinion guidance approach based on motif recognition. It introduces an agent-based polarized news propagation model that simulates interactions among official media, news self-media, and users, incorporating multi-dimensional user attributes and media bias to realistically replicate opinion dynamics. The study enhances the Augmented Multiresolution Network (AMN) method to include multi-attribute node data for more precise clustering and motif discovery, particularly identifying triangular motifs that reflect echo chamber effects. Simulation experiments on real-world social network data demonstrate that motif-based guidance strategies significantly reduce opinion polarization—by approximately 74% compared to no guidance—offering a computational framework with theoretical and practical implications for mitigating polarization and fostering balanced discourse. Limitations include fixed media opinions and single-dataset validation, with future work proposed to develop dynamic opinion models, advanced motif recognition algorithms, and cross-platform analyses.