An Adaptive Structural Balance Method Based on Reinforcement Learning for Signed Social Networks
Bibliographic Data
| ID | 22106954 |
|---|---|
| Authors | Mingzhou Yang (0000-0002-1245-4638, Shenyang University of Technology), Man Yang (0000-0002-7601-6733, Shenyang University of Technology), Qiang He (0000-0001-6481-0924, Northeastern University), Jing Gao (0000-0002-0020-0622, China Mobile (China)), Keping Yu (0000-0001-5735-2507, Hosei University) |
| Year | 2026 |
| Volume | 13 |
| Issue | 1 |
| Pages | 123-135 |
| Publication date | 2026-02-01 |
| Peer Reviewed | Yes |
| Open Access | Yes |
| Type | ARTICLE |
| Venue | IEEE Transactions on Computational Social Systems (JOURNAL) |
| Journal identifiers | ISSN: 2329-924X • E-ISSN: 2373-7476 |
| Publisher | Institute of Electrical and Electronics Engineers (IEEE) (PUBLISHER) |
| DOI | 10.1109/tcss.2025.3583358 |
| OpenAlex | W4412605309 |
| Language | EN |
| References cited | 47 |
The structural balance problem in signed social networks targets at detecting the unbalanced edges and minimizing the cost to make a network balanced. Current studies mainly focus on the cost difference between changing positive and negative edges, ignoring the influence from nodes’ neighbors on the cost to change unbalanced edges. In this article, we propose a novel structural balance model by considering the influences of the edge weights and relationships among nodes’ neighbors on cost of structural balance in networks. To optimize the proposed model, this article combines label propagation algorithm with reinforcement learning. First, a label propagation algorithm for signed social networks is designed by separately considering the labels of positive neighbors and negative neighbors, since the existing label propagation algorithms are mostly used for unsigned social networks. Then, reinforcement learning performs the initialization operation based on the result of the proposed label propagation algorithm. Consequently, the proposed algorithm can not only automatically select the number of clusters for different networks, but also run from a good initial policy and improve the performance. Extensive experiments on five networks demonstrate the performance of our method in terms of convergence, stability and efficiency
Reinforcement · Reinforcement learning · Structural engineering · Computer Science · Engineering · Neuroscience · Opinion Dynamics and Social Influence · Psychology · Artificial Intelligence
Statistical mechanics of complex networks
Collective dynamics of ‘small-world’ networks
Attitudes and Cognitive Organization
ToupleGDD
Structural Reconstruction of Signed Social Networks
An Energy Function for Computing Structural Balance in Fully Signed Network
Misinformation Propagation in Online Social Networks
The influence of structural balance and homophily/heterophobia on the adjustment of random complete signed networks
Optimizing dynamical changes of structural balance in signed network based on memetic algorithm
Clustering and Structural Balance in Graphs
| Citation velocity | historical |
|---|---|
| Highly cited | No |