Signed Graph Representation Learning: A Survey
Zeyu Zhang, Peiyao Zhao, Xin Li, Jiamou Liu, Xinrui Zhang, Junjie Huang, Xiaofeng Zhu
TL;DR
This survey addresses the gap in dedicated reviews for Signed Graph Representation Learning (SGRL) by outlining formal definitions, SGNN architectures, and relevant social theories such as Balance and Status. It proposes a threefold taxonomy—shallow network embedding, deep models, and trustworthy models—and details representative methods across each category, including random-walk, matrix-factorization, attention-based, and adversarial approaches. The review highlights advances to more complex signed networks (e.g., bipartite and temporal graphs), key applications (polarization, stance detection, and negative-feedback recommendation), and practical tools (e.g., PyGSD). It concludes with future directions focused on trustworthy, data-centric, and LLM-assisted SGRL to address robustness, interpretability, and scalability in real-world signed networks.
Abstract
With the prevalence of social media, the connectedness between people has been greatly enhanced. Real-world relations between users on social media are often not limited to expressing positive ties such as friendship, trust, and agreement, but they also reflect negative ties such as enmity, mistrust, and disagreement, which can be well modelled by signed graphs. Signed Graph Representation Learning (SGRL) is an effective approach to analyze the complex patterns in real-world signed graphs with the co-existence of positive and negative links. In recent years, SGRL has witnesses fruitful results. SGRL tries to allocate low-dimensional representations to nodes and edges which could preserve the graph structure, attribute and some collective properties, e.g., balance theory and status theory. To the best of knowledge, there is no survey paper about SGRL up to now. In this paper, we present a broad review of SGRL methods and discuss some future research directions.