Signed Graph Unlearning

2025-10-29

AI Summary

Existing graph unlearning methods are limited to unsigned graphs and neglect structural constraints imposed by positive/negative edges in signed graphs, leading to imbalanced subgraph partitioning and degraded unlearning performance. To address this, we propose SGU—the first unlearning framework tailored for signed graphs. Its core innovations are: (1) a sign-aware balanced subgraph partitioning paradigm that explicitly models edge signs while preserving structural balance; and (2) an incremental GNN update mechanism incorporating signed-graph constraints to jointly mitigate forgetting-induced distortions. Extensive experiments on multiple benchmark datasets demonstrate that SGU significantly outperforms state-of-the-art methods, achieving fast, robust, and accurate unlearning on signed graphs while maintaining high model fidelity. SGU is the first approach to effectively extend graph unlearning to signed network scenarios.

Technology Category

• Machine Learning: Graph-based Machine Learning
• Data Mining & Knowledge Management: Graph Mining, Social Network Analysis & Community
• Reasoning under Uncertainty: Graphical Models

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Scalable techniques for the creation, curation, publication, maintenance, and consumption of large, Web-based, structured, reusable, knowledge graphs and ontologies

Abstract

The proliferation of signed networks in contemporary social media platforms necessitates robust privacy-preserving mechanisms. Graph unlearning, which aims to eliminate the influence of specific data points from trained models without full retraining, becomes particularly critical in these scenarios where user interactions are sensitive and dynamic. Existing graph unlearning methodologies are exclusively designed for unsigned networks and fail to account for the unique structural properties of signed graphs. Their naive application to signed networks neglects edge sign information, leading to structural imbalance across subgraphs and consequently degrading both model performance and unlearning efficiency. This paper proposes SGU (Signed Graph Unlearning), a graph unlearning framework specifically for signed networks. SGU incorporates a new graph unlearning partition paradigm and a novel signed network partition algorithm that preserve edge sign information during partitioning and ensure structural balance across partitions. Compared with baselines, SGU achieves state-of-the-art results in both model performance and unlearning efficiency.

Problem

• Existing unlearning methods ignore signed graph structural properties
• Naive application causes structural imbalance and performance degradation
• Proposes specialized framework to preserve edge sign information

Innovation

• Signed Graph Unlearning framework for networks
• New partition paradigm preserving edge sign information
• Algorithm ensures structural balance across partitions