Signed Graph Autoencoder for Explainable and Polarization-Aware Network Embeddings

2024-09-16

arXiv.org

AI Summary

Existing graph neural network autoencoders lack interpretable modeling capabilities for signed graphs (with positive/negative edges), failing to capture competition-driven adversarial communities and polarization structures. To address this, we propose SGAAE: an interpretable, polarization-aware signed graph autoencoder framework. Methodologically, SGAAE innovatively integrates relational prototype analysis with Skellam-distributed likelihood modeling for signed link generation; it constructs a polarization polyhedral embedding space wherein nodes learn membership degrees to opposing archetypes, enabling formal, interpretable characterization of polarization at both intra- and inter-layer levels. Evaluated on four real-world signed networks, SGAAE achieves significant improvements in directed signed link prediction accuracy, robustly identifies adversarial communities, and supports fine-grained, interpretable quantification of polarization magnitude and direction.

Technology Category

• Machine Learning: Graph-based Machine Learning
• Data Mining & Knowledge Management: Graph Mining, Social Network Analysis & Community
• Game Theory and Economic Paradigms: Adversarial Learning

Application Category

• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Social Networks and Social Media: Media and governance, opinion dynamics, filter bubbles, polarization

Abstract

Autoencoders based on Graph Neural Networks (GNNs) have garnered significant attention in recent years for their ability to extract informative latent representations, characterizing the structure of complex topologies, such as graphs. Despite the prevalence of Graph Autoencoders, there has been limited focus on developing and evaluating explainable neural-based graph generative models specifically designed for signed networks. To address this gap, we propose the Signed Graph Archetypal Autoencoder (SGAAE) framework. SGAAE extracts node-level representations that express node memberships over distinct extreme profiles, referred to as archetypes, within the network. This is achieved by projecting the graph onto a learned polytope, which governs its polarization. The framework employs a recently proposed likelihood for analyzing signed networks based on the Skellam distribution, combined with relational archetypal analysis and GNNs. Our experimental evaluation demonstrates the SGAAEs' capability to successfully infer node memberships over the different underlying latent structures while extracting competing communities formed through the participation of the opposing views in the network. Additionally, we introduce the 2-level network polarization problem and show how SGAAE is able to characterize such a setting. The proposed model achieves high performance in different tasks of signed link prediction across four real-world datasets, outperforming several baseline models.

Problem

• Signed Graphs
• Explainable Models
• Generative Models

Innovation

• SGAAE
• Signed Graph Analysis
• Polarization Identification