Frontiers

User recommendation method integrating hierarchical graph attention network with multimodal knowledge graph

1 Business College, California State University, Long Beach, CA, United States 2 School of Business and Management, Shanghai International Studies University, Shanghai, China In common graph neural ...
IEEE

Community-Aware Role-Based Graph Embedding

Abstract: Graph embeddings are essential for analyzing complex networks across domains like social networks, biology, and ecology. However, existing methods often fail to integrate structural roles ...
GitHub

SIGEM: A Simple yet Effective Similarity based Graph Embedding Method (Best Research Paper, Honorable Mention at KDD'25)

This repository provides the reference implementation of SIGEM and LINOW (linear matrix form, LINOW-sn, LINOW-bn/MP, and LINOW-bn/TF). SIGEM is a self-supervised and contrastive-free embedding method ...
IEEE

SOPA: Sensitivity-Oriented Poisoning Attack for Self-Supervised Graph Embedding Model via Bilevel Evolutionary Optimization

Abstract: Despite the popularity of graph neural networks, perturbed graph data is still a serious threat towards its inherent vulnerabilities. Adversarial examples can still easily manipulate the ...
AlphaGalileo

Adapting to the Stream: An Instance-Attention GNN Method for Irregular Multivariate Time Series Data

DynIMTS replaces static graphs with instance-attention that updates edge weights on the fly, delivering SOTA imputation and P12 classification ...