Tag
This survey comprehensively reviews GNN-based link prediction from a dedicated GNN perspective, categorizing recent advancements by techniques (GCN, GAE, GAT, GFormer) and applications (knowledge graphs, recommendation systems), and discusses challenges and future directions.
This paper proposes GRATE (Gated Rotary Attention for Temporal Encoding), a parameter-free temporal encoding method that enhances inductive knowledge graph foundation models by incorporating relative time differences and query-conditioned gating. It also introduces new inductive temporal knowledge graph benchmarks (GDELTIndT and WIKIIndT) to evaluate cross-dataset transfer, demonstrating improved performance over static base models.
Introduces Drift-Aware Temporal Graph Rewiring (DATGR) to dynamically update co-occurrence edges in biomedical text graphs, capturing semantic drift without full retraining. Evaluated on BIOMRC, it achieves a mean AUROC improvement of 0.066 over static baselines while maintaining precision.
This paper introduces PGRE, a probabilistic model for dynamic knowledge graphs that captures inter-relational dependencies using Poisson-Gamma and Markov processes, achieving competitive link prediction performance especially in sparse settings.
TAG-DLM unifies textual reasoning and graph message passing within a masked diffusion language model, enabling joint reasoning over text and graph topology for node classification and link prediction tasks.
Proposes RelBall, a KGC model that extends Rotate3D with modulus transformation for modeling hierarchies and a tail-centric relation ball to handle one-to-many relations, achieving competitive link prediction performance.
TeRoR introduces a novel temporal knowledge graph embedding method that decouples entity temporal evolution with independent rotation transformations and uses relational circular regions to model diverse relation mapping properties, achieving competitive performance on four datasets.
Proposes KMAS, an adaptive negative sampling method to improve training of knowledge graph foundation models, achieving state-of-the-art results across 44 datasets.
This paper proposes a modular temporal enhancement framework for signed graph neural networks that integrates historical context via a Historical Context Integration Module (HCIM) with LSTM and multi-head temporal attention, achieving consistent improvements on real-world temporal signed networks for dynamic link prediction.
The paper introduces a novel task of fact generation for hyper-relational knowledge graphs (HKGs) and proposes KREPE, a generative representation learning method using masked discrete diffusion that unifies link prediction and fact generation, achieving state-of-the-art performance.
This paper adapts instance discrimination self-supervised learning to link prediction in graphs, proposing new models L-GRACE and L-BGRL that operate on link representations and improve performance especially on unattributed graphs.