Advances in Graph Representation Learning: Methods and Applications

Knowledge graphs (KGs), along with attributed and heterogeneous graphs, underpin a growing share of data-driven artificial intelligence. Recent advances in knowledge graph embeddings, multi-perspective relational modelling, multimodal entity alignment and completion, and tokenised graph transformers have advanced graph representation learning far beyond early shallow embedding methods.At the same time, researchers are coupling large language models with graph structures to improve reasoning. Graph-based methods are also increasingly used in recommendation systems and education, where researchers model learners’ knowledge states and experts’ competencies using concept dependency graphs. However, real-world graphs are often noisy, multimodal and heterogeneous, and models trained on clean benchmarks frequently transfer poorly to these settings.This Research Topic brings together the editors’ complementary expertise in knowledge graph representation learning, multi-perspective relational modelling, completion and entity alignment (Zhifei Li and Zengcan Xue); graph transformers and node representation learning (Jinsong Chen); and graph-based educational data mining, including cognitive diagnosis, knowledge tracing and concept-prerequisite relationship learning (Kui Xiao).We seek contributions that strengthen the foundations of graph learning through robust multimodal knowledge graph embedding and completion, expressive graph transformers, and reasoning under noisy or inductive conditions. We are particularly interested in studies that demonstrate these methods in recommendation systems, question answering and AI-driven education. We also welcome work that integrates graph learning with large language models and multimodal data, and that moves beyond benchmark improvements towards measurable impact in deployable systems.We welcome original research, reviews, perspectives and brief reports on knowledge graph embedding, completion, entity alignment and inductive reasoning; multi-view and multimodal relational learning; graph transformers and graph neural networks for heterogeneous and multimodal graphs; noise-robust and contrastive graph learning; the integration of graphs and knowledge graphs with large language models, including retrieval-augmented reasoning; knowledge graph-enhanced and multimodal recommendation; educational and expert-competence knowledge graphs; cognitive diagnosis, knowledge tracing and concept-prerequisite relationship learning on educational graphs; and reproducible benchmarks and open datasets.Manuscripts should report measurable improvements over strong baselines and, where possible, include open-source code and real-world evaluations. Topic editors: Zhifei Li, Kui Xiao, JInsong Chen Source: https://www.frontiersin.org/research-topics/86587/advances-in-graph-representation-learning-methods-and-applications