A Survey on Heterogeneous Graph Embedding: Methods, Techniques, Applications and Sources

Heterogeneous graphs (HGs) also known as heterogeneous information networks have become ubiquitous in real-world scenarios; therefore, HG embedding, which aims to learn representations a lower-dimension space while preserving the structures and semantics for downstream tasks (e.g., node/graph classification, node clustering, link prediction), has drawn considerable attentions recent years. In this survey, we perform comprehensive review of development on embedding methods techniques. We first introduce basic concepts discuss unique challenges brought by heterogeneity comparison with homogeneous graph representation learning; then systemically survey categorize state-of-the-art based they used learning process address posed heterogeneity. particular, each representative method, provide detailed introduction further analyze its pros cons; meanwhile, explore transformativeness applicability different types industrial environments time. addition, present several widely deployed systems that demonstrated success techniques resolving application problems broader impacts. To facilitate future research applications area, summarize open-source code, existing platforms benchmark datasets. Finally, additional issues forecast directions field.