Abstract: Graph Transformers, emerging as a new architecture for graph representation learning, suffer from the quadratic complexity and can only handle graphs with at most thousands of nodes. To this ...
Abstract: Medical knowledge graph (KG) is sparse KG that contains insufficient information and missing paths. Multi-hop reasoning is an effective approach of medical KG completion, since it offers ...
Foundation = ontology layer with ~71K entities, multi-parent DAG. A disease can belong to multiple branches (e.g., "cutaneous tuberculosis" belongs to both infectious diseases AND skin diseases). MMS ...