Multiple datasets for knowledge graph completion featuring structured triples and corresponding textual descriptions for entities. The data facilitates the training of embedding models that leverage both graph topology and natural language semantics to predict missing links.
Use Cases
- Train a knowledge graph embedding model using the entity descriptions to enrich node representations
- Predict missing relations between entities using the head and tail triple structure
- Evaluate the impact of natural language metadata on link prediction accuracy
Strengths
- Structured triples representing relationships between entities for graph completion tasks
- Natural language descriptions associated with entities to provide semantic context
- Standardized format for knowledge graph embedding (KGE) benchmarking