FactNet Bench: Knowledge Graph Completion Benchmarks

FactNet Benchmarks provides three benchmark datasets derived from FactNet for evaluating AI models. The primary benchmark focuses on Knowledge Graph Completion (KGC), assessing a model's ability to predict missing links within a graph. The dataset was created by the organization 'openbmb' and was last updated on May 7, 2026.

Use Cases

• Benchmarking knowledge graph completion models based on (subject, relation, object) triples.
• Evaluating link prediction performance on a structured knowledge graph.
• Training and testing AI models on tasks requiring missing entity prediction.

Strengths

• Dataset includes three distinct benchmark tasks derived from FactNet.
• The KGC benchmark is explicitly designed for evaluating missing link prediction.
• Data splits (Train/Dev/Test) are defined with careful cross-split collision handling.

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count and dataset size are unknown, which may limit suitability assessment.

Provenance

Source

Derived from FactNet.

Collection Method

Extracted from entity-valued synsets and projected to (S, P, O) triples.

Freshness

Last updated 2026-05-07 00:10:33; freshness should be verified.