ConceptNet Common-Sense Knowledge Graph With 8 Million Entities

ConceptNet 5.7 is a common-sense knowledge graph containing approximately 8 million entities and RDF triples of everyday facts and relationships. It is converted to HuggingFace dataset format for machine learning use, with an extracted size of 1.2 GB.

Use Cases

• Train graph neural networks on RDF triples for common-sense reasoning tasks.
• Build multilingual semantic search systems leveraging the graph's relationships and entities.
• Fine-tune language models using the structured knowledge from the common-sense facts.

Strengths

• Contains approximately 8 million entities, providing a broad base of common-sense knowledge.
• The dataset is 1.2 GB in extracted size, indicating substantial information content.
• Covers multilingual and semantic relation concepts, as indicated by its tags.

Limitations

• Specific column structure, row count, and file formats are unknown, complicating direct analysis.
• The original data source is from 2019, which may limit the recency of the knowledge captured.

Provenance

Source

https://s3.amazonaws.com/conceptnet/downloads/2019/edges/conceptnet-assertions-5.7.0.csv.gz

Collection Method

RDF triples converted to HuggingFace dataset format from an original CSV source.

Time Range

null

Freshness

null

Geography

null