Download the Freebase Easy data dump
Note: based on the original Freebase data, which is provided under a Creative Commmon Attribution License.
freebase-easy-latest.zip (3.3 GB)
The dump includes:
- facts.txt The 362M facts, 22 GB uncompressed. Based on the read-only version of Freebase.
- scores.txt A prominence score for each of the 59.4M entities, 3 GB uncompressed.
- freebase-links.txt A link to the original Freebase resource for each entity, 2.5 GB uncompressed.
freebase-easy-14-04-14.zip (2.5 GB)
The dump includes:
- facts.txt The 242M facts (one fact per line), 15 GB uncompressed.
- scores.txt A prominence score for each of the 48.9M entities, 2.5 GB uncompressed.
- freebase-links.txt A link to the original Freebase resource for each entity, 1.9 GB uncompressed.
Technical Details
Transitive Relations
The transitive closure has been build for the following (pairs of) relations, as described in Section 2.4 of our paper. We have ignored the cause-of-deaths relation, since there are errors in the original freebase data for parent-cause-of-death that would make the transitive hull for that relation unusable.
- ns:biology/organism_classification/higher_classification
- ns:people/person/profession - ns:people/profession/specialization_of
- ns:people/profession/specialization_of
- ns:location/location/containedby
Taxonomy Enrichments
Copies of the facts of the following relations have been added as "is-a" type statements to enrich the taxonomy.
- biology/organism_classification/higher_classification
- people/person/profession
Omitted Prefixes
Freebase relations starting with the following prefixes have not been
extracted into our collection of facts.
They contain duplicate, technical, or experimental data.
- rdf:
- rdfs:
- key:
- ns:user
- ns:base
- ns:freebase
- ns:common
- ns:dataworld
Citing
If you'd like to use this data dump in a publication, please cite our paper.