We convert Freebase to industry-standard RDF, to produce a Rosetta Stone for schema.org, natural language processing and social-semantic systems.
:BaseKB Pro and :BaseKB Lite are created by hydroxide, part of the open-source Infovore framework. Based on SPARQL and the Map/Reduce paradigm, Infovore can do parallel processing of data sets that are too large to fit into a triple store.
:BaseKB knows about people, places, books, movies, medicines, airports, automobiles, historical events and just about anything that would come up in a game of "20 questions".
Based on Freebase, :BaseKB enables applications to interact with DBPedia, schema.org, the Google Knowledge Graph, the Facebook Open Graph and other semantic-social systems.
Knowledge from :BaseKB can also populate databases on a wide range of topics, organize content, reconcile disparate databases and power NLP systems with unprecedented capabilities.
Multiple editions of :BaseKB are available.
:BaseKB Lime is produced from the Freebase RDF Dump. Lime is
refreshingly compatible with RDF standard tools because it removes invalid, unnecessary and dangerous facts. Other than
rdf:type, :BaseKB Lime leaves Freebase vocabulary
A product of the open-source Infovore framework, :BaseKB Lime is ready for commercial and academic use under the same CC-BY license as Freebase.
:BaseKB Lime is a 10 GB file that contains roughly 720 million valid facts, and 13 million invalid facts filtered from a billion raw facts in the Freebase RDF dump.
:BaseKB Lite and :BaseKB Pro can be produced from the old Freebase quad dump. :BaseKB Pro is a 7.8 GB file that contains all topics and significant well-formed facts from Freebase; :BaseKB Pro transforms Freebase by mapping all Freebase vocabulary to unique mid identifiers.
:BaseKB Lite is much smaller, at 2.8 GB, and consists of a consistent subset of Freebase facts concerning the Freebase schema and objects that exist in DBpedia. :BaseKB Lite has lower hardware requirements than :BaseKB Pro and has a better correspondence between system concepts and "common sense" concepts than either Freebase or DBpedia alone.
Research is under way to discover if it possible and desirable to produce a products similar to :BaseKB Pro or Lite based on the new RDF dump.