Friday, April 04, 2008


Freebase is a free, publicly usable database of information created by Metaweb. All content is user generated. An API is provided for developers to access the data and develop mashups or applications on top of it. A recent post on Programmable Web describes how Freebase can be used to create semantic web applications.

Even though the data in Freebase is user generated, it is very structured, with each piece of information having a "type" that can be used to describe the relationships between two pieces of data. For example, the Freebase entry for Baltimore is of the type "Location". The location type has been defined to have a latitude/longitude field, a "contains" field (for describing things in that location) , and a "contained by" field (ie Maryland, USA, North America, etc).

A lot of the data seems to have been automatically imported from Wikipedia. By using a more structured format, though, Freebase allows developers to do more with the available data.

1 comment:

Median Joe said...

I found freebase to be very interesting as well. I mentioned it in my "enabling the semantic web" talk, as sites like Freebase and DBpedia are helping to enable semantic search engines.

I am interested in how the number and diversity of relationships are managed. The DBpedia page for Baltimore seemed to have quite a few spurious and/or repeated relationships e.g. with slight spelling variations. Freebase has far fewer- perhaps they're a bit more restrained in that way.