[Data-modeling] About Freebase dump data and relation between two Types

John Giannandrea jg at metaweb.com
Tue Dec 30 13:05:28 UTC 2008


Nguyen Minh Tuan wrote:
> Firstly about Freebase dump data structure :
> - The next time you export your new dump data is when ?

every quarter, so the next one is due in january.

> - I downloaded your dump data, there is many folder and file, and I do
> not really understand its structure :
>    + Does each folder correspond to a domain ?
>    + Does each domain correspong to a folder ? :D that is two  
> different
> questions, I mean 1-1 correspondance.
>    + Does files correspond to a type ?
>    + Does a type correspond to a file ?

There are two formats, the TSV type files and the Link Export.

In the TSV format, each directory corresponds to a domain, and each  
TSV file to a type in that domain.
It is a 1:1 mapping, there is one file per type.   The data within the  
files is denormalized, that is the same item will show up in multiple  
files if it has multiple types.

-jg



More information about the Data-modeling mailing list