[Freebase-discuss] Frequency of Wikipedia updates?

Brian Karlak zenkat at metaweb.com
Fri Apr 23 18:19:17 UTC 2010


On Apr 22, 2010, at 9:17 PM, Tom Morris wrote:

>> Just to be clear, Wikipedia pipeline runs every two weeks and takes  
>> up to a
>> week to do the full cycle. There is a "cooling off" period of two  
>> days,
>> meaning that the articles that are touched / created during last  
>> two days at
>> the time of update will not be part of the update.
>
> So the maximum normally expected latency is 2+14+7 or 23 days?  I
> think that's the number people were trying to get to.

This is more or less correct.  If an article just misses the threshold  
on one pipeline run, it can take about three weeks for it to get  
created.

Of course, the latency for most articles is much less than this.

> Does "touched" really mean edited in any way?  That seems like it
> could introduced a long delay for actively edited new articles.

No.  All that matters is if the article was created in the two days  
before the pipeline run.  "Touched" is a bit of a misnomer.

Brian


More information about the Freebase-discuss mailing list