[rdfweb-dev] FOAF Schema Validation

[Ian Davis]

On Thursday, 23 January 2003 at 15:16, Doug Ransom wrote:
> Would it be reasonable to simply ignore the errors and related 
> properties in FOAF files and process what you can? If the file as 
> something useful from the foaf namespace, maybe that is good enough?
I'm just trying to catalog and provide additional metadata about the
FOAF and other RDF that exists out there. Hopefully other FOAF
consumers can use that information to make decisions about how they
process the various sources. Also, I hope to be able to encourage
people to make their RDF more consistent and of a higher quality.


> Will you have your stored results from crawling available for public use 
> as a giant FOAF file? 
If you mean will I aggregate the various FOAF files and reserve them,
then no I'm not planning to do that. I am planning to export all the
metadata and validation information as RDF for other robots and
applications to use.

> As FOAF grows, has anyone thought about how crawling agents will know 
> when/if a FOAF file has changed (or does this even matter)? Maybe some 
> form of date-modified indicator in the foaf or beside it. Maybe a 
> number users increment every time the file is saved -- if its not 
> changed, agents would not reprocess the file (other than reading up to 
> the change indicator)? 
We can use the experience gained in RSS to help here. Many aggregators
poll for changes at regular intervals which is reliable but bandwidth
intensive (although a community effort to promote use of existing HTTP
good practices helped there). The alternative is for the publisher to
notify an aggregator that it has changed, i.e. weblogs.com or blo.gs.

However, as you suggest, detecting when a FOAF document has changed
may not be all that relevant. They generally don't contain dynamic
content and most users will rarely change them. In fact they will
probably create one and then forget about it. Simple FOAF won't change
from year to year probably.

The Semantic Planet crawler will list the date the source last changed
as part of the metadata, RSN :).


Ian