another idea: metadata + RDF DB + SPARQL on every Unix OS

Though I think the first way RDF will break out on the web is with  
people opening up valuable databases to SPARQL end points [1], I do  
think that many of the other distributed uses of RDF such as  
publication of FOAF files will also be very successful. It is just  
that for metadata to be widely distributed is a lot more work as a  
lot more people have to agree on things.

Thinking about my mother's trouble with iPhoto recently [2] I do  
think there is another way that will also make a huge impact, though  
this will take some time as there is a lot of infrastructure to  
build. But perhaps this is something for the Linux/Unix crowd to get  
working on.

Essentially for those who are used to OSX Tiger, the idea is that one  
could improve on Apple's
Spotlight by making the following architectural decisions:

	- enable metadata in the file system
	- make the metadata on files be property value pairs
	- make the properties be RDF properties, ie: URIs though URLs would  
be best.
	- have a metadata index that is a SPARQL endpoint on a RDF database  
that keeps
	in sync with the metadata in the file system

I explore what this would mean for end user applications such as  
iPhoto on my blog [2]. The nice thing is that the property value pair  
helps the semantic web get away from the futile debate about XML/RDF,  
which was clearly a dead end. XML/RDF could only loose because it was  
always going to be more complex than XML itself. If you look at  
property/value paris on metadata enabled file systems on the other  
you can clearly see the RDF structure emerge, and the simplicity of  
the idea is bewildering. Much simpler than XML in fact.

I am convinced now that every OS should have an RDF store by default  
that keeps metadata about all the files it ever came across, be they  
on the local file system, on an external hard drive or on the web. By  
making this RDF store searchable with a SPARQL end point we get all  
the functionality of OSX tiger + a lot more + we get open standard  
based system. As the concept is so simple to understand it should be  
easy for all the applications developed in the open source world to  
play well together, and finally get us to the point of a user  
friendly OS we have been waiting for.

Henry Story

[1] http://blogs.sun.com/roller/page/bblfish?entry=know_your_end_user
[2] http://blogs.sun.com/roller/page/bblfish?entry=limitation_of_iphoto

Received on Monday, 19 December 2005 09:32:03 UTC