other formats (Re: [Dev] 4Suite RDF and ZOD)

Paul Snively psnively at earthlink.net
Sun Nov 3 10:02:29 PST 2002

On Sunday, November 3, 2002, at 08:15  AM, Aaron Swartz wrote:

> I understood this to mean that they were adopting the RDF philosophy 
> (flexible data structures, pervasive URIs, fully-extendible 
> descriptions of everything); I certainly hope they aren't planning to 
> store it as RDF/XML.
> RDF is sort of a natural way to think of this data, but that doesn't 
> mean the APIs need look like addTriple, getTriples (indeed, my TRAMP 
> library for Python[1] makes RDF look like regular objects) nor that 
> the storage format encode things in RDF. There are many APIs and 
> backends that can be used without sacrificing the flexibility of the 
> philosophy.
> [1] http://www.aaronsw.com/2002/tramp
> The important features to me are that everything is intertwingled, 
> with links from mail to address book to recent messages to calendar 
> appoints everywhere, and that I can always add more attributes and 
> values to nodes. In this sense, the RDF philosophy is that we make 
> simple statements (this email is about the Chandler project; this 
> folder should show all messages about the Chandler project) rather 
> than placing things in some sort of hierarchy (new folder with name 
> Chandler; move this email to the Chandler folder).
I want to underscore Aaron's point. If you follow the "RDF concepts" 
link that I just posted previously, I think that it can't help but make 
very clear how uncomplicated a commitment to the RDF model (as distinct 
from any particular serialization) is: it almost boils down to "we're 
going to use a directed graph of things referenced by URIs, and these 
things can have properties of various types." There's a bit more to say 
about inheritance, but that's basically it. We should feel free to 
represent this in storage in whatever way makes sense for Chandler's 
usage patterns. People have even modeled RDF in standard relational 
databases as a single table where each row is a triple. It doesn't 
scale worth a darn, but you can do it.

Before I forget, folks who have Java 1.3.1 or later on their machines 
might want to take a look at <http://www.w3.org/2001/11/IsaViz>. It's a 
very nice RDF authoring tool in Java.

Now I need to go check out TRAMP. :-)

> -- 
> Aaron Swartz [http://www.aaronsw.com]
Best regards,

More information about the Dev mailing list