Re: [more on] a bnode URI scheme?!

[afterthoughts, afterthroughts...]

Adam Souzis wrote:

> bnode:///<URL encoding of the model URI>/bNodeIdentifier
> (and a relative bnode like bnode:bNodeIdentifier)

This reminds me of one of the solutions to CWM's broken handling of 
root formulae. CWM's notion of an RDF graph, more or less, is called a 
"formula", and for each RDF document there's an implicit root formula 
in which all statements are contained. The identifier that CWM uses 
for this is <documentURI#_formula>, which is bad because it's 
basically stealing a term from the RDF document owner's URI space.

The proposal was that CWM should use a URI of the form:

    "formula:" base64encode(documentURI) "$" formulaName

Note that CWM used to, and probably still does, steal bNode and 
universally quantified variable names in a similar manner: you see 
fragments such as #_g0 all over the place in CWM's output where they 
shouldn't exist. This, however, I consider just a bug in CWM's 
internal representation of bNodes.

The problem with your suggestion of "bNode:" + modelEncoded + 
Identifier is that the identifier rather depends on the serialization 
at the moment, and serializations don't force you to keep consistent 
local identifiers. I don't see that as a surmountable problem. The 
"formula:" URI scheme suggestion has a similar problem (except worse 
because formulae aren't provided even ephemeral local identifiers).

-- 
Sean B. Palmer, <http://purl.org/net/sbp/>
"phenomicity by the bucketful" - http://miscoranda.com/

Received on Friday, 12 March 2004 05:37:00 UTC