Slash vs Hash URIs redux

I raised this argument with a number of folks at the Cambridge Semantic Web meetup and they suggested that I reiterate  the points I raised on this mailing list regarding the slash vs hash discussion.

At the Tetherless World [1] our websites are built around SPARQL endpoints and graph nodes are primarily slash based because for our use cases we want to get an entire URI from the client so we can execute the appropriate SPARQL query to retrieve data for the entity in question. For example, http://tw.rpi.edu/instances/Deborah_L_McGuinness can be 303 redirected to an HTML page (for browsers) or an RDF document (for sem web applications), where both are generated based on SPARQL queries (the former being transformed using XSLT). We thought it would be an interesting application of Web ID to sign graphs pertaining to when a student completes a MS or PhD degree by using the advisor's WebID. This is where we join the present argument:

If WebID were to enforce that all WebIDs MUST contain fragments it would result in one of two outcomes from our data's point of view:

1) We would have to rename all of our agents to include fragments to be consistent with the proposed definition, breaking any linked data generated by tools we've produced that made use of the original URIs.
2) We double the number of nodes for agents by introducing new hash URIs with #it or #this and assert that those URIs are owl:sameAs the originals. This would require that any agent looking to consume WebID data would need to support some level of the OWL semantics to be able to use any of the WebIDs generated in this way.

So, if this is going to be required, I have some questions:

* Is anyone in the community planning on writing documentation on best practices to bringing non-conforming URLs into conformance for existing implementations?
* If so, how do these two options coincide with what has been proposed thus far?
* What level of semantics is the community looking to provide as part of a WebID agent implementation (e.g. case 2 above)?

Regards,
Evan Patton

[1] http://tw.rpi.edu/

Received on Thursday, 13 December 2012 10:45:55 UTC