- From: Kingsley Idehen <kidehen@openlinksw.com>
- Date: Fri, 20 Mar 2009 10:50:16 -0400
- To: Daniel Schwabe <dschwabe@inf.puc-rio.br>
- CC: Knud Hinnerk Möller <knud.moeller@deri.org>, giovanni.tummarello@deri.org, Tom Heath <Tom.Heath@talis.com>, public-lod@w3.org
Daniel Schwabe wrote: > Kingsley Idehen wrote: >> ... >>> - a stable service with a long-range commitment (as Daniel already >>> said) >> Yes, but is that worth the dislocation of the domain component of the >> HTTP identifier pointing back to me? > Sorry but I'm really confused here. I don't see how this applies to > the scenario where there is NO place where one can reliably publish > the data in a "permanent" form. > I can understand the scenario where I publish data using my own URIs, > and do a owl:sameas to Dogfood server URIs when applicable, but if I > can't guarantee a permanent service this doesn't help in the long run. > If you are referring to the scenario where one does offer this > permanent publishing through one's own server, then it doesn't matter > if it's on the Dogfood server or not. > > Can you give me an example of a scenario of what you mean? > > Cheers > D > > > Daniel, When you say: "No place where one can reliabley publish the data in permanent form", do you mean that this data exists off the Web then? If so, then certainly there is a subtle disconnect as I am referring to data that is on the Web but not in RDF form, or may be in RDF form but not endowed with de-referencable URIs. Otherwise: Musicbrainz is an example, they have URIs but don't currently serve up RDF (they used to in the past). Thus, we processed their PostgreSQL hosted data, produced RDF Linked Data but still point back to their URIs even though as of today they don't resolve. My key point is that the Dog-food server is performing RDFization (in some form) and in doing so it shouldn't mandate that the externally sourced data have URIs minted using its namespace. If on the other hand you are going out and grabbing the data e.g., via some 3rd party API or Feed, the practice should still be to attribute to original data provider by exposing their namespace in the URIs you mint, our add some attribution triples. Generally, my comments here about best practice that we should emphasize, and I am hoping my point by point response to Knud clarifies my concerns about the justification he was making. As stated, my point doesn't differ from Giovanni's. Broader examples of what I am stating here are demonstrated via the proxy-URIs that we've always contructed in the Sponger Cartridges that our Sponger Middleware exposes. -- Regards, Kingsley Idehen Weblog: http://www.openlinksw.com/blog/~kidehen President & CEO OpenLink Software Web: http://www.openlinksw.com
Received on Friday, 20 March 2009 14:51:00 UTC