On 6/8/2010 8:23 PM, Erik Hetzner wrote: [snip]
> In my opinion, RDF is more constraining than XML, because it forces > the designer to think clearly about the underlying model, rather than > presenting a lot of different metadata fields. Then I will leave it to you to explain to the W3C how the RDF specification failed to meet the design goal of "represent[ing] information in a minimally constraining, flexible way." [snip] > If I read [3] correctly, while the “JSON API” is deprecated, the JSON > format of the “RESTful API” is not. So perhaps this conversation will > go nowhere. :) I stand corrected. I had read the name "JSON API" as meaning "an API to return data in JSON format." Upon closer examination, I can now see that it means "an API to perform ad hoc queries against the OL data set where the query is encoded in JSON format." The new RESTful API, by contrast, means "an API to perform ad hoc queries against the OL data set where the query is encoded as a standard HTTP query string." The entire data record is still available in the native JSON format. My interest in OL RDF is dwindling rapidly. [snip] > As I see it, you can just ignore what you don’t want. As long as the > graph has the schema that you understand, you can use that. And if the graph doesn't have a schema I understand, I'm hosed; thus the value of constraints. If I know OL will always represent a person as foaf:person, then I can code for that. But if OL sometimes represents a person as foaf:person, and sometimes as dcterms:agent, and sometimes as rdg2:person, and sometimes as rdf:description and sometimes using some new vocabulary that has not yet been invented that a human being could recognize as conveying "personness" but not a computer algorithm, then I don't know how to deal with that. Of course I can ignore what I don't understand, but what happens if there is nothing left when I do? > The more > the merrier. If OL outputs FOAF& RDA,& it conforms to the semantics > of both, great. If I know what FOAF is, I can use that, but if I only > understand RDA, I can use that instead, and not worry about the > differences between the semantics of RDF& FOAF, because OL has done > that for me. Surely you're not suggesting that OL create RDF records that contain every possible representation of its data in every possible vocabulary... > I really don’t see the problem. A graph can be trimmed > wherever you like. The problem is never a surfeit of data, it is always a paucity thereof. > For the record, XSLT is not very useful for dealing with RDF+XML, > unless one constrains (!) the syntax of RDF+XML. Precisely my point. Thus, OL should constrain its RDF/XML syntax to a limited vocabulary, and rely on XSLT to generate unconstrained vocabularies as needed, as the reverse is not possible. In setting those limits, it should start by trying to determine which vocabulary(ies) are most useful to its consumers and potential consumers. This is, so far, the issue that no one has addressed directly. _______________________________________________ Ol-tech mailing list [email protected] http://mail.archive.org/cgi-bin/mailman/listinfo/ol-tech To unsubscribe from this mailing list, send email to [email protected]
