- From: Richard Newman <r.newman@reading.ac.uk>
- Date: Mon, 7 Aug 2006 09:40:28 -0700
- To: Xiaoshu Wang <wangxiao@musc.edu>
- Cc: "'Semantic Web'" <semantic-web@w3.org>
>> Retrieving >> >> http://example.com/ont/one >> http://example.com/ont/two >> http://example.com/ont/three >> >> requires three separate requests for three separate >> resources, and an HTTP client *does not know* that it should >> -- *perhaps* -- retrieve "http://example.com/ont" to get a >> description of those three terms. >> >> The web architecture does not support retrieval of terms from >> the entire hierarchy. > > Why can't it be cached? For instance, > http://purl.org/dc/elements/1.1/contributor and > http://purl.org/dc/elements/1.1/title, would get back the same > document. If > there is a rule says: all http://purl.org/dc/elements/1.1/* would > return the > same document, why can't it be cached? If you had such a rule in your client, then your client can indeed decide whether to make a request for more data. However, such a rule is not generic and not intrinsic in the web architecture. For instance, your corporate proxy would not do that caching, your browser would not, and so forth. They are separate HTTP resources. The point is that those are two separate resources on the Web, and you are using a heuristic to decide that they have the same representation, and thus you don't need to retrieve more than one representation. That is an assumption that does not hold in most circumstances. (You are also unlikely to know that you can apply such a heuristic without already having the ontology!) > The rest of your message seemed have missed the intension of my > question. > Of course, everything in RDF is processed as a full URI. But > knowing the > relationship between namespace URI and the resource URIs that are > under the > namespace will help the client side processing. > > The relationship between the # namespace URI and its governed URIs is > clearly defined in URI spec. But that for slash namespace URI is > not (at > least from my limited knowledge). It's not, because there is no useful relationship. Namespaces are purely a syntactic convention. It is a useful accident (or design!) of the web architecture that fragIDs in RDF can be used to describe multiple RDF resources within the same representation. Slashes are *meant* to describe multiple distinct resources, and parts of the URL space. Consider translating the following N3: <http://example.com/my/long/ontology/url/here/is/my/resource> rdf:type <http://example.com/my/long/ontology/url/here/is/my/Class> . <http://example.com/my/long/ontology/url> rdf:type owl:Ontology . into RDF/XML. <rdf:RDF xmlns:owl="..." xmlns:rdf="..." xmlns:ns0="http://example.com/my/long/ontology/url/here/is/ my/"> <owl:Ontology rdf:about="http://example.com/my/long/ontology/url"/> <ns0:Class rdf:about="http://example.com/my/long/ontology/url/here/is/ my/resource"/> </rdf> ... dereferencing ns0 (which has to be generated to fill that tag name position) is probably not going to give you useful information; you *might* get useful RDF from the /url URI, but I wouldn't bet on it. Namespaces are syntactic utilities. -R
Received on Monday, 7 August 2006 16:40:46 UTC