I've thrown together my own ad-hoc framework for mapping our domain objects into a Jena model served via Restlet. URLs describing resources in the RDF are valid Restlet patterns and vice versa. I'm very interested in what other people are doing here and seeing if any simple clean re-usable framework comes out of this.
Ryan ... (Unfortunately my experience is that if you throw the word "semantic web" in front of anything it becomes 10x more complicated than it actually is, and throwing a JSR in there sounds even more frightening :-) ... When it comes down to it there's not that much novel in the RDF model -- it's just Codd's relational algebra without types or constraints, binary relations only, about a 1/3rd of the more powerful relational algebraic operators missing, using URIs as primary keys throughout. We use it only as a representational format now, for shuttling down to clients like Mozilla/XUL where it can be used fairly nicely... ) On 2/14/07, Henry Story <[EMAIL PROTECTED]> wrote:
Oh thanks. On the so(m)mer mailing list we have started some discussions already with other groups doing similar mappings, on what types of annotations to use for mapping java objects to the semantic web. We have started converging on some of these annotations, and were thinking of it as a first move to a JSR. The Elmo project for example can take an OWL ontology and produce annotated java classes like the following: http://src.aduna-software.org/viewvc/org.openrdf/projects/elmo/elmo- concepts/src/main/java/org/openrdf/elmo/concepts/atomowl/ Now the interesting question is: how is this related to REST? In a few ways: - the definitions in a good rdf ontology are at the namespace location of the relationships. - it is good practice to make uris dereferenceeable, as in my foaf name [1]: http://bblfish.net/people/henry/card#me This should enable drag and drop behavior [2]. Now since I don't yet know how this mapping is meshing with the new JSR I can't yet say whether we should start another JSR or build together. But one thing's for certain, I need to update the so(m)mer home page a little. Henry [1] http://blogs.sun.com/bblfish/entry/i_have_a_foaf_file [2] http://blogs.sun.com/bblfish/entry/universal_drag_and_drop On 14 Feb 2007, at 12:36, Jerome Louvel wrote: > > Hi Henry, > > Very good points! I have updated my blog post [1] to mention Sommer as > another source of inspiration. > > It would be great if you could participate to this JSR (maybe as a > member of > the expert group?). I think that your Semantic Web point of view and > experience with resource description via Java annotations would be > very > valuable. > > Best regards, > Jerome > > [1] http://blog.noelios.com/2007/02/14/ > >> -----Message d'origine----- >> De : Henry Story [mailto:[EMAIL PROTECTED] >> Envoyé : mercredi 14 février 2007 11:58 >> À : [email protected] >> Cc : [EMAIL PROTECTED]; Peter Mika; James Leigh >> Objet : Re: New JSR to define a high-level REST API for Java >> >> From what I see on JRA [-1] this looks somewhat similar to what we >> have been doing at so(m)mer, where we created an @rdf annotation to >> annotate classes, fields and getters and setters too. >> >> >> The idea of this mapping came from OWL [0] which is a description >> logic [1] vocabulary for RDF. Description logic is essentially >> Declarative Object Oriented notation. >> >> RDF uses Resources to denote everything. Essentially it takes >> REST to >> the limit. >> >> So if one is going to map REST to java then those would clearly be >> the places to look. >> >> Looking at JRA it seems a little ad hoc. It does not seem to >> me to be >> starting from the most solid foundations. >> >> Henry >> >> [-1] http://jra.codehaus.org/ >> [0] http://www.w3.org/2001/sw/WebOnt/ >> [1] http://en.wikipedia.org/wiki/Description_logics >> >> >> >> On 14 Feb 2007, at 11:25, Jerome Louvel wrote: >> >>> >>> Thanks Dave. I also think it is good for REST and for Java. >>> >>> I just want to make it very clear that this JSR will not >>> standardize the >>> Restlet API or a similar API. We still plan to submit the Restlet >>> API to the >>> JCP in 2008. It's purpose is to standardize a higher-level >> API as a >>> set of >>> Java annotations. In my mind, that should address the mapping >>> between the >>> resource-oriented paradigm (as supported by the Restlet API) and the >>> object-oriented paradigm (as expressed by Java domain classes). >>> >>> Right now, we have the Resource class that is the final controller >>> in the >>> Restlet processing chain. This class doesn't constraint in any way >>> how the >>> domain objects should look like or be provided: EJB, persistent >>> POJOs, db4o, >>> Hibernate, JPA, JDO, direct JDBC/SQL calls, etc. This set of >>> annotations >>> will facilitate the mapping between domain POJOs, Restlet >> Resources >>> and URI >>> routing. >>> >>> Best regards, >>> Jerome >>> >>>> -----Message d'origine----- >>>> De : Dave Pawson [mailto:[EMAIL PROTECTED] >>>> Envoyé : mercredi 14 février 2007 10:50 >>>> À : [email protected] >>>> Objet : Re: New JSR to define a high-level REST API for Java >>>> >>>> On 14/02/07, Jerome Louvel <[EMAIL PROTECTED]> wrote: >>>> >>>>> I'm happy to announce that the Restlet project will >>>> contribute to the >>>>> elaboration of a high-level REST API for Java, as a set of Java 5 >>>>> annotations. >>>>> >>>>> This effort, led by Sun Microsystems, will be complimentary >>>> to the Restlet >>>>> API. The Noelios Restlet Engine, will support these new >>>> annotations in a >>>>> future version. >>>> >>>> Nice one Jerome! >>>> good to see your work accepted as a part of Java. >>>> >>>> Good for REST too I think. >>>> >>>> regards >>>> >>>> -- >>>> Dave Pawson >>>> XSLT XSL-FO FAQ. >>>> http://www.dpawson.co.uk
