Re: Klyne Contexts: 3. Statements sets in RDF

At 08:09 PM 11/23/00 +0100, Jonas Liljegren wrote:
>This is the final part of my comment on the DFD [1].
>
>You are defining a new type of container using rdfc:member instead of
>_1, _2, etc.
>
>I think it's redundant to have both systems.

I really have a problem with _1, _2, etc.  Let me try another approach to 
explain my view:

It seems to me that, given an RDF graph, I should be able to extract an 
arbitrary subgraph (i.e. a subset of the statements) and still have a valid 
RDF graph.  The RDF approach to containers doesn't permit this, because (I 
think) the following is not valid per RDF M&S:

    [Foo] --rdf:type--> [rdf:Bag]
    [   ] --rdf:_1----> [Member1]
    [   ] --rdf:_3----> [Member3]

Similarly, I think I should be able to take any two RDF graphs and create a 
valid new graph containin the union of statements from the original 
graphs.  (All this without having to re-write the statements.)

>In Wraf [2] I'm planning to store container members in a special way,
>optimized for compact storage and easy manipulation.  With such
>optimizations, there is no need for another type of
>rdfs:ContainerMembershipProperty.

I'd suggest that you are, in effect, implementing something slightly 
different than that specified by RDF M&S.  I want (need) to make the 
required difference explicit in my model.

[...]
>1. With special methods for conatiner handling, you will not have to
>    bother about the present content.  Wraf will give each container
>    the dynamic property 'size', and methods for adding and quering the
>    container without knowing each propertys number.

This assumes that your implementation knows about all of the members.  I 
think that's a kind of closed-world view.

>2. I would say that you can only be in the container in one way.  But
>    besides the content, there can be multipple properties describing
>    other things, like for a normal resource.  You could also attach
>    extra information to each membership statement.

Well, that's a fair point.  I find it easier to use different 
properties.  My approach also captures the idea that a statement cannot be 
asserted in a context without also being quoted in that context.

>3. Distribution functionality can be added to the standard container.
>    Both distribution of properties to each member and distribution of
>    all the members properties to the container or other resource.

I suppose it could.

[...]
>5. Well. You could explicitly put a statement in each collection.  But
>    that would not be the standard way? :)

I happen to think it's important to be able to deal with statements 
separately from the physical documents from which they may have been 
harvested.  Maybe that's a point I should express more explicitly.

>Anyway.  My point here is that Wraf uses the "contexts" but name them
>selections.  And I found some pointes intresting and will maby
>implemnet them.  Maby we could make our schemas more compatible?  I
>would like to keep the standard rdfs:ContainerMembershipProperty.

I'm all for compatible schema, if we're truly describing the same 
things.  OTOH, as we evolve the semantic web and RDF, we'll need to come to 
terms with the fact that different people will use different words, hence 
different URIs, to refer to the same thing.  Right now, I think it's the 
ideas and structures that matter -- the rest is "mere syntax" ;-)

#g
--

>  [1] http://public.research.mimesweeper.com/RDF/RDFContexts.html
>  [2] http://www.uxn.nu/wraf/



------------
Graham Klyne
(GK@ACM.ORG)

Received on Wednesday, 29 November 2000 16:07:29 UTC