Nesting is usually a nice convenience that makes it easier to read
the document for humans - when you dump your set of statements into
and then out of something like Sesame a lot of the time the
statements end up as standalone triples rather than nested
structures, as thats the way the parser handles them.
You can achieve something like RDF effects without using RDF - a good
example being the copious use of <LINK rel="xxx"> statements in
various APIs (equivalent to triples using qualified dc:related
predicates). I've noted in the FOAF community a sort of movement away
from 'hardcore' RDF into something like 'logical XML', where there is
enough predictable structure for XML parsers, and enough semantic
consistency for RDF tools. For example, the IETF Atom XML binding can
be quite easily converted into RDF and back.
For both RDF and XTM - or even XML or HTML - to represent a set of
statements about other entities only really requires URIs. As long as
the 'thing' is addressable you can use RDF statements, XTM relations,
XML metadata or whatever.
The bottom line, looking at what ePortfolio management systems
actually do today, is to serialize the metadata about the content in
a flat fashion (e.g. Atom/RSS feeds for each portfolio), or provide
presentations (e.g. XHTML + hResume) so that it can be used by other
kinds of processes (such as admissions processing or assessment). The
additional structure is necessary primarily I think for the case of
migration from one EP tool to another mostly similar EP tool - e.g. a
zip of all files plus a structure manifest that could be XTM/RDF/XML.
To play devil's advocate a bit, the lowest common factor for
structuring resources in any system is a HFS, which is supported by
zip natively anyway, so you could get away without a special
relationship schema.
On the RDF front I went through the IMS EP/UK LeaP relationship
matrix with Matthias Palmer at KTH and he confirmed my opinion that
it would be easy to model as an RDFS+OWL ontology and could work very
well in RDF. If indeed this is the underlying standard (that could be
turned into a dc-style common model) then we can play with XTM, RDF
and XML realizations to see what the effects are of the different
binding approaches.
-S
On 21 Sep 2006, at 16:30, Simon Grant wrote:
> Thanks to Pete for correcting the impression I gave, and confirming
> the suspicion that one can indeed take the approach I've
> characterised as "Topic Map" to portfolio relationships, using RDF.
> It would be very interesting to take this forward and work out just
> what would be involved in total. Has anyone done something like
> this in this, or a similar domain? How, exactly? Could it be
> described (or thought of) as 'lite'?
>
> My not understanding the point fully was due to seeing many
> examples of RDF where the relationship triples are indeed put
> inside one of the sides. Given that FOAF has had a rather high
> profile, and this is what seems to be what is normally done there,
> it's good to have it clarified that it is not the only way.
>
> On the Topic Maps front, there is a clear, self-contained format
> (XTM) in which all of the relationships could be represented. I'm
> looking for what would be done with RDF to implement a similar
> approach.
>
> Thanks
>
> Simon
>
> At 15:41 2006-09-21, you wrote:
>> Hi Simon,
>>
>> Just leaping in on one point...
>>
>> > 3. The RDF approach
>> > You embed all relationship information in each appropriate
>> > element, but the related elements stay separate, each with
>> > its own embedded relationship information. Whether you have
>> > both ends of a relationship represented is a good question.
>> > If you don't, then which end is chosen to hold the
>> > relationship information? And what happens if the
>> > relationship is noted at both ends, but the ends don't agree?
>>
>> I'm not sure I agree with your characterisation of the RDF
>> approach, but
>> maybe I am misunderstanding what you are implying by "embed".
>>
>> For the non-initiates, RDF allows you to construct simple three
>> "word"
>> statements where each word is a URI, and the statement is to be
>> read as
>> an assertion that
>>
>> resource-identified-by-first-URI is-related-in-a-way-indicated-by-the
>> second-URI to resource-identified-by-second-URI
>>
>> Using RDF on the Web, "anyone can say anything about anything", and
>> given that the RDF model is fundamentally one of assertions of binary
>> relationships, "anyone can assert the existence of any relationship
>> between any two things".
>>
>> So I am not limited to making assertions about resources I own or
>> which
>> I have write access to.
>> [...]
|