Re: The mentography of reification

> > >Taking that view, I'd always envisioned that a nested or reified triple
> > >would be shown on a graph as arcs originating or terminating on arcs (though
> > >I don't know about the validity of that in graph-speak).
> >
> > It isn't good graph-speak, and it isn't correct RDF either, so don't 
> > think of it that way, I would suggest.
> 
> While it may not reflect graph theoretic purity, I think
> "arcs on arcs" is a very useful model.
> 

Yes, and the only thing you have to allow for that to happen is a
triple ID. If you like, you can even generate it automatically from
the triple (as we do in O-Telos-RDF), though of course this does not
guarantee unique IDs for the whole WWW.

Even the ER model allows attributes on relations, though it would be
great, if we could include this in RDF, and get rid of the current
reification, which is essentially a way to assign an ID to a (triple)
statement anyway.

Wolfgang

> I expect knowledge bases where each "ground" statement has
> at least one statement made about it (source, timestamp,
> confidence, etc.).  Creating 4 additional statements
> (current RDF reification) and still not being able to
> directly associate those statements with the real statement
> is impractical.
> 
> Representative applications include:
> 
>   genealogy databases (e.g. [1]) where the source of each
>   fact is supposed to be documented (birth date and place,
>   mother and father from birth certificate n at courthouse
>   X; death date from tombstone at Cemetery Y; marriage from
>   family Bible in possession of Z as of date d -- hopefully
>   expressed as triples rather than text)
> 
>   web caches [2]
> 
>   intelligence databases (which include source, processing
>   history, uncertainty, conflicting data, etc. potentially
>   on a per-field level)
> 
> The RDF APIs I've used directly support using a statement as
> the subject of another statement (see [3]), although the
> serialized results vary.  I'd like to see this addressed
> directly in the RDF specification so that it becomes a
> generally useful capability.
> 
> I've seen a number of database models that are horrendously
> cluttered by maintaining a timestamp, security
> classification, audit trail, etc. with each field.  A
> simple, general mechanism to accommodate such "tagging" (the
> term some of us are using for the examples above, as opposed
> to more general reification) could well be the most
> compelling feature of RDF for many applications.
> 
> RDF M&S [4] hints at this usage
> 
>   Within propertyElt (production [6.12]), ... The value of
>   the ID attribute, if specified, is the identifier for the
>   resource that represents the reification of the statement.
> 
> but IMHO got hung up on using "quoting" rather than
> "tagging" as the canonical example for reification [5].
> 
> 	Mike
> 
> [1] http://www.daml.org/2001/01/gedcom/
> 
> [2] http://www.daml.org/homework/1/collected/README
> 
> [3] http://www.daml.org/2001/04/reification/
> 
> [4] http://www.w3.org/TR/1999/REC-rdf-syntax-19990222/
> 
> [5] http://www.daml.org/listarchive/joint-committee/0274.html


--
Prof. Dr. Wolfgang Nejdl              tel. +49 511 762-19710
Institut für Technische Informatik    fax. +49 511 762-19712
Rechnergestützte Wissensverarbeitung  http://www.kbs.uni-hannover.de/
Educational Technology Lab            http://www.etl.uni-hannover.de/
Learning Lab Lower Saxony             http://www.learninglab.de/
Universität Hannover, Appelstraße 4, 30167 Hannover, Deutschland

Received on Thursday, 25 October 2001 13:45:52 UTC