Re: Common RDF parser bug?

> The following defect report against the W3C's ARP-based RDF Validator:
> 
>  [1] http://lists.w3.org/Archives/Public/www-rdf-validator/2002Jan/0304.html
> 
> suggests a violation of the following part of M&S:
> 
> [[
> When a resource represents a reified statement; that is, it has an 
> RDF:type property with a value of RDF:Statement, then that resource 
> must have exactly one RDF:subject property, one RDF:object property, 
> and one RDF:predicate property.
> ]]
> 
> Since this defect report appears to be accurate and all of the online RDF 
> parsers (ARP, SWI, CARA, Profium, Raptor, RDFStore) listed at:
> 
>  [2] http://www.w3.org/People/Barstow/#online_parsers
> 
> seem to have this defect, if you are responsible for a parser listed
> at [2], you may want to review [1].  

I think the text from M&S is simply expressing a cardinality
constraint, part of the M&S (informal) ontology.  Therefor this
constraint should be handled when/if other ontology information is
handled, and not in the RDF parser.

The DAML/Description-Logics view on such constraints, by the way, is
that the multiple values of unique properties are inferred to be
equal, instead of saying the input is invalid.  In other words, the
input in [1] is perfectly valid (at first): it just implies that
"http://schema/Contains" and "http://schema/WrittenBy" are symbols
denoting the same thing.  It also implies "Some Content" and "Some
Author" are identical text strings, which leads to a contradiction
which could/should cause the input to be flagged as invalid (not
logically self-consistent).  But that flagging is post-inference, not
in the parser.

   -- sandro

Received on Sunday, 20 January 2002 18:59:29 UTC