Hi Mirko,
well the thing is, it wouldn't really work without a form a of
reification (in my mind). There are use cases, where people prefer a
simple knowledge representation of a semantic relation, and other ones,
where people like to get a more detailed description about the semantic
relation between two particulars. However, it is important to be able to
semantically relate both of them.
When I designed the Cognitive Characteristics Ontology[1], I struggled
(again) with the same issues. Thereby, the Cognitive Characteristics
Ontology includes two opportunities to model cognitive patterns.
The first one is the representation of cognitive characteristics by
using the semantic relation cco:cognitive_characteristic or better its
more specialised sub properties, e.g. cco:interest, to associate the
topics of the cognitive patterns to the users. The second opportunity is
the object-oriented context reification of cco:cognitive_characteristic,
cco:CognitiveCharacteristic, which is a general multiple purpose
cognitive characteristic concept to describe cognitive patterns more in
detail for a specific user or user group.
However, to be able to model the semantic relation between the "shortcut
relation" and its reification statement, one need a further mechanism,
which is included into the Property Reification Vocabulary[2]. This
vocabulary should enable a reasoning engine to apply the implications
between a "shortcut relation" and its reification statement, however not
directly on the RDF Statement however for all possible statement that
uses the defined "shortcut relation" properties and reifications classes
(incl. their related properties).
In case of the Cognitive Characteristics Ontology[3] it enables you to
decide between a skill in soccer player, an expertise in soccer and and
interested in football watching (See [4]).
Alternatively, you can apply Named Graphs, however in my mind, they are
intended to represent 'external context' (especially provenance and
trust), because their semantics are not really clear in that case.
However, from my use case above, I like to represent 'internal context',
as detailed description of a "shortcut relation".
Don't hesitate to ask further question. This all is a work in progress
and suggestions, comments and critics are very welcome.
Cheers,
Bob
[1] http://purl.org/ontology/cco/cognitivecharacteristics.html
[2] http://purl.org/ontology/prv/propertyreification.html
[3] http://purl.org/ontology/prv/propertyreification.html#sec-cco-example
[4]
http://purl.org/ontology/cco/cognitivecharacteristics.html#sec-soccer-example
Am 13.10.2010 15:02, schrieb Mirko:
Hi all,
I try to understand alternatives to reification for Linked Data
publishing, since reification is discouraged. For example, how could I
express the following without reification:
@prefix dc: <http://purl.org/dc/elements/1.1/>.
@prefix foaf: <http://xmlns.com/foaf/0.1/>.
<http://ex.org/stmt>
rdfs:label "Statement that describes user interest in a document"@de;
rdf:subject <http://ex.org/User>;
rdf:predicate foaf:interest;
rdf:object <http://ex.org/Item>;
dc:publisher <http://ex.org/Service>;
dc:created "2010-10-13"^^xsd:date;
dc:license <http://ex.org/License>.
<http://ex.org/User> rdf:type foaf:Agent.
<http://ex.org/Item> rdf:type foaf:Document.
Thanks,
Mirko