- From: Patrick Stickler <patrick.stickler@nokia.com>
- Date: Mon, 28 Jan 2002 13:28:35 +0200
- To: RDF Core <w3c-rdfcore-wg@w3.org>
It has occurred to me that tidy literal nodes seem incompatable with the view that a node in the graph maps to one and only one thing in the universe. The reason why URI labeled nodes can be tidy is because URIs are globally unique, and URIs are presumed to have a consistent global meaning. But literals are like local names, and their meaning is dependent on context. Literals do not have a consistent global meaning. If two literals which have different interpretations are denoted by a single node, then that node maps to more than one thing in the universe, and N:N mappings between resources to things introduces an unacceptable ambiguity into RDF. I agree that it is desirable, from an implementation viewpoint, to be able to compress literal value nodes if possible, and that is what URVs are for -- to encapsulate the TDL pairing of lexical form and datatype identity into a single globally unique resource that can participate in node tidying. Thus, while the literal nodes with label "30" in Bob foo:jerseyNumber "30" . Jane ex:age "30" . foo:jerseyNumber rdfs:range foo:count . ex:age rdfs:range xsd:integer . can never be tidied because they have different interpretations, one could use tdl: URVs to achieve a significant degree of compression Bob foo:jerseyNumber <tdl:(foo:count)30> . Jane ex:age <tdl:(xsd:integer)30> . (forgive the use of qnames in the above tdl: URIs...) where the value nodes, now having URI labels, can participate in tidying, and all occurrences of <tdl:(foo:count)30> are merged to the same node, and that node has a globally consistent interpretation, and likewise all occurrences of <tdl:(xsd:integer)30> are merged to the same node, again having a globally consistent interpretation. C.f. http://ietf.org/internet-drafts/draft-pstickler-tdl-00.txt -- So, in conclusion, while TDL is agnostic about tidy versus untidy literal nodes insofar as interpretation is concerned, it does not appear that tidy literal nodes is compatable with the present MT and the presumed N:1 mapping from nodes in the graph to things in the universe (anyone feel free to set me straight if I'm mistaken about that presumption). Patrick -- Patrick Stickler Phone: +358 50 483 9453 Senior Research Scientist Fax: +358 7180 35409 Nokia Research Center Email: patrick.stickler@nokia.com
Received on Monday, 28 January 2002 06:27:32 UTC