- From: Mike Dean <mdean@bbn.com>
- Date: Fri, 24 Jan 2003 10:02:52 -0800
- To: Brian McBride <bwm@hplb.hpl.hp.com>
- cc: rdf Core <w3c-rdfcore-wg@w3.org>
This message formally summarizes my objection to the datatyping solution adopted [1] by RDF Core [2]. I believe that datatyping is important, but the overwhelming majority of Semantic Web users will want "global" datatyping, where datatypes are associated with properties in the schema/ontology, rather than "local" datatyping, where a datatype must be associated with each value. The current local-only solution forces the hand of users, bloats content, adds to the potential for inconsistencies, and precludes otherwise-compatible schema evolution (e.g. xsd:short to xsd:int or xsd:float). In decreasing order, I'd prefer the following datatyping approaches to the current one: 1) adoption of the datatyping approach used in DAML+OIL (March 2001) [3] 2) an RDF datatyping approach that includes global datatyping (as in Proposal F from [4]) possibly with optional local datatyping 3) a lexical datatyping approach (as in Proposal B from [4]) 4) leaving datatyping out of RDF Given the current local-only approach, I strongly support the following decisions made by RDF Core: 1) incorporating local datatyping within the literal node rather than as additional graph structure (avoiding triple bloat) 2) allowing the use of rdfs:range within a schema to indicate the expected datatype for a property (I hope that this can be extended to form the basis of a future global datatyping solution) Mike [1] http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Oct/0131.html [2] http://www.w3.org/2001/sw/RDFCore/ [3] http://www.daml.org/2001/03/daml+oil-index [4] http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Oct/0049.html
Received on Friday, 24 January 2003 13:03:37 UTC