- From: <MDaconta@aol.com>
- Date: Wed, 13 Nov 2002 13:44:56 EST
- To: jon@spin.ie, www-rdf-interest@w3.org
- Message-ID: <54.22248b4.2b03f7a8@aol.com>
In a message dated 11/13/2002 9:34:17 AM US Mountain Standard Time, jon@spin.ie writes: > 1. You simply may not know (or have a vocabulary to express) the type. > 2. You simply may not have a useful identifier. > I agree that there are valid use cases for blank nodes. > You can still make use of what you do know though. > > The semantics of a blank node convey certain information, and do so in a > manner compatible with statements not using blank nodes. Only if they > failed > to do that could you say that the semantics were weakened. > I don't agree with that. You can certainly compare the semantics of different data representations just as people compare the semantic expressiveness of RDF to XML Schema. I also believe it is intuitive that a typed blank node provides better semantics than an untyped blank node. In terms of best practice, is it not better to avoid using a blank node if possible? > Of course in the example: > > <http://www.example.org/staffid/85740> > <http://www.example.org/terms/address> > _:johnaddress . > > _:johnaddress > <http://www.example.org/terms/street> > "1501 Grant Avenue" . > > It could perhaps be possible to add a type (perhaps from knowing the range > of <http://www.example.org/terms/address>, perhaps from elsewhere) and/or a > URI (perhaps from an unambiguous predicate). So even if your application > required the type or URI the above triples wouldn't necessarily be useless > once they have been merged with others. > Agreed. - Mike ---------------------------------------------------- Michael C. Daconta Director, Web & Technology Services www.mcbrad.com
Received on Wednesday, 13 November 2002 13:45:49 UTC