- From: Richard Cyganiak <richard@cyganiak.de>
- Date: Sun, 21 Oct 2012 19:55:47 +0100
- To: Henry Story <henry.story@bblfish.net>
- Cc: Roger Menday <roger.menday@uk.fujitsu.com>, public-ldp-wg Working Group <public-ldp-wg@w3.org>
Hi Henry, On 17 Oct 2012, at 19:39, Henry Story wrote: > The notion of notion relative-uri-graphs - "direct graphs" in [8] - is not the same as a the notion of a plain rdf-graph. If you want to use a metaphor of child and the mother, the child say being in the context of the mother's womb, and so dependent on it - then the "direct graphs" is such a thing. It is incomplete without its context - as far as certain types of reasoning goes that is. > > (It would be interesting to work out what kind of reasoning one can still do with such a graph, but not one we need to solve for this issue.) The “direct graph” in the RDB2RDF Direct Mapping is composed of relative IRIs simply because that simplified presentation of the concept in the spec. The philosophical implications that you see in the concept of a graph with relative IRIs certainly played no role in the creation of that specification. Graphs with relative IRIs have quite a number of disadvantages compared to all-absolute graphs. For example, unlike normal RDF graphs, two such graphs cannot be merged. Best, Richard >>> [8] http://www.w3.org/TR/rdb-direct-mapping/#emp-addr
Received on Sunday, 21 October 2012 18:57:37 UTC