- From: Wafik Farag <Wafik@Farag.net>
- Date: Wed, 28 Sep 2005 16:04:50 -0400
- To: "Mork, Peter D.S." <pmork@mitre.org>, "Melissa Cline" <cline@pasteur.fr>
- Cc: <helen.chen@agfa.com>, <public-semweb-lifesci@w3.org>, <public-semweb-lifesci-request@w3.org>
Hi Peter, >RDF is a directed, edge-labeled graph; it is not acyclic. Moreover, if >you combine two DAGs, you are not guaranteed to get a DAG. For >example, consider tree T (which is a DAG). Now consider T', in which >every edge in T has been reversed (also a DAG). If you combine T and >T', the result is cyclic (assuming T contains at least one edge). I am glad you clarified this out. This means cyclic graphs are permitted in SW! In fact, the relational model integration problem stems from the fact that the schema must maintain DAG at the end. Hence, the integration effort is consumed in maintaining no cyclic arcs in the combined schema resulting a consistent and correct response to every query presented. A schema MUST be a DAG. I have seen some workarounds in public databases that did not maintain this rule and it was a mess. If this is true - how will an over-arching ontology linking two ontologies render consistent answers with cyclic graphs permitted in the ontology? -Wafik
Received on Wednesday, 28 September 2005 20:05:00 UTC