is rdfg:subGraphOf deductive or assertive


In Carroll et al 2004, 2005 rdfg:subGraphOf is proposed as a statement to be true iff the set of statements in the subject graph is a subset of the statements in the object graph.

>From an earlier discussion on this list I gather that the semantics is meant purely deductive, ie an inferencer could conclude the truth of the statement from inspecting the two graphs.

For my application, where I would like to execute queries over sets of nested subgraphs, I would like an 'assertive' subgraphof statement, much in the way of the subclassof statement, so that

   xxx  {aaa bbb ccc}
&& yyy? {xxx rdfg:SubGraphOf yyy}
-> yyy  {aaa bbb ccc}

[Taking the condition that the subgraphof statement needs to be in the supergraph from (Guha et al 2004); we can also relax this constraint]

- Was this ever intended by the named graphs interest group? (Why not?)
- Is there any reason not to add this to an inferencer? (Such as sesame?)
- Should different vocabulary from rdfg:SubGraphOf be used?


-- Wouter

o Wouter van Atteveldt 
 PhD Student, Free University Amsterdam
 Department of Artificial Intelligence 
 & Department of Communication Science 

Received on Tuesday, 30 May 2006 12:01:26 UTC