This is a simple scenario, I can accomplish the specific thing I am trying to di by other means, but if I can get this to work it leads to a strategy inspired by SPIN, I'm showing it though because it is a simple and clear example.
I have two graphs, G1: [] a :DatasetArtifact ; :name e:tdb ; :path "/lei/testCase/a" . G2: @prefix : <http://rdf.ontology2.com/mogrifier1/> :DatasetArtifact :implementation <java:com.ontology2.rdf.mogrifier.MemoryDatasetFactory> . :GraphArtifact :implementation <java:com.ontology2.rdf.mogrifier.GraphInsertFactory> . :LuceneArtifact :implementation <java:com.ontology2.rdf.mogrifier.LuceneFactory> . Now I can query the union of G1 and G2 and I want to make the graph above look like [] a :DatasetArtifact ; :name e:tdb ; :implementation <java:com.ontology2.rdf.mogrifier.MemoryDatasetFactory> . :path "/lei/testCase/a" . I can do the following CONSTRUCT query prefix : <http://rdf.ontology2.com/mogrifier1/> CONSTRUCT { ?artifact :implementation ?class } WHERE { ?artifact a ?type . ?type :implemenation ?class . } And this will do the right thing if the "root" node is not a bnode but if it is a bnode it creates an output like G3 [ :implementation <java:com.ontology2.rdf.mogrifier.MemoryDatasetFactory> ] and this bnode is not the same bnode as the original bnode so when I take the model I get from execConstruct and append G3 to G1 I don't get the desired result. If I try to have the construct put results back into the original model I get a ConcurrentModificationException so I can't do it that way. Operations like this where you pull out a bnode reference and then add some properties to it are easy to do with the Model interface, but not officially supported in SPARQL. I know that Jena and some other stores support ways to get bnode identifiers out of SPARQL queries so it seems to me that has to be some trick, even if it is a touch evil, that makes it possible to do something like the above. I could skolemize the roots, but I might have 2nd level or 3rd level bnodes in some records. There are a number of other ways I could merge in the :implementation that would work just fine in this case, but I like the simplicity of this solution, the ability to do much more complex things with more complex queries and the composibility of this approach. Is this practical to attain? -- Paul Houle *Applying Schemas for Natural Language Processing, Distributed Systems, Classification and Text Mining and Data Lakes* (607) 539 6254 paul.houle on Skype ontolo...@gmail.com :BaseKB -- Query Freebase Data With SPARQL http://basekb.com/gold/ Legal Entity Identifier Lookup https://legalentityidentifier.info/lei/lookup/ <http://legalentityidentifier.info/lei/lookup/> Join our Data Lakes group on LinkedIn https://www.linkedin.com/grp/home?gid=8267275