RDFS Issues

[Aklakan]

Version

5.6.0-SNAPSHOT

What happened?

Having recently touched jena-geosparql, which uses some RDFS inferencing, brought me back to some issues:

• MatchRDFS can produce literals in subject position: (X p "lit") (p range T) -> ("lit" a T).
• Many duplicate inferences even on small result sets. Perhaps introduce windowed deduplication?
• Inconsistent inferences depending how find() is invoked when there are sub-properties of rdf:type. The following examples fails for example for patterns: :fido ?p ?o, ?s a ?o. I'll open a PR with a test case for all failing patterns.

# Schema
:directType        rdfs:subPropertyOf rdf:type .

# Data
:fido :directType :Dog .

Relevant output and stacktrace

Are you interested in making a pull request?

None