Is geox:transitiveSfOverlap useful?

[dr-shorthair]

geox:transitiveSfOverlap is a sub-property of geo:sfOverlap.

It was added to manage the case where the existence of an overlap has been inferred from the situation where a feature that is contained by the context feature is also within a second feature.

This is axiomatized as follows:

  owl:propertyChainAxiom (
      geo:sfContains
      geo:sfWithin
    ) ;

However, the resulting inferred property differs from a normal geo:sfOverlaps only in terms of the way it was detected - i.e. the provenance of the relationship. IMHO provenance does not belong in a relationship name - it is additional information that might be associated by reification, or named-graph membership, or something.

This approach is related to the way that @nicholascar had used OWL axiomatization in general. The theory was that a lot of relationships could be built by the reasoner, at load-time. This is a fine idea, but does depend on (a) all the ontologies being consistent (b) all the data being correct.

Loading several ontologies developed by different people at different times in different styles, along with several (large) datasets with inferencing turned on is a risky business - you will get way more additional axioms than you need in practice, and there will almost certainly be unintended consequences. Rather I would recommend 'pre-processing' using SPARQL CONSTRUCT/INSERT in order to add specific relationships that are known to be needed. The distinct property based on provenance only is generally unhelpful IMO.