has_role should not be used with information

[jamesaoverton]

I ran into a problem using IAO with SWO together, caused by these two axioms:

• BioPAX RDF/XML format subClassOf has_role some Knowledge representation role
• BioPAX Manchester OWL Syntax format subClassOf has_role some Knowledge representation role

The domain of has_role is independent continuant, but these are both information (generically dependent continuants). SWO does not include an assertion the domain for has_role, so this problem is not apparent when reasoning over SWO alone. IAO does include this assertion, so when IAO and SWO are merged, these two classes are unsatisfiable. I've always found it strange and unfortunate that information can't have roles under BFO, but that's how it works.

The easy fix is to remove these two axioms, but I'm not sure what the best alternative is for expressing that information.

[allysonlister]

I've had a review of swo relations as I've moved to the ODK over the past few weeks and found a few such cases. I think they were hidden by previous release just importing the bare minimum of info from the other ontologies. Now that ODK is properly importing all associated axioms, these crop up.

I've still got the ODK release in a branch because there are some outstanding issues. Hope to merge to master in the next couple of weeks. Feel free to have a look at https://github.com/allysonlister/swo/tree/upgrade-to-odk but note please that this is a WIP and may at any time have a version of the files that is bonkers :-)

Thanks!