tombaker
commented
3 weeks ago @kcoyle Of course I agree with this and would like to add: Disjointness implies a conceptual universe with types that are clearly defined. But if one accepts the point made in Renear and Dubin (2008) (as pointed out by @annakasprzik) -- that three of the four WEMI "entity types" are more like roles than types -- roles specific to "particular social contexts" or "contingent circumstances" rather than "fundamental ontological types" or "genuine types", then that distinction alone should be enough to ensure that they cannot be disjoint.
The main idea to take away from OpenWEMI is that it is defined as a minimally constrained vocabulary on purpose. Because of this, anyone basing their own vocabulary on these concepts is free to add any constraints that they need. A downstream vocabulary can indeed define any or all of the locally defined classes as disjoint. If OpenWEMI defined its classes as disjoint, no one using OpenWEMI could make the decision to have them NOT be disjoint. The solution is to leave constraints to specific applications. This is a model that has worked well for Dublin Core Metadata Terms and OpenWEMI is based on that same philosophy.