RE: SUO: RE: Evaluation Framework for Content Standards
Matthew et al
-----Original Message-----
... in response to John Sowa's comment suggestion:
> Some of these implications will be handled automatically if
> Person4D and Person3D are both defined as subtypes of Person.
> Any properties that are independent of the 3D/4D distinction
> would be defined for the type Person, and they would therefore
> apply equally well to the same person under either description.
MW: The problem is that there is not one John Sowa, but a 3D
John Sowa and a 4D John Sowa, not one that is both 3D and 4D, so
you need something that relates them and says they are equivalent.
I take it that John's proposal *is* predicated on the assumption that there is (in the real world) one John Sowa, who can be identified and described from multiple perspectives (e.g., using a 3D ontology or a 4D ontology). I think that its important here to separate two components within the lattice of theories:
1. When seeking a capability to accommodate different ontologies within a common framework, we have to seek out concepts that are common across these different perspectives. In this example, the common_person_concept is necessarily an abstraction that shares some of the intent of both 3d_person_concept and 4d_person_concept. We need to be able to assert some minimum level of equivalence/correspondence, through abstraction, if we are to have any hope of success then the ability to define a deterministic method by which I can communicate facts from a 3d world view to a 4d world view. Otherwise we might as well start on defining multiple point-to-point mappings between ontologies.
Therefore, within the lattice, there must be relationships between 3d_person_concept and common_person_concept and between 4d_person_concept and common_person_concept ... we shouldn't assume, though, that these relationships are necessarily supertype/subtype.
2. In addition to such relationships that deal with concepts, we need to (separately) address how these concepts are represented in these different world views. Coming back to the process/event discussion, the distinct viewpoints there could be addressed by having a common_event concept that can accommodate both views, and also have necessary relations/constraints that define the necessary and/or preferred representations of these concepts in their respective views. Therefore, the determination of whether a common_event has location in space and/or time is distinct from the determination of whether it is necessary to know the spatial or temporal location of a specific event. This would help to avoid recourse to optional/default relationships at an SUO level, which I see as an unhelpful mixture of ontology and data modeling.
regards
Julian