Re: triple terms are not a universal panacea

On 7/25/2024 12:01 PM, Peter F. Patel-Schneider wrote:
> In the end, it seems to me that the old way is actually a better way 
> of capturing events.

I totally agree, and wonder if you could go further: the "old" way might 
be a better way of capturing any relations where attribution, 
provenance, context, or (bi-)temporality are important.

Coincidentally, I started working on a "relator-oriented" ontology last 
night where all domain relations are modeled with relators. It's more 
verbose, but provides a lot of expressivity and seems to solve a lot of 
problems

I'm also considering avoiding rdfs:Class in the domain model, i.e., 
modelling classes/sets/types as individuals, to avoid the expressiveness 
issues around classes that might hinder a relator-oriented approach. If 
anyone has any thoughts or has been down those roads before, please 
reach out to me privately!

What would it look like to have a "domain-classes-as-individuals" data 
model where relationships are primarily expressed by relators?


instead of

|:mattGroening :hasName "Matt" . :mattGroening :named :bart . :bart 
:hasName "Bart" . :bart a :FictionalCharacter . :bart a :Person . :bart 
a :SimpsonFamily |

we would have

|:matt-groening-hasName-Matt a :NamingRelation ; :named-thing 
:matt-groening ; :name "Matt" . :matt-groening-named-Bart a 
:NamingRelation ; :name-giver :matt-groening ; :named-thing :bart ; 
:name "Bart" . :bart-isa-Simpson a :MembershipRelation ; :member :bart ; 
:group :simpsons-family . :bart-isa-character-in-reality a :KindRelation 
; :context :reality ; :individual :bart ; :kind :fictional-character . 
:bart-isa-person-in-simpsons-world a :KindRelation ; :context 
:simpsons-world :individual :bart :kind :person |

and then you can go to town with relator enhancement (temporality, 
attribution) etc.