Define inverse of properties

[kcoyle]

It is suggested that we define inverse properties for:

• expresses
• manifests
• instantiates

[kcoyle]

Possible terms:

• expressedBy
• manifestedBy
• instantiatedBy

[kcoyle]

Any chance we can limit this to:

• expressed
• manifested
• instantiated

or

• isExpressed
• isManifested
• isInstantiated

[kcoyle]

I'm not sure that we can define these as inverse properties in the OWL sense. From what little I can glean from the OWL documentation, if actually defined as inverse then they both must be object properties. This could explain the choices made by Dublin Core and Schema, to define separate properties that logically are inverse of each other, but not to use the OWL function.
https://www.w3.org/TR/2012/REC-owl2-primer-20121211/#a_ObjectInverseOf

[philbarker]

So we have various (non-exclusive) options,

1. don't have inverse properties
2. have properties but don't declare them as inverse of each other
3. use something other than owl: to declare the inverseness
4. make them object properties and not allow strings
5. make them object properties and allow a "Text" object
6. have different properties for referencing an object and literal / text.

My 2p: inverse properties are useful only if you declare that they are inverse in a meaningful way.

Literals can't have properties so that means both have to be object properties. You can use object properties with Blank Nodes if all you have is, say, a title. So

<example> a :Work ;
    :hasExpressionEntity [ dc:title "Moby Dick"] . 

I don't think option 5 adds anything to this. Assuming appropriate range of :hasExpressionEntity and an inverse property isExpressionOf you can infer that the blank node is an :Expression which has the property :isExpressionOf <example>

Option 6 would allow

<example> a :Work ;
    :hasExpressionNamed  "Moby Dick" .

All this can be done as extensions/profiles of any openWEMI vocabulary we produce, and as the preferences are probably very context dependent perhaps that is the way to go.

[kcoyle]

@philbarker Are you opting for your 1, don't have inverse properties? And figure that if someone wants them they can mint them?

[philbarker]

I don't want to seem too assertive given that I haven't followed all the discussion so far, but it seems to the least prescriptive option, the one that allows the broadest range of practice, and that seems in the spirit of Dublin Core. That said, I wouldn't be against the option at the other extreme, if that was the group will. I think some of the half-way options are maybe compromises that don't help end users as much.

[kcoyle]

Use:

• expressedBy
• manifestedBy
• instantiatedBy

define them as OWL:inverse properties; add to both sets of properties