RE: Relationships and Relations

Hi Pat,

 

Thank you for that very last sentence: “....saying that the relation is a class actually makes good sense.“ !

 

It gives a (partial) answer to a question I couldn’t find an answer for in the impenetrable W3C documentation: What qualifies as owl:Class? 

It is, to me, a general problem that those documents don’t give generic definitions, only examples.

 

After some searching I found:  “In general classes are used to group individuals that have something in common in order to refer to them. Hence, classes essentially represent sets of individuals.”

Who then is the judge which “individuals” qualify to be grouped into a class? The folks who designed an OWL  parser? And where can I find what qualifies as “individual”? Something you can touch? But is Xmas 2018 not an individual, a member of all Xmas’s since the year zero?

 

It seems to me that an official W3C vocabulary should be composed (and maintained!) that defines everything, that you can run into, in formal, consistent, textual definitions, including what is legal in which context.

To illustrate my confusion I digged up a diagram that I made 13 years ago in an attempt to understand that confusing stuff. It was based on what I read in the various W3C douments.

 



 

I guess this is what “organic growth” entails :)

 

Thanks again for your attention, and thanks to those other folks who also responded.

 

Regards,

Hans 

15926.org <http://15926.org/>  

________________________________________

From: Patrick J Hayes <phayes@ihmc.us> 
Sent: zondag 10 november 2019 01:31
To: Hans Teijgeler <hans.teijgeler@quicknet.nl>
Cc: Semantic Web <semantic-web@w3.org>
Subject: Re: Relationships and Relations

 

I  just realized that my first paragraph, below, misses the point completely, because you don't want to talk about the property, but rather a particular case of it. But the more general point remains: you can use the same name for the relation and for the class of ‘relationships'; in fact, they can be the same thing. So in your case you could write

 

Pete isMarriedTo Mary

 

MarriagePeteMary rdf:type isMarriedTo

 

MarriagePeteMary hasHusband Pete

 

where hasHusband is now a ‘role’ property, like timeOf.  Or, of course, you could keep separate names, like Marriage for the class and married for the property. I like the punning version, myself, but YMMV. 

 

I wouldn't call MarriagePeteMary a ‘relationship’, though, as that feels like a category error. It is more like a fact or a circumstance: it has many other preperties than just those linking it to Pete and Mary. For example, it has a lifetime, and it was begun at some definite place, and it might have a certificate, and lots of other things. It's more like a particular instance of a relationship, if relationships had instances, which is why biting that bullet and saying that the relation is a class actually makes good sense. 

 

Pat





On Nov 9, 2019, at 5:19 PM, Patrick J Hayes <phayes@ihmc.us <mailto:phayes@ihmc.us> > wrote:

 

Hi Hans

 

In RDF, RDFS and the ‘full’ versions of OWL, you can use the same IRI as both a property name and as a subject, so you can simply talk about the property without needing to make your distinctions. (The same technique is used in the vastly more expressive language ISO Common Logic, so you can rest assured that doing this does not break anything in the foundations.) The first version of OWL-DL prohobited this, but the more recent OWL2-DL allows it as a form of ‘punning’, so again there is no pressing need to make your distinction: just use the relation name in both roles. 

 

As others have said, there is already quite a literature (not just in Webbish, but going back into philosophical logic and formal linguistics) on this general topic. Senences describing actions are often analyzed as talking about ‘events’ or ‘heppenings’ and some collection of other things which bear relationships such as agentOf, placeOf, timeOf and so on to the single central entity, which is classified by the verb. So “Jim cut the bread in the kitchen quickly and silently” asserts that an event classed as a Cutting (an OWL class of events) exists with JIm as its agent and something classed as Bread as its object and having the location Kitchen and the properties Silent and Fast. This style of representation has many advantages, but it does require everyone to use it consistently if they are going to be able to communicate reliably. But if you don’t want to get deeply into this stuff, there is no principled barrier to just conflating your relation and relationship ideas back into one, and using OWL2 or RDFS directly. 

 

Best wishes

 

Pat

 

 

On Nov 9, 2019, at 6:24 AM, <hans.teijgeler@quicknet.nl <mailto:hans.teijgeler@quicknet.nl> > <hans.teijgeler@quicknet.nl <mailto:hans.teijgeler@quicknet.nl> > wrote:

 

Hi,

 

I would like to hear your opinion about the following.

 

I propose to make a distinction between the terms ‘Relationship’ an ‘Relation’ (‘Property’), not for linguistic reasons but to avoid reification when that is not necessary.

I know that I am on thin ice, so be it.

 

Right now we have  something like 

* Pete isHusbandOf Mary
* Mary isWifeOf Pete.

But these Relations/Properties actually are Roles in a missing Relationship called Marriage.

 

We can also state:

* MarriagePeteMary hasHusband Pete
* MarriagePeteMary hasWife Mary

where MarriagePeteMary is Relationship and an instance of the owl:Class ‘Marriage’, or rather its specialization ‘Hetero Marriage’.

As a consequence we can easily represent information about that Relationship.

 

It appears to me that there are many such Relationships that qualify for being an owl:Class in their own right.

Think about Parenthood, Composition, Employment, etc.

 

Please concur or shoot.

 

Regards,

Hans 

 <http://15926.org/> 15926.org