Re: repair unconventional convention: ... clarified!

Johan De Smedt wrote:

> For the hart of the matter, the SKOS reference specifies  
>  +- skos:broaderTransitive   [JB: declared as "transitive"]
>      |
>      +- skos:broader

> It says the set of triples in skos:broader is a subset of the set of triples in skos:broaderTransitive.
> If in an example, only skos:broader are given, then skos:broaderTransitive can be inferred as the transitive closure of the set of skos:broader triples.
 > And this does not entail that the sub-property skos:broader has the 
transitive property

YES, you are right. Why?

The trap I was caught: "transitive" is *not* a meta-property of a 
property, i.e. it is not sth. which is inherited down a property tree 
(as one might expect from OO thinking). Instead "transitive" is a label 
which tells an inferencing engine that it is ok to derive the transitive 
hull from this property.

http://www.w3.org/TR/2008/WD-skos-reference-20080829/skos.html#broaderTransitive
defines skos:broaderTransitive being transitive.

Does it mean that all subProperties are also transitive?
NO, quite the opposite ist true!

Let's give another example:

  +- ancestor
     +- parent

Where do we have to attach the metaproperty of being transitive?
You are right: to the "ancestor"-property, not to the "parent"-property!



> This clarified it for me.
dito. Thank you for clarification.


yours
Johannes



> 
> best, Johan 
> 
> -----Original Message-----
> From: public-esw-thes-request@w3.org [mailto:public-esw-thes-request@w3.org] On Behalf Of Johannes Busse
> Sent: Friday, 13 February, 2009 16:13
> To: Christophe Dupriez
> Cc: Alistair Miles; Stephen Bounds; public-esw-thes@w3.org
> Subject: repair unconventional convention: broaderTransitive should be subProperty of broader
> 
> 
> Hi SKOS editors,
> 
> today I realized a IMHO severe naming issue in the SKOS schema w.r.t. 
> skos:broader and skos:broaderTransitive.
> 
> My own understandig always was
> 
> Christophe Dupriez wrote:
>  > +- skos:broader (N1)
>  > |   |
>  > |   +- skos:broaderTransitive
> 
> My translation of the N1 visualization into First Order Logic:
> 
>   skos:broaderTransitive(X,Z) <-
>       skos:broaderTransitive(X,Y) AND skos:broaderTransitive (Y,Z).
> 
> and according to th RDF subproperty axioms:
> 
>   skos:broader(X,Y) <-
>       skos:broaderTransitive(X,Y).
> 
> 
> BUT
> http://www.w3.org/TR/2008/WD-skos-reference-20080829/#L4160
> suggests in fact:
> 
> Christophe Dupriez wrote:
>  > +- skos:broaderTransitive
>  > |   |
>  > |   +- skos:broader
> 
> 
> This visualization is coherent to your normal language explanation (c.f. 
> WD-skos-reference-20080829/#L4160 )of skos:broaderTransitive.
> 
> *Both are pretty confusing* (if not wrong, see below), because IMO it is 
> contrary to common naming policies in ontology engineering: You are 
> mixing up the intension and extension of a property.
> 
> In detail: Concatenating a term (i.e. the string "transitive") to the 
> name of a property (here: "has_broader") normally indicates that this 
> term adds an additional characteristic (here: being transitive) to this 
> property. It follows that the extension of the more restricted property 
> is a proper subset of the extension of the less restricted property 
> (which is the def of being a subproperty).
> 
> The implicit policy behind that naming convention is that a speaking 
> object ID (like skos:class, skos:broader etc.) describes informally the 
> respective object, here: the property skos:broaderTransitive. This means 
> that a string like "Transitive" normally is understood being an informal 
> description of an additional characteristics, i.e. the *intension* of 
> the property.
> 
> This habitus of interpreting substrings of an ontology object ID also 
> corresponds with other class naming conventiones, like
> 
> - horse
>    + horseBlack (a horse which is black)
>      . horseBlackMale (a horse which is black and male)
>    + whiteHorse
> 
> According to this habitus an ontology engineer would expect to have
> 
> - broader
>    + broaderTransitive (a broader relation which is transitive)
>      + broaderTransitiveIrreflexive (... and irreflexive)
> 
> What you in SKOS are doing: You are identifying the name of the property 
> skos:relatedTransitive with the *inferrable extension* of the property.
> 
> "Note especially that, by convention, skos:broader and skos:narrower are 
> only used to assert immediate (i.e. direct) hierarchical links between 
> two SKOS concepts. By convention, skos:broaderTransitive and 
> skos:narrowerTransitive are not used to make assertions, but are instead 
> used only to draw inferences."
> 
> This convention is strange; I'd not concede such a convention. What you 
> have defined here:
> 
> - skos:relatedTransitive
>    (pairs X,Y that are related by assertion *or* inference)
>    - skos:related
>      pairs X,Y that are related (only) by assertion
> 
> Transferred to the horse example above this would read as:
> 
> - horseBlackMale : Things that are horses or black or male
>    - horseBlack : Things that are horses or black
>      - horse : Things that are horses
> 
> Of course you are free to define such a semantics. But you should know 
> that this definition IMHO is *very* error prone and will lead to severe 
> misunderstandings and problems in the future.
> 
> ... Hm, and looking up the  SPEC again:
> http://www.w3.org/TR/2008/WD-skos-reference-20080829/skos.html#broaderTransitive
> claims that skos:broaderTransitive should be transitive (which means 
> that all subproperties are also transitive). Thus N1 holds, and the 
> visualization plus the explanation in #L4160 is inconsistent with the 
> schema.
> 
> ?
> 
> 
> yours
> Johannes


-- 
Dr. Johannes Busse, Senior Researcher
An der RaumFabrik 29, D-76227 Karlsruhe
Reg. Office: Karlsruhe, Amtsger. Mannheim, HRB 109540
Managing Directors:    Prof.Dr.J.Angele,  H.P.Schnurr
http://www.ontoprise.de   | phone x49(721) 509 809-62
mailto:busse@ontoprise.de | mobile x49(163) 509 80-62

Received on Saturday, 14 February 2009 06:59:02 UTC