MinCardinality - Revision history

imported>Denny: Reverted edits by Ihonevigefa (Talk); changed back to last version by Denny

2010-12-19T08:35:16Z

Reverted edits by Ihonevigefa (Talk); changed back to last version by Denny

imported>Ihonevigefa at 08:21, 24 November 2010

2010-11-24T08:21:04Z

imported>Denny at 03:17, 28 December 2007

2007-12-28T03:17:02Z

imported>Denny: New page: {{pubdate|5|April|2005}} More on the Unique Name Assumption (UNA), because [http://morenews.blogspot.com/2005/04/uniqueness-and-open-world.html Andrew answered on it], with further arg...

2007-12-27T21:30:27Z

New page: {{pubdate|5|April|2005}} More on the Unique Name Assumption (UNA), because [http://morenews.blogspot.com/2005/04/uniqueness-and-open-world.html Andrew answered on it], with further arg...

New page

{{pubdate|5|April|2005}}
More on the [[Unique Name Assumption]] (UNA), because [http://morenews.blogspot.com/2005/04/uniqueness-and-open-world.html Andrew answered on it], with further arguments. He quotes [http://gearon.blogspot.com/2005/04/more-cardinality-after-prototypos.html Paul]: " The initial problem was cardinality and OWL Flight attempts to solve the problem with cardinality. [http://gearon.blogspot.com/2005/04/more-cardinality-after-prototypos.html Paul put it succinctly]: "So what is the point of statements with the owl:minCardinality predicate? They can't ever be false, so they don't tell you anything! It's kind of like a belt and braces when your belt is unbreakable." "

Again I disagree, this time to Paul: the minimal cardinality axiom does make sense. For what, they ask - well, for saying that there is a minimal cardinality on this relation. Yeah, you are right: this is an axiom which hardly can lead to an inconsisten ontology. But so what? You nevertheless can cut down the number of possible models with it and get more information out of the ontology.

"I would agree - this was my main problem - how do you explain to Joe (and Andrew) that all his CDs are the same rather than different."

That's turning around the argument. If the reasoner would claim that all of Joes CDs are the same, he would be doing a grave mistake. But so would he if he would claim that all are different: the point is, he just doesn't know. Without having someone to state sameness or difference explicitly, well, you can't know.

"I did comment that the resolution, local unique names using AllDifferent, didn't actually seem to solve the problem well enough (without consideration for scalability for example)."

I am not sure why that should be. It seems that Andrew would be happy if there was a file-wide switch claiming "If I use different URIs here I mean different objects. This file makes the UNA." These files would easily be translated to standard OWL files, but there would be less clutter inside (actually, everything that would need to be done is adding an axiom of allDifferent with all the names of the file).

"I have a feeling that context is a better solution to this problem (that might just be my golden hammer though)."

I don't understand this one, maybe Andrew will elaborate a bit on this.

If you imagine an environment with axioms floating around, from repository to repository, being crawled, collected, filtered, mapped and combined, you must not make the Unique Name Assumption. If you remain in your own personal knowledge base, you can embrace UNA. And everything you need between is one more axiom.

Is it that bad?

<noinclude>{{Semantic Nodix post}}</noinclude>
{{Missing comments}}

← Older revision		Revision as of 03:17, 28 December 2007
Line 21:		Line 21:

	<noinclude>{{Semantic Nodix post}}</noinclude>		<noinclude>{{Semantic Nodix post}}</noinclude>
−	~~{{Missing comments}}~~

@@ Line 1: / Line 1: @@
 {{pubdate|5|April|2005}}
-More on the [[Unique Name Assumption]] (UNA), because [http://morenews.blogspot.com/2005/04/uniqueness-and-open-world.html Andrew answered on it], with further arguments. He quotes [http://gearon.blogspot.com/2005/04/more-cardinality-after-prototypos.html Paul]: &quot; The initial problem was cardinality and OWL Flight attempts to solve the problem with cardinality. [http://gearon.blogspot.com/2005/04/more-cardinality-after-prototypos.html Paul put it succinctly]: &quot;So what is the point of statements with the owl:minCardinality predicate? They can't ever be false, so they don't tell you anything! It's kind of like a belt and braces when your belt is unbreakable.&quot; &quot;
+More on the [[Unique Name Assumption]] (UNA), because [http://morenews.blogspot.com/2005/04/uniqueness-and-open-world.html Andrew answered on it], with further arguments. He quotes [http://gearon.blogspot.com/2005/04/more-cardinality-after-prototypos.html Paul]: " The initial problem was cardinality and OWL Flight attempts to solve the problem with cardinality. [http://gearon.blogspot.com/2005/04/more-cardinality-after-prototypos.html Paul put it succinctly]: "So what is the point of statements with the owl:minCardinality predicate? They can't ever be false, so they don't tell you anything! It's kind of like a belt and braces when your belt is unbreakable." "
 Again I disagree, this time to Paul: the minimal cardinality axiom does make sense. For what, they ask - well, for saying that there is a minimal cardinality on this relation. Yeah, you are right: this is an axiom which hardly can lead to an inconsisten ontology. But so what? You nevertheless can cut down the number of possible models with it and get more information out of the ontology.
-&quot;I would agree - this was my main problem - how do you explain to Joe (and Andrew) that all his CDs are the same rather than different.&quot;
+"I would agree - this was my main problem - how do you explain to Joe (and Andrew) that all his CDs are the same rather than different."
 That's turning around the argument. If the reasoner would claim that all of Joes CDs are the same, he would be doing a grave mistake. But so would he if he would claim that all are different: the point is, he just doesn't know. Without having someone to state sameness or difference explicitly, well, you can't know.
-&quot;I did comment that the resolution, local unique names using AllDifferent, didn't actually seem to solve the problem well enough (without consideration for scalability for example).&quot;
+"I did comment that the resolution, local unique names using AllDifferent, didn't actually seem to solve the problem well enough (without consideration for scalability for example)."
-I am not sure why that should be. It seems that Andrew would be happy if there was a file-wide switch claiming &quot;If I use different URIs here I mean different objects. This file makes the UNA.&quot; These files would easily be translated to standard OWL files, but there would be less clutter inside (actually, everything that would need to be done is adding an axiom of allDifferent with all the names of the file).
+I am not sure why that should be. It seems that Andrew would be happy if there was a file-wide switch claiming "If I use different URIs here I mean different objects. This file makes the UNA." These files would easily be translated to standard OWL files, but there would be less clutter inside (actually, everything that would need to be done is adding an axiom of allDifferent with all the names of the file).
-&quot;I have a feeling that context is a better solution to this problem (that might just be my golden hammer though).&quot;
+"I have a feeling that context is a better solution to this problem (that might just be my golden hammer though)."
 I don't understand this one, maybe Andrew will elaborate a bit on this.
@@ Line 28: / Line 20: @@
 Is it that bad?
-&lt;noinclude&gt;{{Semantic Nodix post}}&lt;/noinclude&gt;
+<noinclude>{{Semantic Nodix post}}</noinclude>