Adding multiple domains to objectProperty in Protege 5

Question

I have created an ontology using Protege 5-beta-17. In my ontology I have some classes:

Mountain, Lake, Location etc...

I also have an object property:

hasLocation.

For this object property I have set the range the "Location" class, and the domain the "Mountain" and "Lake" classes.

When I try to view the ontology using the CMap tool it shows that only the

"Mountain" "hasLocation" "Location".

The "Lake" class is presented without the "hasLocation" object property.

Did I do something wrong? Ore do I have do something else in Protege?

Compagnie

Did you have a look at the OWL/XML (or whatever you save it as) and checked if the two domains are there? – Marathon

Yes I have seen the file. The domains where there. – Compagnie

But in Cmap just one of them appears. – Compagnie

Answer 1

I found out what the problem was.

When adding a domain/range to object property in protege you have to click the following buttons and select one of your classes:

If you want to add another domain/range you simply click one of the buttons again and add another class. If you are doing it like this your telling Protege that the domain/range of your object property is an INTERSECTION of two classes. This means that the individual that will take the domains/ranges place is an INSTANCE OF BOTH CLASSES and NOT EXCLUSIVELY OF ONE OF THEM. This was my mistake. I was adding the classes to the domain in the wrong way.

So... The correct way for adding multiple distinct domains for an object property is the following:

Simply click the domain/range button again and select the "Class expression editor" tab:

And in the "Class expression editor" type in your classes like this: "ClassA or ClassB or ClassC or ...".

In my case it was "Mountain or Lake".

After that click "ok" and thats it.

Compagnie

This is the right way to accomplish what you're trying to do, but I'd quibble with the description. If a property P has a a class D as a domain, then whenever we have an assertion P(x,y), we can infer that x is an instance of D. If P has multiple domains, say D1 and D2, then from P(x,y) we can infer that x is both a D1 and is a D2. That's what happens if P has multiple domains. What you're actually trying to do is add a single domain that is a union class. If P has the union class "(D1 or D2)" as a domain, then from P(x,y) we can infer that x is a "(D1 or D2)". That's a single domain. – Livy