names in webprotege

classic Classic list List threaded Threaded
5 messages Options
Reply | Threaded
Open this post in threaded view
|

names in webprotege

Mark Longo
Hi All, 

Is there some reason it would be a bad idea to have two classes that share the same rdfs:label?  In webprotege, if I try to create two classes with the same name, they are automatically mapped to the same IRI. 

I'm trying to capture the fact that there are different meanings of words. So for instance "posterior" can be both a spatial region (e.g. "posterior aspect of the X") or a relation (e.g. "posterior to the X").  So I have two classes called "aspect descriptor" and "relation descriptor."  If I create "posterior" subclasses under these two classes, they get linked to the same IRI. 

I can think of some clunky workarounds, but was wondering why Webprotege works this way. Maybe Protege does as well...haven't checked. 

Cheers,
Mark



_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user
Reply | Threaded
Open this post in threaded view
|

Re: names in webprotege

marksem
I don't use Web Protege, but do not think that should be so.

Regular Protege allows 2 Classes to have the same rdfs:label value.  

Attached example file was generated via Desktop Protege 5.5.0.

From: protege-user <[hidden email]> on behalf of Mark Longo <[hidden email]>
Sent: Thursday, January 23, 2020 9:11 PM
To: User support for WebProtege and Protege Desktop <[hidden email]>
Subject: [protege-user] names in webprotege
 
Hi All, 

Is there some reason it would be a bad idea to have two classes that share the same rdfs:label?  In webprotege, if I try to create two classes with the same name, they are automatically mapped to the same IRI. 

I'm trying to capture the fact that there are different meanings of words. So for instance "posterior" can be both a spatial region (e.g. "posterior aspect of the X") or a relation (e.g. "posterior to the X").  So I have two classes called "aspect descriptor" and "relation descriptor."  If I create "posterior" subclasses under these two classes, they get linked to the same IRI. 

I can think of some clunky workarounds, but was wondering why Webprotege works this way. Maybe Protege does as well...haven't checked. 

Cheers,
Mark



_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user

labeltest.owl (1K) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: names in webprotege

Michael DeBellis-2
I just tried it in Web Protege and I was able to set the rdfs:label annotation of two different classes and two different instances to be the same string with no problem. They are still different IRIs. Here is a screen snapshot with two classes from a small demo Car ontology I have. I set the label of the Car class and the label of the Engine class to both be the string Foo. The IRI's were not changed. 

Although, in general I don't think it's a good idea. It seems like you may be setting yourself up for confusion from others who use the ontology but it seems like you have a special use case that may justify it. 

DifferentLabels.PNG

On Thu, Jan 23, 2020 at 7:17 PM Mark Wallace <[hidden email]> wrote:
I don't use Web Protege, but do not think that should be so.

Regular Protege allows 2 Classes to have the same rdfs:label value.  

Attached example file was generated via Desktop Protege 5.5.0.

From: protege-user <[hidden email]> on behalf of Mark Longo <[hidden email]>
Sent: Thursday, January 23, 2020 9:11 PM
To: User support for WebProtege and Protege Desktop <[hidden email]>
Subject: [protege-user] names in webprotege
 
Hi All, 

Is there some reason it would be a bad idea to have two classes that share the same rdfs:label?  In webprotege, if I try to create two classes with the same name, they are automatically mapped to the same IRI. 

I'm trying to capture the fact that there are different meanings of words. So for instance "posterior" can be both a spatial region (e.g. "posterior aspect of the X") or a relation (e.g. "posterior to the X").  So I have two classes called "aspect descriptor" and "relation descriptor."  If I create "posterior" subclasses under these two classes, they get linked to the same IRI. 

I can think of some clunky workarounds, but was wondering why Webprotege works this way. Maybe Protege does as well...haven't checked. 

Cheers,
Mark


_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user

_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user
Reply | Threaded
Open this post in threaded view
|

Re: names in webprotege

Mark Longo
Thanks very much Michael and Mark. 
Looks like I am able to assign the same rdfs:label to different classes once I've already created them (i.e. change the labels once the classes are created).  However, if I try assigning duplicate labels at class creation I run into the mapping-to-one IRI issue. So for instance, this results in the creation of one class:

image.png
image.png

In any case, I have an easy workaround (create classes using whatever labels and then go back and change them to be identical or not). 

It sounds like the main reason I wouldn't want the same rdfs:labels is potential confusion among users?  We should be leveraging the IRIs to differentiate classes, so I'm not so worried about the confusion factor.  But I wonder if there are other reasons to avoid the practice. 

 

On Thu, Jan 23, 2020 at 8:27 PM Michael DeBellis <[hidden email]> wrote:
I just tried it in Web Protege and I was able to set the rdfs:label annotation of two different classes and two different instances to be the same string with no problem. They are still different IRIs. Here is a screen snapshot with two classes from a small demo Car ontology I have. I set the label of the Car class and the label of the Engine class to both be the string Foo. The IRI's were not changed. 

Although, in general I don't think it's a good idea. It seems like you may be setting yourself up for confusion from others who use the ontology but it seems like you have a special use case that may justify it. 

DifferentLabels.PNG

On Thu, Jan 23, 2020 at 7:17 PM Mark Wallace <[hidden email]> wrote:
I don't use Web Protege, but do not think that should be so.

Regular Protege allows 2 Classes to have the same rdfs:label value.  

Attached example file was generated via Desktop Protege 5.5.0.

From: protege-user <[hidden email]> on behalf of Mark Longo <[hidden email]>
Sent: Thursday, January 23, 2020 9:11 PM
To: User support for WebProtege and Protege Desktop <[hidden email]>
Subject: [protege-user] names in webprotege
 
Hi All, 

Is there some reason it would be a bad idea to have two classes that share the same rdfs:label?  In webprotege, if I try to create two classes with the same name, they are automatically mapped to the same IRI. 

I'm trying to capture the fact that there are different meanings of words. So for instance "posterior" can be both a spatial region (e.g. "posterior aspect of the X") or a relation (e.g. "posterior to the X").  So I have two classes called "aspect descriptor" and "relation descriptor."  If I create "posterior" subclasses under these two classes, they get linked to the same IRI. 

I can think of some clunky workarounds, but was wondering why Webprotege works this way. Maybe Protege does as well...haven't checked. 

Cheers,
Mark


_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user
_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user


--
Mark D. Longo, Ph.D.  |  Head of Data Science
Sirona Medical
236 8th Street, San Francisco, CA 94103
Cell: (650) 644-5647

_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user
Reply | Threaded
Open this post in threaded view
|

Re: names in webprotege

Michael DeBellis-2
Mark, Glad we could help and you found a work around. I think the reason you may be having that issue when
you create multiple classes is that when you do that the names you enter are not the rdfs:label rather they
are the last part of the IRI of each class you create.  I.e., when you do that I think you are telling Web 
Protege to create two classes with the same IRI which makes no sense. The default in Protege and 
I think Web Protege is the name you give is just the last part of the IRI (that typically comes after a "#" in the IRI) 
and the rdfs:label is blank by default, but if you set the rdfs:label then Protege uses that rather than the last part 
of the IRI to display the entity but it doesn't change the IRI. 

Regarding multiple labels for different IRIs, as far as I know the only potential problem is that
it may confuse others who try to use your ontology. 

Cheers,
Michael

On Fri, Jan 24, 2020 at 9:51 AM Mark Longo <[hidden email]> wrote:
Thanks very much Michael and Mark. 
Looks like I am able to assign the same rdfs:label to different classes once I've already created them (i.e. change the labels once the classes are created).  However, if I try assigning duplicate labels at class creation I run into the mapping-to-one IRI issue. So for instance, this results in the creation of one class:

image.png
image.png

In any case, I have an easy workaround (create classes using whatever labels and then go back and change them to be identical or not). 

It sounds like the main reason I wouldn't want the same rdfs:labels is potential confusion among users?  We should be leveraging the IRIs to differentiate classes, so I'm not so worried about the confusion factor.  But I wonder if there are other reasons to avoid the practice. 

 

On Thu, Jan 23, 2020 at 8:27 PM Michael DeBellis <[hidden email]> wrote:
I just tried it in Web Protege and I was able to set the rdfs:label annotation of two different classes and two different instances to be the same string with no problem. They are still different IRIs. Here is a screen snapshot with two classes from a small demo Car ontology I have. I set the label of the Car class and the label of the Engine class to both be the string Foo. The IRI's were not changed. 

Although, in general I don't think it's a good idea. It seems like you may be setting yourself up for confusion from others who use the ontology but it seems like you have a special use case that may justify it. 

DifferentLabels.PNG

On Thu, Jan 23, 2020 at 7:17 PM Mark Wallace <[hidden email]> wrote:
I don't use Web Protege, but do not think that should be so.

Regular Protege allows 2 Classes to have the same rdfs:label value.  

Attached example file was generated via Desktop Protege 5.5.0.

From: protege-user <[hidden email]> on behalf of Mark Longo <[hidden email]>
Sent: Thursday, January 23, 2020 9:11 PM
To: User support for WebProtege and Protege Desktop <[hidden email]>
Subject: [protege-user] names in webprotege
 
Hi All, 

Is there some reason it would be a bad idea to have two classes that share the same rdfs:label?  In webprotege, if I try to create two classes with the same name, they are automatically mapped to the same IRI. 

I'm trying to capture the fact that there are different meanings of words. So for instance "posterior" can be both a spatial region (e.g. "posterior aspect of the X") or a relation (e.g. "posterior to the X").  So I have two classes called "aspect descriptor" and "relation descriptor."  If I create "posterior" subclasses under these two classes, they get linked to the same IRI. 

I can think of some clunky workarounds, but was wondering why Webprotege works this way. Maybe Protege does as well...haven't checked. 

Cheers,
Mark


_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user
_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user


--
Mark D. Longo, Ph.D.  |  Head of Data Science
Sirona Medical
236 8th Street, San Francisco, CA 94103
Cell: (650) 644-5647
_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user

_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user