Re: [protege-user] How to Create Sub-Property

Michael,

Thank you for the detailed explanations.

It seems the issue is what I expected: that some ontologies are not 
available at their URLs. That's too bad. But as you point out many are 
available and I think it would be very convenient if perhaps Protege 
tried to fetch the ontologies that had prefix mappings and make their 
terms available for use and only force you to add them if they were not 
available. Additionally, it would be nice if Protege just included the 
common ones by default. For me, those would include Dublin Core and FOAF.

I did just try manually adding dct:creator as an annotation property so 
that I could make my creator property a subproperty of that. I see that 
it adds this to the ontology file

###http://purl.org/dc/terms/creator
dct:creator rdf:type owl:AnnotationProperty .

This is /definitely/ not what I want! I wouldn't add that if I were 
creating my ontology by hand, why would I want it there when I use Protege?

I suppose I could try your other suggestion of importing the ontology 
and then deleting it at the end (your step 2.2 below), but honestly, 
that's terrible UX.

I'm going to take your other advice of not even including creator in my 
ontology and just use the dct one, but the issues I have still stand, in 
general.

I think there needs to be a distinction between things I want to 
/reference/ in my ontology as one usually does when using prefix 
mappings and which should not end up in the ontology file, and things I 
want imported and which would end up in the file.

Tim

On 9/12/25 15:41, Michael DeBellis wrote:
>
>     Yes, but /why/ is it not correct? 
>
>
> I didn't participate in defining the standard so I don't know but I 
> suspect the reason is that just because you add a mapping to your 
> ontology doesn't mean you want every entity in that mapped ontology in 
> your ontology. In fact, you almost never do. Consider SNOMED. Adding a 
> snomed: prefix doesn’t import SNOMED itself, it just gives you a 
> shorthand for writing IRIs. SNOMED CT is enormous — over 350,000 
> concepts — and covers diagnoses, procedures, medications, anatomy, and 
> more. It would be very unusual for any ontology to want or need to 
> reuse all of those concepts; in practice you only reference the 
> handful that are relevant to your domain.
>
> If you ever programmed in Python they do the same thing. At the top of 
> your file you have definitions like:
>
>     from rdflib import Graph, URIRef, Namespace, BNode, Literal
>
>
> There are a lot of other useful functions and other things in rdflib 
> but you only get the ones you need so you don't clutter up the 
> namespace and memory. You can do:
>
>     from rdflib import *
>
>
> Which gets you everything but that is considered a bad practice and 
> almost no one ever does it.
>
>     What I'm trying to understand is why Protege doesn't fetch the
>     ontology description of those ontologies I've created prefix
>     mappings for behind the scenes and make available to me all the
>     individuals, classes, properties, etc. so that I can reference
>     them in my ontology? Why am I forced to do it manually?
>
>
> For the same reason you should declare explicitly which code in a 
> Python library you want to use from a Python library. Because while 
> requiring you to explicitly add them is a bit more work it is far less 
> work than what would happen if it worked the other way and for example 
> you only wanted skos:prefLabel, skos:altLabel in your ontology but 
> once you put the SKOS mapping you got the whole thing.
>
>     I suppose a related question is this: why am I able to reference
>     some terms like rdf:type and rdfs:label without doing anything
>     special?
>
>
> Those are fundamentally different. Entities like rdf:type, rdfs:label, 
> and owl:Class are part of the meta-model that defines an OWL ontology. 
> You need to have them all to create an ontology to work. Most other 
> ontologies are domain ontologies where it is much less likely you will 
> ever want the whole thing.
>
>     I assume this is the case because I don't understand what you mean
>     by ". . . it can slow down your loading because every time you
>     load it has to go out to the Internet." If that's the case, isn't
>     it always true, even for references to rdf and rdfs?
>
>
> Most people get confused about this as well, I know I did. When you 
> use an IRI like http://www.w3.org/2000/01/rdf-schema#label The IRI is 
> a unique locator and there is no need to go to the W3C site when you 
> load an ontology that has rdfs:label as one of its entities. You CAN 
> go there because the IRI has been mapped to an Internet domain but 
> often with ontologies that people build that isn't the case. E.g., in 
> an ontology I'm working on now there are classes like: 
> https://www.michaeldebellis.com/climate_obstruction/Belief If you type 
> that into a browser you will get a 404 Error because I haven't mapped 
> the ontology IRIs to pages on my web site. A lot of ontologies are 
> like this. Although the ones that get reused the most are mapped. 
> E.g., if you enter: http://www.w3.org/2000/01/rdf-schema#label in your 
> browser, it will take you to the file where that entity is defined. A 
> more sophisticated example is if you enter: 
> https://www.w3.org/2009/08/skos-reference/skos.html#example in your 
> browser it will take you to the place in the SKOS documentation where 
> that entity is defined. But in all cases just because the IRIs can 
> sometimes be entered into a browser doesn't mean that when you load 
> them the system has to go to their site... as long as they are saved 
> in your ontology which is why I usually do the Merge Ontologies. That 
> way the IRIs are part of my ontology and no need to go over the 
> Internet to find them..
>
> On Fri, Sep 12, 2025 at 11:51 AM Tim McIver <[email protected]> wrote:
>
>>         My expectation is that properties from the ontologies for
>>         which I've created prefix mappings would show up there.
>>
>>
>>     That's not correct.
>
>     Yes, but /why/ is it not correct? I created the following small
>     ontology by hand:
>
>     @prefix owl:<http://www.w3.org/2002/07/owl#> <http://www.w3.org/2002/07/owl#> .
>     @prefix rdf:<http://www.w3.org/1999/02/22-rdf-syntax-ns#> <http://www.w3.org/1999/02/22-rdf-syntax-ns#> .
>     @prefix xml:<http://www.w3.org/XML/1998/namespace> <http://www.w3.org/XML/1998/namespace> .
>     @prefix xsd:<http://www.w3.org/2001/XMLSchema#> <http://www.w3.org/2001/XMLSchema#> .
>     @prefix rdfs:<http://www.w3.org/2000/01/rdf-schema#> <http://www.w3.org/2000/01/rdf-schema#> .
>     @prefix dct:<http://purl.org/dc/terms/> <http://purl.org/dc/terms/> .
>     @base<http://timmciver.com/file-ontology/> <http://timmciver.com/file-ontology/> .
>
>     <http://timmciver.com/file-ontology> <http://timmciver.com/file-ontology> rdf:type owl:Ontology ;
>                                           rdfs:label "An ontology for describing file data and metadata."@en .
>
>     :OWLDataProperty_f2f87965_60f0_4405_a160_36fe9bcb23e2 rdf:type owl:DatatypeProperty ;
>                                                            rdfs:subPropertyOf dct:creator ;
>                                                            rdfs:comment "The dateTime on which the FileData object was created."@en ;
>                                                            rdfs:label "created"@en .
>
>     (You can probably tell I cheated a bit and used part of a file
>     produced by Protege). I'm free to simply reference dct:creator
>     without "importing" anything.
>
>     What I'm trying to understand is why Protege doesn't fetch the
>     ontology description of those ontologies I've created prefix
>     mappings for behind the scenes and make available to me all the
>     individuals, classes, properties, etc. so that I can reference
>     them in my ontology? Why am I forced to do it manually?
>
>     I suppose a related question is this: why am I able to reference
>     some terms like rdf:type and rdfs:label without doing anything
>     special?
>
>     I'm fairly new to Protege but have quite a bit of familiarity with
>     linked data and RDF. It's possible that I just don't understand a
>     nuance here and would love to be enlightened. I assume this is the
>     case because I don't understand what you mean by ". . . it can
>     slow down your loading because every time you load it has to go
>     out to the Internet." If that's the case, isn't it /always/ true,
>     even for references to rdf and rdfs?
>
>     Thanks,
>     Tim
>
>
>>     Just because you have a mapping in Protege doesn't mean you can
>>     see all the entities that exist in an ontology that has the base
>>     IRI as the mapping. You still need to bring some or all of those
>>     entities into your ontology. There are a few ways to do this:
>>
>>     1) Manually. Don't recommend this but you can literally copy an
>>     IRI like:  IRI http://purl.org/dc/terms/creator and add that as a
>>     property in your ontology.This is more for special cases where
>>     you may only need a few entities and you need them ASAP to
>>     continue a task. If you do this you won't get all the annotations
>>     that are associated with that entity. E.g., things like
>>     Equivalent Property = http://xmlns.com/foaf/0.1/maker. You can
>>     copy some or all of those annotations by hand as well and I've
>>     done that in a few situations but that is not a good way to do it.
>>
>>     2) Import the ontology you want to use. This is the more common
>>     way. You probably know there is an Import Ontology subtab in the
>>     Active Ontology tab in Protege. After you import the ontology you
>>     can then delete entities that you don't need (at least at that
>>     time). What you do after that is debatable and the way I do it
>>     probably isn't something everyone would agree on but what I
>>     usually do after importing an ontology is use: Refactor>Merge
>>     Ontologies and merge the ontology I just imported into my main
>>     ontology. What I do after that is:
>>     2.1) Add documentation to an annotation property on the ontology
>>     itself that says something like "reused ontology Foo from
>>     https://FooBar.com <https://FooBar.com>
>>     2.2) Delete the ontology I imported from that same subtab on
>>     Active Ontologies.
>>
>>     The reason I do 2.2 is if I don't and I send the ontology to
>>     someone else, actually even for my own use, it can slow down your
>>     loading because every time you load it has to go out to the
>>     Internet. If you do the merge, the IRIs for the reused vocabulary
>>     don't change, its just that you have explicitly add them to your
>>     ontology rather than requiring others to wait for Protege to load
>>     the ontology from somewhere on the Internet.
>>
>>     Also, an issue that comes up is how do you handle changes to the
>>     ontology you imported? Do you want to add changes from the
>>     vocabulary to the version you have in yours or are you fine just
>>     using the entities as they existed and don't care if the reusable
>>     vocabulary changes.  I think there are some tools in the OBO
>>     foundry https://obofoundry.org/resources that help you do this
>>     process when you want to incorporate one vocabulary and keep
>>     consistent with it. I think the people at Semantic Arts also have
>>     tools for this for their Gist Upper Model:
>>     https://www.semanticarts.com/gist/   . But I've never used either
>>     of them.
>>
>>     Since you are a new Protege user, you might find this tutorial
>>     useful. It is a revision of the original Pizza tutorial that is
>>     one of the most common simple examples used to teach people how
>>     to build ontologies. I also added sections on SWRL, SHACL, and
>>     SPARQL that weren't in the original tutorial:
>>     https://www.michaeldebellis.com/post/new-protege-pizza-tutorial
>>
>>     One last thing: I don't understand your domain of course but I
>>     just thought I would mention you might want to re-think if you
>>     even need your own property called creator and can't just use the
>>     DCT one.
>>
>>     Cheers,
>>     Michael
>>     https://www.michaeldebellis.com/blog
>>
>>
>>     On Fri, Sep 12, 2025 at 7:29 AM Tim McIver <[email protected]> wrote:
>>
>>         Hello,
>>
>>         I'm a new Protege user. I'm attempting to design a custom
>>         ontology for a
>>         linked data application.
>>
>>         My current goal is to create a property `creator` that I wish
>>         to be a
>>         sub-property of `dct:creator`, the term from the Dublin Core
>>         Terms
>>         ontology. I have added `dct` as prefix under "Ontology
>>         Prefixes" with a
>>         mapping to http://purl.org/dc/terms/. I created my `creator`
>>         property as
>>         a Data Property and I attempted to make it a sub-property of
>>         `dct:creator` but when I attempt to do this, there is no
>>         `dct:creator`
>>         option in the dialog box. My expectation is that properties
>>         from the
>>         ontologies for which I've created prefix mappings would show
>>         up there.
>>
>>         My research suggests that I must import the ontologies that
>>         I'd like to
>>         use this way. Is this correct? I found this SO answer
>>         describing the
>>         process: https://stackoverflow.com/a/44266590. Is this still
>>         the method
>>         I should use?
>>
>>         Tim
>>
>>         _______________________________________________
>>         protege-user mailing list
>>         [email protected]
>>         https://mailman.stanford.edu/mailman/listinfo/protege-user
>>
>>
>>     _______________________________________________
>>     protege-user mailing list
>>     [email protected]
>>     https://mailman.stanford.edu/mailman/listinfo/protege-user
>     _______________________________________________
>     protege-user mailing list
>     [email protected]
>     https://mailman.stanford.edu/mailman/listinfo/protege-user
>
>
> _______________________________________________
> protege-user mailing list
> [email protected]
> https://mailman.stanford.edu/mailman/listinfo/protege-user

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