Ontograf and inference display and property filter

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

Ontograf and inference display and property filter

Julien FAURE
Hi there,

I'm new to this forum, protégé software and ontology in general so I
apologize in advance if the following looks naive to all of you.

I've created a small ontology (following a beginner tutorial) that is:Thing
----> Animal
----> Plant
----> Carnivore

Where a Carnivore is a "(Animal or Plant) Eat Animal" and "Eat" an object
property.

1) I have one Individual named Simba which is explicitly a Carnivore.I
wonder why no line links Simba to Animals with "Eat" label in Ontograf?
While such arrow links Carnivore to Animal. I guess this is the same reason
why I do not have "Simba Eat Animal" as inferred information (yellow
background).

2) I have added a subclass of Animal named Bipede. I added an Individual
named Human that has a "Data property assertion" saying pawsNumber
"2"^^xsd:int. I would like to have Human selected automatically in Bipede
Class. I mean by "automatically" that if I change pawsNumber of Human to 3
then it will be excluded from Bipede. I tried to use "Target for Key" in
Bipede class but failed. Is it possible to write such filter?

Thanks in advance
Julien



--
Sent from: http://protege-project.136.n4.nabble.com/Protege-User-f4659818.html
_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user
Reply | Threaded
Open this post in threaded view
|

Re: Ontograf and inference display and property filter

marksem
Hi Julien.  It is always helpful if you attach the example ontology file to your email.  This allows others to help you more quickly, and shows us your model explicitly. 

From: protege-user <[hidden email]> on behalf of Julien FAURE <[hidden email]>
Sent: Thursday, February 27, 2020 7:36 AM
To: [hidden email] <[hidden email]>
Subject: [protege-user] Ontograf and inference display and property filter
 
Hi there,

I'm new to this forum, protégé software and ontology in general so I
apologize in advance if the following looks naive to all of you.

I've created a small ontology (following a beginner tutorial) that is:Thing
----> Animal
----> Plant
----> Carnivore

Where a Carnivore is a "(Animal or Plant) Eat Animal" and "Eat" an object
property.

1) I have one Individual named Simba which is explicitly a Carnivore.I
wonder why no line links Simba to Animals with "Eat" label in Ontograf?
While such arrow links Carnivore to Animal. I guess this is the same reason
why I do not have "Simba Eat Animal" as inferred information (yellow
background).

2) I have added a subclass of Animal named Bipede. I added an Individual
named Human that has a "Data property assertion" saying pawsNumber
"2"^^xsd:int. I would like to have Human selected automatically in Bipede
Class. I mean by "automatically" that if I change pawsNumber of Human to 3
then it will be excluded from Bipede. I tried to use "Target for Key" in
Bipede class but failed. Is it possible to write such filter?

Thanks in advance
Julien



--
Sent from: http://protege-project.136.n4.nabble.com/Protege-User-f4659818.html
_______________________________________________
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: Ontograf and inference display and property filter

Julien FAURE
I attached my .owl file to this reply.
PlantAnimal.owl
<http://protege-project.136.n4.nabble.com/file/t377201/PlantAnimal.owl>  



--
Sent from: http://protege-project.136.n4.nabble.com/Protege-User-f4659818.html
_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user
Reply | Threaded
Open this post in threaded view
|

Re: Ontograf and inference display and property filter

Michael DeBellis-2
Julien, I think I see the problems. 

1) You have asserted that Simba is an instance of Carnivore. That means Simba must have the value of it's Eat property be  some instance of Animal. But you have not asserted any values for Simba's Eat property. Your ontology is really small now but imagine it was a more realistic size. There would likely be hundreds or more instances of Animal. There is no way the reasoner can know WHICH instances of Animal that Simba has eaten. So that value can't be deduced by the reasoner, you or some program has to fill it in explicitly (or you could have some rules that explicitly describes which Animals Simba eats... although that's not typically how rules are used, they usually are used to describe things more at the class level or TBox level which I'll explain next). 

That is a second issue with how I think you are understanding OWL. You asked " why I do not have "Simba Eat Animal" as inferred information " There are two kinds of knowledge in an ontology what are called (for historical reasons and to make things confusing... just kidding on that last part) TBox and ABox. TBox means Terminology it is about classes and properties and general things that can be said about them. ABox is for assertions, it's about the individuals in your ontology and the specific values they have. I think you are confusing TBox and Abox. Saying a Carnivore eats Animals is TBox information. It describes general things about classes. Saying Simba ate part of SamAntelope who was recently killed by Simba's tribe is ABox information.  

Btw,, the convention is to make properties start with a lowercase letter and then use upper case if there are subsequent words. So typically a property would be named "eats" rather than "eats" or "eatsAnimals" if you need 2 or more words for the property name.  You won't get any errors if you don't do this and some organizations have their own standards but other things being equal it is a good idea to follow the standard unless you have a good reason not to. 

2) First, I think you may be again confusing TBox and ABox. Human would typically be a class not an instance. JohnDoe would be an instance. This isn't always straight forward though, often when to make something a class vs. an individual requires some thought about the model. In this case though I think Human should be a class. However, I'll ignore that for now and treat Human as an individual. If you want Human to be automatically classified to be a Biped that is easy. You need to make Biped what is called a Defined Class. Add the axiom: "Animal and pawsNumber value "2"^^xsd:integer"  to the "Equivalent to" definition of Biped. Also, you need to make Human an instance of Animal so that it satisfies both conditions of that axiom (or if you want to include things like robots in Biped then make Biped not be a subclass of Animal and then just make the axiom: " pawsNumber value "2"^^xsd:integer". I've included a revised version of your ontology with those changes. As you can see Human is now an instance of Biped. 

Cheers,
Michael

On Thu, Feb 27, 2020 at 7:55 AM Julien FAURE <[hidden email]> wrote:
I attached my .owl file to this reply.
PlantAnimal.owl
<http://protege-project.136.n4.nabble.com/file/t377201/PlantAnimal.owl



--
Sent from: http://protege-project.136.n4.nabble.com/Protege-User-f4659818.html
_______________________________________________
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

PlantAnimalRevised.owl (7K) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: Ontograf and inference display and property filter

Julien FAURE
Hi,

You actually saw the problem, I did not thought that Box considerations was
that important. Thanks for these instructive explanations.

Second point is crystal clear.
For the first example, I understand that I should get the expected result
with a Class rather than an Individual. I then added the "lion" class as a
subclass of Animal but I still not get inference like "lion eats Animal"
and the visual line in Ontograf either (see attached file updated).
PlantAnimal.owl
<http://protege-project.136.n4.nabble.com/file/t377201/PlantAnimal.owl>  

Actually I don't need to get the inference written somewhere. I'm mostly
interested in visualisation so having it display on the graph would be
enough. May I should use an external visualisation app? or a dedicated
plugin?

Regards

Julien



--
Sent from: http://protege-project.136.n4.nabble.com/Protege-User-f4659818.html
_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user
Reply | Threaded
Open this post in threaded view
|

Re: Ontograf and inference display and property filter

Michael DeBellis-2
I don't use OntoGraph that much so I'm not completely certain but I think what you are observing is just the way OntoGraph decides to draw graphs which when you give it some thought makes some sense. So when showing properties that link between classes OntoGraph (and I think this is probably true of all OWL graphing packages) doesn't show you the links for all the subclasses. I.e., if Person hasMother some Woman and AmericanPerson is a subclass of Person it won't also show you the hasMother link from AmericanPerson to Woman. It only shows you the highest class in the hierarchy that the relation holds for because you can also see in the graph what classes are subclass of that class and you know by the semantics of OWL that if the relation (relation is just another name for a property) holds between a Class and another Class then that relation also holds for all the subclasses of those classes as well. For very small learning ontologies why you would want this isn't that obvious but when you think of larger ontologies where a class can have tens, hundreds or even thousands of subclasses you can start to see why. The graph would get very complex very quickly if it showed the relation between all the subclasses. Also, since the whole point of the visualization is to help you understand the model showing all those extra links isn't really needed because once you understand how OWL works you know that if some relation holds for a class it also holds for all its subclasses. 

I tried an example where I added a new sub-property of eats called eatsAntelope, added a subclass of Animal called Antelope,  and I gave Lion the axiom: "Carnivore and (eatsAntelope some Antelope)". When I graph it in OntoGraph I see the eats relation between Carnivore and Animal and the eatsAntelope relation between Lion and Antelope which is what I expect. The  "eats some Animal" axiom still holds for Lion but OntoGraph doesn't bother showing that because it shows that relation between Carnivore and Animal and Lion is a subclass of Animal (which is also something it shows) But it does show the eatsAntelope relation between Lion and Antelope. I've attached the updated ontology.

BTW, if you want to try another graphing package you might want to try WebVOWL: http://vowl.visualdataweb.org/webvowl.html  But I think you would find it behaves the same way. 

On Thu, Feb 27, 2020 at 1:32 PM Julien FAURE <[hidden email]> wrote:
Hi,

You actually saw the problem, I did not thought that Box considerations was
that important. Thanks for these instructive explanations.

Second point is crystal clear.
For the first example, I understand that I should get the expected result
with a Class rather than an Individual. I then added the "lion" class as a
subclass of Animal but I still not get inference like "lion eats Animal"
and the visual line in Ontograf either (see attached file updated).
PlantAnimal.owl
<http://protege-project.136.n4.nabble.com/file/t377201/PlantAnimal.owl

Actually I don't need to get the inference written somewhere. I'm mostly
interested in visualisation so having it display on the graph would be
enough. May I should use an external visualisation app? or a dedicated
plugin?

Regards

Julien



--
Sent from: http://protege-project.136.n4.nabble.com/Protege-User-f4659818.html
_______________________________________________
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

PlantAnimalRevised2.owl (10K) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: Ontograf and inference display and property filter

Julien FAURE
Thank you very much, it is clear.
Since I have some other questions about visualisation I may open a new
thread on this.

To continue a bit the discussion:
I tried 2 things:
1) Set Lion SubClass with "Carnivore and (Eat some Animal)"
I get the inference that the Lion is a SubClass of Carnivore. That looks
trivial to me (actually I don't even know why it has to be written since it
is already clear in hierarchy).

2) Add a new Class Tiger and set the Tiger SubClass with "Animal and (Eat
some Animal)"
As in test 1, I get the inference saying Tiger is a SubClass of Animal
(well, I know...). I expected to get also the less trivial inference saying
Tiger is a SubClass of Carnivore. But it is not.

Is there anything wrong with what I did or is there anything else I do not
understand yet?


Thanks in advance

Julien

PlantAnimal.owl
<http://protege-project.136.n4.nabble.com/file/t377201/PlantAnimal.owl>  



--
Sent from: http://protege-project.136.n4.nabble.com/Protege-User-f4659818.html
_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user
Reply | Threaded
Open this post in threaded view
|

Re: Ontograf and inference display and property filter

Lorenz Buehmann

On 28.02.20 23:28, Julien FAURE wrote:

> Thank you very much, it is clear.
> Since I have some other questions about visualisation I may open a new
> thread on this.
>
> To continue a bit the discussion:
> I tried 2 things:
> 1) Set Lion SubClass with "Carnivore and (Eat some Animal)"
> I get the inference that the Lion is a SubClass of Carnivore. That looks
> trivial to me (actually I don't even know why it has to be written since it
> is already clear in hierarchy).
because it also follows from the intersection. and yes, it could be
filtered but isn't
>
> 2) Add a new Class Tiger and set the Tiger SubClass with "Animal and (Eat
> some Animal)"
> As in test 1, I get the inference saying Tiger is a SubClass of Animal
> (well, I know...). I expected to get also the less trivial inference saying
> Tiger is a SubClass of Carnivore. But it is not.
>
> Is there anything wrong with what I did or is there anything else I do not
> understand yet?

why do you think Tiger should be a Carnivore?

In your ontology you have

Tiger SubClassOf Animal and (Eat some Animal)

and

Carnivore SubClassOf (Animal or Plant) and (Eat some Animal)

So, I don't see how a reasoner should infer this. SubClassOf is "from
left to right", i.e. it means - and we have to be a bit formally strict
here "IF anything is a carnivore THEN it is either a plant or an animal
and it eats animals"


If you do

Carnivore EquivalentTo (Animal or Plant) and (Eat some Animal)

then also the other direction holds because it's syntactic sugar for

Carnivore SubClassOf (Animal or Plant) and (Eat some Animal)
(Animal or Plant) and (Eat some Animal)  SubClassOf Carnivore



By the way, a better practice is to use lower case for properties names,
i.e. "eat"

>
>
> Thanks in advance
>
> Julien
>
> PlantAnimal.owl
> <http://protege-project.136.n4.nabble.com/file/t377201/PlantAnimal.owl>  
>
>
>
> --
> Sent from: http://protege-project.136.n4.nabble.com/Protege-User-f4659818.html
> _______________________________________________
> 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: Ontograf and inference display and property filter

Julien FAURE
Hi,

Ok it's clear. I was trying to describe "Carnivore" as a part of "Animal" or
"Plant" that have the specific feature to "eat some Animal". I understand I
was doing wrong.
So now I have the "Carnivore" is just EquivalentTo "eat some Animal".

Thank you for the explanation. Espacially the
"IF anything is a carnivore THEN it is either a plant or an animal and it
eats animals"
It really helps me to figure out how I have to organize things.

I guess this also applies to "object properties". Am I right?
If so, I'm a bit lost to add that the "eat" property is a combination "chew"
and "swallow".
I have set the "eat" property as a SubProperty of the two (see .owl
attached) but I'm not very satisfied.
This induces that "eat" appears two times in the hierarchy (under "chew" and
under "swallow"). Next, if I break down "chew" into "OpenMouth" and
"CloseMouth", it will then appear 3 times...
Is that the way people deal with that kind of decomposition?
If so, is there any method that help to filter out that complexity? Maybe
something like Venn diagram?

best wishes

Julien

PlantAnimal.owl
<http://protege-project.136.n4.nabble.com/file/t377201/PlantAnimal.owl>  



--
Sent from: http://protege-project.136.n4.nabble.com/Protege-User-f4659818.html
_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user
Reply | Threaded
Open this post in threaded view
|

Re: Ontograf and inference display and property filter

Michael DeBellis-2
I would encourage you to read up a bit on basic set theory. If you look here:  https://www.michaeldebellis.com/post/owl-theoretical-basics  you will find (among other things) a PDF that is the first couple of chapters from a book on Computational Linguistics. Those chapters have nothing to do with Linguistics they are an overview of set theory and logic. This is just my personal opinion, others on the group may not agree but I think understanding some basic set theory really helps when using OWL because everything in OWL is based on set theory.  There is at least one other good overview on the page I link to above there that describes some OWL basics from the University of Manchester. 

In this case (and when I first started using OWL I found this confusing because I was used to properties in OOP which are different) the important thing to remember is that a property (or a relation in set theory) is essentially just a set so saying that one property is a sub-property of another says that all the binary tuples in the super property are also in the sub property but not necessarily vice versa.  Just as saying that one class is a subclass of another means that all the individuals in the superclass are in the subclass but not necessarily vice versa. Note: I use the terms property/relation and class/set as synonyms because in OWL they are. 

Back to your question: the way you describe it is the way it is supposed to work but you don't always have to make properties sub-properties of each other. I've seen many ontologies that were very useful where everything was just a sub-property of owl:topObjectProperty (or owl:topDataProperty). Also, whether you see all those inferences can be to some extent controlled by the reasoning parameters. I think you can set them so the reasoner doesn't do all those inferences, not sure about that last part though because whenever I use a sub-property I want to see all the inferences. 

Here is a real example from an ontology I'm working on right now. Suppose I'm describing a software activity. I want to have a property called hasTeamMembers   for all the team members involved in that Activity. I also want a sub-property called hasTeamLeader for the person who has overall responsibility for the Activity. In my programs or rules I have things that say for all the members in the Activity do X.  On the other hand I also have rules that say for the team leader do Y but I still want the things that apply to all the team members to apply to the team leader because he is also a team member. That's when you use sub-properties when you want to record inferences like superProperty is a relation between set X and set Y  and subProperty is a relation between some subset of X and/or Y. 

For example, another way to model carnivores would be using domains and ranges and sub-properties. eats could be a property with domain Organism and range Organism. eatsMeat could be a sub-property of eats with Domain Carnivore and Range Animal. eatsPlants could be a sub property of eats with a domain of Herbivore and range of Plant. I'm not saying this is better than using DL axioms, that depends on what else you want to model, how your ontology will be used, etc. 

But the one thing I would emphasize is don't use sub-properties just because they are there. Use them when you have a specific use case (like my example of team members and leader) where you want to have reasoning about the different refinements of a property. Hope that helps. 

Cheers,
Michael

On Sun, Mar 1, 2020 at 1:56 PM Julien FAURE <[hidden email]> wrote:
Hi,

Ok it's clear. I was trying to describe "Carnivore" as a part of "Animal" or
"Plant" that have the specific feature to "eat some Animal". I understand I
was doing wrong.
So now I have the "Carnivore" is just EquivalentTo "eat some Animal".

Thank you for the explanation. Espacially the
"IF anything is a carnivore THEN it is either a plant or an animal and it
eats animals"
It really helps me to figure out how I have to organize things.

I guess this also applies to "object properties". Am I right?
If so, I'm a bit lost to add that the "eat" property is a combination "chew"
and "swallow".
I have set the "eat" property as a SubProperty of the two (see .owl
attached) but I'm not very satisfied.
This induces that "eat" appears two times in the hierarchy (under "chew" and
under "swallow"). Next, if I break down "chew" into "OpenMouth" and
"CloseMouth", it will then appear 3 times...
Is that the way people deal with that kind of decomposition?
If so, is there any method that help to filter out that complexity? Maybe
something like Venn diagram?

best wishes

Julien

PlantAnimal.owl
<http://protege-project.136.n4.nabble.com/file/t377201/PlantAnimal.owl



--
Sent from: http://protege-project.136.n4.nabble.com/Protege-User-f4659818.html
_______________________________________________
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: Ontograf and inference display and property filter

Michael DeBellis-2
In reply to this post by Julien FAURE
One last thing. If you want to understand properties and sub-properties (and other OWL concepts) more you might find this tutorial from Manchester very useful:   http://owl.cs.manchester.ac.uk/publications/talks-and-tutorials/fhkbtutorial/  Also, you've probably done it already but if not you should do the Stanford Pizza tutorial (I would encourage you to do that before the Manchester one). 

On Sun, Mar 1, 2020 at 1:56 PM Julien FAURE <[hidden email]> wrote:
Hi,

Ok it's clear. I was trying to describe "Carnivore" as a part of "Animal" or
"Plant" that have the specific feature to "eat some Animal". I understand I
was doing wrong.
So now I have the "Carnivore" is just EquivalentTo "eat some Animal".

Thank you for the explanation. Espacially the
"IF anything is a carnivore THEN it is either a plant or an animal and it
eats animals"
It really helps me to figure out how I have to organize things.

I guess this also applies to "object properties". Am I right?
If so, I'm a bit lost to add that the "eat" property is a combination "chew"
and "swallow".
I have set the "eat" property as a SubProperty of the two (see .owl
attached) but I'm not very satisfied.
This induces that "eat" appears two times in the hierarchy (under "chew" and
under "swallow"). Next, if I break down "chew" into "OpenMouth" and
"CloseMouth", it will then appear 3 times...
Is that the way people deal with that kind of decomposition?
If so, is there any method that help to filter out that complexity? Maybe
something like Venn diagram?

best wishes

Julien

PlantAnimal.owl
<http://protege-project.136.n4.nabble.com/file/t377201/PlantAnimal.owl



--
Sent from: http://protege-project.136.n4.nabble.com/Protege-User-f4659818.html
_______________________________________________
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: Ontograf and inference display and property filter

Lorenz Buehmann

And one comment from my side: in OWL 2 DL, you don't have the ability to express the intersection of two properties. That would need a more expressive Description Logic than SROIQ being the logic behind OWL.

SWRL rules should possibly work here.

On 02.03.20 16:26, Michael DeBellis wrote:
One last thing. If you want to understand properties and sub-properties (and other OWL concepts) more you might find this tutorial from Manchester very useful:   http://owl.cs.manchester.ac.uk/publications/talks-and-tutorials/fhkbtutorial/  Also, you've probably done it already but if not you should do the Stanford Pizza tutorial (I would encourage you to do that before the Manchester one). 

On Sun, Mar 1, 2020 at 1:56 PM Julien FAURE <[hidden email]> wrote:
Hi,

Ok it's clear. I was trying to describe "Carnivore" as a part of "Animal" or
"Plant" that have the specific feature to "eat some Animal". I understand I
was doing wrong.
So now I have the "Carnivore" is just EquivalentTo "eat some Animal".

Thank you for the explanation. Espacially the
"IF anything is a carnivore THEN it is either a plant or an animal and it
eats animals"
It really helps me to figure out how I have to organize things.

I guess this also applies to "object properties". Am I right?
If so, I'm a bit lost to add that the "eat" property is a combination "chew"
and "swallow".
I have set the "eat" property as a SubProperty of the two (see .owl
attached) but I'm not very satisfied.
This induces that "eat" appears two times in the hierarchy (under "chew" and
under "swallow"). Next, if I break down "chew" into "OpenMouth" and
"CloseMouth", it will then appear 3 times...
Is that the way people deal with that kind of decomposition?
If so, is there any method that help to filter out that complexity? Maybe
something like Venn diagram?

best wishes

Julien

PlantAnimal.owl
<http://protege-project.136.n4.nabble.com/file/t377201/PlantAnimal.owl



--
Sent from: http://protege-project.136.n4.nabble.com/Protege-User-f4659818.html
_______________________________________________
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

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

Re: Ontograf and inference display and property filter

Julien FAURE
Hi,

Thank you for the refs and for the time you spent looking at my questions.
I will definitely go through the tuto.
To be honest, for the moment I still wonder if learning the ontology method
is worth the effort for me.

To make things clearer for you, I should tell you a bit more about what I am
trying to achieve.

I would like to draw some connections between the two distinct fields of
physics and economics. Both have their own semantic field. Actually, this
approach has already been taken
(https://www.editions-parole.net/produit/de-la-thermodynamique-a-leconomie/)
but I'd like to make the connections more explicit.

I wonder if the ontology method could help me do so. Before considering the
"ontology option" I've already started  with a pen and paper, and tried to
build my own method (a dictionnary with links between words)... not very
convincing. Based on this, I now look at ontology with a few things that I
may need in mind (defining synonyms, homonyms, etc.). So the point for me is
not to build an ontology that could be "used" in future applications. I'd
rather do it to help me clarify things related to these two semantic fields.
That's why I asked for Object Property categorization. That's also why I'm
very curious about visu tools and reasonner for inference. I hope this make
sense.

Any opinion on the relevance of the ontology method for what I'm trying to
achieve is welcome.
Maybe the ontology method has already been used in such a way?
For now i'm gonna play with the tuto.

Cheers,

Julien



--
Sent from: http://protege-project.136.n4.nabble.com/Protege-User-f4659818.html
_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user
Reply | Threaded
Open this post in threaded view
|

Re: Ontograf and inference display and property filter

Jean-Louis Janin-2
Hello,

I have used successfully Protege desktop 5.5.0 to set up the model of a
bilingual (French-English) lexicon on textual and geophysical data. It
has been the main argument of a thesis in linguistics presented in
January 2020 at Bordeaux Montaigne University, with a lot of examples
introduced as instances of the classes of the ontology. It took me 6
years to implement the model ("maquette", in French).

Protege is an excellent tool to developp the concepts of the domain,
transfer them into the lexicon, which happens to be stratified, and
connect the lexical units to published texts treated by textometry.

Yours

Jean-Louis Janin.

Le 04/03/2020 à 15:20, Julien FAURE a écrit :

> Hi,
>
> Thank you for the refs and for the time you spent looking at my questions.
> I will definitely go through the tuto.
> To be honest, for the moment I still wonder if learning the ontology method
> is worth the effort for me.
>
> To make things clearer for you, I should tell you a bit more about what I am
> trying to achieve.
>
> I would like to draw some connections between the two distinct fields of
> physics and economics. Both have their own semantic field. Actually, this
> approach has already been taken
> (https://www.editions-parole.net/produit/de-la-thermodynamique-a-leconomie/)
> but I'd like to make the connections more explicit.
>
> I wonder if the ontology method could help me do so. Before considering the
> "ontology option" I've already started  with a pen and paper, and tried to
> build my own method (a dictionnary with links between words)... not very
> convincing. Based on this, I now look at ontology with a few things that I
> may need in mind (defining synonyms, homonyms, etc.). So the point for me is
> not to build an ontology that could be "used" in future applications. I'd
> rather do it to help me clarify things related to these two semantic fields.
> That's why I asked for Object Property categorization. That's also why I'm
> very curious about visu tools and reasonner for inference. I hope this make
> sense.
>
> Any opinion on the relevance of the ontology method for what I'm trying to
> achieve is welcome.
> Maybe the ontology method has already been used in such a way?
> For now i'm gonna play with the tuto.
>
> Cheers,
>
> Julien
>
>
>
> --
> Sent from: http://protege-project.136.n4.nabble.com/Protege-User-f4659818.html
> _______________________________________________
> 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: Ontograf and inference display and property filter

Michael DeBellis-2
In reply to this post by Julien FAURE
Well for what it's worth and just my opinion of course but what you described sounds very interesting and something that I think OWL, SWRL, and SPARQL could be very useful for. It's actually something that I've tried (without much success) to do, is to get people who are interested in the "soft sciences" (and that includes work like seeing issues that span disciplines such as economics and physics) to investigate using OWL to bring more rigor to discussions. I'm going to send you a longer message actually because I want to go into more detail but what I'm going to say is no longer about Protege support so not of interest to most of the rest of the list. But before I end this reply you might find the following group interesting:  https://groups.google.com/forum/#!forum/ontolog-forum  This is a group that discusses ontologies as well but more at a conceptual level than at a "how do I do X in OWL" level. I don't think the topic of physics and economics has come up but we've had other discussions on related topics such as the definition of science and how that relates to ontologies. 

On Wed, Mar 4, 2020 at 6:20 AM Julien FAURE <[hidden email]> wrote:
Hi,

Thank you for the refs and for the time you spent looking at my questions.
I will definitely go through the tuto.
To be honest, for the moment I still wonder if learning the ontology method
is worth the effort for me.

To make things clearer for you, I should tell you a bit more about what I am
trying to achieve.

I would like to draw some connections between the two distinct fields of
physics and economics. Both have their own semantic field. Actually, this
approach has already been taken
(https://www.editions-parole.net/produit/de-la-thermodynamique-a-leconomie/)
but I'd like to make the connections more explicit.

I wonder if the ontology method could help me do so. Before considering the
"ontology option" I've already started  with a pen and paper, and tried to
build my own method (a dictionnary with links between words)... not very
convincing. Based on this, I now look at ontology with a few things that I
may need in mind (defining synonyms, homonyms, etc.). So the point for me is
not to build an ontology that could be "used" in future applications. I'd
rather do it to help me clarify things related to these two semantic fields.
That's why I asked for Object Property categorization. That's also why I'm
very curious about visu tools and reasonner for inference. I hope this make
sense.

Any opinion on the relevance of the ontology method for what I'm trying to
achieve is welcome.
Maybe the ontology method has already been used in such a way?
For now i'm gonna play with the tuto.

Cheers,

Julien



--
Sent from: http://protege-project.136.n4.nabble.com/Protege-User-f4659818.html
_______________________________________________
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