Search Mailing List Archives


Limit search to: Subject & Body Subject Author
Sort by: Reverse Sort
Limit to: All This Week Last Week This Month Last Month
Select Date Range     through    

[protege-owl] Re : How to use an Ontology

James Howison james at howison.name
Tue Oct 21 09:44:49 PDT 2008


On 21 Oct 2008, at 4:25 AM, Filipe Pinto wrote:

>
> Hi,
> At this moment I'm starting with ontologies too.
> By the opposite to Joseph I'm interested in database marketing  
> ontology (focus on
> database use within .marketing activities)
> Joseph, if you want to, I can send to you my ontology and may be we  
> could help each
> other.
>
> James,
> Could you help me to this question:
> -Is it possible to model in a ontology a process step  order (to  
> data mining process:
> 1rst- data selection; 2nd) data pre-processing; 3rd) modeling,...)?  
> How can I do this?

I can think of a few ways:

# A class to describe Steps
:Step rdf:type owl:Class .

# three individuals of Class Step, which know their :nextStep
:dataSelection rdf:type :Step ;
                rdfs:label "Data Selection Step" ;
                :nextStep :dataPreProcessing .

:dataPreProcessing rdf:type :Step ;
                    rdfs:label "Data Pre-processing Step" ;
                    :nextStep :modeling .

:modeling rdf:type :Step ;
           rdfs:label "Modeling Step" .

But you are almost certainly going to be better served by  
understanding and using rdf:Seq:

http://www.w3.org/TR/rdf-schema/#ch_containervocab

> - After a selection made (e.g., attribute dataset selection) how can  
> I model, in a
> second level the dataset created?

Sorry, not sure what you mean here. More info?

> - Is it possible to "calculate" values, and use them after?

Yes, you can add statements to describe the calculated values in at  
least two ways.  The first I'd describe as procedural, using a library  
such as Jena or Protege's OWLAPI to query the properties needed for  
the calculation, then inserting the results to your model as a new  
statement.  This is very similar to how many programs use an SQL  
database.

Or you could model using Semantic Web Rule Language, which is a  
declarative method.  This means, provided you use an appropriately  
capable reasoner such as Pellet, you can add statements to your  
ontology which perform the calculations at reasoning time.  This means  
you don't have to run external code.  Perhaps, in effect if not  
method, it is like a db with triggered procedures.

You might find this thread interesting:

https://mailman.stanford.edu/pipermail/protege-owl/2008-October/008354.html

hmmm, as usual email thread is borked, you'll have to reconstruct the  
thread, it's Micheal Lodeman, myself and Thomas Russ with the subject  
"Modeling Problem".

--J



More information about the protege-owl mailing list