Re: [ontology-summit] Ontolog Ontology of Ontologies

[Andrea Westerinen <arwesterinen@xxxxxxxxx>]

John, John and Hans, I will try to respond to all your individual points in this email ...

> Hans : One concern I have with this line of thinking is that it appears as if it considers that only 

> an ontology can be used to characterize other ontologies for retrieval and reuse purposes. 

I was not making this distinction at all (or talking about characterizing ontologies).  I was only saying that I put all my mappings between concepts and properties into a separate "integrating ontology" and leave each small, individual module as a stand-alone, reusable entity. 

> Hans :  One concern I have with this line of thinking is that it appears as if it considers that only an 

> ontology can be used to characterize other ontologies for retrieval and reuse purposes. 

Actually, I don't think that ONLY an ontology can be used, but an ontology COULD be used.  Characteristics of an ontology ("ontology metadata" as Gary Berg-Cross noted in another thread) are important and should be formally defined.  Whether they are defined as an individual ontology, reuse SKOS concepts or defined using some other mechanism is less important than whether they exist, are used and are supported by tooling. 

> Hans : Another concern I have is this notion that an integrating ontology should not be “open-ended”. 

> What happens if someone develops an ontology that doesn’t fit into the integrating ontology in a way 

> that makes it easy to discover for its intended purpose?

I think that I stated my position poorly.  My integrating ontology is part of my overall application or domain ontology.  It is not the reusable part (unless you are reusing my application and have my use cases).  The reusable components are the individual semantics that I need in my application (or domain), that I am bringing together, and that I need to integrate into my "operational" whole.  I do this integration (logical mapping including equivalence) in what I call an "integrating ontology".  The complete set of modules and semantic concepts that I bring together is dependent on my application and use cases (and theoretically "open-ended").

> JohnM : If you're saying ontologies should be specifying attributes like dc:subject, referencing some controlled 

> SKOS vocabulary(s) of functional domains, and or other provenance information such as dc:coverage, 

> then I'm all for that as a specific item in the communique.  

Yes, I would agree that metadata for finding semantic content would be a valuable output of this summit and a goal for Track A.

> JohnM : I'm hearing an "integrating" ontology is composed of nothing but owl:imports, does not define its own 

> classes, properties or individuals.

Not exactly.  For me, an "integrating" ontology imports the relevant ontologies (including reused modules but also any new ontologies that are defined for the application/domain) and does a logical mapping between the concepts.  New classes or properties that define the axioms for mapping the concepts may be included.  But, in my experience, I have tried to keep the purpose of each of my ontologies (including the "integrating" one) and their contents very focused (and cohesive).

> JohnY : Entity Relationship Combinations such as I. Entity E-OA = E-OB, II. Entity E-OA sub of E-OB, ...

Indeed, these are the mappings of my "integrating" ontology (they are at the entity and property level). I am not talking about integrating or mapping full-blown, complete, application/domain ontologies with 100s of concepts and properties.  I am talking about integrating ontologies defined as small, reusable modules.  Therefore, your set theoretic concepts (where Ontology A reuses some concepts from Ontology B) are aligned with my example ... My application/domain ontologies reuse content from modules that are defined independently or defined as part of other, large, application/domain  ontologies.  Cohesive piece parts (modules) of Ontology B could be picked up, reused and integrated in Ontology A - where the integration is defined in its own "module" (its "integrating ontology").

Hopefully, this helps.  If not, I am sure that we will be writing more.  :-)

Andrea

On Wed, Feb 5, 2014 at 9:29 PM, John Yanosy Jr. <jyanosyjr@xxxxxxxxx> wrote:

Hans, John, Andrea,

Some not very well investigated ideas, for possible investigation.  Basically I am looking for the ability to represent and query the type of Entity relationships between ontologies as well as compute a metric that would help classify the aggregation of these relationships in a meaningful manner.

It may be useful to consider some possible set theoretic and specialization/generalization relationships between ontologies when considering their reuse. Though I believe that metadata about the ontology could prove useful, i.e., Dublin Core like, it might also be useful to clarify the relationships between the internal elements of these ontologies in some mathematical form, that lends itself to logical queries as well as numerical metrics.

Some possible relationships are:

Entity Relationship Combinations

I. Entity E-OA = E-OB

II. Entity E-OA sub of E-OB

III. Entity E-OB sub of E-OA

IV. Entity E-OA not related to E-OB either sub or equal

Set Theoretic Relationship

Note: (The following are easy to consider when the Entities under analysis follow rule I, or rule IV above, i.e., are equal or not related. But if the entities are related semantically but are not equivalent than the question is whether the set theoretic relationship are still useful.)

1. Ontology A reuses some Entities of ontology B (OB). Neither OA or OB subsumes the other. In this case I believe that the entity relationships under consideration(E-OA, E-OB) could either be II or III, and in fact there may be a mix of these all three relationship types, I, II, or III.

2. OA is a subset of OB (All Entities OA are in OB - possibly OA) (I would assume that in most cases the entity relationships would be a mix of I and II. Some sort of domain specialization is occurring among most of the entities reused from OB, but some are also used fully.

3. OB is a subset of OA (This could be like 2, but also adds additional entities to further domain specialize OA.)

4. OA = OB (Fully equivalent - not sure why this would occur except when some renaming or specializing occurs between some classes and properties)

A numerical metric might be defined based on these set theoretic and semantic relationships.

On Wed, Feb 5, 2014 at 6:44 PM, Hans Polzer <hpolzer@xxxxxxxxxxx> wrote:

John,

 

Yes, that’s what I was trying to articulate, with a particular emphasis on the provenance/coverage attributes. And I agree that this community should be a good source of SME’s for developing an O3.

 

Hans

 

From: ontology-summit-bounces@xxxxxxxxxxxxxxxx [mailto:ontology-summit-bounces@xxxxxxxxxxxxxxxx] On Behalf Of John McClure
Sent: Wednesday, February 05, 2014 6:59 PM
To: ontology-summit@xxxxxxxxxxxxxxxx

Subject: Re: [ontology-summit] Ontolog Ontology of Ontologies

 

Hi Hans -
If you're saying ontologies should be specifying attributes like dc:subject, referencing some controlled SKOS vocabulary(s) of functional domains, and or other provenance information such as dc:coverage, then I'm all for that as a specific item in the communique. 

At the same time, I'd love to see folks here at Ontolog develop a small vocabulary of TYPES of ontologies, such as that Sowa suggested, that can be referenced by ontologies inventoried, for instance, in the Ontolog Repository that one of the tracks has responsibility for.

Much agree that any given ontology that hasn't incorporated any rdf:type pointing to an "O3 type" should not be penalized in search results -- I definitely hope the repository permits one to independently annotate any resource (ontology) with any property, including <rdf:type="an O3 type"/>!

The key though is to tap into SMEs' knowledge right here -- the Subject Matter Experts where the 'subject' is "ontologies". Surely there should be willingness to develop an O3 among these experts!!!!

/jmc
 

On 2/5/2014 3:31 PM, Hans Polzer wrote:

Andrea, John:

 

One concern I have with this line of thinking is that it appears as if it considers that only an ontology can be used to characterize other ontologies for retrieval and reuse purposes. I’m not an ontology expert or a logician, and maybe one can capture all the salient attributes regarding the scope of applicability of any given ontology with such an ontology of/about ontologies, but I believe it would be smart to consider other/additional means of characterizing ontologies for retrieval and utilization decision purposes. Such means may well be grounded in ontologies of specific knowledge and human institutional domains, but could be expressed in simpler domain-specific attribute terms. I believe this would simplify initial searches for relevant ontologies for someone in a particular application or institutional domain. It could also broaden the potential audience for discovering relevant ontologies.

 

Another concern I have is this notion that an integrating ontology should not be “open-ended”. What happens if someone develops an ontology that doesn’t fit into the integrating ontology in a way that makes it easy to discover for its intended purpose? What’s wrong with open-endedness for such a purpose? Do we already know all the possible domains for which we will have ontologies?

 

fHans

 

From: ontology-summit-bounces@xxxxxxxxxxxxxxxx [mailto:ontology-summit-bounces@xxxxxxxxxxxxxxxx] On Behalf Of Andrea Westerinen
Sent: Wednesday, February 05, 2014 5:58 PM
To: Ontology Summit 2014 discussion
Subject: Re: [ontology-summit] Ontolog Ontology of Ontologies

 

John and John,

 

In as much as the "integrating" ontology is an "ontology about ontologies (and their concepts)" ... yes, it is a meta-ontology.  However, I find that people handle the term, "integrating ontology", much better.  And, it conveys the purpose (as opposed to being open-ended).

 

Given that some of my customers view ontologies with some skepticism (they call ontology, the "o-word"), I much prefer avoiding adding scary modifiers onto a scary word.  (That is just my experience.)

 

Andrea

 

On Wed, Feb 5, 2014 at 3:38 PM, John McClure <jmcclure@xxxxxxxxxxxxxx> wrote:

Thanks John,
"Metalevel Ontology" sounds like a candidate class for an Ontolog Ontology of Ontologies (O3).

Any other types of ontologies come to mind for anyone?
thanks/jmc

On 2/5/2014 11:11 AM, John F Sowa wrote:

The term "integrating ontology" is good.  But it's important to note

that this is a *metalevel* ontology.

 

_________________________________________________________________
Msg Archives: http://ontolog.cim3.net/forum/ontology-summit/
Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/ontology-summit/
Unsubscribe: mailto:ontology-summit-leave@xxxxxxxxxxxxxxxx
Community Files: http://ontolog.cim3.net/file/work/OntologySummit2014/
Community Wiki: http://ontolog.cim3.net/cgi-bin/wiki.pl?OntologySummit2014
Community Portal: http://ontolog.cim3.net/wiki/

--

Andrea Westerinen

T: 425.891.8407

arwesterinen@xxxxxxxxx or andreaw@xxxxxxxxxxx

http://ontolog.cim3.net/cgi-bin/wiki.pl?AndreaWesterinen

organizingknowledge.blogspot.com

_________________________________________________________________
Msg Archives: http://ontolog.cim3.net/forum/ontology-summit/   
Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/ontology-summit/  
Unsubscribe: mailto:ontology-summit-leave@xxxxxxxxxxxxxxxx
Community Files: http://ontolog.cim3.net/file/work/OntologySummit2014/
Community Wiki: http://ontolog.cim3.net/cgi-bin/wiki.pl?OntologySummit2014  
Community Portal: http://ontolog.cim3.net/wiki/     (01)