W3C home > Mailing lists > Public > public-owl-dev@w3.org > April to June 2009

Re: Graphical notation standard for OWL (ODM UML profile)

From: Guntis Barzdins <guntis@latnet.lv>
Date: Fri, 01 May 2009 23:42:03 +0300
Message-ID: <49FB5E9B.5000903@latnet.lv>
To: public-owl-dev@w3.org
Cc: Elisa Kendall <ekendall@sandsoft.com>, Marcel Ferrante <marcelf@gmail.com>, Igor Istochnick <igor.istochnick@gmail.com>
I just wanted to bring attention of those interested in ODM profile 
for UML visualisation of OWL to the recently released plugin for
Protege 4.0, which implements large part the ODM profile:

   http://protegewiki.stanford.edu/index.php/OWL2UML

This is surely an early version of the plugin with lot of missing features, 
but any feedback from those interested would be welcome. To our knowledge,
this is the only fully functional (although incomplete) implementation of ODM profile for UML/OWL currently. 
The plugin is implemented by Igor Istocnik, my student at University of Latvia.

  -- Guntis Barzdins


Elisa Kendall wrote:
> 
> Hi Marcel,
> 
> There is a much more recent version of the ODM than the one you used for 
> your comparison, with added graphical notation.  For example, there are 
> now multiple mechanisms for representing properties -- as associations, 
> as association classes, and as reified classes in the current profile.  
> Your comparison only mentions the associations.  Compositions and 
> aggregations do not have equivalent semantics to the UML semantics, 
> which is why we shied away from those.
> There will be some additional graphical modifications to the profile(s) 
> in an ODM 1.1, currently planned for September, fyi.  We were torn when 
> developing the profile(s) between providing a notation that would be 
> intuitive for UML power users and profiles that would be helpful to RDF 
> and OWL developers. The compromise does include some text 
> representation, notably for multiplicities, but there are also more 
> graphics in this latest version than you might realize.  It has also 
> been formalized by OMG as a 1.0 standard, vetted by language authors on 
> both the OMG and W3C sides of the proverbial fence, with publication of 
> that version (no technical changes from what is currently posted), 
> forthcoming.
> 
> I'll take a careful look at your comparison, and will attempt to respond 
> later this week with a more complete comparison for you, but I would 
> suggest that you take a look at http://www.omg.org/spec/ODM/1.0/Beta3/.  
> The cmof versions of both the metamodels and profiles are also available 
> at that link, and I have equivalent MagicDraw and EMF versions if you 
> would like me to send those to you.
> Best regards,
> 
> Elisa
> 
> Marcel Ferrante wrote:
>>
>> Hi Elisa, how are you?
>>
>>  
>>
>> Just now we could feedback you (my appologies)…We need time to diggest 
>> the information that you sent to me and it was absoluty necessary for 
>> my study, thank you.
>>
>>
>> So, we take the ontology elements from the ontology specification 
>> (http://www.w3.org/TR/owl-guide) mapping to the graphical 
>> representation defined in the ODM specification. This mapping is 
>> summarized in this document below. Please, feel free to correct me…
>>
>> http://www.uniportal.com.br/media/marcel/UML-to-OWL-EN.pdf
>>
>>
>> The conclusion of this work is:
>>
>> Most of ontology elements do not have a specific graphic symbol for 
>> their representation or the element is represented in a literal and 
>> graphic form.
>>
>>
>> So, the problem continues because we couldn't define graphic symbols 
>> to represent these elements.
>>
>>
>> We understand the ODM representation form (based in UML) is not fully 
>> suitable for information visualization interfaces, because the literal 
>> representation needs much bigger area than a graphic symbol. In some 
>> cases, it is not clear what iproperty is being represented.
>>
>>
>> Do you know why some UML symbols like composition and aggregation 
>> aren't used until now (they could be mapped to some kind ontology 
>> properties)? And there is another approach for this kind of 
>> representation (not based on uml)?
>>
>>
>> Thank you very much,
>>
>> Marcel
>>
>>
>>
>> 2008/12/26 Elisa Kendall <ekendall@sandsoft.com 
>> <mailto:ekendall@sandsoft.com>>
>>
>>
>>     Hi Marcel,
>>
>>     Merry Christmas!
>>     There are really several components of the specification of a
>>     mapping between UML and OWL.  First, for model interchange (so
>>     that UML tools, Eclipse/EMF, MOF tools, and others can exchange
>>     RDF vocabularies and OWL ontologies consistently), there are
>>     metamodels for RDF (chapter 10) and OWL (chapter 11).  The OWL
>>     metamodel reflects OWL 1.0, although there are plans to extend it
>>     to support OWL 2 as part of a revision task force activity.  This
>>     may entail development of (1) an independent, standalone metamodel
>>     that is disconnected from RDF, (2) a metamodel that extends the
>>     current OWL metamodel in the specification, and (3) a mapping
>>     between the two, where the standalone metamodel may be derived
>>     from one originally developed by Peter Haase, Saartje Brockmans,
>>     and Boris Motik as part of the OWL 2 language development process.
>>
>>     Secondly, and more to your point, there is a graphical notation,
>>     called a profile from the UML perspective, for extending UML with
>>     specific notation to support ontology development in UML tools
>>     (chapter 14).  Chapter 14 is where you should find a number of
>>     stereotypes and tagged values (both could be considered symbols)
>>     supporting OWL ontology development in tools such as IBM's
>>     Rational Software Architect, No Magic's MagicDraw, Enterprise
>>     Architect from Sparx Systems, and others.  The OWL profile may
>>     also be extended to support additional features of OWL 2 as part
>>     of the ODM revision process.
>>
>>     UML is both more expressive than and less expressive than OWL -
>>     the mapping is not straightforward.  Chapter 16 attempts to cover
>>     some of the issues in a direct mapping between the logical subset
>>     of UML and OWL (ignoring behavioral features, for example), and
>>     contains some "code", developed by IBM China Research using the
>>     MOF (Meta-Object Facility) Query/View/Transformations (QVT)
>>     language.  This section is not normative, as we were not convinced
>>     that "all mappings are created equal" depending on the purpose of
>>     the mapping, the models themselves, and other constraints on the
>>     mapping processes.
>>
>>     You're correct that UML does not have an underlying model theory,
>>     although one may be developed for a subset of UML over the next
>>     year or so to support FUML (Functional UML, also called Executable
>>     UML).  There is discussion ongoing in the FUML community about
>>     developing such a model theory, and testing it via CLIF/IKL, which
>>     is a dialect of Common Logic.  I'm not a direct participant in
>>     that work, though I plan to follow it and have friends who are
>>     working on it if you're interested.
>>
>>     A number of us are in the process of launching a new Eclipse/MDT
>>     project to support ODM-based development, with a target initial
>>     release date of June, coincident with the Galileo release of the
>>     umbrella Eclipse project.  Let me know if you're interested and
>>     I'd be happy to point you to the work and/or enlist support :).
>>
>>     I hope this is helpful.  Happy Holidays!
>>
>>     Elisa
>>
>>     Marcel Ferrante wrote:
>>
>>         Hi Elisa,
>>
>>         Im reading the ODM specification and a important point is the
>>         ODM is necessary because UML doesnt have semantic enough
>>         to represent completely a ontology model like OWL DL.
>>
>>         But, I read the OWLbase metamodel and I cant see a /new
>>         symbol/ from UML created to these situations. Maybe because
>>         I dont expert in UML either...So can you indicate it for me ?
>>
>>         Thanks a lot,
>>         Marcel
>>
>>
>>
>>         2008/12/20 Elisa Kendall <ekendall@sandsoft.com
>>         <mailto:ekendall@sandsoft.com> <mailto:ekendall@sandsoft.com
>>         <mailto:ekendall@sandsoft.com>>>
>>
>>
>>
>>
>>                The closest thing to a standardized graphical notation
>>         is the
>>                OMG work:
>>                       http://www.w3.org/2007/OWL/wiki/UML_Concrete_Syntax
>>
>>                This may be revised in light of OWL 2.
>>                           Thanks for the plug, Bijan :).  This 
>> prompted me to revise the
>>            page, linked above, which now reflects links to the latest
>>         version
>>            of the ODM, now a formally adopted OMG standard, etc.
>>
>>            Evan Wallace and I gave a tutorial on this awhile back at
>>         an OMG
>>            meeting, and are planning to update it for the March 2009 OMG
>>            meeting in Washington DC.  If anyone is interested, let me 
>> know
>>            and I can provide details on the meeting or the old slides, as
>>            desired.
>>
>>            Best regards,
>>
>>            Elisa
>>
>>
>>
>>
>>         --         Marcel Ferrante Silva
>>         Doutorando em Ciência da Informação - ECI/UFMG
>>         www.finaltec.com.br <http://www.finaltec.com.br>
>>         <http://www.finaltec.com.br>
>>
>>         +55 31 8851-9069 2626-2859
>>         MSN: marcelferrante@hotmail.com
>>         <mailto:marcelferrante@hotmail.com>
>>         <mailto:marcelferrante@hotmail.com
>>         <mailto:marcelferrante@hotmail.com>>
>>
>>
>>
>>
>>
>>
>> -- 
>> Marcel Ferrante Silva
>> Doutorando em Ciência da Informação - ECI/UFMG
>> www.finaltec.com.br <http://www.finaltec.com.br>
>> +55 31 8851-9069 2626-2859
>> MSN: marcelferrante@hotmail.com <mailto:marcelferrante@hotmail.com>
> 
> 
> 
> 
Received on Friday, 1 May 2009 20:24:02 GMT

This archive was generated by hypermail 2.3.1 : Wednesday, 27 March 2013 09:32:57 GMT