ontolog-forum
[Top] [All Lists]

Re: [ontolog-forum] Plural taxonomies?

To: "[ontolog-forum]" <ontolog-forum@xxxxxxxxxxxxxxxx>
From: "doug foxvog" <doug@xxxxxxxxxx>
Date: Thu, 17 Jun 2010 10:49:55 -0400 (EDT)
Message-id: <54478.74.96.159.4.1276786195.squirrel@xxxxxxxxxxxxxx>
Sorry, this late reply was stuck in a Drafts buffer, not having been sent
a couple weeks ago when i wrote it.  I hope it still can add something to
the discussion.    (01)

-- df    (02)

On Mon, May 31, 2010 15:16, Mike Bennett said:
> Hi Deborah,
>
> Thanks for the pointer to Omniclass, it looks interesting. It looks as
> though you are dealing with the same issues we are dealing with in the
> financial securities industry. The industry recognises the need to
> define terms globally across a whole industry, and as such those terms
> are a superset of what you would find within a single application's data
> model. People in different parts of our industry use the same words in
> different ways, for instance we have "Over the Counter" as a means to
> sell debt securities, which are essentially contracts that can be bought
> and sold, but derivatives traders use the term "Over the Counter" to
> signify bilateral contracts between two parties (struck, rather than
> traded, over the counter), and are often not aware of the other usage.
> The ontology gives context to the term, so that it should be clear from
> looking at the diagram, whether it is a diagram of a kind of bilateral
> contract or not.    (03)

When there are multiple meanings for a natural language term, there should
be multiple terms in the ontology, one for each meaning to be represented.
This would allow programs, not just people looking at a diagram, to deal
with the multiple meanings appropriately.  The two properties in this
case could be named "Struck_Over_The_Counter" and "Traded_Over_The_Counter"    (04)


> There is an interesting by-product of this.    (05)

The following seems to me to be a different issue.  The above issue is
how to deal with the same NL term having multiple meanings (in different
contexts), while the below issue is that different aspects of a class/
type or property are important in different contexts.    (06)

> In order to have a model in
> which each term used within the industry is modeled semantically, we
> have a taxonomy with multiple inheritance, but for any one application
> the data model of that application would be constrained to have a single
> inheritance taxonomy.    (07)

Why restrict a context that might deal with multiple aspects of a type
of thing?    (08)

> As an example, securities traders would be
> intersted in the cashflow of securities since this directly affects the
> cashflow of their portfolio, so they would classify things in terms of
> fixed v floating interest and amortizing versus "Bullet" (single
> payment) principal repayment. Meanwhile a risk management application in
> the "Middle office" would be more interested in the issuer and / or the
> collateralization of a security.
>
> Here's where it gets interesting: it should be possible to formally
> identify the different facets by which sub-classes of a class of thing,
> as "Classification Facets" (my term, but it might catch on).    (09)

This takes either the use of meta-classes, allowing predicates to have
classes (the facets) as arguments, or both.    (010)

> What if one
> could write a program that could parse those classification facet
> indications, and build a single taxonomy for a given application, based
> on a single facet? Then one ontology can be used to formally specify
> interoperable and meaningful terms across the enterprise.    (011)

An alternative would be to create multiple ontologies that combine to form
the whole, and let individual departments select a subset of these to use
for their own purposes.    (012)

-- doug foxvog    (013)

> The
> alternative, which I suspect goes on, is when people apply the
> limitations of the technology onto the ontology, which of course is
> right if one is documenting the ontology of a single, existing
> application, but offers no new benefits to the enterprise other than
> better application documentation. Since an ontology sits at the level of
> "Business Conceptual Model" in the Zachman Framework and other
> development frameworks, it should of course not reflect any technical
> design limitations.
>
> In terms of your definition, I would suggest that including taxonomy in
> the definition makes it circular. I would adapt Schwartz's definition to
> say something like "A taxonomy is a structure which models concepts in a
> domain from abstract to specific". I would also lose the word "Data"
> since both a data model and a conceptual business model may embody a
> taxonomy (one being of data elements, the other of real world entities).
>
> Cheers,
>
> Mike
>
> Deborah MacPherson wrote:
>> Hi Mike
>>
>> Thanks for this explanation. Your statements could apply to this
>> situation because of Omniclass <http://www.omniclass.org/>, a
>> multi-faceted classification strategy for the built environment.
>> Omniclass includes facility types, space types, properties,
>> organizational roles etc. in 15 related tables. Any object in the
>> built environment could be classified and processed any number of
>> different ways on multiple levels, using Building Information Modeling
>> (BIM) parameters, which some BIM vendors are beginning to support.
>> When BIM or CAD data needs to work within exchange models, with
>> pre-defined taxonomies like NIEM, an ideal ability would be setting up
>> exchanges and processing not only BIM parameters but also Geographic
>> Information Systems (GIS), financial data, regional and weather data,
>> health and human services, public safety - a huge assortment of
>> natural and man made activities that can be tied to specific buildings
>> and building types (ie hospitals).
>>
>> Based on your feedback, perhaps the definition could be +/-
>>
>> "A conceptual data model that represents relationships and rules
>> among nodes in a polyhierarchical taxonomy."
>>
>> Or, considering the suggestions from Alex S. and Chris M could be
>> simplified further
>>
>> "A conceptual data model that represents relationships and rules among
>> entities in exchange specific taxonomies."
>>
>> Maybe something like that. Thanks also David E for your comments. To
>> some extent data about buildings and infrastructure does need
>> handle terms having multiple meanings, more often though it is the
>> opposite where multiple terms have the same meaning.  There is a lot
>> of slang for example "drywall" versus "gypsum board".  Also the issue
>> of translating between with different natural languages - on the
>> simpler side English and French for projects in Canada, on the larger
>> side for the International Framework for Dictionaries
>> <http://dev.ifd-library.org/index.php/Ifd:IFD_in_a_Nutshell> (IFD)
>> which software developers should be able to use for many purposes.
>>
>> Another problem is a consistent, reliable method for getting back to
>> the simplest version of a term for exchanges. For example architects
>> and engineers could spend a lot of time determining the exact
>> requirements for a concrete mix or structural steel - project specific
>> properties. By the time that information gets to the point of
>> specifications and real world testing results - BIMs and construction
>> documents can lose sight of the fact a project-specific material needs
>> to revert to a generic, non-technical description for first responders
>> as simply "concrete" or "steel". Where it really gets complex is
>> trying to apply multi-faceted classification to the IFD to serve the
>> detailed needs of architects, engineers, software developers and data
>> modelers - but also the general needs of cost estimators, fire
>> departments, insurance agencies that only need 1 or 2 high levels to
>> be consistent across the entire spectrum of potential exchanges.
>>
>> So - relationships and rules among entities in exchange
>> specific taxonomies - might work
>>
>> Thanks again
>>
>> Deborah
>>
>> On Mon, May 31, 2010 at 10:23 AM, Mike Bennett
>> <mbennett@xxxxxxxxxxxxxxx <mailto:mbennett@xxxxxxxxxxxxxxx>> wrote:
>>
>>     I tend to start from the definition given in Schwartz 2005
>>     
>http://homepages.cwi.nl/~media/publications/masterthesis_kat_domainmodel_2005.pdf
>>     
><http://homepages.cwi.nl/%7Emedia/publications/masterthesis_kat_domainmodel_2005.pdf>
>>
>>     "A taxonomy is essentially a hierarchical tree structure which
>>     models a
>>     domain from abstract to specific."
>>
>>     However, she goes on to say that a taxonomy "should" not be
>>     polyhierarchical, which may be good advice for an individual
>>     application
>>     but I think supports the wider definition of a taxonomy as any set
>> of
>>     terms disposed according to transitive "Is A" relationships. So as
>>     such
>>     I think her definition is too specific.
>>
>>     Many developers take the technical (or common sense) limitation of
>>     single hierarchy and assume that this must apply to taxonomies
>>     generally. I don't go along with this. Some taxonomies (like
>> Linnaeus)
>>     are by definition monohierarchical because they classify entities
>>     according to one classification facet alone. Others like the ISO
>> 10962
>>     Classification of Financial Securities fall down precisely because
>>     they
>>     try to shoehorn entities into a single hierarchy while classifying
>>     them
>>     according to more than one classification criterion. In the EDM
>>     Council
>>     semantics repository ontology, we have built the model around a
>>     polyhierarchical taxonomy, in order to formally model each of the
>>     terms
>>     that is considered meaningful in the industry. One thing I am
>>     looking at
>>     for a future version is to formally identify the classification
>>     criteria
>>     against which each sub-set of something is defined. For example debt
>>     instruments are frequently classified according to their issuer
>> types
>>     (Corporate, Sovereign, Municipal) and separately against their
>>     cashflow
>>     behaviour (fixed, floating etc.) and these are all meaningful. One
>>     would
>>     expect any individual data application to base its data model around
>>     only one of those classification facets.
>>
>>     So my advice would be to describe something as "a" taxonomy in the
>>     singular if it contains a single coherent sest of entities disposed
>>     according to "Is A" relationships, whether that taxonomy is
>>     monohierarchical or polyhierarchical. That I think would be the
>>     simplest
>>     descriptive framework around which to dicuss the nature of any given
>>     taxonomy. I've started to standardise on the term "Classification
>>     Facet"
>>     for the different monohierarchical sets of content within that, and
>> I
>>     think others are converging on similar terms but I'm open to ideas.
>>
>>     Also note that this usage supports the creation and description of
>>     taxonomies which are themselves partitioned according to a lattice
>>     such
>>     as the KR Lattice, since one taxonomy may have e.g. Independent,
>>     Relative, Mediating as well as Continuant v occurrent at the top
>> level
>>     with "Thing" above that and multiply classified intersections below
>>     (classifying something as a Continuant Independent etc.). Though one
>>     could describe as a taxonomy any coherent sub-set of that whole, for
>>     instance a taxonomy of types of contract.
>>
>>     Mike
>>
>>     Deborah MacPherson wrote:
>>     >
>>     > Dear Ontolog Forum
>>     >
>>     >
>>     >
>>     > Since last July I've been talking with the National Information
>>     > Exchange Model (NIEM) Business Architecture Committee (NBAC) about
>>     > facilities information, and looking at NIEM documentation in more
>>     > detail to figure out what needs to be done with facility classes
>> and
>>     > xml schemas for re-use outside the building industry.
>>      Currently, NBAC
>>     > is looking at the upcoming Information Exchange Model (IEM)
>>     > Specification. An appendix lists definitions for IEM Artifacts,
>> the
>>     > following definition is used for Ontology
>>     >
>>     >
>>     >
>>     > "A conceptual data model that represents relationships and rules
>>     among
>>     > nodes in taxonomy"
>>     >
>>     >
>>     >
>>     > Please temporarily disregard previous conversations on this forum
>>     > about appropriate definitions for ontology - this seems to be OK
>> for
>>     > purposes of this exchange model - even if it may not be correct
>> for
>>     > other purposes. However, grammatically there seems to be a problem
>>     > with what is singular and what is plural
>>     >
>>     >
>>     >
>>     > ·         A conceptual data model
>>     >
>>     > ·         represents
>>     >
>>     > ·         relationships and rules
>>     >
>>     > ·         nodes
>>     >
>>     > ·         taxonomy
>>     >
>>     >
>>     >
>>     > My inclination is this should say "a" taxonomy. But that is why
>> I'm
>>     > writing, would it be more conceptually and technically correct
>>     to say
>>     > "multiple" or "related" or "a set of" taxonomies? Feedback would
>> be
>>     > appreciated on exactly how this short definition should be written
>>     > accurately. Also, the definition does need to stay very short
>>     >
>>     >
>>     > Thank you
>>     >
>>     >
>>     >
>>     > Deborah MacPherson
>>     >
>>     >
>>     >
>>     > --
>>     > ********************************************************
>>     >
>>     > Deborah L. MacPherson CSI CCS, AIA
>>     > Specifications and Research Cannon Design
>>     > Projects Director, Accuracy&Aesthetics
>>     >
>>     > ********************************************************
>>     >
>>     ------------------------------------------------------------------------
>>     >
>>     >
>>     > _________________________________________________________________
>>     > Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
>>     > Config Subscr:
>>     http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
>>     > Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
>>     <mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx>
>>     > Shared Files: http://ontolog.cim3.net/file/
>>     > Community Wiki: http://ontolog.cim3.net/wiki/
>>     > To join:
>> http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
>>     > To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx
>>     <mailto:ontolog-forum@xxxxxxxxxxxxxxxx>
>>     >
>>
>>
>>     --
>>     Mike Bennett
>>     Director
>>     Hypercube Ltd.
>>     89 Worship Street
>>     London EC2A 2BF
>>     Tel: +44 (0) 20 7917 9522
begin_of_the_skype_highlighting              +44 (0) 20 7917
9522      end_of_the_skype_highlighting
>>     Mob: +44 (0) 7721 420 730
>>     www.hypercube.co.uk <http://www.hypercube.co.uk>
>>     Registered in England and Wales No. 2461068
>>
>>
>>     _________________________________________________________________
>>     Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
>>     Config Subscr:
>> http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
>>     Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
>>     <mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx>
>>     Shared Files: http://ontolog.cim3.net/file/
>>     Community Wiki: http://ontolog.cim3.net/wiki/
>>     To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
>>     To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx
>>     <mailto:ontolog-forum@xxxxxxxxxxxxxxxx>
>>
>>
>>
>>
>> --
>> ********************************************************
>>
>> Deborah L. MacPherson CSI CCS, AIA
>> Specifications and Research Cannon Design
>> Projects Director, Accuracy&Aesthetics
>>
>> ********************************************************
>> ------------------------------------------------------------------------
>>
>>
>> _________________________________________________________________
>> Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
>> Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
>> Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
>> Shared Files: http://ontolog.cim3.net/file/
>> Community Wiki: http://ontolog.cim3.net/wiki/
>> To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
>> To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx
>>
>
>
> --
> Mike Bennett
> Director
> Hypercube Ltd.
> 89 Worship Street
> London EC2A 2BF
> Tel: +44 (0) 20 7917 9522
> Mob: +44 (0) 7721 420 730
> www.hypercube.co.uk
> Registered in England and Wales No. 2461068
>
>
> _________________________________________________________________
> Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
> Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
> Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
> Shared Files: http://ontolog.cim3.net/file/
> Community Wiki: http://ontolog.cim3.net/wiki/
> To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
> To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx
>
>    (014)


=============================================================
doug foxvog    doug@xxxxxxxxxx   http://ProgressiveAustin.org    (015)

"I speak as an American to the leaders of my own nation. The great
initiative in this war is ours. The initiative to stop it must be ours."
    - Dr. Martin Luther King Jr.
=============================================================    (016)


_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/  
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/  
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/ 
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx    (017)

<Prev in Thread] Current Thread [Next in Thread>