Loading ...
Sorry, an error occurred while loading the content.

3363Re: [TaxoCoP] data modeling and taxonomy

Expand Messages
  • Keipat Patkei
    Jan 5, 2010
    • 0 Attachment
      John, great response.  Actually, this is terrific thread over all, and thank you, Heather, Gabriel, and Bob for providing terrific information getting at the crux of defining parts of the whole. 

      As some of you know, one of my frustrations over the years has been the ongoing view or use of the term "taxonomy" to describe sort of the beginning and the end of all enterprise management of data everywhere, i.e. "the be all and end all" or "THE solution" rather than being viewed as "A solution" or part of a larger system of models and decision-making depending on the nature of the enterprise, required need, expected outcomes, and so on.  I suppose my real frustration comes when "taxonomy" limitations are then considered a "trap" or "failure" by those who think it can work around or make up for deficiencies in other systems with which it must co-exist/integrate. 

      Getting at just where taxonomy, data modeling, and ontology specification begin, end, and overlap is really welcome. I've been actively engaged in all points raised in this thread in the course of my work of the last ten years; but you've provided me with some new ways of thinking about and describing/communicating the work that I do.  It's particularly welcome at the beginning of what I know is going to be a great year for all of us.  Thanks, again.

      Keith DeWeese

      Central Administration/Tribune Technology-Architecture

      Tribune Company-Tribune Interactive


      +1-312-527-8740 (w)/+1-312-286-3568 (c)

      --- On Tue, 1/5/10, John O'Gorman <jogorman@...> wrote:

      From: John O'Gorman <jogorman@...>
      Subject: Re: [TaxoCoP] data modeling and taxonomy
      To: TaxoCoP@yahoogroups.com
      Date: Tuesday, January 5, 2010, 10:30 AM


      Excellent question, and as an expert in neither activity (data modelling or taxonomy creation) I can give you a layman's perspective that may (or may not) add to the discussion.
      Individually, both exercises are designed to create a portable (meaning: for the purpose of communication) snapshot of someone's (or a group of someones) understanding of a problem space at a given point in time and in a defined 'locale'. From my point of view, these conditions (relatively narrow constraints of what, when and where) are prerequisites for the creation of a data model, be it conceptual, logical or to a lesser extent physical, and a taxonomy. To press the point a little, see what happens when one or more of the constraints are lifted: a data model for the insurance industry vs health care or now vs the 1900s or in Canada vs the United Arab Emirates (or for that matter between IBM and MicroSoft) would all be more or less different.
      So, the first thing a given data model and a given taxonomy have in common is that they are purpose-built to solve a particular problem. They have a whole range of assumptions built in that - depending on the skill of the creator - make them more or less extensible. A potential problem may exist if/when one of each is used to describe the same application, but that's a topic for another day.
      The second thing is given the constraints previously mentioned they should contain almost all of the same 'primary entities' but in different conformations. Data models and taxonomies do not serve the same function: the former is one of several classes of artifacts used in the process of capturing the information management lifecycle in a given enterprise. The latter is a standalone representation of the conceptual classification categories that normally serve only that function. Again, this is not without precedent or value: the alternator in my car only serves one purpose.
      The third thing is related to the previous point: Data models have almost unlimited degrees of freedom, while a taxonomy (at least in the traditional sense) only ever has two. Almost by definition, a data model must include any entity in which the particular problem space is 'interested' while a taxonomy must make binary decisions from top to bottom: in or out (of this category); yes or no. While the creative taxonomist can invent interesting splits, each level of a taxonomic hierarchy can only accommodate one choice.
      Lastly, I believe that there could be a much more organic relationship between data models and taxonomies. For example, if all of the 'entities' that a data modeller wanted to use were already classified by a taxomonist and resided in a master data managment inventory, then a sort of symbiotic relationship could exist between the necessarily narrow application of the data and the universal 'connectivity' of a fully faceted business vocabulary.
      It sounds like this is what Gabriel is close to (if nor exactly) describing, so apologies if I've stated the obvious.
      John O'
      -----Original Message-----
      From: Gabriel Tanase [mailto:gabtanase@ gmail.com]
      Sent: Tuesday, January 5, 2010 03:08 AM
      To: TaxoCoP@yahoogroups .com
      Subject: Re: [TaxoCoP] data modeling and taxonomy



      I agree with you and I already knew it, however I didn't want to complicate matters in my first response to Heather.

      Apologies if the below will sound like a product plug. It is not intended as such, it is only intended to give you brief information on something that people seem to have an interest for and illustrate one of the approaches that exist.

      As it happens, the set of models I am lead developer for (the IBM Insurance Information Warehouse - IIW, part of the IBM Industry Models product set) does have at its "conceptual" apex an ontology / taxonomy of business terms, relationships between these and a classification hierarchy. It is named the "Conceptual Model" (CM). It started originally - many years ago - as a flat glossary.
      (I said  "ontology / taxonomy" in the above because I'm not clear myself whether our CM does satisfy a full definition of "ontology"; for example as yet we have no mechanisms for making inferences).

      All the elements in the downstream IIW data models (below the taxonomy) do have mappings back to terms / concepts and relationships in the CM taxonomy.
      Also, the supporting toolset can generate a vocabulary / glossary from our CM that can be exported to the IBM InfoSphere Business Glossary product (part of the InfoSphere Metadata Server). From here it can be used as a reference vocabulary / glossary for other IBM products dealing with the data lifecycle (Information Analyzer, FastTrack, DataStage, QualityStage, Cognos).

      I believe that, currently, IBM's Industry Models is the only product on the market that supports data models with an ontology / taxonomy in the same model set and maintains mappings between these. There might be other similar products out there, however I am not knowledgeable of any myself.

      Kind regards,
      http://www.linkedin .com/in/gabrielt anase

      2010/1/4 Bob Bater <bbater@infoplex- uk.com>

      I think Gabriel captures the overlap between data modelling and taxonomy precisely. My only comment would be that the missing component is an over-arching ontology which expresses the context of both data model and taxonomy.






      From: TaxoCoP@yahoogroups .com [mailto:TaxoCoP@yahoogroups .com] On Behalf Of Gabriel Tanase
      Sent: 04 January 2010 22:31
      To: TaxoCoP@yahoogroups .com

      Subject: Re: [TaxoCoP] data modeling and taxonomy




      A data modeler is expected to create blueprint designs (known as 'logical data models') for data structures that are then physically implemented in databases, for use by software application( s).

      As part of logical data modeling a data modeler may create a generalization/ specialization hierarchy of types (classes, entities), which - IMHO - is the area where data modeling and taxonomy meet most intensely.

      http://www.linkedin .com/in/gabrielt anase

      2010/1/4 Heather Hedden <heather@hedden. net>

      How do data modeling and taxonomy relate?
      I'm curious to hear from those of you who have done data modeling how you would describe what a data modeler does.

      -- Heather

      Heather Hedden
      Hedden Information Management
      www.Hedden-Informat ion.com


    • Show all 21 messages in this topic