[Sigia-l] Taxonomy/Classification IA and Enterprise IA

[John O'Donovan]

Hi Donna,

Thanks for your comments. The SMEF standard has been a good stake in the
ground for us at the BBC but there are other areas as well that have got
their head above water long enough to think at this level.

For example, in work I have done on Educational systems, I found good
emerging standards there for data interchange. E.G. SCORM:

http://www.adlnet.org/index.cfm?fuseaction=scormabt&cfid=204974&cftoken=3091
6792

The problem is competing standards in these areas and also ensuring lack of
bias towards aspects of the area under consideration (such as vendor
specific issues).

The metameta-model comes in when you have different metamodels to deal with,
providing a language and context in which to describe an infinite variety of
metamodels each with a different viewpoint. As you say, the level of
abstraction is important but it does not have to be exclusively object
oriented - relational data models also handle these structures well.

Cheers,

jod


----- Original Message -----
From: "Donna M. Fritzsche" <donnamarie at amichi.info>
To: "John O'Donovan" <jod at adito.net>; "Donna M. Fritzsche"
<donnamarie at amichi.info>; <Sigia-l at asis.org>
Cc: "John O'Donovan" <jod at badhangover.net>
Sent: Thursday, March 20, 2003 4:29 PM
Subject: Re: [Sigia-l] Taxonomy/Classification IA and Enterprise IA


>
> Hi John,
> This a great example!  I made some comments below.
>
>
> At 4:28 AM +0000 3/20/03, John O'Donovan wrote:
> >Hi Donna / Chiara,
> >
> >Providing end to end semantic data models is gaining considerable
ground -
> >and not just from a technology perspective. Using some examples from the
> >broadcast industry you might like to take a look at these:
> >
> >SMEF
> >http://www.bbc.co.uk/guidelines/smef/
>
> I think this quote from their site captures some of the important goals:
>
> "The BBC intends to use the SMEF-DM as a means of integrating key
> information systems; with an appropriate systems architecture, and
> harmonising other related systems as appropriate to improve
> commonality of data. SMEF-DM will also be used to manage data
> definitions across media applications, and embedded metadata
> definitions in media formats."
>
> Specifically the need to "harmonize" data is key.  We will not get
> rid of the legacy application problem for a long time, if ever,  and
> we will be dealing with legacy and various industry structures and
> semantics for a long time.
>
> >
> >and
> >
> >P-Meta
> >http://www.ebu.ch/tech_32/tech_t3295.html
>
> I took this quote form there site, I think it is very relevant to the
> earlier Enterprise discussions:
>
>   "The goal is to make a standard metadata exchange framework
> available for adoption in a business-to-business scenario without
> interfering in the internal structures, workflows, and concepts of
> the participating organisations. It offers a way to share the meaning
> of electronic information which is deemed to be necessary or useful
> for the business-to-business exchange of programme-related
> information and content"
>
> >
> >
> >This is not just about getting data from server to server, it is about
> >building metadata models that provide exciting new features by wrapping
> >content. In fact these features are becoming a key differentiator in
> >Broadcast Enterprise Systems.
>
> It sounds like they are ahead of the curve. Very Impressive!
>
> >The point about namespace is an interesting one. I found when dealing
with a
> >client in South Africa they had a particular problem with many
applications
> >and multiple languages (11) and multiple formats (Radio, web, TV) with
> >multiple contexts. Namespaces are a serious issue and integrity of
> >information from one application to the next is a possible point of
massive
> >complexity.
> >
> >The approach I use is to start thinking about higher levels of
abstraction.
> >Look at the metamodel - the framework that describes the metadata itself
and
> >ensure that this supports all the semantic concepts. In essence a
metamodel
> >is the collection of concepts that are the vocabulary with which you are
> >talking about a certain domain. Without being concerned with specific
> >technologies.
> >
> >You may also then find yourselves dealing with metameta-models...but I am
> >learning not to make big posts here so I'll leave that for now...:) I'll
be
> >glad to know anyone got this far...
> >
> >
>
> I think that we are very close to agreeing here.
> I think that the application profile/namespace model that I gave a
> link to encompasses the
> functionality you are talking about.  It basically provides a toolkit
> of different metamodels that can be mixed and matched depending on
> the application's need.  Each namespace then represents a set of
> metamodels.  It is a very object-oriented approach.  I agree, it is
> preferred that the top level model be the most abstract and that is
> should contain attributes that map across all subdomains when
> possible.
>
> Its hard to hold this conversation without a whiteboard!
>
> Thanks for the interesting conversation!
>
> Donna M. Fritzsche
> Partner
> Amichi, LLC
> Strategic Knowledge Architecure and Information Design
> www.amichi.info
> donnamarie at amichi.info
> (773) 680-2188
>
>