Re: "thou shalt not conflate meta-data with data"

From: Paul <paul_at_test.com>
Date: Fri, 04 Mar 2005 10:50:09 +0000
Message-ID: <42283d61$0$16766$ed2619ec@ptn-nntp-reader02.plus.net>

Neo wrote:

>>Please read my previous post. If I know the context, I can easily
>> distinguish data from metadata. If I don't know the context, I can't.

>
> That context is "data that imposes some condition (structure,
> constraints, significance, etc) upon other data" which is similar to
> your statement "the derivation rule and the constraint are metadata"
> which is also similar to Paul's alternative "meta-data is data about
> the representation of data". Can you refinement it?

I'm not sure that I'd say that all constraints were meta-data. It depends whether the constraints are expressible "inside" the model or not.

For example a foreign key constraint like "every person has a gender" makes sense inside the model, so I'd argue that it isn't meta-data (at least from a relational perspective).

But a "constraint" like "values in the 'person' domain are stored as varchar(50)" would be meta-data, because this is meaningless inside of the model.

Say your DBMS only allowed strings called "varchar" of unspecified length, then you *could* have a constraint like "all persons have names of less than 50 characters" which would be classed as data.

I think the grey area comes about because of strings being lists of characters, and the fact that the representation of the name "John" as a list of four symbols *does* have meaning inside the model. So a constraint expressed inside the model that all names are less than 50 characters can be data. But an artificial constraint of varchar(50) imposed from outside the model is meta-data. Quite a subtle difference though.

From a less rigorous perspective you might want to search for semiotics, semiology, Roland Barthes, signifier and signified, etc.

Paul. Received on Fri Mar 04 2005 - 04:50:09 CST