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

Neo wrote:
>>John is a single entity in the set of persons.
>>Persons have an attribute of gender, which takes on the value male.
>>The representation of the value, say ISO codes (0,1,2,9), is drawn
>>from a domain, which have a data type and constraints.
>>All of these "levels" are completely different.

> 
> I still didn't get the fundamental distinction. What definition, rule
> or step-by-step method does one use to determine whether something is
> data or metadata?

There is no way to do it - something isn't intrinsically data or metadata. It can only be classed as metadata or data when you put it in some form of context. i.e. it's a relative concept.

To take an analogy that might be relevant to you :) consider the Matrix films. If you think of the people in the Matrix as "reality", then the people outside the Matrix are a meta-reality, because they define the inner reality. But these terms are only relative, because who's to say that the people outside the Matrix aren't actually in a larger meta-Matrix, and there is a further layer outside them? Etc.

To talk about systems you have a language. You can talk about most things to do with that system using that language. But sometimes you need to step outside the language to talk about the language itself. That's when the prefix "meta-" applies.

But of course you could just extend the language to include your meta-language statements. And then the meta-language ceases to be a meta-language, it's just part of your plain language. But of course you then need a further meta-language to talk about *that* extended language, and so on, ad infinitum.

In the database context, you are modelling some system using logical propositions. When you need to ask things about the implementation details e.g what domains are being used etc. you need to step outside of that data world into the meta-data world.

That should have made things as clear as mud.

Paul. Received on Tue Mar 01 2005 - 19:09:09 CET