Re: some information about anchor modeling

On Jul 19, 9:53 pm, Eric <e..._at_deptj.eu> wrote:
> Suppose I am designing a database and I have decided (rightly or wrongly,
> but I _have_ decided) to use a surrogate key for a table whose rows are
> about some real-world object. We have to communicate with other people
> about these real-world objects, so we suggest it would be easier for
> all concerned if everyone used the same arbitrary numbers (i.e. our
> surrogate key) to refer to each particular real-world object. In fact,
> since we are a government department, we can tell them that they have to
> use it. Oh, we seem to have just invented something very like a VIN!

Note that if one uses the name of an identifier of an entity in a communication, then it does not mean that this one can identify this entity in the real world.

Imagine that a Honda dealer has 2000 identical new Hondas, and that neither of them has a VIN number. In this situation, a database application that uses surrogate keys will not work at all. All the Hondas will have the same attributes and the unique surrogates. This database will be a total confusion and a database disaster. This is a very clear example that surrogates are bad solution. However, if I apply a database based on VIN numbers, then everything will be ok, and I do not need surrogate keys at all.

Note that many products have all the same intrinsic attributes. For this case, I introduced the law which is a generalization of Leibniz’s Law. (see my paper, Semantic Databases and Semantic Machines, section 5.6 at http://www.dbdesign11.com )

> > The VIN is an intrinsic property.
>
> No it is not. It is not even an acceptable candidate key. If you think
> that it is, consider a database about car crime and insurance fraud.

The VIN is an intrinsic property, you can see the surrogate on each car. The VIN is based on international standards. For example in US it is used by important institutions to identify individual motor vehicle.

> Are you trying to say that whoever makes something can assign it an
> identifier? Of course they can, but that doesn't necessarily make it
> intrinsic; painting or tattooing a number on something does not make
> that number an intrinsic property of what would still be the same thing
> without the number.

If you are interested in intrinsic properties and the identification of attributes, then you can see my paper "Database design and data model founded on concept and knowledge constructs" at http://www.dbdesign11.com . In section 2, I introduce intrinsic properties and in section 3.3 identification of attribute is defined in (3.3.3).

Vladimir Odrljin Received on Sun Jul 22 2012 - 01:59:15 CEST