Re: How to represent category, subcategory, product

paul c wrote:

> JOG wrote:
> ...
> > * kermit is a member of the set of frogs and of the set of amphibians.
> ...
> > By normalizing to 3NF it becomes clear that you are actually dealing
> > with two distinct predicate structures:
> ...
> > * All frogs are amphibians.
> ...
>
> I like your point that making up tables without first forming the
> predicates is jumping the gun but in this example I don't see how the
> frogs are necessarily a subset of the amphibians (ignoring outside
> information).
>
> p

See bob's response, but also consider that we must always use outside information. A relation value is just a single member of a (perhaps infinite) set of all values for a relvar. And one certainly cannot construct an ideal schema from considering a single relation value in isolation (my point is that it can give you a lot of pointers nonetheless). But yes, because enumerating the complete set of possible rel values is also often unachievable, one finds the middleground, using heuristics and external knowledge to make our best effort of organizing the data. All best, J. Received on Thu Nov 23 2006 - 16:35:43 CET