Re: Modeling Address using Relational Theory

On 9 Sep 2005 05:20:41 -0700, "dawn" <dawnwolthuis_at_gmail.com> wrote:

>I'd rather see what your conceptual model is.

In my experience, people don't learn much if they don't do their own thinking.

>I still think that the address I'm trying to model is an id for a
>location.

In my own work, I sometimes come across addresses like this fictionalized one.

 123 Flintstone St
 Mountain Pass, CA 12345

This mailbox isn't in Mountain Pass, Calif. (Neither is the building we might associate with it, but that's a slightly different issue.) It's in a city on the far side of the ridge.

I also have addresses like this fictionalized one.

 RR 2 BOX 1999
 BEAVER FALLS ID 54321 The mailbox (and the building we might associate with it) aren't in Idaho. The post office that handles its mail is in Idaho; the mailbox (and the building) are on the Wyoming side of the Idaho-Wyoming border.

Mail sent to the fictional ship Wibble might be addressed like this.

 USS Wibble
 FPO AE 99999 "USS Wibble" is neither a street address nor a post office box. "FPO" is not a city. "AE" is neither a state nor a possession. The Wibble itself is certainly not "in" AE, which refers to Europe, the Middle East, Africa, and Canada. It's probably in the water, and today it could be in the Pacific Ocean.

If you want to develop a conceptual model, you're probably better off starting with representative sample data and thinking about it, rather than starting with assertions or hypotheses about what addresses identify.

You can get a lot of good sample data *for the US* from USPS Pub 28. Don't expect other postal authorities to do things the way the USPS does.

-- 
Mike Sherrill