Quote:
Originally Posted by Pat Phelan
I can think of so many possible answers, and I can see that my original answer leaves out some important possibilities.
Why don't you take a stab at what you think is correct, then I'll offer my observations about it. You probably know more about your needs than I do, so you'll almost certainly get closer on the first try than I did.
-PatP
|
The problem is that each model I came into, seems to be not lossless or not the best one at least...
For example
The ternary relationship (with just one ternary table) could be
Crash Person Vehicle
1 1 1
1 2 1
2 1 1
2 2 1
1 3 2
3 4 1
3 5 1
Then I need a way to make it binary tables
From the theory I have if I call
Crash = X Person = Y Vehicle = Z
(XYZ) = (M:N:1)
Then the two posible decompositions are:
1) (XY)(XZ) with a binary imposition of (X:Z)=(M:1)
And the other one is
2) (XY)(XZ) -or- (XY)(YZ) with binary impositions of
(X:Z)=(M:1) and
(Y:Z)=(M:1)
Using 1)
Crash-Person Crash-Vehicle
1 1 1 1
1 2 2 1
2 1 1 2
2 2 3 1
1 3
3 4
3 5
I don't get what should I do with the binary imposition of (X:Z)=(M:1) here.
Now if I want to get all the people involved in crash 1 that were in the vehicle 1....
And it brings different results between the ternary option and this one...
Please run the attached script (inside the zip) and you'll be able to see what I'm talking about...
Thanks!!! Guillermo.