Re: Why spurious tuples with fifth normal form?

Good questions. Please correct me if I'm wrong, but isn't a join dependency when you are able to join decomposed tables on common keys to get the original table? In my case, I thought I had found a join dependency because most of the time, if I joined the decomposed tables together I got the original back. But every so often a spurious tuple showed up. Would this be an example of checking if a JD held? (and it didn't hold when the spurious tuple showed up)? Is the only way to check this by trial-and-error?

So if there would have been no spurious results, then I assume the JD would have held and thus I should have decomposed. I'm just not sure why my data set didn't have the JD when I have seen similar sets of data (with three inter-dependant columns) that did have JD's. Are all JD's cyclic? Thanks, I appreciate the explanations. Received on Thu Jun 16 2005 - 21:39:37 CEST