Re: functional dependencies tht do not hold
From: <lauri.pietarinen_at_atbusiness.com>
Date: 4 Jun 2005 22:09:24 -0700
Message-ID: <1117948164.832298.177230_at_g14g2000cwa.googlegroups.com>
Date: 4 Jun 2005 22:09:24 -0700
Message-ID: <1117948164.832298.177230_at_g14g2000cwa.googlegroups.com>
Nabil wrote:
> hey,
> i have been studying databases for a while now . There exist some
> cases where it is usefull to store even those FD's that do not hold.
> Ofcourse we save these as not holding and do not use them while
> normalizing. Does ne1 know when it is helpfull to store fd's that do
> not hold??
An interesting case might be an "FD" which holds with some probability (it would not be an actual FD, of course, since they always hold 100%). This would be of help to the optimiser. Maybe FD's are just special cases of probabilities?
BTW, does anybody know if there has been any research exploring the relationship between Bayes networks and the RM?
regards,
Lauri Pietarinen
Received on Sun Jun 05 2005 - 07:09:24 CEST