Re: ?? Functional Dependency Question ??

From: paul c <>
Date: Tue, 21 Oct 2008 17:20:07 GMT
Message-ID: <bboLk.2947$fF3.1778_at_edtnps83>

Bob Badour wrote:
> paul c wrote:

>> Bob Badour wrote:
>>> paul c wrote:
>>>> David BL wrote:
>>>> ...
>>>>> Consider that in the FD world symbol X represents a set of attributes
>>>>> from some relation R.  Let some tuple of R be given.  Then as a
>>>>> proposition we interpret X as implying that we are given or can deduce
>>>>> (for the given tuple) the values of all the attributes associated with
>>>>> X.   This interpretation makes it obvious that unions of attributes
>>>>> map to logical conjunctions, and that an FD maps to a logical
>>>>> implication.
>>>> Thanks, but how does that interpretation work when R has no attributes?
>>> The same way DEE works with join.
>> I'm not sure about that.  Maybe I'm being too literal in asking about 
>> this, but David B is talking about a mapping that involves values of 
>> attributes.  I don't know what the value of "no attribute" is.

> Regardless, it is a value. It can either exist in the relation or not.
> We can test it for equality. What else does one need to know about it?

I can see that the tuple in DEE must have a value, the empty set I presume. But when I think of an empty set of <A,T,V> triples, ie., {}, I'm darned if I can imagine a mental device for preserving a V, concrete or not, that corresponds with an A that isn't present. That's what I was quibbling about in David B's interpretation.

(Maybe I shouldn't drift the topic, but I'm guessing there's no problem with DUM! I guess empty relations never have any problem satisfying relational closure as long as their constraints aren't effectively negations, eg., "IS_EMPTY".) Received on Tue Oct 21 2008 - 19:20:07 CEST

Original text of this message