Re: Mixing OO and DB

From: rpost <rpost_at_pcwin518.campus.tue.nl>
Date: Mon, 17 Mar 2008 20:26:06 +0100
Message-ID: <5762a$47dec5ce$839b4533$1977@news1.tudelft.nl>


Brian Selzer wrote:

>
>> If you can define "meaning" without reference to behavior
>> or operations, please do.
>
>Suppose you have a symbol, "Red." Under an interpetation, meaning is
>assigned to a symbol by associating it with an object in the universe of
>discourse, in this case the color Red. So what part does behavior or
>operations play in that association? None that I can see.

Yes, you can state the data, the extensions of your predicates, without any reference to what they mean. But in order to understand the data, you need to say how Redness is verified, and this will involve something like: a human observer has looked at it and called the color Red, or: we've subjected it to a test and the color spectrum satisfied this-and-this criterion, or something similar. I wanted to stress this point because in my experience sloppy data models often arise from modellers just labeling their relations and attributes with names that sound meaningful enough, without being specific enough about how the attribute values are obtained or verified.

-- 
Reinier
Received on Mon Mar 17 2008 - 14:26:06 CDT

Original text of this message