Re: A simple situation shows confusion about basic concepts

From: Ney André de Mello Zunino <zunino_at_inf.ufsc.br>
Date: Mon, 25 Oct 2004 18:33:55 -0300
Message-ID: <2u59kcF27bu54U1_at_uni-berlin.de>


Alan wrote:

> 0. It strikes me that KB is incorrect in that you have uc_id1 and uc_id2.
> There should only be uc_id, and this is a foreign key to UC.

The KB is a collection of entries, where each of which establishes a relationship between *two* UCs. There are different types of relationships, e.g. a /car/ *is a* /vehicle/ or /Brazil/ *is part of* /South America/. I thought of two foreign keys because the definition of a KB entry requires two UCs. I apologize if I have misunderstood your remark.

> 1. You only need uc_id at all if you have more than one uc with the same
> name(description?) or you will be adding additional attributes to UC. So...

Yeah, I considered that as well. However, because the UCs are going to be referenced from some other entities and because their "description" can get very large, I chose to use an artificial key, as mentioned in the original post.

> 3. need uc_id, and you therefor have a 1:m relationship between UC and KB.
> uc_id is the PK in UC and the FK in KB.
>
> You end up with:
>
>
> UC (Universal Concept)
> ======================
> uc_id (PK)
> uc
>
> KB (Knowledge Base)
> ===================
> entry_id (PK)
> uc_id (FK)
> relation_type (whatever that is)
>
> You are making this much more complex than it really is. This is a simple
> 1:m situation. E.g., one word has one or more definitions.

I don't see how the above scheme would allow me to model the binary relationships between UCs. I mean, each entry *must* refer to two UCs.

I hope to have clarified the situation a little better and thank you already for your response.

Regards,

-- 
Ney André de Mello Zunino
Received on Mon Oct 25 2004 - 23:33:55 CEST

Original text of this message