Some context would help but some (possibly relevant) possibilities:
1: If you have relation with 5 attributes you can generate tuples with all 5 attributes OR you can generate tuples with a subset of those attributes. Seems obvious but often these kinds of definitions are.
2: When looking at a relation you want to consider whether any subset of the attributes are dependent upon another attribute that is not a key for the relation. If so then that subset can likely be broken out into its own relation with the attribute that the subset relates to acting as a foreign key on the original relation.
Without further context as to where and why the definition is being given further conclusions are tough to draw. My first point is simply an obvious property of how relations and attributes and the second relates to normalization of a relation schema.
My $0.03 (inflation ya know)
David J
From: pgsql-general-owner@postgresql.org [mailto:pgsql-general-owner@postgresql.org] On Behalf Of Alpha Beta
Sent: Tuesday, February 15, 2011 5:01 PM
To: pgsql-general@postgresql.org
Subject: [GENERAL] subset of attributes
Hi,
while reading about databases, I didn't understand what can be subset of attributes, see the following :
Let Ri (Xi) be a relation sheme, where Xi is a set of attributes. and if t is a tuple for Ri (Xi) and Y is a subset of Xi, then t[Y] denote the subtuple of t corresponding to Y.
does a subset of attributes just means the instances that exist in each tuple?
Best,