Database Reference
In-Depth Information
Algorithm 19.1 Composing relational schema mappings
Require: on the source side variables are reused only in equalities
Σ 13 := 0
m := max ψ) Σ 23 ϕ
for all
ϕ 1 −→ π 1 ,
ϕ 2 −→ π 2 , ...,
ϕ k −→ π k Σ 12 , k
m do
{
in case of repetitions, rename variables
}
ρ
π 1 ∧···∧ π k
for all
:=
π α −→ π Σ 23 and all homomorphisms h :
π ρ
do
ϕ
π )
Σ
∧···∧ ϕ k
h (
α
)
−→
h (
13
1
end for
end for
return Σ 13
Proof Each constraint in
Σ 13 is of the form
ϕ 1 ∧···∧ ϕ k
h (
α
)
−→
h (
ψ
) ,
π α −→ π Σ 23 ,and h is a
where
ϕ 1 −→ π 1 ,
ϕ 2 −→ π 2 , ...,
ϕ k −→ π k Σ 12 and
)[ f , a ].
homomorphism from
π
into
π 1 ∧···∧ π k .Pick a such that S 1 |
=
ϕ 1 ∧···∧ ϕ k
h (
α
π 1 ∧···∧ π k [ f , a ] and we get S 2 |
)[ f , a ].As h (
)[ f , a ] holds, we have
Then S 2 |
=
= h (
π
α
)[ f , a ] and hence S 3 |
)[ f , a ].
S 2 |
= h (
π
)
h (
α
= h (
ψ
f , there exists an interme-
Lemma 19.10 If ( S 1 , S 3 )
M 13
with a witnessing valuation
diate instance S 2 such that ( S 1 , S 2 )
M 12
and ( S 2 , S 3 )
M 23
with the same witnessing
f.
valuation
Proof Consider
S 2 = R ( t 1 ,..., t )[ f , a ]
,
[ f , a ],
R ( t 1 ,..., t ) is an atom in
ϕ −→ π Σ 12 , S 1 |
=
ϕ
π
where R ( t 1 ,..., t )[ f , a ] is the tuple obtained from R ( t 1 ,..., t ) by interpreting the function
symbols according to f and the variables according to a . Obviously, ( S 1 , S 2 )
M 12
with
the witnessing valuation f , so it remains to see that ( S 2 , S 3 )
M 23
with the witnessing
valuation f .
Suppose that S 2 |
[ f , a ] for some
π α −→ π Σ 23 and let
=
π α
π
= A 1 ∧···∧
A k with
m . By definition of S 2 , A 1 [ f , a ] ,..., A k [ f , a ] are equal to
B 1 [ f , c 1 ] ,..., B k [ f , c k ] ,
k
ϕ i [ f , c i ] for i = 1 , 2 ,..., k
where B i is an atom of some
π i such that
ϕ i −→ π i Σ 12 and S 1 |
=
(we can assume the rules use disjoint sets of variables).
Since
π
uses no function symbols, nor reuses variables, we can define a homomorphism
g :
π π 1 ∧···∧ π k
Search WWH ::




Custom Search