Database Reference
In-Depth Information
If
V
=
t then
B is an exact translation (i.e., side-effect free) for
t . Other-
wise
t .
The problem of translating general select-project-join (SPJ) view deletions to
underlying database deletions has been studied in [ 11 ] which gives an algorithm
that exploits lineage information to find an exact (i.e., side-effect free) deletion-to-
deletion translation whenever possible.
The lineage information is used as a starting point to enumerate all candidate
witnesses for a deletion. In [ 8 ], it was shown that this minimization is NP-hard
in general for monotonic select-project-join-union (SPJU) fragment of relational
queries.
B is inexact, that is, with side-effects (or extra deletions) E
=
V
\
Proposition 31
Let us consider a database mapping system ( graph ) G
=
(V G ,E G )
with a current mapping-interpretation α :
Sch (G)
DB ( i . e ., from Definition 11
and Proposition 16 , α
DB Sch (G) ) which is a model of each database
schema in V G ( but not necessarily satisfies all inter-schema mappings in G ) with
A
Int(G)
α (
=
A
) obtained by deleting some tuples in
A
and there exists
M BA ={
Φ
}∈
E G , where Φ is a normalized SOtgd formula
f
x 1 q B 1 ( x 1 )
r A, 1 ( t 1 ) ∧···∧∀
x m q B m ( x m )
r A,m ( t m )
with M BA =
. Let
( here S is defined in the algorithm 'Cmp' in Definition 11 for the relational symbols
r i 1 ,...,r i k in q B i ( x i ) )
S B = (V i ,t i )
MakeOperads( M BA ) ={ v 1 · q B, 1 ,...,v m · q B,m , 1 r }: B A
V i = q B i ( x i ) α ( B ) ,t i = Cmp(S, d 1 ,..., d k )
|
|
d m
α(r i m ),
= =∅
=
=
=
m
1 ,...,k, b
α(q B,i )( d 1 ,..., d k )
,α(v i )( b )
,
for v i ·
q B,i
MakeOperads(
M BA ),
O(r q ,r A,i )
q B,i
O(r i 1 ,...,r i k ,r q ),v i
, t i t i
is a subset of tuples accepted to be consecutively deleted from the view V i of a
database
be a nonempty set where for each element (V i ,t i ) ∈S B , 1
i n =|S B |
. Then the concatenation of updates transitions ( in Definition 49 ) of the
instance-databases of
B
B
M P n 1
M P 1
M P n
α ===
α n ===
α n 1 ···===
α 1
where P i is a sequential composition of RA arrows in the Application Plan consid-
ered as a program of categorial machine M RC which executes the deleting of the
tuples t i
OP
BA
M
, will be simply denoted by α ===
α 1 ,
t i in
B
for each (V i ,t i )
∈S B
OP
where
M
BA = M P n ◦···◦ P 1 is associated to the inverted arrow w . r . t .
M BA , so that
α 1 (
with minimal side-effects . Hence ,
we extend these changes to the mapping-interpretation ( a functor ) α 1 :
B
) is the final updated database instance of
B
Sch (G)
in V G , α 1 (
α (
DB such that for each
C = B
C
)
C
) .
Search WWH ::




Custom Search