Database Reference
In-Depth Information
each arrow f in DB I represents the PO relation dom(f )
cod(f ) . Let us show
that each PO relation A
B is represented by a monomorphism in DB I as well: if
0
A
then it is represented by a composition of the isomorphism (monic as well)
0 and initial arrow from
0 into B which is monic as well. Otherwise A
is
B is generally representable as A A S B S B where A S is a strictly-complex
object obtained by eliminating all objects
:
A
→⊥
0 from A (analogously for B S ). Thus
A B is represented in DB I as a composition of these two isomorphisms (that are
also monic) and a monomorphism in S I between two strictly-complex objects (the
composition of monic arrows is monic as well). Consequently, between any two
given objects in DB I there can exist at most one arrow, so this is a PO category. The
power-view operator T satisfies the following:
(i) TA
=
T(TA) , as explained in the introduction, Sect. 1.4.1 .
(ii) A
B implies TA
TB : T is a monotonic operator w.r.t.
(the free
SPRJU algebra of terms
L A is monotonic with respect to its carrier set A ;
see Sect. 1.4.1 ).
(iii) A
TA , each element of A is also a view of A .
Thus, T is a closure operator and an object A such that A
=
TA is a closed
object.
For any two simple morphisms f,g
B such that f
:
A
g , it is easy to show
: f
T e (f OP
f OP
in
=
f ep ;
f in
that there are monomorphisms β
)
g . In fact, for
ep
ep f ep : A A with h 1 = f OP
ep f ep = f f = f and h 2 = f in f OP
h 1 = f OP
in :
B with h 2 = f in f OP
f
f
= f , we obtain that h 2
= f
f
= f and
B
f
in
g
f
= f (for f
h 1 =
g
g ), so that h 2
f
=
g
h 1 , and, consequently, the
: f
morphism T e (h 1 ;
T e (f OP
f OP
in
h 2 )
=
ep
f ep ;
f in
)
g is well defined. Thus,
β = T e (f OP
ep f ep ; f in f OP
in OP
g T(f in f OP
in ) =
in f)
in
f , so that (from
T f OP
in = T f in = T f = f ), β = in OP
g T f in T f OP
in Tf in
f = g f = f ,so
T e (f OP
f OP
that β
in ) is a monomorphism (from Corollary 9 ).
It is easy to verify that DB is a 2-category with 0-cells (its objects), 1-cells
(its ordinary morphisms) and 2-cells (“inclusions” arrows between mappings). The
horizontal and vertical composition of 2-cells is just the comp os ition of PO re-
lations. Given f,g,h : A −→ B w ith 2-cells β : f −→ g , δ : g −→ h , their
vertical composition is γ
=
ep
f ep ;
f in
= δ
β
f −→
:
h .Given f,g
:
A
−→
B and
g , δ
C with 2-cells β
f −→
h −→
h,l
:
B
−→
:
:
l , for a given composition
functor
•:
DB (A,B)
×
DB (B,C)
−→
DB (A,C) , their horizontal composition is
δ
β
γ
f
=
:
−→
h
l
g .
For example, the equivalence q A i q B j (i.e., q A i q B j and q B j q A i )oftwo
view mappings q A i : A −→ TA and q B j : B −→ TB , for the simple databases A
and B , is obtained when they produce the same view.
The categorial symmetry operator T e J :
Ob DB for any morphism
f in DB produces its information flux f (i.e., the “conceptualized” database of this
mapping). Consequently, we can define a “mapping between mappings” (which are
Mor DB −→
Search WWH ::




Custom Search