Information Technology Reference
In-Depth Information
~
. Meanwhile, all regular points of
x
ij
are in 0,1.
G
definition 12.7 belong to
p
1
n
So the regular points of
Ã
∏
=
[
a
(
x
)]
are in [0,1] also. That is, the maxium
ij
n
i
1
j
=
1
regular point of C will not be less than the maxium regular point of S, so C will
not be less than S. Because S, C,
ŋ
,
Ȳ
are in
~
, S can be compared with
ŋ
, C
and
Ȳ
. When S≥α and S≥β, the rule “if
X
is
D
then
Y
is
E
” will be suitable.
G
12.6 Distributed Mining Algorithm For Association Rules
The typical application of asociation rules mining is market-basket analysis,
where the items represent products, and the records represent point-of-sales data
at large grocery stores or department stores. Other application domains for
asociation rules mining include customer segmentation, catalog design, store
layout, and telecommunication alarm prediction. Because data is increasing in
terms of both the dimensions and size, asociation rules mining algorithm should
handle massive data stores. Sequential algorithms cannot provide scalability, in
terms of the data dimension, size, or runtime performance, for such large
databases. Therefore, we must rely on high-performance parallel and distributed
computing.
Two dominant approaches for using multiple processors have emerged:
distributed memory and shared memory. A shared-memory architecture has
many desirable properties. Each processor has direct and equal access to all the
system's memory. Parallel programs are easy to implement on such a system. In
a distributed-memory architecture, each processor has its own local memory,
which only that processor can access directly. For a processor to access data in
the local memory of another processor, message passing must send a copy of the
desired data elements from one processor to the other. Although a shared
memory architecture offers programming simplicity, a common bus's finite
bandwidth can limit scalability. A distributed memory, message-passing
architecture cures the scalability problem by eliminating the bus, but at the
expense of programming simplicity. A very popular paradigm combines the
best of the distributed- and shared-memory approaches. Included in this
paradigm are hardware or software distributed shared-memory systems. These
