文档介绍：Generalization
Based Data Mining in Ob ject
Orien ted Databases
Using an Ob ject Cub e Mo del
x y z x
Jia w ei Han Shojiro Nishio Hiro yuki Ka w ano W ei W ang
x
Sc ho ol puting Science
Simon F raser Univ ersit y
Burnab y
BC
Canada V
A
S
y
Departmen t of Informatio n Systems Engineering
Osak a Univ ersit y
Osak a
Japan
z
Departmen t of Applied Systems Science
Ky oto Univ ersit y
Ky oto
Japan
f han
cs
sfu
ca
nishio
ise
osak a
u
a c
jp
k a w ano
kuamp
k y o to
u
ac
jp
w eiw
cs
sfu
ca g
Abstract
Data mining is the disco v ery of kno wledge and useful information from the large amoun ts of data stored in
databases
With the increasing p opularit y of ob ject
orien ted database systems in adv anced database applications
it is imp ortan t to study the data mining metho ds for ob ject
orien ted databases b ecause mining kno wledge from
suc h databases ma y impro v e understanding
organization
and utilization of the data stored there
In this pap er
issues on generalization
ba sed data mining in ob ject
orien ted databases are in v estigated in three
asp ects
generalization plex ob jects
class
based generalization
and
extraction of di
eren tkinds
of rules
An ob ject cub e mo del is prop osed for class
based generalization
on
line analytical pro cessing
and data
mining
The study sho ws that
i
a set of sophisticated generalization op erators can b e constructed for generalization
plex data ob jects
ii
a dimension
based class generalization mec hanism can b e dev elop ed for ob ject cub e
construction
and
iii
sophisticated rule formation metho ds can b e dev elop ed for extraction of di
eren t kinds of
kno wledge from data
includin g c haracteristic rules
discriminant rules
asso ciation rules
and classi
cati on rules
urthermore
the applicatio n of suc h disco v ered kno wledge ma y substan tiall y enhance the p o w er and
exibilit yof
F
bro wsing database
or