Indonesi
an
Journa
l
of
El
ect
ri
cal Engineer
ing
an
d
Comp
ut
er
Scie
nce
Vo
l.
12
,
No.
3
,
Decem
ber
201
8
, p
p.
1
39
4
~
1
400
IS
S
N: 25
02
-
4752, DO
I: 10
.11
591/ijeecs
.v1
2
.i
3
.pp
1
3
9
4
-
1
400
1394
Journ
al h
om
e
page
:
http:
//
ia
es
core.c
om/j
ourn
als/i
ndex.
ph
p/ij
eecs
A Conce
ptu
al Model
of R
ole Base
d Access
Co
nt
rol
Using R
ole
Mini
ng Alg
or
ith
m
Na
z
ir
ah
Abd
Ha
mi
d
, Ra
bia
h Ahma
d, Si
ti
Rah
ayu S
el
am
at
Facul
t
y
of
Infor
m
at
ic
s a
nd
Com
puti
ng,
Univer
sit
i
Sulta
n
Z
ai
n
al
Abidin,
Ma
lay
si
a
Facul
t
y
of
Infor
m
at
ion
&
Com
muni
cation Techn
olog
y
,
Univ
ersit
i
Te
kn
ika
l
Mal
a
ysia
Mel
aka,
Ma
l
a
y
si
a
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
Sep
12
, 201
8
Re
vised
Oct
2
2
, 2
01
8
Accepte
d
Oct
3
1
, 201
8
Num
ero
us
studie
s
have
show
n
t
hat
cu
rre
nt
l
y
,
ro
le
-
base
d
acce
ss
cont
rol
h
as
bec
om
ing
one
of
the
succ
essful
a
cc
ess
cont
ro
l
m
odel
be
ca
use
of
i
t
s
princ
iple
tha
t
coul
d
sim
pli
fie
s
the
wor
k
of
sec
urity
a
dm
ini
strat
ors.
How
eve
r,
to
construc
t
a
con
ci
se,
role
-
b
ase
d
ac
c
ess
cont
rol
s
y
stem,
a
good
role
m
ini
ng
al
gorit
hm
structure
is
nee
ded
therefore
the
obj
ect
ive
s
of
thi
s
pape
r
are
first
l
y
,
to
provide
a
ge
ner
al
over
v
ie
w
on
phase
s
tha
t
invol
ved
in
des
igni
ng
and
deve
lop
ing
the
a
lgori
thm
and
sec
ondl
y
,
to
int
rodu
ce
a
concept
u
a
l
m
odel
tha
t
construc
t
ed
base
d
on
the
an
aly
sis
and
thi
s
m
odel
r
epr
ese
nts
a
g
eneral
proc
ess
in
rol
e
m
ini
ng
m
odel
.
Th
is
m
odel
invol
ved
ser
ie
s
of
ph
ase
s
th
at
b
egi
n
wi
th
the
inpu
t
of
d
at
a
,
pre
-
pro
ce
ss
ing
stage
,
ca
nd
ida
t
e
role
gen
erati
on
phase
,
ro
l
e
sele
c
ti
on
a
nd
rol
e
assignm
ent
proc
ess
and
la
stl
y
n
um
ber
of
role
s
as
gene
rat
e
d
output
.
Ke
yw
or
d
s
:
Con
ce
ptu
al
m
od
el
RB
AC
Role m
ining
Copyright
©
201
8
Instit
ut
e
o
f Ad
vanc
ed
Engi
n
ee
r
ing
and
S
cienc
e
.
Al
l
rights re
serv
ed
.
Corres
pond
in
g
Aut
h
or
:
Nazira
h Abd
H
a
m
id
,
Faculty
of In
form
atics and
C
om
pu
ti
ng
,
Un
i
ver
sit
i S
ultan Zai
nal Abid
in,
Ma
la
ysi
a.
Em
a
il
: ow
enira
@g
m
ai
l.co
m
1.
INTROD
U
CTION
Role
-
Ba
sed
A
ccess
Co
ntro
l
(RBAC)
has
dem
on
strat
ed
to
be
a
pro
pe
r
acce
ss
c
on
t
r
ol
m
od
el
to
m
anag
e
a
uthor
iz
at
ion
s
as
pect
especial
ly
for
the
secu
rity
adm
inist
rati
on
du
e
t
o
it
s
flex
ibil
it
y
and
abili
ty
to
captu
re
an
org
anizat
ion’s
str
uctu
re
an
d
obje
ct
ives.
Acc
or
ding
to
NIST
[
1]
,
RB
AC
is
a
ccepte
d
as
a
s
ecur
it
y
sta
nd
a
rd
i
n
num
ero
us
dom
ain
s,
s
uc
h
as
Indu
st
rial
,
Mi
li
t
ary
an
d
Healt
hcar
e
.
Role
e
ng
i
neer
i
ng
has
bee
n
introd
uced
by
[2]
an
d
has
be
en
ap
plied
to
de
fine
a
requisi
te
and
co
rr
ect
s
et
of
ro
le
s
a
nd
per
m
issi
on
s
an
d
ro
le
m
ining
is
a
con
cept
in
the
r
ol
e
eng
inee
rin
g
that
popu
la
r
am
on
g
the
rese
arch
e
rs
due
to
the
nature
of
a
pp
ly
in
g
com
pu
ti
ng
-
inte
ns
ive
a
ppr
oac
hes
that
co
uld
decr
ease
the
cost
of
m
ai
ntaining
t
he
secu
rity
featur
es
a
nd
al
s
o
si
m
plifie
s the
work of se
cu
rity
ad
m
inist
rator
s.
The
obj
ect
ives
of
t
his
pa
pe
r
are
firstly
,
to
a
naly
ze
an
d
cl
assify
on
so
m
e
of
t
he
prese
nt
ro
le
m
ining
al
gorithm
s
fr
om
20
13
to
20
17
an
d
the
n
pro
vid
e
a
gen
e
r
al
over
view
on
ph
a
ses
o
r
st
ages
t
hat
in
volved
i
n
desig
ning
a
nd
dev
el
op
i
ng
the
m
and
seco
ndl
y,
to
pro
pose
a
co
nce
ptu
al
m
od
el
that
co
nst
ru
ct
ed
based
on
the
analy
sis o
f
the
aforem
entione
d ph
a
ses.
The
rem
ai
nd
er
of
the
pa
per
is
structu
re
d
as
f
ollows.
T
he
an
al
ysi
s
of
the
ba
ckgr
ound
stu
dy
is
sho
w
n
in
Sect
ion
2.
Sect
ion
3
pres
ents
a
m
at
he
m
at
ic
al
back
gro
und
of
this
st
ud
y
wh
il
e
Sec
ti
on
4
i
ntr
oduc
es
the
gen
e
ral
pr
oces
s
in
r
ole
m
ining
m
od
el
.
Last
ly
,
Sect
ion
5
discuss
es
t
he
c
oncl
us
i
on
s
a
nd
the
f
uture
w
ork
that
can lead
to
fur
t
her en
ha
ncem
e
nt o
f
t
his f
ie
ld
.
2.
LIT
ERATUR
E REVIE
W
Nu
m
erous
stu
di
es
hav
e
re
port
ed
that
present
ly
,
ro
le
-
based
acce
ss
co
ntro
l
(RBAC)
has
be
com
ing
the
pr
e
dom
inant
acce
ss
co
ntr
ol
m
od
el
becau
se
of
it
pr
i
ncipl
e
that
co
uld
s
ign
ific
a
ntly
sim
pl
ifie
s
the
w
ork
of
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci
IS
S
N:
25
02
-
4752
A Co
ncep
t
ua
l
Mo
del o
f R
ole
Base
d
Acces
s
Con
tr
ol Usi
ng
Role Mini
ng Al
go
rit
hm
(
Nazir
ah A
bd Ha
mid
)
1395
secur
it
y
adm
inist
ra
tors
[3
-
6]
.
The
a
bovem
entione
d
pri
nci
ple
of
RB
AC
cou
l
d
be
def
i
ne
d
as
e
ve
ry
r
ol
e
is
a
gro
up of
per
m
i
ssion
s
, a
nd eac
h user o
btains
the
per
m
issi
on
s
only
thro
ugh
t
he roles.
Accor
ding
to
Ye
et
al
.
[
4]
,
RB
AC
syst
e
m
cou
ld
be
i
m
ple
m
ented
thr
ough
tw
o
ap
proache
s
sp
eci
fical
ly
the
top
-
dow
n
a
nd
the
bott
om
-
up
m
et
ho
d.
The
a
uthors
ha
ve
ex
plained
t
hat
th
e
top
-
dow
n
a
ppr
oac
h
bu
il
ds
a
RB
AC
syst
e
m
with
the
involv
em
ent
of
e
xp
e
rts’
a
naly
sis
on
the
bu
si
ness
proce
sses
ye
t,
this
appr
oach
consum
es
a
lot
of
ti
m
e
because
of
hu
m
an
pa
rtic
ipati
on
[
7]
.
The
bo
tt
om
-
up
ap
proac
h,
ac
cordin
g
to
Hu
et
al
.
[8]
can
unc
ov
e
r
r
oles
f
ro
m
the
existi
ng
us
e
r
-
pe
rm
issi
on
assignm
ents
(UP
A)
a
uto
m
at
ic
ally
that
is
kn
ow
n
a
s
ro
le
m
ining
and
beca
us
e
of
it
s
natur
e
that
base
d
on
c
ompu
ti
ng
-
intensi
ve
app
r
oac
h,
it
is
widely
app
li
ed
to
bu
il
d a R
BAC
m
od
el
.
Howe
ver,
to
buil
d
an
d
su
sta
i
n
a
RB
AC
m
od
el
,
r
ole
m
ining
is
beco
m
ing
a
gr
eat
interest
[9
-
10]
and
the
a
uthors
ha
ve
i
den
ti
fie
d
t
he
nee
d
of
r
ol
e
m
ining
to
de
sign
an
d
de
vel
op
a
n
al
go
rith
m
to
determ
in
e
r
oles
base
d
on
data
m
ining
m
et
hods
beca
us
e
it
cou
l
d
reduce
the
c
os
t
of
al
lo
cat
ing
r
oles
m
anu
al
ly
th
us
a
ble
to
const
ru
ct
a
con
ci
se RB
AC sys
tem
. Th
e
nex
t
sect
ion
w
ou
l
d pro
vid
e i
n
-
dep
t
h
a
naly
sis o
n m
et
ho
dolo
gy to bu
il
d
a RB
AC m
od
e
l usin
g ro
le
m
inin
g
al
go
rithm
.
2.1
.
Ro
le
Mi
ning M
od
el
In
ge
ne
ral,
Fuchs
an
d
Me
ie
r
[11]
has
int
r
oduce
d
a
general
Role
Mi
nin
g
P
ro
ces
s
Mod
el
as
in
Figure
1
a
nd th
e autho
r has
de
scribe
d
the
pha
ses in Fi
gure
1
as the
fo
ll
owin
g:
Figure
1. Role
m
ining
process
m
od
el
2.1.1 In
pu
t Data
In
m
os
t
of
the
ro
le
m
ining
al
gorithm
s,
us
er
-
per
m
issi
on
ass
ign
m
ent
(U
P
A
)
m
a
trix
can
be
con
si
der
e
d
as
input
data,
corres
pondin
gly,
an
act
ive
re
search
sho
uld
be
done
t
o
e
xplore
t
he
po
ssi
bi
li
t
y
of
oth
e
r
t
ypes
of
data to
be use
d i
n ro
le
m
ining p
ro
ces
s.
2.1.2 Pre
-
pr
oc
essi
ng
Ma
ny
researc
he
rs
ha
ve
em
ph
asi
zed
on
the
im
po
rtance
of
this
sta
ge
[
12
-
13]
par
ti
cularly
to
gen
e
rate
a
cl
ean
a
nd
qual
it
y
data
an
d
usual
ly
pre
-
pro
c
essing
sta
ge
involve
the
pro
cess
to
cl
ean
t
he
no
ise
s
that
m
igh
t
aff
ect
th
e re
su
l
ts.
2.1.3
Ro
le
De
t
ection
This
sta
ge
is
sign
ific
a
nt
in
the
ro
le
m
ining
proces
s
m
o
del
beca
us
e
it
inv
ol
ves
the
disco
ver
y
of
appr
opriat
e
cand
i
date
r
oles f
r
om
ob
ta
inable set
of
in
p
uts pr
efera
bly
a
cl
ean d
at
a
us
i
ng
da
ta
m
ining
te
ch
niqu
e
s
or h
e
ur
ist
ic
s al
gorithm
s call
e
d
as
role m
ining
alg
ori
thm
.
2.1.4 P
ost
-
pr
oc
essing
In
t
his
sta
ge
,
t
he
ac
qu
i
red
ca
nd
i
date
r
oles
f
ro
m
the
previ
ous
sta
ge
ar
e
be
ing
sel
ect
e
d
a
nd
assi
gn
e
d
op
ti
m
al
l
y by usi
n
g y
et
o
the
r
s
uitable
alg
or
it
hm
s.
2.1.5 O
upu
t D
ata
The
outp
ut
of
these
processe
s
is
norm
al
l
y
a
set
of
r
oles
and
a
RB
AC
sta
te
su
c
h
as
hierar
c
hy
or
involvin
g
c
ons
trai
nts.
2.2
.
Ro
le
Mi
ning Alg
orit
h
m Ph
as
es
Nu
m
erous
stu
dies
ha
ve
pro
po
s
ed
ro
le
m
i
ning
al
gorith
m
s
that
cou
l
d
s
olv
e
Role
Mi
ni
ng
P
roblem
(RMP)
in
R
ole
Ba
sed
Access
Con
tr
ol
(RBA
C)
syst
e
m
and
this
fo
ll
owin
g
sect
ion
w
ould
analy
ze
and
cl
assify
on
so
m
e
of
the
pr
ese
nt
r
ole
m
ining
al
gorit
hm
s
fr
om
20
13
to
2017
acc
ordi
ng
t
o
the
ph
ases
that
sta
te
d
in
2.1
sect
io
n
a
nd the
d
et
ai
l i
s prese
nted
i
n
Ta
ble
1.
Inp
u
t Data
Pre
-
Proces
sin
g
Ro
le Detec
tio
n
Po
st
-
Proces
sin
g
Ou
tp
u
t Data
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2502
-
4752
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci,
Vo
l.
12
, N
o.
3
,
Dece
m
ber
2
01
8
:
1
3
9
4
–
1
400
1396
Table
1.
R
ole
m
ining
alg
ori
thm
p
hases
No
Au
th
o
rs (
Year
)
&
Title
Can
d
id
ate Ro
les (CR) Phas
e and
Ro
le Selection
(
RS
)
&
Assig
n
m
en
t
Ph
ase (A
P
)
Inp
u
t &
Outp
u
t Da
ta
1.
14]
An
E
f
f
icien
cy
App
roach
f
o
r
RB
AC
Reco
n
f
ig
u
ration
with Mini
m
al
Ro
les an
d
Per
tu
rba
tio
n
CR
: gen
erate
the c
an
d
id
ate r
o
les
u
sin
g
FastMiner al
g
o
rith
m
[
1
5
]
RS & AP:
sel
ect a
su
b
set f
ro
m
CR
th
at
m
in
i
m
i
zes th
e
p
erturbatio
n
Inp
u
t: acce
ss
his
to
ry
log
&
U
PA
m
a
trix
Ou
tp
u
t: GR as the
g
en
erate
d
r
o
les
an
d
Q
R as th
e qu
al
if
ied
r
o
les
2.
[
1
6
]
An
App
roach
f
o
r
Hierarch
ical
RB
AC Reco
n
f
ig
u
r
atio
n
with
Mini
m
al
Per
tu
rbati
o
n
CR
: gen
erate
the c
an
d
id
ate r
o
les
u
sin
g
FastMiner al
g
o
rith
m
[
1
5
]
RS & AP:
sel
ect th
e si
m
ila
r
roles
wh
ich
is t
h
e
m
o
st s
i
m
ilar
the set o
f
q
u
alif
ied
r
o
les
Inp
u
t: acce
ss
his
to
ry
log
&
U
PA
m
a
trix
Ou
tp
u
t: hierarc
h
ic
al RBAC
state
(RH)
3.
[
1
0
]
The RB
AC S
y
ste
m
Bas
ed
on
Ro
le
Ris
k
and
User
T
rus
t
CR
: clus
tering
r
o
le
s b
ased
on
r
isk
RS & AP:
f
o
r
ea
ch
r
o
le iden
tif
y
the
p
er
m
iss
io
n
(
P)
>
a
ss
ig
n
ed
tr
u
st
th
resh
o
ld
Ph
ase 1
: clus
tering
r
o
les b
ased
on
risk
Inp
u
t: UPA
m
atrix
Ou
tp
u
t: RBAC
r
o
les &
R
H stab
le
Ph
ase 2
: tr
u
st
Inp
u
t: RBAC
r
o
les
Ou
tp
u
t: r
o
les
4.
[
1
7
]
Mutu
al E
x
clu
sio
n
Ro
le Co
n
strain
t
Minin
g
Based
o
n
W
eig
h
t in Ro
le
-
Bas
ed
Acce
ss
Co
n
trol Sy
ste
m
CR
: gen
erate
the c
an
d
id
ate
p
er
m
iss
io
n
sets b
a
sed
on
weigh
t
RS & AP:
gen
er
at
e all
co
m
b
in
atio
n
s o
f
per
m
iss
io
n
sets
wh
o
se weig
h
ted
su
p
p
o
rt
is g
reate
r
th
an
the u
ser sp
eci
f
ied
m
in
i
m
u
m
weig
h
ted
Inp
u
t: UPA
m
atrix
Ou
tp
u
t: r
o
les
5.
[
1
8
]
Sco
tt D.
Sto
ller
an
d
T
h
an
g
Bu
i
(20
1
6
)
Minin
g
Hier
ar
ch
ic
al T
e
m
p
o
ral
Ro
les with
M
u
ltip
le M
et
rics
CR
: gen
erate
s in
iti
al r
o
les an
d
t
h
en
crea
tes
ad
d
itio
n
al cand
id
ate r
o
les
b
y
intersectin
g
sets
of
initial roles
u
sin
g
FastMiner
[
1
5
]
RS & AP:
Co
n
stru
ct r
o
le hier
arch
y
Inp
u
t: ACL
p
o
licy
Ou
tp
u
t: r
o
le hierar
ch
y
6.
[
1
9
]
Minin
g
App
rox
i
m
ate Ro
les u
n
d
er
I
m
p
o
rtant Ass
ig
n
m
e
n
t
CR
: UPA
is
deco
m
p
o
sed
into
two
ass
ig
n
m
en
ts,
NUP
A & IUP
A
RS
&
AP:
(1) N
U
PA is pro
cess
ed
b
y
the δ
-
Ap
p
rox
RM
alg
o
rith
m
an
d
gen
erate
s NRo
les (2)
IUPA
is
p
rocess
ed
by
an
y
a
lg
o
rith
m
to
g
en
erate
I
Ro
les
Inp
u
t: UPA
m
atrix
Ou
tp
u
t: r
o
les
7.
[
2
0
]
Minin
g
T
e
m
p
o
ral
Ro
les u
sin
g
Many
-
Valu
ed
Co
n
cept
s
CR
: con
stru
ct the
co
n
cept
s o
n
ly
f
o
r
a pre
-
d
ete
r
m
in
ed
v
alu
e of
θ (
m
an
y
-
v
alu
e con
cept
).
RS & AP:
m
ax
i
m
u
m
a
rea
o
f
co
v
erage is selecte
d
and
is add
ed
to
th
e f
in
al set of
conce
p
ts
Inp
u
t: T
UPA
m
at
ri
x
Ou
tp
u
t: UA,
PA
& REB
8.
[
2
1
]
Towards
Use
r
-
o
rie
n
ted
RBAC
Mod
el
CR
: three dif
f
erent way
s
of
g
en
erating
cand
id
ate r
o
les (1
)
itself
(2) intersectio
n
(3)
asso
ciatio
n
RS & AP:
enf
o
rce m
a
x
i
m
a
l r
o
le
ass
ig
n
m
en
t con
stra
in
ts (1) greed
y
(2) fewes
t (
3
)
m
o
st
(
4
)
rand
Inp
u
t: UPA
m
atrix
Ou
tp
u
t: UA
&
PA
9.
[
2
2
]
The Genera
lized
T
e
m
p
o
ral
Ro
le
Minin
g
Pr
o
b
le
m
CR
: cr
eatio
n
of
ca
n
d
id
ate r
o
le set
b
y
takin
g
un
io
n
o
f
the sets
of
un
its,
in
itial and
gen
erate
d
r
o
les
RS & AP:
it
erative
selectio
n
of
a
m
in
i
m
al
car
d
in
alit
y
su
b
set o
f
the
cand
id
ate r
o
le set
u
sin
g
any
o
n
e of
th
e f
o
u
r
g
reedy
h
e
u
ristics
Inp
u
t: T
UPA
m
at
ri
x
Ou
tp
u
t: R,
UA,
P
A & REB
10.
[
2
3
]
Ro
le M
in
in
g
Based
o
n
Per
m
iss
io
n
Card
in
ality
Co
n
str
ain
t and
User
Card
in
ality
Co
n
str
ain
t
CR
: initial role:
(1) on
e is f
ro
m
the
p
rer
eq
u
isite role se
t (
2
)
th
e initial
role set g
en
eration
alg
o
rith
m
RS & AP:
role sele
ctio
n
algo
rith
m
&
r
o
le
state gen
er
a
tio
n
algo
rith
m
Inp
u
t: UPA
m
at
rix
Ou
tp
u
t: UA
&
PA
11.
[
7
]
Ro
le M
in
in
g
bas
ed
on
Card
in
ality
Co
n
strain
ts
CR
: gen
erating
the in
itial r
o
le set
b
ased
car
d
in
ality
c
o
n
strain
ts o
f
roles
and
per
m
iss
i
o
n
s
RS & AP:
sel
ectin
g
r
o
le pair fo
r
role u
p
d
ate algo
rith
m
(
h
ier
archical
relation
sh
ip
s) &
u
p
d
atin
g
the in
itial
role state
(graph
op
ti
m
i
zatio
n
alg
o
rit
h
m
)
Inp
u
t: UPA
m
atrix
Ou
tp
u
t: r
o
les
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci
IS
S
N:
25
02
-
4752
A Co
ncep
t
ua
l
Mo
del o
f R
ole
Base
d
Acces
s
Con
tr
ol Usi
ng
Role Mini
ng Al
go
rit
hm
(
Nazir
ah A
bd Ha
mid
)
1397
12.
[
2
4
]
Mi
g
rating
f
ro
m
D
AC to
RBAC
CR
: each iter
atio
n
us
es th
e
DEM
in
e
r
alg
o
rith
m
to
gen
erat
es a
cand
id
ate r
o
le
RS & AP:
at
eac
h
i
teration
,
th
e new
cand
id
ate r
o
le is in
tersected with
th
e r
o
les in
R and
UPA a
re
p
erfo
r
m
ed
at
ea
ch
iteration
to
reflect th
e up
d
ates
in
R.
Inp
u
t: UPA
m
atrix
Ou
tp
u
t: UA
&
r
o
le
s
13.
[
5
]
Han
d
lin
g
L
east Pri
v
ileg
e Pr
o
b
le
m
an
d
Ro
le M
in
in
g
in RB
AC
CR
: (
1
)
The
f
irst kin
d
of
cand
i
d
ate
roles
set (
FCR
)
(2) T
h
e se
co
n
d
kind o
f
ca
n
d
id
ate
roles
set (
SCR
)
RS & AP:
le
ast p
ri
v
ileg
es p
rincip
le
Inp
u
t: a
set o
f
us
ers,
a set of
p
rivileg
es, and
a
se
t of
us
er
-
p
rivileg
e assig
n
m
e
n
t r
elatio
n
Ou
tp
u
t: r
o
les
14.
[
2
5
]
Perf
o
r
m
an
ce of
A
I
Algo
rith
m
s f
o
r
Minin
g
M
eani
n
g
f
u
l Ro
les
CR
: gen
erating
r
o
les u
sin
g
eli
m
in
atio
n
algo
rit
h
m
RS & AP:
in e
ach
g
en
eration
,
th
e
elitist selectio
n
sch
e
m
e is
app
lied
to
gu
arantee that t
h
e f
ittest
m
e
m
b
er
o
f
eac
h
gen
eration
is cop
ied
d
irectly in
to
the n
ex
t gen
eration
(GA)
Inp
u
t: RH
Ou
tp
u
t: r
o
les
15.
[
26]
An
Opti
m
izat
io
n
Fr
a
m
e
wo
rk f
o
r
Ro
le M
in
in
g
CR
: three dif
f
erent way
s
of
g
en
erating
cand
id
ate r
o
les (1
)
itself
(2) intersectio
n
(3)
asso
ciatio
n
RS & AP:
gr
eedy
alg
o
rith
m
Inp
u
t: UPA
Ou
tp
u
t: UA
&
PA
16.
[
2
7
]
Visu
al E
licitatio
n
o
f
Ro
les: u
sing A
Hy
b
rid Ap
p
roach
CR
: Ran
d
o
m
Data
Gen
erator
(RDG)
RS & AP:
Matr
ix
s
o
rting
algo
rith
m
Inp
u
t: UPA
Ou
tp
u
t: r
o
le sets
17.
[
2
8
]
Toward
Minin
g
of
Te
m
p
o
ral
Ro
les
CR
: enu
m
erate
s
the set o
f
cand
id
ate r
o
les f
rom
an
inp
u
t
TUPA
m
at
rix
Ro
le Selection
and
assig
n
m
en
t:
elects th
e least
po
ss
ib
le nu
m
b
er
of
roles
f
ro
m
the c
an
d
id
ate r
o
les u
sin
g
a greedy
heu
ristic
Inp
u
t: T
UPA
m
at
ri
x
Ou
tp
u
t: UA,
PA
& REB
18.
[
3
]
Ro
le M
in
in
g
Usin
g
Bo
o
lean
M
atrix
Deco
m
p
o
sitio
n
with Hierarch
y
CR
: the cand
id
ate
roles
thro
u
g
h
f
o
r
m
al
con
cept
ana
ly
sis
RS & AP:
redu
n
d
an
t r
o
les can b
e
re
m
o
v
ed
acc
o
rdin
g
to co
st
-
u
tility
an
aly
sis
Inp
u
t: UPA
m
atrix
Ou
tp
u
t:
UA,
PA
,
RH, UA
′
&
PA
′
m
a
trix
19.
[
2
9
]
Minin
g
Par
a
m
eteri
zed R
o
le
-
b
ased
Po
licies
CR:
u
se Co
m
p
le
te
Miner
[
1
5
]
to
g
en
erate
cand
id
ate
r
o
les.
RS & AP:
(1) I
t
se
lects roles
f
ro
m
h
ig
h
est q
u
ality
to
lo
west (2)
Co
m
p
u
te Ro
le
Hie
rar
ch
y
(
3
)
Inp
u
t: UPA
m
atrix
Ou
tp
u
t: UA,
PA
& RH
20.
[
3
0
]
Evo
lv
in
g
r
o
le def
i
n
itio
n
s th
rough
p
er
m
iss
io
n
in
v
o
catio
n
pattern
s
CR
: which
ar
e s
ele
cted
to o
p
ti
m
ize
an
ob
jectiv
e f
u
n
cti
o
n
that b
alan
ces
d
istan
ce fro
m
the
o
rigin
al r
o
les
with
beh
av
io
rial
sim
i
larit
y
in
th
e
f
o
r
m
of
per
m
iss
io
n
RS & AP:
assig
n
e
d
to roles
accord
in
g
to a cr
it
erion
that
m
i
tig
ates
redu
n
d
an
cy
Inp
u
t: acce
ss
his
to
ry
log
Ou
tp
u
t: r
o
les
3.
MA
T
HEM
AT
ICA
L
BA
CKGRO
U
ND
This
sect
ion
pr
esents
so
m
e
of
the
fo
rm
al
de
finiti
on
s
that
r
el
at
ed
to
the
R
ole
Ba
sed
Acc
ess
Con
tr
ol
(RBAC)
a
s
we
ll
as
Role
Mi
ni
ng
Prob
le
m
(RMP)
an
d
it
s
var
ia
nts
an
d
s
om
e
of
te
rm
s
are
ass
ociat
ed
to
th
e
con
ce
ptu
al
m
od
el
.
Def
ini
ti
on
1. (
RBA
C Model
)
The
RB
AC m
od
el
h
a
s the
foll
ow
i
n
g ba
sic
ele
m
ents
[14]
:
a)
U,
R
an
d P are
sign
ify
in
g
t
he se
t of u
sers
, rol
es an
d perm
issi
on
s
.
b)
UA
⊆
U
×
R is
r
e
pr
ese
ntin
g
t
he user
-
r
ole as
sign
m
ents.
c)
PA
⊆
P
× R is
def
i
ning the
ro
le
-
per
m
issi
on
a
ssign
m
ents.
d)
UPA
⊆
U
×
P is
the
us
er
-
per
m
i
ssion assi
gnm
e
nts.
e)
RH
⊆
R × R,
a
p
a
rtia
l order
on roles
desc
rib
ed
the
in
her
it
a
nce
relat
ion
s
hi
ps
.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2502
-
4752
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci,
Vo
l.
12
, N
o.
3
,
Dece
m
ber
2
01
8
:
1
3
9
4
–
1
400
1398
Def
ini
ti
on
2. (
RBA
C
s
tate)
RB
AC
sta
te
cou
l
d
be
e
xpre
ssed
as
<
R,
U
A,
P
A,
R
H>
that
is
co
ns
ist
ent
with
a
n
acc
ess
co
ntr
ol
config
ur
at
io
n
=<
,
,
>, where U defines a set o
f a
ll
u
sers,
P
is a set
o
f
al
l per
m
issi
on
s a
nd
⊆
×
a
s the
u
se
r
-
pe
rm
issi
on
r
el
at
ion
[10]
.
Def
ini
ti
on
3. (
Basic R
MP
)
Assum
ed
a
set
of
us
e
rs
(
U
),
a
set
of
pe
rm
i
ssion
s
(P)
a
nd
a
us
e
r
-
per
m
is
sion
as
sig
nm
e
nt
(
UPA)
is
giv
e
n,
ac
quire
a
set
of
r
oles
(
R),
a
us
er
-
r
ole
assignm
ent
(
UA)
a
nd
a
r
ole
-
pe
rm
issi
on
assignm
ent
(PA
)
w
hile
reducin
g
t
he n
um
ber
of
ro
le
s
|R
| = k.
In m
at
r
ix notat
ion, it
c
an be
form
ulated
as
[
12
]
:
∥
UA
P
A
−
UPA
∥
1 =
0
(1)
Def
ini
ti
on
4. (
-
Ap
pr
ox
RMP)
δ
-
a
ppr
ox
RM
P
is
a
par
t
of
Ba
sic
-
RM
P
that
al
lows
a
par
ti
al
m
a
tc
h
between
the
us
e
r
-
pe
rm
issi
on
assignm
ent
(UPA)
a
nd
the
ge
ner
at
e
d
us
e
r
-
r
ole
assignm
ent
(U
A
)
an
d
a
rol
e
-
per
m
issi
on
assignm
ent
(PA)
an
d
can
occasio
na
ll
y
decr
ease
t
he
total
nu
m
ber
of
ro
le
s
,
k,
subst
a
ntial
ly
.
It
can
be
f
orm
ulate
d
in
m
at
rix
represe
ntati
on
,
su
c
h
t
hat
[
31
]
:
∥
UA
PA −
U
PA
∥
1 ≤
δ
(2)
Def
ini
ti
on
5. (
MinNoise
R
M
P)
Fo
r
Mi
nNoise
RM
P,
the
num
ber
of
ro
le
s
(k)
is
bounde
d
s
o
that
the
num
ber
of
m
is
m
at
ches
betwee
n
the
UPA
an
d
t
he
ge
ner
at
e
d
UPA
is
m
ini
m
iz
ed.
S
o,
pres
um
ed
a
set
of
use
rs
(
U),
a
set
of
per
m
issi
on
s
(P
)
,
a
us
er
-
per
m
issi
on
assi
gn
m
ent
(
UPA
)
a
nd
a
num
ber
of
ro
le
s
(
k)
is
gi
ven,
di
scov
e
r
a
set
of
k
r
oles
(R),
a
us
e
r
-
ro
le
as
sig
nm
ent U
A
and a
rol
e
-
pe
rm
issi
on
assign
m
ent PA
by m
ini
m
iz
ing
[31]
:
∥
UA
PA −
U
PA
∥
1
(3)
Def
ini
ti
on
6. (
User
-
Permi
ssion Assi
gn
men
t)
The
use
r
-
pe
rm
issi
on
assi
gn
m
ent
(UPA
)
m
atr
ix
is
an
m
×
n
bin
a
ry
m
at
rix
UPA,
m
is
rep
resen
ti
ng
the
nu
m
ber
of
us
e
rs,
w
hile
n
can
be
def
i
ned
as
the
nu
m
ber
of
per
m
issi
on
s.
T
he
el
em
ent
UP
A
(i,
j)
=
1
in
di
cat
es
the assig
nm
ent of
per
m
issi
on
j
to
user i
[
14
]
.
Def
ini
ti
on
7. (
Access
Hist
or
y
L
og)
Access
histor
y
log
is
a
se
ries
of
quat
ern
i
on
(U
,
P,
R,
t
)
an
d
this
se
ries
in
dicat
es
an
acc
ess
eve
nt
i
n
the
sy
ste
m
and
re
presents
the
us
er
(
U)
i
nvocate
the
pe
rm
issi
on
(P
)
by
act
ivati
ng
the
r
ole
(R)
at
the
tim
e t
[16]
.
4.
CONCE
PT
U
AL
MO
DEL
A
co
nce
ptu
al
m
od
el
determ
i
nes
a
c
om
pr
eh
ensive
unde
rst
and
i
ng
a
nd
sc
op
e
s
of
a
pro
pose
d
so
l
ution
us
in
g
the
or
ga
nized
c
on
ce
pt
s
that
are
li
nked
toge
ther
[32]
and
for
thi
s
pap
e
r,
the
c
on
ce
ptu
al
m
od
el
is
const
ru
ct
e
d
ba
sed
on
li
te
rature
re
view
i
n
Se
ct
ion
2
a
nd
thi
s
m
od
el
repres
ents
a
ge
ner
al
process
in
ro
le
m
inin
g
m
od
el
as shown
in
Fi
gu
re
2.
Figure
2. Ge
ne
ral proce
ss in
role m
ining
m
od
el
The
c
om
po
ne
nt
s
of
the
m
odel
and
the
rela
ti
on
s
hip
s
betw
een
them
are
descr
i
bed
as
f
ollows.
T
he
process
sta
rts
by
insertin
g
in
pu
t
data
an
d
as
li
ste
d
in
Table
1,
m
os
t
of
the
ro
le
m
ining
al
gorithm
s
a
re
util
iz
ing
var
ia
nts
of
us
e
r
-
per
m
issi
on
a
ssign
m
ent
(UP
A)
m
at
rix
as
a
n
in
put
data
a
nd
s
om
e
of
the
r
esearche
rs
al
s
o
us
in
g
acce
ss
histor
y
log
.
T
he
n
the
data
is
transf
e
r
red
to
pr
e
-
pr
oc
essing
sta
ge
and
the
ge
ner
al
act
ivit
y
in
this
sta
ge
Inpu
t
Da
ta
UPA
m
atrix
o
r
h
isto
ry
lo
g
Stag
e 1
Pre
-
Proces
sin
g
Stag
e 2
Can
d
id
ate Ro
les
Gen
eration
Stag
e 3
Ro
le Selection
&
Ass
ig
n
m
en
t
Ph
ase
O
utput
Da
ta
Ro
les o
r
UA and
PA
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci
IS
S
N:
25
02
-
4752
A Co
ncep
t
ua
l
Mo
del o
f R
ole
Base
d
Acces
s
Con
tr
ol Usi
ng
Role Mini
ng Al
go
rit
hm
(
Nazir
ah A
bd Ha
mid
)
1399
include
s
the
da
ta
cl
ean
ing
a
nd
data
norm
al
iz
at
ion
sp
eci
fical
ly
the
acts
of
rem
ov
ing
no
ise
a
nd
ha
nd
li
ng
m
issi
ng
d
at
a
a
nd the
pur
pose
of this sta
ge
is
to cr
eat
e a
set
of co
rr
ect
data
for
the
n
e
xt
process.
Nex
t,
t
he
cl
ean
data
is
ad
va
nced
t
o
the
ne
xt
sta
ge
nam
ely
cand
idate
r
ol
es
ge
ne
rati
on
ph
a
se.
T
his
sta
ge
is
the
m
os
t
m
eaning
f
ul
process
beca
us
e
it
involve
s
the
disc
ov
e
ry
of
a
ppr
opriat
e
cand
i
date
r
ol
es
by
exp
l
oiti
ng
a
ny
su
it
able
data
m
ining
te
ch
niques
or
he
ur
ist
i
cs
al
go
rit
hm
s
or
known
as
r
ole
m
ining
al
gorithm
s.
This
sta
ge
usu
al
ly
pr
o
duce
a
big
po
ol
of
c
and
i
date
r
oles,
there
fore
i
n
t
he
nex
t
ste
p,
r
ole
sel
ect
ion
a
nd
r
ole
assignm
ent
phase,
m
or
e
sp
e
ci
fic
and
sm
aller
ro
le
s
are
pro
duced
acc
ordin
g
to
the
de
sired
obj
ect
ive
s
and
ou
t
pu
ts
by
util
iz
ing
any
ap
pro
pr
ia
te
data
or
r
ole
m
ining
al
gorith
m
s.
Last
ly,
as
m
entioned
b
ef
or
e
a
nd
b
as
ed
on
inf
or
m
at
ion
in
Table
1,
t
he
ou
tpu
ts
c
ould
be
var
ie
d
bu
t
m
ost
of
the
resea
rc
her
s
prefe
r
nu
m
ber
of
r
oles
as
the
m
ai
n
ou
t
pu
t
.
5.
CONCL
US
I
O
N AND F
UT
U
RE W
ORKS
We
ha
ve
propose
d
a
c
oncept
ual
m
od
el
that
is
co
ns
tr
ucted
ba
se
d
on
t
he
li
te
ratur
e
r
evie
w
a
nd
thi
s
m
od
el
rep
re
se
nts
a
ge
ner
al
process
i
n
ro
le
m
ining
m
od
el
.
T
his
m
od
el
involve
s
se
ries
of
phases
t
hat
beg
i
n
with
the
in
put
of
data,
pre
-
proce
ssin
g
st
age,
ca
ndidate
ro
le
ge
ner
at
i
on
phase,
r
ol
e
sel
ect
ion
a
nd
r
ole
assignm
ent
process
an
d
la
stl
y
nu
m
ber
of
ro
le
s
as
ge
nerat
ed
ou
t
pu
t.
F
or
the
f
uture
works,
we
int
end
t
o
i
m
pr
ove the
concept
ual m
od
el
w
it
h
a
co
m
pr
e
hensi
ve
a
nd c
om
ple
te
m
od
el
.
ACKN
OWLE
DGME
NT
This
resea
rch
i
s
fu
ll
y
spon
s
or
ed
by
Re
searc
h
Ma
na
gem
ent,
Inn
ov
at
io
n
&
Com
m
ercial
izati
on
Ce
ntre
(RMIC),
U
nive
rsity
Su
lt
an
Z
ai
nal
A
bid
i
n
(
Un
iS
ZA
)
a
nd
Ce
ntre
for
Re
s
earch
an
d
I
nnov
at
io
n
Ma
na
gem
ent
(CRIM),
U
nive
rsiti
Tekn
i
kal
Ma
la
ysi
a
Melaka
(
UteM
)
with
the
Proj
e
ct
Cod
e
of
T
RGS/1/
2016/F
TMK
-
CACT/
01/D
00006.
REFERE
NCE
S
[1]
“
Role
Based
Acc
ess
Control
,
”
Nati
on
al
Institute
of
Standa
rd
s
and
Te
chnol
o
g
y
",
2016.
[On
li
ne]
.
Avail
ab
le:
htt
ps://
csrc
.
n
ist.gov/project
s/rol
e
-
base
d
-
acce
ss
-
co
ntrol
#rba
c
-
stand
ard
.
[2]
E.
J.
Co
y
ne
,
“
Role
Engi
n
ee
r
ing,”
in
Proceedi
ng
s
of
the
fi
rs
t
AC
M
Workshop
on
Rol
e
-
based
ac
c
ess
cont
rol
,
199
6,
no.
4
,
pp
.
15
–
16
.
[3]
W
.
Ye,
R
.
L
i,
a
nd
H.
L
i,
“
Rol
e
Mining
Us
ing
Boole
an
M
at
rix
Dec
om
positi
on
with
Hie
rar
ch
y,
”
in
12th
IEEE
Inte
rnational
Co
nfe
renc
e
on
Tr
ust,
S
ec
urit
y
and
P
rivac
y
in
Co
mputing
and
Comm
unic
ati
ons
,
2013
,
pp.
805
–
812
.
[4]
W
.
Ye,
R.
L
i,
X
.
Gu,
Y.
Li,
and
K.
W
en,
“
Role
Mining
using
Ans
wer
Set
Program
m
ing,
”
Fut
ur.
Gene
r.
Comput.
Syst.
,
vol
.
55
,
pp
.
336
–
343
,
Feb
.
2016.
[5]
H.
Huang,
F.
Shang,
J.
L
iu,
and
H.
Du,
“
Handli
ng
Le
ast
Privi
le
g
e
Problem
and
Role
Mining
in
RBAC
,
”
J.
Com
b
.
Optim.
,
vol. 30,
no.
1
,
pp
.
63
–
86
,
Jul.
2015
.
[6]
Y.
R.
More
and
S.
V
Gu
m
aste
,
“
Perfor
m
anc
e
Eva
lu
at
ion
of
A
Role
Based
Acc
ess
Control
Constrai
nts
in
Role
Mining
Us
ing
C
ard
inality
,
”
Int
.
J.
Adv. Re
s.
Sci. Manag.
Tec
h
nol.
,
vol
.
2
,
no
.
7
,
pp
.
1
–
7
,
2016
.
[7]
R.
Li
,
H.
L
i,
X.
Gu,
Y.
Li
,
W
.
Ye,
and
X.
Ma,
“
Role
Mining
bas
ed
on
Cardi
nalit
y
Constrai
n
ts,”
Concurr.
Comput.
Pract
.
E
xp.
,
vol
.
27,
pp
.
3126
–
31
44,
2015
.
[8]
J.
Hu,
K.
M.
Kh
an,
Y.
Zha
ng
,
Y
.
Bai,
and
R.
Li
,
“
Role
Updati
ng
in
Inform
at
ion
S
y
stems
using
Model
Chec
k
ing
,
”
Knowl.
In
f. Sy
st
.
,
vol
.
51
,
no
.
1
,
p
p.
187
–
234
,
Apr.
2017.
[9]
M.
Frank,
J.
M.
Buhm
ann,
and
D.
Basin,
“
On
the
Defi
nition
of
Role
Mining,”
in
Proce
ed
ing
o
f
the
15th
ACM
sympos
ium
on
A
cc
ess c
on
trol
mo
del
s and te
chnologi
es
-
S
ACMA
T
’10
,
2010,
p.
35
.
[10]
C.
Jin,
A.
Shen
,
and
W
.
Yu
,
“
T
he
RBAC
S
y
st
e
m
Based
on
Rol
e
Risk
and
Us
er
Trust,”
Int
.
J.
Comput.
Comm
un.
Eng.
,
vol
.
5
,
no
.
5,
pp
.
374
–
380
,
2016.
[11]
L.
Fuchs
and
S.
Meie
r,
“
The
Ro
le
Mining
Proc
e
ss
Model
-
Underl
ini
ng
the
Ne
e
d
for
a
Com
pre
hensive
Rese
arch
Perspec
ti
v
e,”
in
2011
Sixth
Int
er
nati
onal
Con
fe
re
nce
on
A
vai
lab
ility,
R
eliability a
nd
Sec
uri
ty
,
201
1,
pp
.
35
–
42
.
[12]
S.
Vavil
is,
A.
I
.
Egne
r,
M.
Petk
ovic
,
and
N.
Zannone,
“
Role
Mining
with
Miss
ing
Value
s,”
in
1
1th
Inte
rnation
a
l
Confe
re
nc
e
on
A
vai
labilit
y, Rel
ia
bil
ity
and
Se
curi
ty
(
AR
ES)
,
20
16,
pp.
167
–
176.
[13]
J.
Lu
and
Q.
Zh
u,
“
An
Eff
ec
ti
v
e
Algorit
hm
Based
on
Densit
y
C
luste
ring
Fram
e
work,”
IEEE
Acce
ss
,
vol.
5
,
pp.
4991
–
5000,
201
7.
[14]
N.
Pan,
Z
.
Zhu,
L.
He,
and
L
.
Sun,
“
An
Eff
ic
i
en
c
y
Appro
ac
h
for
RBAC
Rec
onfi
gura
ti
on
with
M
ini
m
al
Role
s
an
d
Perturba
t
ion,
”
C
oncurr.
Comput.
Pract. Ex
p
.
,
no.
Dec
ember, pp. 1
–
15,
De
c. 2017.
[15]
J.
Vaid
y
a
,
V.
At
luri
,
and
J.
W
ar
ner
,
“
Role
Min
er
:
Mining
Rol
es
using
Subs
et
En
um
era
ti
on,
”
in
P
roce
edi
ngs
of
th
e
13th
ACM
confe
ren
ce
on
Compu
te
r and comm
uni
cat
ions se
curit
y
-
CCS ’06
,
2006
,
no.
Janua
r
y
,
p
.
144.
[16]
N.
Pan,
L.
Sun,
L.
He,
and
Z
.
Zhu,
“
An
Ap
proa
ch
for
Hier
arc
hi
ca
l
RBAC
Rec
onfiguratio
n
with
Minim
al
Perturba
t
ion,
”
I
EE
E
Acce
ss
,
vol
.
2169
–
3536
,
no
.
c, pp. 1
–
11,
201
7.
[17]
X.
Ma,
J.
W
ang
,
L
.
Zh
ao,
and
R.
Li,
“
Mutual
Exc
lusion
Ro
le
Constrai
nt
Min
i
ng
base
d
on
W
e
ight
in
Role
-
B
ase
d
Acc
ess Cont
rol
S
y
stem,”
Int
.
J. I
nnov.
Comput
.
I
nf.
Con
trol
,
vol.
12,
no
.
1
,
pp
.
91
–
101,
2016
.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2502
-
4752
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci,
Vo
l.
12
, N
o.
3
,
Dece
m
ber
2
01
8
:
1
3
9
4
–
1
400
1400
[18]
S
.
D.
Stoll
er
and
T.
Bui,
“
Mining
Hier
arc
h
ic
a
l
T
e
m
pora
l
Role
s
with
Multi
pl
e
Metr
ic
s,”
in
IFI
P
An
nual
Confe
renc
e
on
Data
and
Ap
pli
cations Se
curi
ty
and
Pri
vac
y
,
2016,
pp
.
79
–
95
.
[19]
N.
Pan,
Z.
Zhu
,
L.
He
,
L.
Sun,
a
nd
H.
Su,
“
Mining
Approxim
at
e
Role
s
und
er
Im
port
ant
As
signme
nt,”
in
2nd
I
EEE
Inte
rnational
Co
nfe
renc
e
on
Co
mputer
and
Com
municat
ions
,
20
16,
pp
.
1319
–
13
24.
[20]
B.
Mitr
a,
S
.
Sur
al
,
J.
Va
id
y
a
,
a
nd
V.
Atlur
i,
“
Mining
Te
m
por
al
Rol
es
using
Man
y
-
Va
lue
d
C
once
pts,
”
Comp
ut.
Sec
ur.
,
vol
.
60
,
p
p.
79
–
94
,
Jul.
20
16.
[21]
H.
L
u,
Y.
Hong,
Y.
Yang,
L.
Duan,
and
N.
Badar,
“
Towa
rds
Us
e
r
-
orie
nt
ed
RBAC
Model,
”
J
.
Comput.
Sec
ur.
,
vol
.
23,
no
.
1
,
pp
.
10
7
–
129,
Mar
.
201
5.
[22]
B.
Mitra,
S.
Sural,
V.
Atlu
ri,
an
d
J.
Vaid
y
a,
“
Th
e
Gene
ra
li
z
ed
T
emporal
Role
M
ini
ng
Problem,”
J.
Comput.
S
ec
u
r.
,
vol.
23
,
no
.
1
,
pp
.
31
–
58
,
2015
.
[23]
X.
Ma,
R.
Li,
H.
W
ang,
and
H.
L
i,
“
Role
Mining
base
d
on
Perm
is
sion
Cardi
nalit
y
Constrai
nt
and
Us
er
Cardi
nal
i
t
y
Constrai
nt
,
”
S
ecur
.
Comm
un.
Ne
tworks
,
vol
.
8
,
n
o.
13
,
pp
.
2317
–
2328,
2015
.
[24]
E.
Uzun,
D.
Lo
ren
zi,
V.
Atluri
,
J.
Vaid
y
a
,
and
S.
Sural,
“
Migrat
ing
from
DA
C
to
RBAC
,
”
in
Lect
ure
Not
es
in
Computer
Scien
ce
,
vol
.
9149
,
P.
Sam
ara
ti
,
Ed
.
C
ham:
Springer
In
te
rna
ti
ona
l
Publi
shing,
2015,
pp.
69
–
84.
[25]
X.
Du
and
X.
C
hang,
“
Perform
a
nce
of
AI
Algor
it
hm
s
for
Minin
g
Mea
ningfu
l
R
ole
s,”
in
201
4
I
EE
E
Congress
on
Ev
olutionar
y
Co
mputati
on
(
CEC
)
,
2014,
pp
.
207
0
–
2076.
[26]
H.
Lu,
J.
Vaid
y
a
,
and
V.
Atluri
,
“
An
Optimiza
ti
o
n
Fram
ework
fo
r
Role
Mining,
”
J.
Comput.
Sec
u
r.
,
vol.
22,
no.
1,
pp.
1
–
31
,
Jan
.
2
014.
[27]
A.
A.
Eu
cha
rist
a
and
K.
Har
iba
s
kar
,
“
Visual
El
i
ci
t
at
ion
of
Role
s
:
using
A
H
y
br
i
d
Approac
h,
”
O
rient
.
J.
Comput
.
Sci
.
Techno
l.
,
vo
l.
6
,
no
.
1
,
pp
.
10
3
–
110,
2013
.
[28]
B.
Mitra,
S.
Sur
al
,
V
.
Atluri,
an
d
J.
Vaid
y
a
,
“
T
oward
Mining
o
f
Te
m
pora
l
Rol
e
s,”
in
Lec
ture
N
ote
s
in
Compute
r
Sci
en
ce
,
vol
.
79
64
LNCS,
IFIP
I
nte
r
na
ti
ona
l
Fed
era
t
ion
for
Infor
m
at
ion
Proce
ss
i
ng,
2013
,
pp
.
65
–
80.
[29]
Z.
Xu
and
S.
D.
Stoll
er
,
“
Mining P
ara
m
et
erize
d
Role
-
base
d
Polici
es,
” in
Data
and
appli
cation
se
cu
rity
and
pri
vacy
,
2013,
p
.
255
.
[30]
W
.
Zha
ng,
Y.
Chen,
C.
Gunte
r,
D.
Li
ebovitz
,
and
B.
Mali
n,
“
Evo
lvi
ng
Role
Defi
nit
ions
Th
rough
Perm
issio
n
Invoc
ation
Patt
e
rns,”
in
Proceed
ings
of
the
18th
ACM
sympos
ium
on
Ac
ce
ss
cont
rol
models
and
te
chnol
og
ie
s
-
SACMA
T ’13
,
2
013,
no
.
June
,
p
.
37.
[31]
B.
Mitr
a,
S.
Sura
l,
J.
Vaid
y
a
,
and
V.
Atluri,
“
A
Surve
y
of
Rol
e
Mi
ning,
”
ACM
Co
mput.
Surv.
,
vo
l.
48,
no
.
4,
pp.
1
–
37,
2016
.
[32]
Y.
Jaba
re
en,
“
Buil
ding
a
Con
ceptua
l
Fr
amework:
Philosoph
y
,
Defi
nit
ions
,
and
Proce
dure
,
”
Int
.
J.
Qual.
Me
tho
ds
,
vol.
8
,
no
.
4
,
pp
.
49
–
62,
Mar
.
200
9.
Evaluation Warning : The document was created with Spire.PDF for Python.