I
nte
rna
t
io
na
l J
o
urna
l o
f
E
lect
rica
l a
nd
Co
m
p
ute
r
E
ng
in
ee
ring
(
I
J
E
CE
)
Vo
l.
7
,
No
.
6
,
Dec
em
b
er
201
7
,
p
p
.
3
6
9
2
~
3
6
9
9
I
SS
N:
2088
-
8708
,
DOI
: 1
0
.
1
1
5
9
1
/
i
j
ec
e
.
v
7
i
6
.
pp
3
6
9
2
-
3699
3692
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ia
e
s
jo
u
r
n
a
l.c
o
m/o
n
lin
e/in
d
ex
.
p
h
p
/I
JE
C
E
Clustering
i
n Agg
rega
ted
User
P
ro
f
iles
a
cros
s
M
ultip
le Socia
l
Netw
o
rk
s
Cha
ru
Vir
m
a
ni
1
,
A
nu
ra
dh
a
P
illa
i
2
,
Di
m
ple J
un
ej
a
3
1
Re
se
a
rc
h
S
c
h
o
lar,
Ym
c
a
Un
iv
e
r
s
it
y
o
f
S
c
ien
c
e
a
n
d
T
e
c
h
n
o
l
o
g
y
,
F
a
rid
a
b
a
d
2
De
p
a
rtme
n
t
o
f
Co
m
p
u
ter S
c
ien
c
e
,
Ym
c
a
Un
iv
e
rsit
y
o
f
S
c
ien
c
e
a
n
d
T
e
c
h
n
o
lo
g
y
,
F
a
rid
a
b
a
d
3
De
p
a
rtme
n
t
o
f
Co
m
p
u
ter A
p
p
li
c
a
ti
o
n
,
Na
ti
o
n
a
l
In
stit
u
te o
f
T
e
c
h
n
o
lo
g
y
,
Ku
ru
k
sh
e
tra
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
A
p
r
1
8
,
2
0
1
7
R
ev
i
s
ed
Ma
y
30
,
2
0
1
7
A
cc
ep
ted
J
u
n
1
5
,
2
0
1
7
A
so
c
ial
n
e
tw
o
rk
is
in
d
e
e
d
a
n
a
b
stra
c
ti
o
n
o
f
re
late
d
g
ro
u
p
s
in
tera
c
ti
n
g
a
m
o
n
g
st
th
e
m
s
e
lv
e
s
to
d
e
v
e
lo
p
re
latio
n
s
h
ip
s.
H
o
w
e
v
e
r,
to
a
n
a
l
y
z
e
a
n
y
re
latio
n
sh
i
p
s
a
n
d
p
sy
c
h
o
lo
g
y
b
e
h
in
d
i
t,
c
l
u
ste
rin
g
p
lay
s
a
v
it
a
l
ro
le.
Clu
ste
rin
g
e
n
h
a
n
c
e
s
th
e
p
re
d
ic
tab
il
it
y
a
n
d
d
isc
o
v
e
r
y
o
f
li
k
e
m
in
d
e
d
n
e
ss
a
m
o
n
g
st
u
se
rs.
T
h
is
a
rti
c
le’s
g
o
a
l
e
x
p
lo
it
s
th
e
tec
h
n
iq
u
e
o
f
En
se
m
b
le
K
-
m
e
a
n
s
c
lu
ste
rs
to
e
x
trac
t
th
e
e
n
ti
ti
e
s
a
n
d
t
h
e
ir
c
o
rre
sp
o
n
d
in
g
i
n
ter
e
stsa
s
p
e
r
th
e
sk
il
ls
a
n
d
lo
c
a
ti
o
n
b
y
a
g
g
re
g
a
ti
n
g
u
se
r
p
ro
f
il
e
s
a
c
ro
ss
th
e
m
u
lt
ip
le
o
n
l
in
e
so
c
ial
n
e
tw
o
rk
s.
T
h
e
p
ro
p
o
se
d
e
n
se
m
b
le
c
lu
ste
rin
g
u
ti
li
z
e
s
k
n
o
wn
K
-
m
e
a
n
s
a
lg
o
rit
h
m
to
im
p
ro
v
e
re
su
lt
s
f
o
r
t
h
e
a
g
g
re
g
a
ted
u
se
r
p
ro
f
il
e
s
a
c
ro
ss
m
u
lt
ip
le
so
c
ial
n
e
tw
o
rk
s.
T
h
e
a
p
p
ro
a
c
h
p
r
o
d
u
c
e
s
a
n
e
n
se
m
b
le
si
m
il
a
rit
y
m
e
a
su
re
a
n
d
p
ro
v
id
e
s 7
0
%
b
e
tt
e
r
re
su
lt
s th
a
n
t
a
k
in
g
a
f
ix
e
d
v
a
lu
e
o
f
K o
r
g
u
e
ss
i
n
g
a
v
a
lu
e
o
f
K
w
h
il
e
n
o
t
a
lt
e
rin
g
th
e
c
lu
ste
rin
g
m
e
th
o
d
.
T
h
is
p
a
p
e
r
sta
tes
th
a
t
g
o
o
d
e
n
se
m
b
les
c
lu
ste
rs
c
a
n
b
e
sp
a
w
n
e
d
to
e
n
v
isa
g
e
th
e
d
isc
o
v
e
ra
b
il
it
y
o
f
a
u
se
r
f
o
r
a
p
a
rti
c
u
lar i
n
ter
e
st.
K
ey
w
o
r
d
:
C
lu
s
ter
i
n
g
,
E
n
s
e
m
b
le
c
lu
s
ter
K
-
Me
a
n
s
So
cial
n
et
w
o
r
k
Co
p
y
rig
h
t
©
2
0
1
7
I
n
stit
u
te o
f
Ad
v
a
n
c
e
d
E
n
g
i
n
e
e
rin
g
a
n
d
S
c
ien
c
e
.
Al
l
rig
h
ts
re
se
rv
e
d
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
C
h
ar
u
Vir
m
a
n
i
,
Ass
o
ciate
P
r
o
f
ess
o
r
Dep
ar
m
e
n
t o
f
co
m
p
u
ter
Scie
n
ce
an
d
E
n
g
i
n
ee
r
i
n
g
Facu
lt
y
o
f
E
n
g
i
n
ee
r
i
n
g
a
n
d
T
e
ch
n
o
lo
g
y
Ma
n
av
R
ac
h
n
a
I
n
ter
n
atio
n
al
Un
i
v
er
s
it
y
E
m
ail:
c
h
ar
u
.
f
et
@
m
r
i
u
.
ed
u
.
i
n
1.
I
NT
RO
D
UCT
I
O
N
As
th
e
n
u
m
b
er
o
f
s
o
cial
n
et
wo
r
k
u
s
er
s
i
n
cr
ea
s
e
s
,
a
tr
e
m
en
d
o
u
s
a
m
o
u
n
t
o
f
d
ata
is
g
e
n
er
ated
b
y
t
h
e
s
h
ar
i
n
g
o
f
i
n
f
o
r
m
atio
n
.
T
h
e
in
tu
iti
v
e
n
a
tu
r
e
o
f
t
h
ese
s
o
cial
n
et
w
o
r
k
s
i
s
t
h
e
cr
ea
tio
n
o
f
r
elate
d
g
r
o
u
p
s
(
o
r
clu
s
ter
s
)
[
1
]
.
T
h
is
h
as
b
ec
o
m
e
an
ar
ea
o
f
in
ter
est
in
th
e
d
is
co
v
er
y
o
f
co
m
m
u
n
i
ties
in
r
ec
en
t
t
i
m
e
s
.
T
h
ese
p
atter
n
s
ar
e
u
s
ed
to
m
in
e
a
v
a
r
iet
y
o
f
in
f
o
r
m
at
io
n
,
w
h
ic
h
i
s
th
en
u
s
ed
in
v
ar
io
u
s
f
ie
ld
s
[
2
].
C
lu
s
ter
an
al
y
s
is
,
o
r
clu
s
ter
in
g
i
n
a
s
o
cial
n
et
w
o
r
k
co
n
tex
t,
i
s
t
h
e
g
r
o
u
p
i
n
g
o
f
a
s
et
o
f
d
ata
o
b
j
ec
ts
(
f
o
r
ex
a
m
p
le,
f
r
ie
n
d
s
,
co
n
n
ec
tio
n
s
,
co
m
m
u
n
ities
,
o
r
p
er
s
o
n
al
in
f
o
r
m
atio
n
)
in
s
u
ch
a
w
a
y
t
h
at
o
b
j
ec
ts
in
t
h
e
s
a
m
e
g
r
o
u
p
(
o
r
clu
s
ter
s
)
ar
e
m
o
r
e
s
i
m
ilar
to
e
ac
h
o
th
er
th
an
to
th
o
s
e
in
o
t
h
er
g
r
o
u
p
s
(
o
r
clu
s
ter
s
)
.
T
h
e
id
en
tif
ica
tio
n
o
f
th
e
s
e
p
atter
n
s
in
to
clu
s
ter
s
h
as n
u
m
er
o
u
s
ap
p
licatio
n
s
i
n
th
e
f
ield
o
f
d
ata
s
cien
ce
.
Ma
n
y
al
g
o
r
ith
m
s
ca
n
b
e
u
s
e
d
to
clu
s
ter
d
ata
[
3
]
.
Po
p
u
la
a
r
clu
s
ter
s
in
cl
u
d
e
g
r
o
u
p
s
with
s
m
all
d
is
tan
ce
s
b
et
w
ee
n
cl
u
s
ter
m
e
m
b
er
s
,
d
en
s
e
ar
ea
s
o
f
th
e
d
ata
s
p
ac
e,
i
n
ter
v
al
s
,
o
r
p
ar
ticu
lar
s
tatis
t
ical
d
is
tr
ib
u
tio
n
[
4
]
.
T
h
er
ef
o
r
e,
c
lu
s
ter
i
n
g
ca
n
b
e
f
o
r
m
u
lated
as
a
m
u
lti
-
o
b
j
ec
tiv
e
o
p
tim
i
za
tio
n
p
r
o
b
lem
.
A
s
u
itab
le
cl
u
s
ter
i
n
g
alg
o
r
it
h
m
an
d
p
ar
am
e
ter
s
etti
n
g
s
v
ar
y
f
r
o
m
t
h
e
in
d
i
v
id
u
al
i
n
p
u
t
an
d
ex
p
ec
ted
r
esu
lts
.
Nu
m
er
o
u
s
atte
m
p
ts
w
er
e
m
ad
e
to
im
p
r
o
v
e
t
h
e
q
u
alit
y
o
f
cl
u
s
ter
s
u
s
i
n
g
en
s
e
m
b
li
n
g
tech
n
iq
u
es
[
5
]
[6
]
[
7
]
[
8
]
[
9
]
[
10
]
[
11
]
[
1
2
]
.
T
h
e
m
a
in
co
n
ce
r
n
o
f
m
a
n
y
o
f
t
h
e
s
e
alg
o
r
ith
m
s
is
to
el
u
cid
ate
la
b
el
co
r
r
es
p
o
n
d
en
ce
p
r
o
b
lem
.
T
h
e
li
m
itatio
n
o
f
m
a
n
y
o
f
t
h
ese
al
g
o
r
ith
m
s
is
t
h
e
a
s
s
u
m
p
tio
n
o
f
t
h
e
s
a
m
e
n
u
m
b
e
r
o
f
clu
s
ter
in
ea
c
h
p
ar
titi
o
n
an
d
m
a
y
p
er
f
o
r
m
p
o
o
r
l
y
w
h
e
n
t
h
e
in
f
o
r
m
at
io
n
ab
o
u
t o
u
tp
u
t c
l
u
s
ter
is
n
o
t k
n
o
w
n
i
n
ad
v
an
ce
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
E
C
E
I
SS
N:
2088
-
8708
C
lu
s
ter
in
g
in
A
g
g
r
eg
a
ted
User
P
r
o
files
a
cro
s
s
Mu
ltip
le
S
o
cia
l Netw
o
r
k
s
(
C
h
a
r
u
V
ir
ma
n
i
)
3693
L
iter
at
u
r
e
p
er
tain
in
g
to
clu
s
t
er
in
g
o
n
a
g
g
r
e
g
ated
p
u
b
licl
y
-
av
a
ilab
le
u
s
er
p
r
o
f
ile
d
ata
o
f
v
ar
io
u
s
s
o
cial
n
et
w
o
r
k
s
w
a
s
th
o
r
o
u
g
h
l
y
d
w
elled
an
d
it
w
a
s
d
is
co
v
er
ed
th
at
k
-
m
ea
n
s
al
g
o
r
ith
m
a
n
d
en
s
e
m
b
le
clu
s
ter
i
n
g
ar
e
th
e
m
o
s
t
p
o
p
u
la
r
alg
o
r
ith
m
s
to
clu
s
ter
th
e
d
ata
to
o
b
tain
r
esu
lts
T
h
e
s
tu
d
y
t
h
u
s
ai
m
s
to
ap
p
l
y
k
-
m
ea
n
s
cl
u
s
ter
in
g
o
n
ag
g
r
eg
ate
d
s
o
cial
n
et
w
o
r
k
d
ata
an
d
e
n
s
e
m
b
le
cl
u
s
ter
s
th
u
s
f
o
r
m
ed
b
y
g
r
o
u
p
in
g
d
i
f
f
er
en
t
p
ar
am
eter
s
an
d
i
n
ter
p
r
et
r
esu
l
ts
.
I
n
t
h
is
w
o
r
k
,
th
e
p
ar
titi
o
n
s
ar
e
g
en
er
ated
w
i
th
v
ar
y
i
n
g
n
u
m
b
er
o
f
cl
u
s
ter
s
an
d
th
u
s
i
m
p
r
o
v
in
g
th
e
q
u
ali
t
y
a
n
d
s
tab
ilit
y
o
f
t
h
e
co
n
s
e
n
s
u
s
p
ar
titi
o
n
.
Go
o
d
en
s
e
m
b
le
c
lu
s
ter
ar
e
p
r
o
v
id
ed
b
y
eli
m
i
n
ati
n
g
th
e
d
ep
en
d
en
c
y
o
f
th
e
i
n
p
u
t
p
ar
a
m
eter
lik
e
k
,
th
e
n
u
m
b
er
o
f
c
lu
s
ter
s
.
H
u
n
g
ar
ia
n
alg
o
r
ith
m
[
1
3
]
an
d
cu
m
u
lati
v
e
v
o
ti
n
g
s
ch
e
m
e
[
14
]
ar
e
u
s
ed
to
o
b
tain
f
in
a
l
cl
u
s
ter
s
.
T
h
e
p
ap
er
o
f
f
er
s
t
w
o
-
f
o
ld
co
n
tr
ib
u
tio
n
i.e
.
id
en
ti
f
y
i
n
g
t
h
e
s
k
ill
o
f
a
u
s
er
f
o
r
p
ar
ticu
lar
lo
ca
tio
n
ac
r
o
s
s
m
u
ltip
le
s
o
cial
n
et
w
o
r
k
s
a
n
d
eli
m
i
n
ati
n
g
th
e
d
ep
en
d
en
c
y
o
f
in
p
u
t p
ar
a
m
eter
li
k
e
K.
T
h
e
cu
r
r
en
t
w
o
r
k
u
n
iq
u
el
y
co
n
tr
ib
u
tes to
th
e
li
m
itat
io
n
o
f
th
e
r
eq
u
ir
e
m
e
n
t
o
f
eq
u
a
l
n
u
m
b
er
o
f
cl
u
s
ter
i
n
i
n
p
u
t
p
ar
titi
o
n
a
n
d
th
e
k
n
o
w
led
g
e
o
f
t
h
e
n
u
m
b
er
o
f
clu
s
ter
s
to
b
e
k
n
o
w
n
i
n
ad
v
a
n
ce
.
T
h
is
p
ap
er
is
s
tr
u
ctu
r
ed
in
to
f
i
v
e
s
ec
tio
n
s
:
Sect
io
n
2
th
r
o
w
s
l
ig
h
t
o
n
th
e
w
o
r
k
o
f
e
m
i
n
e
n
t
r
esear
ch
er
s
h
ig
h
li
g
h
ti
n
g
t
h
eir
s
u
b
s
tan
t
ial
co
n
tr
ib
u
tio
n
s
.
T
h
e
d
is
cu
s
s
io
n
in
s
ec
tio
n
2
in
d
icate
s
th
e
li
m
i
tatio
n
s
o
f
k
m
ea
n
s
alg
o
r
ith
m
.
T
h
e
cu
r
r
en
t
w
o
r
k
t
h
u
s
f
in
d
s
m
o
ti
v
atio
n
an
d
r
eso
l
v
es t
h
e
c
h
alle
n
g
e
l
is
ted
ab
o
v
e.
Sectio
n
3
u
n
iq
u
el
y
co
n
tr
ib
u
tes
a
n
e
n
s
e
m
b
le
cl
u
s
ter
to
id
en
tify
g
r
o
u
p
s
o
f
clu
s
ter
s
o
n
a
m
ea
s
u
r
e
o
f
s
i
m
ilar
i
t
y
.
T
h
is
h
as
b
ee
n
estab
lis
h
ed
w
i
th
a
d
ata
s
et
i
n
t
h
e
ev
al
u
atio
n
s
ec
tio
n
g
i
v
en
i
n
s
ec
tio
n
4
.
Sectio
n
5
f
i
n
all
y
co
n
clu
d
e
s
.
2.
RE
L
AT
E
D
WO
RK
T
r
a
d
itio
n
all
y
,
s
o
cial
n
e
t
w
o
r
k
clu
s
ter
i
n
g
is
eit
h
er
h
ier
ar
ch
ica
l
o
r
p
ar
titi
o
n
in
g
w
h
er
e
v
er
tice
s
j
o
in
in
to
g
r
o
u
p
s
o
f
s
i
m
ilar
it
y
[
1
5
]
.
C
o
m
m
u
n
it
y
d
etec
tio
n
in
s
o
cial
n
et
w
o
r
k
s
h
a
s
b
ee
n
an
in
ter
es
t
f
o
r
w
h
ic
h
a
s
u
cc
es
s
f
u
l
alg
o
r
ith
m
i
s
d
ep
icted
in
[
16
]
[1
7
]
[
1
8]
[1
9
]
.
A
s
o
n
e
o
f
th
e
s
i
m
p
le
s
t
u
n
s
u
p
er
v
is
ed
cl
u
s
te
r
in
g
tech
n
iq
u
e
s
,
k
-
m
ea
n
s
d
i
s
co
v
er
s
th
e
d
eg
r
ee
o
f
s
i
m
ilar
it
y
a
m
o
n
g
k
g
r
o
u
p
s
a
s
s
u
m
in
g
k
ce
n
tr
o
id
s
.
K
-
ce
n
te
r
s
ar
e
d
ef
i
n
ed
a
n
d
p
lace
d
s
p
atiall
y
as
f
ar
as
p
o
s
s
ib
le.
E
ac
h
s
p
atial
p
o
in
t
is
m
ar
k
ed
to
a
g
i
v
en
d
ata
s
et
an
d
ass
o
ciate
d
to
th
e
n
ea
r
est
ce
n
ter
.
Ne
w
ce
n
tr
o
id
s
ar
e
ca
lcu
lated
as
b
ar
y
ce
n
ter
o
f
th
e
clu
s
te
r
s
a
n
d
r
eb
o
u
n
d
ed
b
et
w
ee
n
s
a
m
e
d
ata
s
et
p
o
in
ts
to
t
h
e
n
ea
r
es
t
n
e
w
ce
n
ter
.
T
h
u
s
,
k
ce
n
ter
s
c
h
a
n
g
e
its
lo
ca
tio
n
ai
m
i
n
g
at
m
i
n
i
m
izi
n
g
a
n
o
b
j
ec
tiv
e
f
u
n
ctio
n
k
n
o
w
n
a
s
s
q
u
ar
ed
er
r
o
r
f
u
n
ct
io
n
[]
b
y
:
J
(
V)
=
∑
∑
(
|
|
x
i
−
v
j
|
|
)
2
c
i
j
=
1
c
i
=
1
(
1
)
w
h
er
e
|
|
x
i
-
v
j
|
|
is
t
h
e
E
u
cl
id
ea
n
d
is
ta
n
ce
b
et
w
ee
n
x
i a
n
d
v
j
.
ci
is
th
e
n
u
m
b
er
o
f
d
ata
p
o
in
ts
in
i
th
cl
u
s
ter
.
c
is
th
e
n
u
m
b
er
o
f
cl
u
s
ter
ce
n
ter
s
.
T
h
e
em
er
g
in
g
f
ield
o
f
s
o
cial
an
al
y
s
is
u
s
e
s
d
ata
m
in
in
g
as
th
e
k
e
y
in
p
u
t
f
o
r
an
al
y
zin
g
d
ata.
C
lu
s
ter
i
n
g
is
a
n
i
m
p
o
r
tan
t
f
ac
t
o
r
in
th
is
an
al
y
s
is
.
I
t
is
ap
p
r
o
a
ch
ed
b
y
v
ar
io
u
s
cl
u
s
ter
i
n
g
alg
o
r
ith
m
s
,
in
cl
u
d
in
g
:
k
-
m
ea
n
s
,
f
u
zz
y
c
-
m
ea
n
,
a
n
d
ta
b
le
m
o
d
elin
g
[
20
]
[
2
1
]
.
W
h
ile
k
-
m
ea
n
s
i
s
v
er
y
f
as
t,
its
ce
n
te
r
v
al
u
e
d
ep
en
d
s
o
n
th
e
v
al
u
e
o
f
k
.
Di
f
f
er
en
t
v
alu
es
o
f
k
w
ill
r
es
u
lt
in
d
i
f
f
er
e
n
t
clu
s
ter
s
[
2
2
]
[
2
3
]
.
Yan
g
et
al
[
24
]
o
b
s
er
v
ed
th
at
th
e
K
-
m
ea
n
s
lear
n
i
n
g
alg
o
r
it
h
m
r
eq
u
ir
es
s
p
ec
i
f
icatio
n
o
f
th
e
n
u
m
b
er
o
f
clu
s
ter
ce
n
t
er
s
.
I
f
t
w
o
h
i
g
h
l
y
-
o
v
er
lap
p
in
g
d
ata
ex
i
s
t,
th
e
n
k
-
m
ea
n
s
w
ill
n
o
t
b
e
ab
le
to
r
eso
lv
e
th
e
p
r
ese
n
ce
o
f
t
w
o
cl
u
s
te
r
s
an
d
also
it
is
n
o
t
in
v
ar
ia
n
t to
n
o
n
-
lin
ea
r
tr
a
n
s
f
o
r
m
atio
n
s
.
Z
h
an
g
et
al
[4
]
p
r
o
p
o
s
ed
th
e
m
ap
p
in
g
o
f
n
et
w
o
r
k
n
o
d
es
to
id
en
ti
f
y
t
h
e
o
v
er
lap
p
in
g
co
m
m
u
n
it
y
b
y
E
u
clid
ea
n
s
p
ac
e
a
n
d
f
u
zz
y
c
-
m
ea
n
s
cl
u
s
ter
i
n
g
.
Ma
n
y
r
e
s
ea
r
ch
er
s
h
a
v
e
s
o
u
g
h
t
co
m
m
u
n
it
y
in
s
o
cia
l
n
et
w
o
r
k
s
,
as
w
e
ll
as
p
r
o
p
o
s
ed
m
etr
ics
f
o
r
ev
alu
ati
n
g
th
e
s
tr
u
ct
u
r
e
[
2
5
]
[
2
6
]
[
27
]
[
28
]
.
Yan
g
et
al
[
24
]
p
r
o
p
o
s
ed
f
in
d
in
g
p
eo
p
le
b
y
u
s
i
n
g
m
o
b
ile
p
h
o
n
e
u
s
a
g
e
p
atter
n
s
in
a
s
o
cial
n
et
w
o
r
k
.
An
o
t
h
er
r
esear
ch
er
p
r
o
p
o
s
ed
a
h
y
b
r
id
s
tu
d
y
to
r
eta
in
c
u
s
to
m
er
s
u
s
in
g
cl
u
s
ter
in
g
[
2
8
]
.
T
h
e
au
th
o
r
s
u
s
e
d
ag
g
r
e
g
ated
d
ata
o
n
u
s
er
p
r
o
f
iles
f
r
o
m
v
ar
io
u
s
s
o
cial
n
et
w
o
r
k
s
.
W
it
h
v
ar
ian
c
e
clu
s
ter
in
g
,
t
h
e
y
u
s
ed
k
-
m
ea
n
s
a
n
d
e
n
s
e
m
b
le
cl
u
s
ter
i
n
g
to
g
r
o
u
p
u
s
er
s
as
p
er
th
eir
p
u
b
lic
i
n
f
o
r
m
atio
n
.
T
h
e
s
tu
d
y
w
as
r
estricte
d
to
cl
u
s
t
er
th
e
u
s
er
o
f
a
lo
ca
tio
n
w
h
o
h
as
i
n
ter
est
in
a
s
p
ec
if
ic
s
k
ill.
B
u
s
in
e
s
s
e
s
an
d
m
ar
k
eti
n
g
s
tr
ateg
ie
s
ca
n
als
o
u
s
e
th
is
tec
h
n
iq
u
e
f
o
r
p
r
o
m
o
tio
n
al
b
en
e
f
it
s
b
y
ap
p
ly
i
n
g
it to
o
th
er
attr
ib
u
tes t
o
f
in
d
u
s
er
s
i
m
ilar
itie
s
.
Nu
m
er
o
u
s
tec
h
n
iq
u
e
s
f
o
r
g
e
n
er
atin
g
cl
u
s
ter
r
esu
l
ts
a
n
d
co
m
b
in
i
n
g
t
h
e
m
h
a
v
e
b
ee
n
s
ee
n
in
liter
at
u
r
e
[
5
]
[6
]
[
7
]
[
8
]
[
9
]
[
10
]
[
11
]
[
1
2
]
.
Gen
er
atio
n
o
f
in
p
u
t
p
ar
titi
o
n
f
o
llo
w
ed
b
y
i
n
te
g
r
atio
n
o
f
all
th
e
p
ar
titi
o
n
s
t
o
o
b
tain
f
i
n
al
p
ar
titi
o
n
is
a
t
w
o
w
a
y
p
r
o
ce
s
s
g
i
v
e
n
b
y
v
e
g
a
-
p
o
n
s
et
al.
[
2
9
].
Me
d
ian
p
ar
titi
o
n
an
d
o
b
j
ec
t
co
-
o
cc
u
r
r
en
ce
ar
e
th
w
t
w
o
w
a
y
s
to
g
en
er
ate
a
co
n
s
en
s
u
s
.
I
n
m
ed
ian
p
ar
titi
o
n
,
th
e
f
i
n
al
p
ar
titi
o
n
m
a
x
i
m
izes
t
h
e
s
i
m
ilar
it
y
w
i
th
all
t
h
e
g
e
n
er
ated
s
et
in
th
e
en
s
e
m
b
le.
T
h
is
ap
p
r
o
ac
h
is
n
o
t
co
n
s
id
er
ed
f
o
r
clu
s
ter
in
g
a
s
d
ef
in
i
n
g
th
e
M
ir
k
i
n
Dis
tan
c
e
[
3
0
]
h
av
e
b
ee
n
p
r
o
v
en
NP
-
h
ar
d
an
d
co
m
p
u
tatio
n
all
y
ex
p
en
s
iv
e.
Ob
j
ec
t
-
co
o
cc
u
r
r
en
ce
is
a
n
o
t
h
er
ap
p
r
o
ac
h
t
h
at
o
b
tai
n
s
th
e
f
in
al
p
ar
titi
o
n
f
r
o
m
th
e
g
e
n
er
atio
n
s
et
d
ep
en
d
in
g
u
p
o
n
t
h
e
f
r
eq
u
en
c
y
o
f
o
cc
u
r
r
en
ce
o
f
o
b
j
ec
t
to
g
eth
er
o
r
an
o
b
j
ec
t
to
o
n
e
cl
u
s
ter
f
o
llo
w
e
d
b
y
s
i
m
ilar
it
y
b
ased
cl
u
s
ter
in
g
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2088
-
8708
I
J
E
C
E
Vo
l.
7
,
No
.
6
,
Dec
em
b
er
201
7
:
3
6
9
2
–
3
6
9
9
3694
alg
o
r
ith
m
.
Co
-
as
s
o
ciatio
n
Ma
t
r
ix
f
o
llo
w
ed
b
y
cl
u
s
ter
i
n
g
m
ec
h
an
i
s
m
is
a
w
a
y
to
g
e
n
er
ate
th
e
o
cc
u
r
r
en
ce
o
f
a
n
o
b
j
ec
t.
R
elab
ellin
g
an
d
cu
m
m
u
lati
v
e
v
o
ti
n
g
is
a
n
o
th
er
c
h
o
ice
f
o
r
attai
n
in
g
t
h
e
f
in
al
p
ar
titi
o
n
f
r
o
m
t
h
e
g
en
er
atio
n
s
et
d
ep
en
d
i
n
g
u
p
o
n
t
h
e
f
r
eq
u
e
n
c
y
o
f
o
c
cu
r
r
en
ce
o
f
o
b
j
ec
ts
.
R
elab
ellin
g
s
o
lv
e
lab
el
co
r
r
esp
o
n
d
en
ce
p
r
o
b
l
em
u
s
in
g
H
u
n
g
ar
ia
n
A
l
g
o
r
it
h
m
[
1
3
]
f
o
llo
w
i
n
g
v
o
ti
n
g
p
r
o
ce
s
s
b
y
u
s
i
n
g
c
u
m
u
lat
iv
e
v
o
tin
g
[
1
4
]
to
o
b
tain
f
in
a
l
p
ar
titi
o
n
.
Ot
h
er
f
i
n
al
p
ar
titi
o
n
s
c
an
b
e
o
b
tai
n
ed
b
y
Gen
e
tic
al
g
o
r
ith
m
[
3
0
]
,
NM
F
[
3
1
]
an
d
k
er
n
el
Me
th
o
d
[
3
2
]
u
n
d
er
o
b
j
ec
t c
o
-
o
cc
u
r
r
en
ce
th
at
is
b
e
y
o
n
d
th
e
co
n
s
id
er
atio
n
o
f
th
is
p
ap
er
.
I
t
h
as
b
ee
n
o
b
s
er
v
ed
d
u
r
i
n
g
th
e
r
esear
ch
t
h
at
n
o
w
o
r
k
h
as
b
ee
n
d
e
v
o
ted
to
ap
p
l
y
i
n
g
e
n
s
e
m
b
le
clu
s
ter
i
n
g
m
et
h
o
d
s
i
n
a
n
a
l
y
zi
n
g
a
u
s
er
’
s
p
u
b
licl
y
a
v
ailab
le
in
f
o
r
m
at
io
n
.
Ho
w
e
v
er
,
d
if
f
er
en
t
s
tr
ateg
ies
h
a
v
e
b
ee
n
u
t
ilized
to
r
ec
o
g
n
ize
c
o
m
m
u
n
it
y
a
n
d
m
er
g
e
co
m
m
u
n
i
t
y
s
tr
u
ctu
r
e
s
[
3
3
]
.
A
s
d
ata
clu
s
ter
in
g
a
n
d
co
m
m
u
n
it
y
d
etec
tio
n
ar
e
v
e
r
y
co
m
p
ar
ati
v
e,
it
o
u
g
h
t
to
b
e
co
n
ce
i
v
ab
le
to
m
er
g
e
co
m
m
u
n
it
y
i
n
a
n
in
d
is
ti
n
g
u
is
h
ab
le
w
a
y
f
r
o
m
e
n
s
e
m
b
les
o
f
clu
s
ter
s
w
it
h
g
r
e
at
o
u
tco
m
es.
T
h
e
p
r
o
p
o
s
ed
al
g
o
r
ith
m
p
er
f
o
r
m
ed
clu
s
ter
i
n
g
o
n
ag
g
r
eg
a
ted
u
s
er
p
r
o
f
iles
f
r
o
m
v
ar
io
u
s
s
o
cial
n
et
w
o
r
k
s
b
y
c
h
an
g
i
n
g
t
h
e
v
al
u
e
o
f
k
f
o
r
d
i
f
f
er
e
n
t
p
ar
am
eter
s
.
T
h
en
,
p
ar
titi
o
n
s
wer
e
co
m
b
i
n
ed
to
o
v
er
co
m
e
cl
u
s
ter
in
s
tab
ilit
y
.
3.
P
RO
P
O
SE
D
WO
RK
A
p
eo
p
le
g
r
o
u
p
o
r
co
m
m
u
n
it
y
is
a
s
u
b
s
et
o
f
h
u
b
s
in
s
id
e
a
s
y
s
te
m
s
u
c
h
t
h
at
as
s
o
ciatio
n
s
b
et
w
ee
n
h
u
b
s
i
n
t
h
e
s
u
b
s
e
t
ar
e
d
en
s
er
th
an
a
s
s
o
ciatio
n
s
w
it
h
r
est
o
f
th
e
s
y
s
te
m
.
Dete
cti
n
g
a
co
m
m
u
n
it
y
is
a
f
o
r
m
o
f
clu
s
ter
i
n
g
o
f
th
e
i
n
f
o
r
m
atio
n
w
h
ic
h
is
s
i
m
ilar
a
m
o
n
g
n
ei
g
h
b
o
r
s
.
T
h
e
aim
o
f
th
i
s
s
ec
tio
n
is
to
p
r
o
p
o
s
e
m
et
h
o
d
f
o
r
co
m
b
i
n
i
n
g
s
e
v
er
al
clu
s
ter
s
an
d
g
en
er
alize
t
h
is
f
o
r
th
e
u
s
er
’
s
in
f
o
r
m
a
tio
n
.
T
h
e
p
r
o
p
o
s
ed
s
tr
ateg
y
cr
ea
tes
a
n
e
w
f
ea
tu
r
e
s
p
ac
e
u
t
ili
zi
n
g
t
h
e
y
ield
s
o
f
i
n
itial
k
m
ea
n
s
al
g
o
r
ith
m
.
T
h
e
p
h
ases
o
f
th
e
p
r
o
p
o
s
ed
m
et
h
o
d
o
lo
g
y
ar
e:
1.
Gen
er
ate
I
n
itia
l c
lu
s
ter
s
u
s
i
n
g
K
-
m
ea
n
s
f
o
r
v
ar
y
i
n
g
v
al
u
e
o
f
k
.
2.
Gen
er
ate
n
e
w
co
m
p
o
n
en
ts
b
y
Hu
n
g
ar
ia
n
alg
o
r
it
h
m
.
3.
E
n
s
e
m
b
le
f
i
n
al
cl
u
s
ter
s
o
n
th
e
n
e
w
g
e
n
er
ated
co
m
p
o
n
en
ts
.
Un
s
u
p
er
v
i
s
ed
tr
ain
i
n
g
i
s
u
s
ed
to
p
ar
titi
o
n
d
ata
o
n
th
e
b
asis
o
f
s
i
m
ilar
it
y
u
s
i
n
g
k
-
m
ea
n
s
.
M
o
r
e
s
i
m
ilar
u
s
er
s
ar
e
g
r
o
u
p
ed
in
to
a
cl
u
s
t
er
u
s
i
n
g
E
u
clid
ea
n
d
is
ta
n
ce
i
n
th
i
s
tech
n
iq
u
e
ac
r
o
s
s
all
th
e
p
r
o
f
iles
ag
g
r
eg
a
ted
b
y
th
e
n
et
w
o
r
k
.
T
h
is
r
es
u
lt
s
i
n
a
clu
s
ter
b
elo
n
g
in
g
to
a
p
ar
ticu
lar
lo
ca
tio
n
.
A
p
ar
ticu
lar
s
k
i
l
l
w
ill b
e
f
o
u
n
d
an
d
ap
p
lied
f
o
r
th
at
lo
ca
tio
n
.
Ho
w
ev
er
,
a
w
ei
g
h
ted
E
u
clid
ea
n
d
i
s
tan
ce
i
s
u
s
ed
to
clu
s
ter
t
h
e
d
ata
o
f
m
o
r
e
s
i
m
ilar
b
elo
n
g
i
n
g
to
lo
ca
tio
n
an
d
s
k
il
l.
A
w
e
ig
h
t
w
a
s
as
s
ig
n
ed
to
o
n
e
p
ar
a
m
et
er
an
d
g
r
o
u
p
;
th
e
u
s
er
w
a
s
ass
ig
n
ed
b
ased
o
n
a
d
i
f
f
er
en
t
p
ar
a
m
et
er
.
Fo
r
m
i
n
i
n
g
th
e
s
k
ill
f
r
o
m
t
h
e
u
s
er
-
g
en
er
ated
p
o
s
t,
t
h
e
p
o
s
t
e
x
tr
ac
ted
i
s
clea
n
ed
an
d
co
n
v
er
ted
in
to
a
k
e
y
p
air
.
T
h
e
p
air
in
clu
d
es
t
h
e
p
o
s
t
I
D
(
o
r
u
s
er
n
a
m
e)
an
d
th
e
p
o
s
t’
s
lis
t
o
f
w
o
r
d
s
s
e
r
v
i
n
g
as
t
h
e
s
k
il
ls
lis
t
th
at
t
h
e
u
s
er
ap
p
lies
i
n
t
h
e
p
o
s
t.
T
h
e
li
s
t
i
s
co
n
v
er
ted
i
n
to
a
n
u
m
er
ical
v
ec
to
r
;
w
ei
g
h
ts
ar
e
d
eter
m
i
n
ed
u
s
in
g
s
o
f
t T
F
-
I
DF.
K
-
m
ea
n
s
cl
u
s
ter
i
n
g
m
o
d
els
a
r
e
ap
p
lied
o
n
th
e
co
n
v
er
ted
l
is
t
w
h
er
e
k
=
3
to
1
2
f
o
r
s
k
i
ll
an
d
b
y
-
v
ar
ian
ce
cl
u
s
ter
s
f
o
r
s
k
i
ll
a
n
d
lo
ca
tio
n
to
g
en
er
ate
in
p
u
t
p
ar
t
ito
n
s
.
T
h
ese
tec
h
n
iq
u
e
s
ar
e
ap
p
lied
s
ep
ar
at
el
y
o
n
th
e
d
if
f
er
en
t
v
ar
iab
les,
t
h
u
s
r
esu
lt
in
g
p
ar
titi
o
n
s
in
t
o
d
if
f
er
e
n
t
n
u
m
b
er
o
f
cl
u
s
ter
s
.
T
h
e
r
esu
lt
s
o
f
cl
u
s
ter
s
ar
e
th
en
co
m
b
i
n
ed
u
s
i
n
g
H
u
n
g
ar
i
an
al
g
o
r
ith
m
a
n
d
c
u
m
u
lat
iv
e
v
o
tin
g
f
o
r
ea
ch
cl
u
s
ter
.
H
u
n
g
ar
ian
al
g
o
r
ith
m
i
s
a
m
u
lti
-
o
b
j
ec
tiv
e
clu
s
ter
in
g
co
m
p
r
i
s
i
n
g
o
f
m
u
ltip
le
cl
u
s
ter
i
n
g
p
ar
titi
o
n
s
w
it
h
o
b
j
ec
tiv
e
f
u
n
ctio
n
s
.
I
t
e
n
s
e
m
b
les
m
u
ltip
le
p
ar
titi
o
n
s
b
y
co
m
b
i
n
i
n
g
i
n
d
iv
id
u
a
l c
l
u
s
ter
i
n
g
p
ar
titi
o
n
an
d
g
i
v
i
n
g
a
f
i
n
al
p
ar
titi
o
n
.
Fi
n
al
p
ar
titi
o
n
s
o
f
clu
s
ter
s
ca
n
b
e
f
o
u
n
d
b
y
ap
p
ly
in
g
th
e
v
o
tin
g
s
c
h
e
m
e
[
1
6
]
.
C
o
n
f
u
s
io
n
m
atr
ix
i
s
u
s
e
d
to
co
m
p
o
u
te
th
e
s
i
m
ilar
it
y
b
et
w
ee
n
clu
s
ter
s
.
T
o
co
m
p
u
te
t
h
e
co
n
f
u
s
io
n
m
atr
i
x
o
f
t
w
o
d
i
f
f
er
e
n
t
n
u
m
b
er
o
f
clu
s
ter
,
th
e
r
e
m
ain
in
g
cl
u
s
ter
o
f
t
h
e
s
m
all
er
n
u
m
b
er
o
f
clu
s
ter
w
i
ll
b
e
k
ep
t
as
e
m
p
t
y
.
C
o
n
f
u
s
io
n
m
atr
ix
f
o
r
t
w
o
cl
u
s
ter
s
(
A
,
B
)
is
o
f
s
ize
Ax
B
.
T
h
e
(
i,
j
)
th
in
d
e
x
o
f
t
h
e
m
atr
i
x
co
r
r
esp
o
n
d
s
to
th
e
o
b
j
ec
t th
at
ar
e
in
clu
s
ter
i o
f
A
a
n
d
i
n
clu
s
ter
j
o
f
B
.
Ma
x
i
m
u
m
ele
m
en
t
i
s
s
e
lecte
d
u
s
i
n
g
Hu
n
g
a
r
ian
A
l
g
o
r
ith
m
.
I
n
te
g
r
atio
n
o
f
E
le
m
e
n
t
is
d
o
n
e
b
y
ag
g
r
e
g
ati
n
g
t
h
e
ali
g
n
ed
p
ar
ti
tio
n
s
b
y
s
elec
ti
n
g
t
h
e
e
le
m
e
n
t
t
h
at
ta
k
es
th
e
m
aj
o
r
ity
cl
u
s
ter
lab
el
f
o
r
ea
ch
o
b
s
er
v
ed
p
ar
titi
o
n
.
Ma
j
o
r
ity
V
o
tin
g
an
d
p
l
u
r
alit
y
v
o
tin
g
ar
e
th
e
m
at
h
o
d
s
to
g
e
n
er
ate
t
h
e
f
i
n
al
cl
u
s
ter
s
t
h
a
t
in
v
o
l
v
es
s
elec
ti
n
g
a
n
o
b
j
ec
t
w
h
o
s
e
co
u
n
t
i
s
g
r
ea
ter
th
a
n
a
t
h
r
esh
o
ld
v
al
u
e
w
h
er
ea
s
p
l
u
r
alit
y
v
o
ti
n
g
co
n
s
id
er
s
th
e
m
aj
o
r
it
y
cl
u
s
ter
lab
el
f
o
r
e
ac
h
o
b
s
er
v
ed
v
al
u
e.
T
h
e
p
r
o
p
o
s
ed
alg
o
r
ith
m
is
s
h
o
u
m
i
n
A
l
g
o
r
ith
m
1
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
E
C
E
I
SS
N:
2088
-
8708
C
lu
s
ter
in
g
in
A
g
g
r
eg
a
ted
User
P
r
o
files
a
cro
s
s
Mu
ltip
le
S
o
cia
l Netw
o
r
k
s
(
C
h
a
r
u
V
ir
ma
n
i
)
3695
A
l
g
o
r
it
h
m
1
1.
P
ass
th
e
en
tire
d
ataset
an
d
id
en
ti
f
y
t
h
e
p
o
in
t
w
i
th
t
h
e
w
ei
g
h
t
ass
i
g
n
ed
to
it.
2.
C
o
m
p
ar
e
th
e
o
b
j
ec
ts
an
d
co
n
s
i
d
er
it a
s
p
er
k
(
k
=
3
to
1
2
)
.
3.
C
h
ec
k
t
h
e
s
i
m
ilar
it
y
an
d
ca
lc
u
late
th
e
m
ea
n
v
al
u
e
f
r
o
m
ea
c
h
ce
n
tr
o
id
to
th
e
clu
s
ter
f
o
r
th
e
o
b
j
ec
t.
4.
E
ac
h
o
b
j
ec
t m
a
y
r
es
id
e
in
th
e
clu
s
ter
it
w
i
n
s
th
e
s
i
m
ilar
it
y
.
5.
R
ep
ea
t step
s
2
to
4
if
th
er
e
is
n
o
ch
an
g
e.
6.
R
ep
ea
t step
f
o
r
an
o
th
er
v
al
u
e
o
f
k
u
n
ti
l K
=1
2
7.
C
o
m
p
u
te
co
n
f
u
s
io
n
m
atr
ix
b
a
s
ed
o
n
m
u
ltip
le
d
ata
p
ar
titi
o
n
s
f
r
o
m
s
tep
5
.
8.
Fin
d
i
ts
m
a
x
i
m
u
m
ele
m
e
n
t,
a
s
s
o
ciate
t
h
e
t
w
o
cl
u
s
ter
as
p
er
th
e
m
a
x
i
m
u
m
o
b
j
ec
t.
T
h
u
s
,
r
ed
u
ce
t
h
e
m
atr
i
x
u
p
o
n
r
e
m
o
v
al
o
f
t
h
ese
clu
s
ter
s
.
E
r
r
o
r
r
ate
,
J
ac
ar
d
I
n
d
ex
an
d
R
A
ND
i
n
d
ex
ar
e
co
n
s
id
er
ed
to
m
ea
s
u
r
e
t
h
e
q
u
alit
y
o
f
cl
u
s
ter
s
.
E
r
r
o
r
r
ate
d
ep
icts
t
h
e
a
v
er
ag
e
n
u
m
b
er
o
f
m
i
s
clas
s
if
ied
ele
m
en
ts
.
P
ar
titi
o
n
s
ar
e
m
o
r
e
s
i
m
i
lar
i
f
t
h
e
er
r
o
r
r
ate
is
le
s
s
.
E
r
r
o
r
r
ate
is
u
s
ed
to
v
alid
ate
th
e
ac
cu
r
ac
y
o
f
t
h
e
f
in
al
p
ar
tit
io
n
.
R
A
ND
[
34
]
p
r
o
p
o
s
es
a
m
ea
s
u
r
e
to
v
alid
at
e
th
e
q
u
alit
y
o
f
t
h
e
cl
u
s
ter
as:
r
(
A
,
B
)
=
x
+
y
x
+
y
+
z
+
w
(
2
)
W
h
er
e:
U:
s
et
o
f
n
clu
s
ter
s
A
: p
ar
titi
o
n
in
U
h
av
in
g
r
s
u
b
s
ets
B
: p
ar
titi
o
n
in
U
h
a
v
i
n
g
q
s
u
b
s
ets
x
: n
u
m
b
er
o
f
p
air
o
f
ele
m
e
n
ts
f
r
o
m
U
w
h
ic
h
o
cc
u
r
in
A
a
n
d
B
y
: n
u
m
b
er
o
f
p
a
ir
s
o
f
ele
m
e
n
t
s
f
r
o
m
U
w
h
ich
ar
e
d
if
f
er
en
t in
A
a
n
d
B
z:
n
u
m
b
er
o
f
p
air
o
f
ele
m
e
n
t
s
f
r
o
m
U
w
h
ic
h
o
cc
u
r
in
A
b
u
t
n
o
t in
B
w
:
n
u
m
b
er
o
f
p
air
o
f
ele
m
e
n
t
s
f
r
o
m
U
w
h
ich
o
cc
u
r
in
B
b
u
t
n
o
t in
A
T
h
e
J
ac
ar
d
in
d
ex
[
3
5
]
to
m
ea
s
u
r
e
th
e
s
i
m
i
lar
it
y
i
s
co
m
p
u
ted
as:
J
(
A
,
B
)
=
y
y
+
z
+
w
(
3
)
4.
RE
SU
L
T
AND
DI
SCUS
SI
O
N
Var
io
u
s
s
o
cial
n
et
w
o
r
k
s
ar
e
cr
a
w
led
to
cr
ea
te
r
a
w
d
ata
o
n
u
s
er
p
r
o
f
ile
in
f
o
r
m
atio
n
,
i
n
cl
u
d
i
n
g
:
n
a
m
e,
d
escr
ip
tio
n
,
lo
ca
tio
n
,
in
ter
es
ts
an
d
t
w
ee
ts
/
n
e
w
s
f
ee
d
.
T
h
e
co
llected
d
ata
w
as
a
g
g
r
e
g
ated
o
n
th
e
v
ec
to
r
<U
s
er
I
D,
n
a
m
e>
.
T
h
is
s
et
o
f
r
a
w
d
ata
is
cr
ea
ted
in
Mo
n
g
o
DB
.
T
o
c
r
ea
te
en
r
ich
ed
d
ata,
th
e
d
ata
w
a
s
clea
n
ed
f
o
r
n
o
is
e
r
e
m
o
v
al
an
d
s
to
r
ed
in
th
e
j
s
o
n
d
o
cu
m
e
n
ts
.
T
h
e
p
r
o
p
o
s
ed
clu
s
ter
in
g
alg
o
r
it
h
m
w
as
a
p
p
lied
to
th
e
d
ata
to
cr
ea
te
d
esire
d
clu
s
ter
s
.
Fi
g
u
r
e
1
s
h
o
w
s
t
h
e
ar
ch
itec
t
u
r
e
f
o
r
v
i
s
u
aliz
in
g
u
s
er
in
f
o
r
m
at
io
n
.
Fig
u
r
e
1
.
A
r
ch
itectu
r
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2088
-
8708
I
J
E
C
E
Vo
l.
7
,
No
.
6
,
Dec
em
b
er
201
7
:
3
6
9
2
–
3
6
9
9
3696
T
w
it
ter
p
u
b
lic
s
ea
r
c
h
an
d
B
i
n
g
s
ea
r
ch
A
P
I
ac
ts
a
s
t
h
e
s
o
u
r
ce
o
f
d
ata
co
llectio
n
.
W
h
il
e
T
w
it
ter
s
ea
r
ch
o
u
tp
u
t
s
r
elev
a
n
t
u
s
er
-
g
en
er
ated
p
o
s
ts
w
h
e
n
s
ea
r
ch
ed
w
ith
a
n
in
p
u
t
q
u
er
y
.
T
h
e
B
in
g
s
ea
r
ch
A
P
I
allo
w
ed
cr
ea
ti
n
g
t
h
e
m
i
x
ed
in
p
u
ts
o
f
u
s
er
-
v
ar
iab
les.
Fo
r
ex
a
m
p
le,
u
s
er
-
n
a
m
e
+
u
s
er
-
lo
ca
ti
o
n
+
u
s
er
-
g
en
d
er
+
u
s
er
-
d
escr
ip
tio
n
-
k
e
y
w
o
r
d
s
.
T
h
is
u
s
er
i
n
f
o
r
m
a
tio
n
i
s
u
s
ed
t
o
ex
tr
ac
t
in
f
o
r
m
atio
n
f
r
o
m
o
t
h
er
s
o
cial
n
et
w
o
r
k
s
lik
e
F
u
ll
C
o
n
n
ec
t,
Go
o
g
le,
an
d
B
in
g
b
y
cr
a
w
li
n
g
o
r
u
s
in
g
a
p
i’
s
o
f
t
h
e
r
esp
ec
ti
v
e
n
et
w
o
r
k
s
.
T
o
tal
2
7
,
9
5
6
u
s
er
p
r
o
f
iles
ex
tr
ac
ted
;
co
m
p
lete
d
ata
co
n
s
is
ted
o
f
4
5
,
8
9
9
u
s
er
-
g
e
n
er
ated
p
o
s
ts
.
T
h
e
d
ata
is
clea
n
ed
i.e
.
w
h
ite
s
p
ac
es,
s
to
p
w
o
r
d
s
,
an
d
co
m
m
o
n
ter
m
s
(
i.e
.
,
a,
a
n
,
a
n
d
t
h
e)
ar
e
r
e
m
o
v
ed
a
n
d
co
n
v
er
ted
i
n
to
lo
w
er
ca
s
e.
User
p
r
o
f
iles
w
er
e
ag
g
r
eg
a
ted
b
y
m
atc
h
in
g
u
s
er
I
D
an
d
n
a
m
e
(
p
u
b
lic
attr
ib
u
tes
u
s
i
n
g
J
ar
o
-
w
r
i
n
k
ler
)
.
Ou
t
o
f
2
7
,
9
5
6
u
s
er
p
r
o
f
iles
,
1
8
,
8
9
7
u
s
er
p
r
o
f
iles
ar
e
ag
g
r
eg
ated
.
T
h
e
co
m
p
lete
d
ata
s
tatis
tics
is
s
h
o
w
n
in
tab
le
1
an
d
th
e
p
s
eu
d
o
co
d
e
to
ag
g
r
eg
ate
t
h
e
p
r
o
f
ile
is
d
ep
icted
in
alg
o
r
i
th
m
2
.
T
ab
le
1
.
Statis
tics
o
f
i
n
p
u
t d
at
a
#
I
n
p
u
t
Q
u
e
r
i
e
s
12
#
R
a
w
D
o
c
u
me
n
t
s
2
7
,
9
5
6
#
u
n
i
q
u
e
u
se
r
s
–
T
w
i
t
t
e
r
1
5
,
5
3
0
#
u
se
r
s
-
En
r
i
c
h
e
d
P
r
o
f
i
l
e
1
8
,
8
9
7
#
se
a
r
c
h
e
n
g
i
n
e
t
o
t
a
l
l
i
n
k
s
5
6
,
8
9
6
#
se
a
r
c
h
e
n
g
i
n
e
u
se
r
l
i
n
k
s
2
1
,
6
7
4
A
l
g
o
r
ith
m
2
:
P
r
o
f
ile
A
g
g
r
eg
at
io
n
1
.
I
n
itia
liz
e
Do
c1
<
-
S
o
u
r
ce
1
R
a
w
Do
cu
men
t
2
.
I
n
itia
liz
e
Do
c2
<
-
S
o
u
r
ce
2
R
a
w
Do
cu
men
t
3
.
I
n
itia
liz
e
Do
cN<
-
S
o
u
r
ce
N
R
a
w
Do
cu
men
t
4
.
I
n
itia
liz
e
P
a
ir
s
<
-
ca
r
tesi
a
n
_
p
a
ir
s
o
f a
ll d
o
c
u
men
ts
a
.
P
a
ir
s
<
-
N
*
N
d
o
cu
men
ts
5
.
I
tera
te
in
ev
ery
P
a
ir
a
.
R
el_
va
r
1
<
-
o
n
e
o
f th
e
r
elev
a
n
t v
a
r
ia
b
le
ex
–
n
a
me
b
.
th
r
esh
o
ld
_
s
co
r
e<
-
Ja
r
o
_
w
r
in
kler(
r
e
l_
va
r
1
,
p
a
ir
)
c.
I
f sco
r
e
>th
r
esh
o
ld
_
s
co
r
e:
merg
e_
en
r
ich
(
r
el_
va
r
1
,
p
a
ir
)
d
.
else :
p
a
s
s
&
ig
n
o
r
e
6
.
Up
d
a
te
f
o
r
ev
ery
p
a
ir
a
.
P
ick
o
r
r
ep
la
ce
r
el_
va
r
va
l
u
es a
cc
n
to
p
r
io
r
ity.
4
.
1
.
Sk
ill Wi
s
e
Clu
s
t
er
s
o
f
K
ey
w
o
rds
by
Users
T
h
e
s
y
s
te
m
h
a
s
ch
o
s
e
n
v
alu
e
o
f
k
v
ar
y
i
n
g
f
r
o
m
3
to
1
2
to
g
en
er
ate
th
e
p
ar
titi
o
n
s
,
f
ir
s
t
ex
p
er
im
e
n
t
i
s
ca
r
r
ied
b
y
p
ass
i
n
g
v
a
lu
e
o
f
k
as
3
r
esu
lti
n
g
i
n
th
r
ee
cl
u
s
te
r
s
f
o
r
ea
ch
o
f
th
e
1
2
q
u
er
ies:
No
d
e,
NL
P
,
J
av
a,
m
ac
h
in
e
lear
n
i
n
g
,
d
atab
ase,
P
y
th
o
n
,
j
av
ascr
ip
t,
b
ig
d
ata
,
d
ee
p
lear
n
in
g
,
SQ
L
,
Had
o
o
p
an
d
Data
s
cie
n
ce
.
T
h
ese
m
o
d
el
s
id
en
ti
f
y
r
ep
ea
tin
g
p
att
er
n
s
in
d
ata
an
d
o
r
g
an
ize
th
e
m
in
to
b
u
c
k
ets
k
n
o
w
n
o
r
“
d
at
a
clu
s
t
er
s
”
an
d
ar
e
d
ep
icted
in
tab
le
2
.
Si
m
ilar
r
e
s
u
lt
s
ar
e
o
b
tai
n
ed
f
r
o
m
k
-
m
ea
n
cl
u
s
ter
in
g
v
ar
y
i
n
g
k
f
r
o
m
4
to
1
2
.
Hen
ce
,
th
e
s
i
m
ilar
r
esu
lt
s
ar
e
o
m
itted
.
T
ab
le
2
.
K
-
m
ea
n
s
clu
s
ter
s
f
o
r
k
=
3
f
o
r
th
e
th
r
ee
s
k
ill
s
d
a
t
a
b
a
se
2
2
1
0
T
o
p
t
e
rm
s p
e
r
c
l
u
st
e
r:
d
a
t
a
b
a
se
C
l
u
st
e
r
0
:
j
o
b
a
d
m
i
n
i
s
t
r
a
t
o
r
sq
l
h
i
r
e
d
a
t
a
b
a
se
se
rve
r
d
e
r
b
y
o
ra
c
l
e
d
b
a
d
i
sk
C
l
u
st
e
r
1
:
h
t
t
p
t
u
n
g
st
e
n
d
a
c
u
se
f
u
l
e
j
n
e
t
w
o
rk
o
n
l
i
n
e
d
e
l
e
t
e
8
i
l
o
a
d
s
e
r
v
e
r
C
l
u
st
e
r
2
:
d
a
t
a
b
a
se
s
q
l
l
o
o
k
d
b
m
n
o
sq
l
9
i
sa
n
c
t
i
o
n
o
p
m
e
x
p
e
c
t
d
b
2
j
a
v
a
s
c
ri
p
t
2
2
4
4
6
T
o
p
t
e
rm
s p
e
r
c
l
u
st
e
r:
j
a
v
a
s
c
ri
p
t
C
l
u
st
e
r
0
:
j
a
v
a
s
c
ri
p
t
i
n
sp
i
r
a
t
e
e
b
o
o
k
n
j
a
v
a
scr
i
p
t
k
o
m
o
p
e
n
s
o
u
rce
d
i
sp
o
n
i
b
l
e
e
s
t
a
C
l
u
st
e
r
1
:
j
a
v
a
s
c
ri
p
t
d
e
v
e
l
o
p
e
r
d
e
v
o
p
s
j
o
b
l
i
b
r
a
ry
j
q
u
e
r
y
k
n
o
w
u
se
l
i
n
u
x
d
e
si
g
n
C
l
u
st
e
r
2
:
n
c
e
rt
i
f
i
c
a
t
i
o
n
d
m
o
z
w
e
b
m
a
st
e
r
l
e
a
zys
u
n
n
y
p
h
p
j
a
v
a
sc
ri
p
t
j
a
v
a
s
c
r
i
p
t
d
f
ra
n
f
o
rm
v
a
l
i
d
a
t
i
o
n
d
a
t
a
sci
e
n
c
e
3
6
3
6
T
o
p
t
e
rm
s p
e
r
c
l
u
st
e
r:
d
a
t
a
s
c
i
e
n
c
e
C
l
u
st
e
r
0
:
d
a
t
a
sc
i
e
n
c
e
d
a
t
a
b
i
g
d
a
t
a
m
a
c
h
i
n
e
l
e
a
rn
i
n
g
a
n
a
l
y
t
i
c
s
i
o
t
p
y
t
h
o
n
b
u
si
n
e
ss
s
t
a
t
i
s
t
i
c
l
e
a
r
n
C
l
u
st
e
r
1
:
b
i
g
d
a
t
a
c
a
n
c
e
r
b
e
a
t
u
se
a
r
t
i
f
i
c
i
a
l
i
n
t
e
l
l
i
g
e
n
c
e
d
e
e
p
l
e
a
rn
i
n
g
d
a
t
a
sc
i
e
n
c
e
i
o
t
c
h
a
t
b
o
t
f
i
n
t
e
c
h
C
l
u
st
e
r
2
:
r
o
n
a
l
d
v
a
n
l
o
o
n
l
e
a
rn
m
a
c
h
i
n
e
t
e
a
m
m
i
x
e
x
p
e
r
t
r
i
g
h
t
e
n
g
i
n
e
e
r
k
n
o
w
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
E
C
E
I
SS
N:
2088
-
8708
C
lu
s
ter
in
g
in
A
g
g
r
eg
a
ted
User
P
r
o
files
a
cro
s
s
Mu
ltip
le
S
o
cia
l Netw
o
r
k
s
(
C
h
a
r
u
V
ir
ma
n
i
)
3697
Fo
r
in
p
u
t
q
u
er
ies,
u
s
er
’
s
i
n
f
o
r
m
at
io
n
i
s
co
llected
a
n
d
d
if
f
er
en
tia
ted
o
n
t
h
e
b
as
is
o
f
i
n
ter
est
a
n
d
lo
ca
tio
n
.
Data
w
a
s
co
llected
f
o
r
th
r
ee
d
if
f
er
en
t
lo
ca
tio
n
s
U
n
ited
Kin
g
d
o
m
,
U
n
ited
States
a
n
d
L
o
n
d
o
n
.
I
t
w
a
s
an
al
y
ze
d
o
n
th
e
b
asis
o
f
j
av
a,
n
lp
,
P
y
th
o
n
,
j
av
as
cr
ip
t,
etc.
Di
f
f
er
en
t
p
ar
a
m
e
ter
s
ar
e
an
al
y
ze
d
to
th
e
m
o
d
el
v
ia
k
-
m
ea
n
s
cl
u
s
ter
i
n
g
o
n
th
e
d
ata
s
et
(
d
o
cu
m
e
n
ts
r
ela
ted
to
u
s
er
-
s
k
ills
a
n
d
u
s
er
-
le
v
el
v
ar
iab
l
es
s
u
c
h
as
lo
ca
tio
n
,
d
escr
ip
tio
n
s
,
etc.
)
.
I
n
o
r
d
er
to
id
en
ti
f
y
t
h
at
th
e
u
s
er
o
f
a
p
ar
ticu
lar
lo
ca
tio
n
h
a
s
a
p
ar
tic
u
lar
s
k
il
l,
a
n
ap
p
r
o
ac
h
m
u
s
t
b
e
f
o
u
n
d
to
id
en
tify
th
e
s
k
ill
s
et
o
f
t
h
e
u
s
er
o
f
t
h
e
p
ar
tic
u
lar
lo
ca
tio
n
.
T
h
e
p
ar
ticu
lar
lo
ca
tio
n
cl
u
s
ter
ca
n
b
e
cr
ea
ted
th
r
o
u
g
h
th
e
k
-
m
ea
n
s
alg
o
r
ith
m
b
ec
au
s
e
o
f
it
s
q
u
ic
k
co
n
v
er
g
e
n
ce
to
s
i
m
ilar
it
y
.
T
h
e
s
k
ill cl
u
s
ter
s
h
o
u
ld
d
ef
in
e
t
h
e
b
o
u
n
d
ar
ies o
f
th
e
s
k
ill s
e
t; th
is
e
n
d
s
in
a
co
m
p
le
x
task
.
T
o
o
b
tain
th
e
s
k
il
l set o
f
th
e
u
s
er
,
o
n
e
n
ee
d
s
to
k
n
o
w
t
h
e
in
ter
e
s
t
f
r
o
m
t
h
e
in
ter
est
attr
ib
u
te
(
if
a
v
ailab
l
e
f
r
o
m
th
e
s
o
cial
n
et
w
o
r
k
)
,
as
w
ell
as
t
h
e
u
s
er
-
g
en
er
ated
p
o
s
t to
m
i
n
e
in
f
o
r
m
atio
n
f
o
r
t
h
e
p
ar
ticu
lar
s
k
i
ll.
I
n
t
h
is
s
tu
d
y
,
clu
s
ter
s
w
er
e
o
b
t
ain
ed
f
o
r
k
=
3
to
1
2
o
n
s
k
ill
w
i
s
e
u
s
er
p
u
b
lic
d
ata
co
llected
f
r
o
m
v
ar
io
u
s
s
o
cial
n
et
w
o
r
k
s
.
K
p
ar
titi
o
n
s
ar
e
g
en
er
ated
o
p
ti
m
all
y
r
ep
r
esen
tin
g
M
p
ar
titi
o
n
s
b
y
v
o
tin
g
s
ch
e
m
e
to
g
e
n
er
at
e
a
s
k
i
lled
p
u
b
lic
g
r
o
u
p
f
o
r
th
at
p
ar
ti
cu
lar
lo
ca
tio
n
.
T
h
e
d
etailed
alg
o
r
ith
m
is
s
h
o
w
n
in
alg
o
r
ith
m
1
.
I
n
p
u
t
p
ar
titi
o
n
s
to
t
h
e
co
n
f
u
s
i
o
n
m
a
tr
ix
ar
e
t
h
e
cl
u
s
ter
s
o
b
tain
ed
f
r
o
m
t
h
e
p
r
ev
io
u
s
l
y
d
is
cu
s
s
ed
k
-
m
ea
n
s
(
i.e
.
,
k
=
3
to
1
2
)
.
I
n
t
h
is
p
h
ase,
t
h
e
cl
u
s
ter
i
n
g
r
es
u
l
ts
ar
e
co
m
b
i
n
ed
an
d
b
est
cl
u
s
ter
s
b
y
i
s
c
h
o
s
en
b
y
co
m
p
u
ti
n
g
s
i
m
ilar
it
y
m
ea
s
u
r
e
u
s
i
n
g
co
n
f
u
s
io
n
m
a
tr
ix
a
n
d
v
o
tin
g
s
ch
e
m
e.
Fi
g
u
r
e
2
r
ep
r
ese
n
ts
n
u
m
b
er
o
f
u
s
er
s
p
ec
ialized
in
s
k
i
ll
f
o
r
d
if
f
er
e
n
t
lo
ca
tio
n
an
al
y
ze
d
f
r
o
m
t
h
e
p
ar
titi
o
n
s
.
T
a
b
le
3
s
h
o
w
s
th
e
to
p
f
iv
e
ter
m
s
o
f
ea
ch
clu
s
ter
b
y
co
m
b
i
n
in
g
t
h
e
r
esu
lt
s
f
o
r
a
p
ar
ticu
lar
lo
ca
tio
n
L
o
n
d
o
n
.
Fig
u
r
e
2
.
C
o
u
n
t o
f
u
s
er
f
o
r
d
if
f
er
en
t
s
k
i
lls
f
o
r
d
if
f
er
e
n
tlo
ca
ti
o
n
T
ab
le
3
.
T
o
p
f
iv
e
clu
s
ter
s
D
a
t
a
S
c
i
e
n
c
e
J
a
v
a
sc
r
i
p
t
D
a
t
a
b
a
se
mac
h
i
n
e
l
e
a
r
n
i
n
g
Jq
u
e
r
y
D
a
t
a
b
a
se
d
a
t
a
sc
i
e
n
c
e
F
o
r
mv
a
l
i
d
a
t
i
o
n
N
o
sq
l
B
i
g
d
a
t
a
N
o
d
e
j
s
S
q
l
d
e
e
p
l
e
a
r
n
i
n
g
L
i
b
r
a
r
y
M
o
n
g
o
d
b
a
n
a
l
y
t
i
c
s
R
e
a
c
t
j
s
P
y
m
o
n
g
o
I
t
h
as
b
ee
n
o
b
s
er
v
ed
t
h
at
th
e
r
esu
lt
s
p
r
o
d
u
ce
d
b
y
e
n
s
e
m
b
le
clu
s
ter
i
n
g
i
s
7
0
%
b
etter
t
h
an
t
h
at
ca
n
b
e
p
r
o
d
u
ce
d
b
y
g
u
e
s
s
i
n
g
v
a
lu
e
o
f
k
o
r
ta
k
in
g
a
f
i
x
ed
v
alu
e
o
f
k
.
T
h
e
co
m
p
le
x
it
y
o
f
k
-
m
ea
n
s
i
s
O(
KNI
d
)
w
h
er
e
k
b
elo
n
g
s
to
n
u
m
b
er
o
f
c
lu
s
te
r
s
,
N
b
elo
n
g
s
to
n
u
m
b
er
o
f
s
a
m
p
les,
I
b
elo
n
g
s
to
iter
ati
o
n
s
o
f
k
m
ea
n
s
to
co
n
v
er
g
e
an
d
d
b
elo
n
g
to
n
u
m
b
er
o
f
co
m
p
o
n
en
ts
.
T
h
e
co
m
p
lex
it
y
o
f
p
r
o
p
o
s
ed
en
s
e
m
b
le
c
lu
s
ter
is
O(
k
3
)
.
T
h
e
co
m
p
ar
is
o
n
b
et
w
ee
n
K
-
m
ea
n
s
an
d
en
s
e
m
b
le
K
-
m
ea
n
s
clu
s
te
r
in
g
w
a
s
ev
al
u
ated
u
s
i
n
g
er
r
o
r
r
ate,
J
ac
a
r
d
in
d
ex
an
d
R
A
ND
s
co
r
e
w
h
ich
i
s
an
ex
ten
t to
ev
al
u
ate
clu
s
ter
q
u
ali
t
y
as s
h
o
w
n
in
tab
le
4
.
T
ab
le
4
.
s
im
i
lar
it
y
b
et
w
ee
n
k
-
m
ea
n
s
an
d
E
n
s
e
m
b
le
K
-
m
ea
n
s
D
a
t
a
se
t
M
e
t
h
o
d
Er
r
o
r
r
a
t
e
Jac
a
r
d
I
n
d
e
x
R
A
N
D
sco
r
e
A
g
g
r
e
g
a
t
e
d
u
se
r
’
s p
u
b
l
i
c
i
n
f
o
r
ma
t
i
o
n
K
-
m
e
a
n
s
45
0
.
4
9
0
.
6
8
A
g
g
r
e
g
a
t
e
d
u
se
r
’
s p
u
b
l
i
c
i
n
f
o
r
ma
t
i
o
n
En
se
mb
l
e
K
-
me
a
n
s
15
0
.
9
7
0
.
9
5
0
10
0
20
0
30
0
40
0
No
of Us
er
s
Loc
ation
Dat
a
Sci
e
n
ce
J
av
aSc
rip
t
Dat
ab
ase
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2088
-
8708
I
J
E
C
E
Vo
l.
7
,
No
.
6
,
Dec
em
b
er
201
7
:
3
6
9
2
–
3
6
9
9
3698
5.
CO
NCLU
SI
O
N
AND
SCO
P
E
O
F
F
URTHER R
E
S
E
AR
CH
T
h
is
ar
ticle
an
al
y
ze
d
t
w
o
cl
u
s
t
er
in
g
al
g
o
r
ith
m
s
i
n
th
e
co
n
te
x
t
o
f
clu
s
ter
i
n
g
s
o
cial
n
et
w
o
r
k
d
ata
w
h
e
n
co
llected
f
r
o
m
d
if
f
er
en
t
s
o
cia
l
n
et
w
o
r
k
s
.
I
t
h
a
s
b
ee
n
r
ep
o
r
ted
th
at
it
i
s
p
o
s
s
ib
le
to
d
etec
t
co
m
m
u
n
it
y
u
s
in
g
en
s
e
m
b
le
t.
T
h
is
p
ap
er
p
r
o
v
es
th
at
th
e
e
n
s
e
m
b
le
K
-
m
ea
n
s
c
lu
s
ter
i
n
g
p
r
o
d
u
ce
s
b
etter
r
es
u
l
ts
i
n
ter
m
o
f
er
r
o
r
r
ate,
R
A
N
D
s
co
r
e
an
d
J
ac
ar
d
i
n
d
ex
.
T
h
is
o
p
en
s
u
p
t
h
e
s
co
p
e
o
f
f
u
r
t
h
er
r
esear
ch
i
n
r
eg
ar
d
s
to
ef
f
icie
n
t
u
s
e
f
o
r
b
u
s
i
n
ess
a
n
d
m
ar
k
eti
n
g
s
tr
ate
g
ies
RE
F
E
R
E
NC
E
S
[1
]
Otte,
E.
,
&
Ro
u
ss
e
a
u
,
R.
(2
0
0
2
).
S
o
c
ial
n
e
tw
o
rk
a
n
a
l
y
sis:
A
p
o
we
rf
u
l
stra
te
g
y
,
a
lso
f
o
r
th
e
in
f
o
rm
a
ti
o
n
sc
ien
c
e
s.
J
o
u
rn
a
l
o
f
in
fo
rm
a
t
io
n
S
c
ien
c
e
,
28
(
6
),
4
4
1
-
4
5
3
.
[2
]
Kre
b
s,
V
.
E.
(
2
0
0
2
,
A
p
ril
).
Un
c
l
o
a
k
in
g
terro
rist
n
e
tw
o
rk
s.
Fi
rs
t
M
o
n
d
a
y
,
7
(
4
).
[3
]
Est
iv
il
l
-
Ca
stro
,
V
.
(2
0
0
2
).
W
h
y
so
m
a
n
y
c
lu
ste
rin
g
a
lg
o
rit
h
m
s:
A
p
o
siti
o
n
p
a
p
e
r.
S
IGKD
D
Exp
lo
ra
ti
o
n
s
Ne
wsle
tt
e
r
,
4
(
1
),
6
5
-
75.
[4
]
Zh
a
n
g
,
S
.
,
W
a
n
g
,
R.
S
.
,
&
Zh
a
n
g
,
X
.
S
.
(2
0
0
7
)
.
Id
e
n
t
if
ica
ti
o
n
o
f
o
v
e
rlap
p
in
g
c
o
m
m
u
n
it
y
stru
c
t
u
re
in
c
o
m
p
lex
n
e
tw
o
rk
s u
sin
g
f
u
z
z
y
c
-
m
e
a
n
s clu
ste
rin
g
.
Ph
y
sic
a
A:
S
t
a
ti
stica
l
M
e
c
h
a
n
ics
a
n
d
it
s A
p
p
li
c
a
ti
o
n
s
,
3
7
4
(
1
),
4
8
3
-
4
9
0
.
[5
]
S
treh
l,
A
.
,
&
G
h
o
sh
,
J.
(2
0
0
2
)
.
Clu
ste
r
e
n
se
m
b
les
---
a
k
n
o
w
le
d
g
e
re
u
se
f
ra
m
e
w
o
rk
f
o
r
c
o
m
b
in
in
g
m
u
lt
ip
le
p
a
rti
ti
o
n
s.
J
o
u
rn
a
l
o
f
ma
c
h
in
e
le
a
rn
in
g
re
se
a
rc
h
,
3
(De
c
),
5
8
3
-
6
1
7
.
[6
]
T
o
p
c
h
y
,
A
.
,
M
in
a
e
i
-
Bid
g
o
li
,
B.
,
Ja
in
,
A
.
K.,
&
P
u
n
c
h
,
W
.
F
.
(2
0
0
4
,
A
u
g
u
st).
A
d
a
p
ti
v
e
c
lu
ste
rin
g
e
n
se
m
b
les
.
In
Pa
tt
e
rn
Rec
o
g
n
it
io
n
,
2
0
0
4
.
ICP
R
2
0
0
4
.
Pr
o
c
e
e
d
in
g
s
o
f
t
h
e
1
7
th
I
n
t
e
rn
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
o
n
(
V
o
l
.
1
,
p
p
.
2
7
2
-
2
7
5
).
IEE
E.
[7
]
T
o
p
c
h
y
,
A
.
,
Ja
in
,
A
.
K.,
&
P
u
n
c
h
,
W
.
(2
0
0
3
,
N
o
v
e
m
b
e
r).
Co
m
b
i
n
in
g
m
u
lt
ip
le
w
e
a
k
c
lu
ste
rin
g
s.
In
Da
t
a
M
in
i
n
g
,
2
0
0
3
.
ICDM
2
0
0
3
.
T
h
ird
IEE
E
In
ter
n
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
o
n
(p
p
.
3
3
1
-
3
3
8
).
IEE
E.
[8
]
T
o
p
c
h
y
,
A
.
,
Ja
in
,
A
.
K.,
&
P
u
n
c
h
,
W
.
(
2
0
0
4
,
A
p
ril
).
A
m
ix
tu
re
m
o
d
e
l
f
o
r
c
l
u
ste
rin
g
e
n
se
m
b
les
.
In
Pro
c
e
e
d
in
g
s
o
f
th
e
2
0
0
4
S
IAM
In
ter
n
a
ti
o
n
a
l
C
o
n
fer
e
n
c
e
o
n
Da
ta
M
i
n
in
g
(
p
p
.
3
7
9
-
3
9
0
)
.
S
o
c
iety
f
o
r
In
d
u
s
tri
a
l
a
n
d
A
p
p
li
e
d
M
a
th
e
m
a
ti
c
s.
[9
]
Ay
a
d
,
H.
G
.
,
&
Ka
m
e
l,
M
.
S
.
(
2
0
0
8
).
Cu
m
u
lativ
e
v
o
ti
n
g
c
o
n
se
n
s
u
s
m
e
th
o
d
f
o
r
p
a
rti
ti
o
n
s
w
it
h
v
a
riab
le
n
u
m
b
e
r
o
f
c
lu
ste
rs.
IEE
E
tr
a
n
sa
c
t
io
n
s
o
n
p
a
tt
e
rn
a
n
a
lys
is
a
n
d
ma
c
h
i
n
e
i
n
telli
g
e
n
c
e
,
30
(1
),
1
6
0
-
1
7
3
.
[1
0
]
S
in
g
h
,
V
.
,
M
u
k
h
e
rjee
,
L
.
,
P
e
n
g
,
J.,
&
X
u
,
J.
(
2
0
1
0
)
.
En
se
m
b
le
c
lu
ste
rin
g
u
sin
g
se
m
id
e
f
in
it
e
p
ro
g
ra
m
m
in
g
w
it
h
a
p
p
li
c
a
ti
o
n
s.
M
a
c
h
i
n
e
lea
r
n
in
g
,
79
(1
-
2
),
1
7
7
-
2
0
0
.
[1
1
]
Bh
a
tn
a
g
a
r,
V
.
,
&
A
h
u
ja,
S
.
(2
0
1
0
,
Ju
ly
).
Ro
b
u
st
c
lu
ste
rin
g
u
sin
g
d
i
sc
ri
m
in
a
n
t
a
n
a
ly
sis.
In
In
d
u
stria
l
Co
n
fer
e
n
c
e
o
n
Da
ta
M
in
in
g
(p
p
.
1
4
3
-
1
5
7
)
.
S
p
rin
g
e
r
Be
rli
n
He
id
e
lb
e
rg
.
[1
2
]
Du
d
o
i
t,
S
.
,
&
F
rid
ly
a
n
d
,
J.
(2
0
0
3
).
Ba
g
g
in
g
to
im
p
ro
v
e
th
e
a
c
c
u
ra
c
y
o
f
a
c
lu
ste
rin
g
p
ro
c
e
d
u
re
.
Bi
o
i
n
fo
rm
a
ti
c
s
,
19
(
9
),
1
0
9
0
-
1
0
9
9
.
[1
3
]
M
a
h
e
sh
,
O.
,
&
S
ri
n
iv
a
sa
n
,
G
.
(2
0
0
2
).
I
n
c
re
m
e
n
tal
c
e
ll
f
o
r
m
a
ti
o
n
c
o
n
si
d
e
rin
g
a
lt
e
rn
a
ti
v
e
m
a
c
h
in
e
s.
In
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
Pro
d
u
c
ti
o
n
Res
e
a
rc
h
,
40
(
1
4
),
3
2
9
1
-
3
3
1
0
.
[1
4
]
Dim
it
riad
o
u
,
E.
,
W
e
in
g
e
ss
e
l,
A
.
,
&
Ho
rn
ik
,
K.
(2
0
0
1
,
A
u
g
u
st).
Vo
ti
n
g
-
m
e
rg
in
g
:
A
n
e
n
se
m
b
le
m
e
th
o
d
f
o
r
c
lu
ste
rin
g
.
In
In
ter
n
a
ti
o
n
a
l
Co
n
fe
re
n
c
e
o
n
Arti
f
icia
l
Ne
u
ra
l
Ne
tw
o
rk
s
(p
p
.
2
1
7
-
2
2
4
)
.
S
p
rin
g
e
r
Be
rli
n
He
id
e
lb
e
rg
.
[1
5
]
F
o
rtu
n
a
to
,
S
.
(2
0
1
0
).
C
o
m
m
u
n
it
y
d
e
tec
ti
o
n
in
g
ra
p
h
s
.
Ph
y
sic
s R
e
p
o
rts
,
4
8
6
(3
)
,
7
5
-
1
7
4
.
[1
6
]
Clau
se
t,
A
.
,
Ne
wm
a
n
,
M
.
E.
,
&
M
o
o
re
,
C.
(
2
0
0
4
).
F
i
n
d
i
n
g
c
o
m
m
u
n
it
y
stru
c
tu
re
in
v
e
r
y
larg
e
n
e
tw
o
rk
s.
Ph
y
sic
a
l
Rev
iew E
,
70
(
6
),
0
6
6
1
1
1
.
[1
7
]
Ne
wm
a
n
,
M
.
E.
,
&
G
ir
v
a
n
,
M
.
(
2
0
0
4
).
F
i
n
d
i
n
g
a
n
d
e
v
a
lu
a
ti
n
g
c
o
m
m
u
n
it
y
stru
c
tu
re
in
n
e
t
w
o
rk
s.
Ph
y
sic
a
l
Rev
iew
E
,
69
(2
),
0
2
6
1
1
3
.
[1
8
]
L
a
n
c
ich
in
e
tt
i,
A
.
,
&
F
o
rtu
n
a
to
,
S
.
(2
0
0
9
).
Co
m
m
u
n
it
y
d
e
tec
ti
o
n
a
lg
o
rit
h
m
s:
A
c
o
m
p
a
ra
ti
v
e
a
n
a
ly
sis.
Ph
y
sic
a
l
re
v
iew E
,
80
(
5
)
,
0
5
6
1
1
7
.
[1
9
]
P
a
ll
a
,
G
.
,
De
ré
n
y
i,
I.
,
F
a
rk
a
s,
I.
,
&
V
ics
e
k
,
T
.
(2
0
0
5
).
Un
c
o
v
e
rin
g
th
e
o
v
e
rlap
p
in
g
c
o
m
m
u
n
it
y
stru
c
tu
re
o
f
c
o
m
p
lex
n
e
tw
o
rk
s in
n
a
tu
re
a
n
d
so
c
iety
.
Na
tu
re
,
4
3
5
(7
0
4
3
),
8
1
4
-
8
1
8
.
[2
0
]
Ha
n
,
J.,
Ka
m
b
e
r,
M
.
,
&
P
e
i,
J.
(
2
0
1
1
).
D
a
ta
mi
n
in
g
:
Co
n
c
e
p
ts
a
n
d
tec
h
n
i
q
u
e
s
(3
rd
e
d
.
)
.
T
h
e
Ne
th
e
rlan
d
s:
M
o
rg
a
n
Ka
u
fm
a
n
n
.
[2
1
]
Orm
e
,
B.
,
&
Jo
h
n
s
o
n
,
R
.
(2
0
0
8
).
Imp
r
o
v
in
g
k
-
me
a
n
s
c
lu
ste
r
a
n
a
lys
is:
En
se
mb
le
a
n
a
lys
is
i
n
s
tea
d
o
f
h
i
g
h
e
st
re
p
ro
d
u
c
i
b
il
it
y
re
p
l
ica
tes
(
S
a
w
to
o
th
S
o
f
tw
a
re
Re
se
a
rc
h
P
a
p
e
r
S
e
ri
e
s)
.
S
e
q
u
im
,
WA
:
S
a
w
to
o
th
S
o
f
t
w
a
r
e
,
In
c
.
[2
2
]
S
a
h
u
,
M
.
,
P
a
rv
a
th
i,
K.
,
&
Krish
n
a
,
M
.
V
.
(2
0
1
7
).
P
a
ra
m
e
tri
c
C
o
m
p
a
riso
n
o
f
K
-
m
e
a
n
s
a
n
d
A
d
a
p
ti
v
e
K
-
m
e
a
n
s
Clu
ste
rin
g
P
e
rf
o
rm
a
n
c
e
o
n
Diffe
re
n
t
Im
a
g
e
s.
In
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
E
lec
trica
l
a
n
d
C
o
mp
u
te
r
En
g
i
n
e
e
rin
g
(
IJ
ECE
)
,
7
(2
).
[2
3
]
S
a
h
u
,
M
.
,
P
a
rv
a
th
i,
K.
,
&
Kr
ish
n
a
,
M
.
V
.
(2
0
1
7
).
P
a
ra
m
e
tri
c
C
o
m
p
a
riso
n
o
f
K
-
m
e
a
n
s
a
n
d
A
d
a
p
ti
v
e
K
-
m
e
a
n
s
Clu
ste
rin
g
P
e
rf
o
rm
a
n
c
e
o
n
Diffe
re
n
t
Im
a
g
e
s.
In
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
E
lec
trica
l
a
n
d
C
o
mp
u
te
r
En
g
i
n
e
e
rin
g
(
IJ
ECE
)
,
7
(2
).
[2
4
]
Ya
n
g
,
X
.
,
W
a
n
g
,
Y.,
W
u
,
D.,
&
M
a
,
A
.
(2
0
1
0
,
N
o
v
e
m
b
e
r).
K
-
m
e
a
n
s
b
a
se
d
c
lu
ste
rin
g
o
n
m
o
b
il
e
u
sa
g
e
f
o
r
so
c
ial
n
e
tw
o
rk
a
n
a
l
y
sis
p
u
rp
o
se
.
I
n
2
0
1
0
6
th
In
ter
n
a
ti
o
n
a
l
C
o
n
fer
e
n
c
e
o
n
Ad
v
a
n
c
e
d
In
fo
rm
a
ti
o
n
M
a
n
a
g
e
m
e
n
t
a
n
d
S
e
rv
ice
(
IM
S
)
(p
p
.
2
2
3
-
2
2
8
).
I
EE
E.
[2
5
]
Ole
iw
i,
W
.
K.
(2
0
1
6
)
.
Us
in
g
t
h
e
F
u
z
z
y
L
o
g
ic
to
F
in
d
Op
ti
m
a
l
Ce
n
ters
o
f
Cl
u
ste
rs
o
f
K
-
m
e
a
n
s.
In
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
El
e
c
trica
l
a
n
d
C
o
mp
u
t
e
r E
n
g
i
n
e
e
rin
g
,
6
(
6
),
3
0
6
8
.
[2
6
]
Ja
in
,
A
.
K.,
&
Du
b
e
s,
R
.
C.
(
1
9
8
8
).
Al
g
o
rith
ms
fo
r clu
ste
ri
n
g
d
a
ta
.
P
re
n
ti
c
e
-
Ha
ll
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
E
C
E
I
SS
N:
2088
-
8708
C
lu
s
ter
in
g
in
A
g
g
r
eg
a
ted
User
P
r
o
files
a
cro
s
s
Mu
ltip
le
S
o
cia
l Netw
o
r
k
s
(
C
h
a
r
u
V
ir
ma
n
i
)
3699
[2
7
]
S
tein
h
a
e
u
se
r,
K.,
&
Ch
a
w
la,
N.
V
.
(
2
0
1
0
).
I
d
e
n
ti
f
y
in
g
a
n
d
e
v
a
lu
a
ti
n
g
c
o
m
m
u
n
it
y
stru
c
tu
re
in
c
o
m
p
lex
n
e
tw
o
rk
s.
Pa
tt
e
rn
Rec
o
g
n
it
io
n
L
e
tt
e
rs
,
31
(
5
),
4
1
3
-
4
2
1
.
[2
8
]
L
i,
X
.
,
Hu
a
n
g
,
Y.,
L
i,
S
.
,
&
Zh
a
n
g
,
Y.
(2
0
1
1
,
M
a
y
).
H
y
b
rid
re
ten
ti
o
n
stra
teg
y
f
o
rm
u
latio
n
in
tele
c
o
m
b
a
se
d
o
n
k
-
m
e
a
n
s
c
lu
ste
rin
g
a
n
a
l
y
si
s.
In
2
0
1
1
In
ter
n
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
o
n
E
-
Bu
sin
e
ss
a
n
d
E
-
Go
v
e
rn
me
n
t
(
ICEE
)
(p
p
.
1
-
4
)
.
IEE
E.
V
e
g
a
-
P
o
n
s,
S
.
,
Co
rre
a
-
M
o
r
ris,
J.,
&
Ru
iz
-
S
h
u
lclo
p
e
r,
J.
(2
0
0
8
,
S
e
p
tem
b
e
r).
Weig
h
ted
c
lu
ste
r
e
n
se
m
b
le
u
sin
g
a
k
e
rn
e
l
c
o
n
se
n
su
s
f
u
n
c
ti
o
n
.
In
Ib
e
ro
a
me
ric
a
n
C
o
n
g
re
ss
o
n
Pa
tt
e
rn
Rec
o
g
n
it
i
o
n
(
p
p
.
1
9
5
-
2
0
2
).
S
p
rin
g
e
r
Be
rli
n
He
id
e
lb
e
rg
.
[2
9
]
M
irk
in
,
B.
(
1
9
9
6
)
.
M
a
th
e
m
a
ti
c
a
l
Clas
sif
ic
a
ti
o
n
a
n
d
Clu
ste
ri
n
g
,
No
n
c
o
n
v
e
x
Op
ti
miza
ti
o
n
a
n
d
Its
Ap
p
l
ica
ti
o
n
s
,
V
o
l
u
m
e
1
1
,
P
a
r
d
a
lo
s,
P
.
a
n
d
Ho
rs
t,
R.
,
e
d
it
o
rs.
[3
0
]
Yo
o
n
,
H.
S
.
,
A
h
n
,
S
.
Y.
,
L
e
e
,
S
.
H.,
Ch
o
,
S
.
B.
,
&
Kim
,
J.
H.
(2
0
0
6
,
A
p
ril
).
He
tero
g
e
n
e
o
u
s
c
l
u
st
e
rin
g
e
n
se
m
b
le
m
e
th
o
d
f
o
r
c
o
m
b
in
in
g
d
if
fe
re
n
t
c
lu
ste
r
re
su
lt
s.
In
I
n
ter
n
a
ti
o
n
a
l
W
o
rk
sh
o
p
o
n
D
a
ta
M
in
in
g
f
o
r
Bi
o
me
d
ic
a
l
Ap
p
li
c
a
ti
o
n
s.
S
p
ri
n
g
e
r B
e
rli
n
He
i
d
e
lb
e
rg
.
(
p
p
.
8
2
-
9
2
)
.
[3
1
]
L
i,
T
.
,
Din
g
,
C.
,
&
Jo
rd
a
n
,
M
.
I.
(
2
0
0
7
,
Oc
to
b
e
r).
S
o
lv
in
g
c
o
n
se
n
s
u
s an
d
se
m
i
-
su
p
e
rv
ise
d
c
lu
ste
rin
g
p
ro
b
lem
s u
sin
g
n
o
n
n
e
g
a
ti
v
e
m
a
tri
x
f
a
c
to
riza
ti
o
n
.
In
Da
t
a
M
in
in
g
,
2
0
0
7
.
ICDM
2
0
0
7
.
S
e
v
e
n
t
h
IEE
E
In
ter
n
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
o
n
(p
p
.
5
7
7
-
5
8
2
).
IEE
E.
[3
2
]
W
e
in
g
e
s
se
l,
A
.
,
Di
m
i
tri
a
d
o
u
,
E.
,
&
Ho
rn
ik
,
K.
(2
0
0
3
).
A
n
e
n
se
m
b
le
m
e
th
o
d
f
o
r
c
lu
ste
rin
g
.
In
P
r
o
c
e
e
d
in
g
s
o
f
th
e
3
rd
I
n
tern
a
ti
o
n
a
l
W
o
rk
sh
o
p
o
n
Di
strib
u
te
d
S
tatisti
c
a
l
C
o
m
p
u
ti
n
g
.
[3
3
]
Da
h
li
n
,
J.,
&
S
v
e
n
so
n
,
P
.
(2
0
1
3
).
En
se
mb
le
a
p
p
r
o
a
c
h
e
s
fo
r
imp
ro
v
in
g
c
o
mm
u
n
it
y
d
e
tec
ti
o
n
me
th
o
d
s
.
a
rX
iv
p
re
p
rin
t
a
rX
iv
:1
3
0
9
.
0
2
4
2
.
[3
4
]
Ra
n
d
,
W
.
M
.
(
1
9
7
1
)
.
O
b
jec
ti
v
e
c
rit
e
ria
f
o
r
th
e
e
v
a
lu
a
ti
o
n
o
f
c
lu
ste
rin
g
m
e
th
o
d
s.
J
o
u
rn
a
l
o
f
th
e
Ame
ric
a
n
S
t
a
ti
stica
l
a
ss
o
c
ia
ti
o
n
,
6
6
(3
3
6
)
,
8
4
6
-
8
5
0
.
[3
5
]
Be
n
-
Hu
r,
A
.
,
El
isse
e
ff
,
A
.
,
&
G
u
y
o
n
,
I.
(2
0
0
1
,
De
c
e
m
b
e
r).
A
sta
b
il
it
y
b
a
se
d
me
th
o
d
fo
r
d
isc
o
v
e
rin
g
str
u
c
tu
re
in
c
lu
ste
re
d
d
a
t
a
.
I
n
P
a
c
if
ic sy
mp
o
si
u
m o
n
b
i
o
c
o
m
p
u
ti
n
g
(Vo
l.
7
,
p
p
.
6
-
1
7
).
Evaluation Warning : The document was created with Spire.PDF for Python.