I
nte
rna
t
io
na
l J
o
urna
l o
f
I
nfo
rm
a
t
ics a
nd
Co
m
m
un
ica
t
io
n T
ec
hn
o
lo
g
y
(
I
J
-
I
CT
)
Vo
l.
1
4
,
No
.
1
,
A
p
r
il
20
2
5
,
p
p
.
216
~
22
8
I
SS
N:
2252
-
8
7
7
6
,
DOI
:
1
0
.
1
1
5
9
1
/iji
ct
.
v
1
4
i
1
.
pp
216
-
22
8
216
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ict.
ia
esco
r
e.
co
m
Srv
y
cite:
a hy
bri
d scienti
fic
a
r
ticle
rec
o
mm
enda
tion
sy
stem
Siv
a
s
a
nk
a
ri
R,
J
.
Dhili
pa
n
D
e
p
a
r
t
me
n
t
o
f
C
o
mp
u
t
e
r
A
p
p
l
i
c
a
t
i
o
n
s (M
C
A
)
,
S
R
M
I
n
st
i
t
u
t
e
o
f
S
c
i
e
n
c
e
a
n
d
Te
c
h
n
o
l
o
g
y
-
R
a
ma
p
u
r
a
m
C
a
mp
u
s,
C
h
e
n
n
a
i
,
I
n
d
i
a
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Ma
y
7
,
2
0
2
4
R
ev
is
ed
Au
g
2
9
,
2
0
2
4
Acc
ep
ted
Sep
2
2
,
2
0
2
4
A
re
c
o
m
m
e
n
d
a
ti
o
n
sy
ste
m
is
b
e
c
o
m
in
g
p
a
rt
o
f
e
v
e
r
y
wo
rk
d
o
n
e
to
d
a
y
t
o
re
d
u
c
e
th
e
e
ffo
rt
o
f
wo
rk
d
o
n
e
b
y
th
e
u
se
rs
in
se
a
rc
h
in
g
fo
r
it
e
m
s
in
n
e
e
d
b
y
re
c
o
m
m
e
n
d
in
g
n
e
w
it
e
m
s
th
a
t
m
a
y
b
e
u
se
fu
l.
Th
is
th
e
m
e
h
a
s
a
lso
b
e
e
n
u
se
d
in
re
se
a
rc
h
a
rti
c
le
re
c
o
m
m
e
n
d
a
ti
o
n
s
y
ste
m
s
fo
r
re
c
o
m
m
e
n
d
in
g
a
rti
c
les
o
f
in
tere
st
to
re
se
a
rc
h
e
rs
fro
m
a
b
u
lk
o
f
d
ig
it
a
l
re
se
a
rc
h
d
o
c
u
m
e
n
ts
sp
re
a
d
a
c
ro
ss
d
iffere
n
t
d
a
tab
a
se
s
o
n
th
e
in
tern
e
t.
To
e
a
se
th
e
tas
k
o
f
t
h
is
a
rti
c
le
re
c
o
m
m
e
n
d
a
ti
o
n
p
r
o
c
e
ss
,
we
h
a
v
e
p
ro
p
o
se
d
a
n
o
v
e
l
a
p
p
ro
a
c
h
,
S
r
v
y
c
it
e
,
b
y
u
ti
li
z
i
n
g
t
h
e
su
rv
e
y
a
rti
c
le
c
it
a
ti
o
n
n
e
two
r
k
a
lo
n
g
with
t
h
e
o
ri
g
in
a
l
re
se
a
rc
h
a
rti
c
le
n
e
two
rk
.
Th
e
p
u
r
p
o
se
o
f
u
ti
li
z
i
n
g
t
h
e
su
r
v
e
y
a
rti
c
le
c
it
a
ti
o
n
n
e
two
r
k
is
to
d
e
tec
t
th
e
m
o
st
i
n
fl
u
e
n
ti
a
l
a
rti
c
les
th
a
t
a
re
c
o
n
si
d
e
re
d
t
o
b
e
imp
o
r
tan
t
b
y
o
th
e
r
re
se
a
rc
h
e
rs i
n
th
e
sa
m
e
field
.
T
h
e
S
rv
y
c
it
e
a
p
p
r
o
a
c
h
u
t
il
iz
e
s th
e
tex
t
a
n
d
m
e
ta
fe
a
tu
re
s
o
f
a
rti
c
les
to
r
e
c
o
m
m
e
n
d
p
a
p
e
rs.
To
p
re
p
r
o
c
e
ss
th
e
tex
t
fe
a
tu
re
s
u
ti
li
z
e
d
,
we
h
a
v
e
e
m
p
l
o
y
e
d
W
o
rd
2
Ve
c
a
n
d
b
id
irec
ti
o
n
a
l
e
n
c
o
d
e
r
re
p
re
se
n
tatio
n
s
fro
m
tran
sf
o
rm
e
rs
(
BERT
)
fo
r
v
e
c
t
o
riza
ti
o
n
.
Th
e
n
c
it
a
ti
o
n
g
ra
p
h
a
n
d
su
r
v
e
y
c
it
a
ti
o
n
g
ra
p
h
s
a
re
g
e
n
e
ra
ted
t
o
fi
n
d
th
e
m
o
st
i
n
flu
e
n
ti
a
l
n
o
d
e
s.
Th
e
we
ig
h
ted
te
x
t
sim
il
a
ri
ty
sc
o
re
is
fi
n
a
ll
y
c
o
m
p
u
ted
b
y
c
o
m
b
in
i
n
g
th
e
c
it
e
d
b
y
v
a
lu
e
s
a
n
d
t
h
e
tex
t
si
m
il
a
rit
y
sc
o
re
fro
m
t
h
e
c
it
a
ti
o
n
a
n
d
s
u
rv
e
y
c
it
a
ti
o
n
g
ra
p
h
to
li
st
a
rti
c
les
a
s
re
c
o
m
m
e
n
d
a
ti
o
n
s
fo
r
t
h
e
u
se
r.
T
h
is
sy
ste
m
i
s
p
ro
v
e
n
to
i
n
c
re
a
se
th
e
a
c
c
u
ra
c
y
o
f
th
e
a
rti
c
le rec
o
m
m
e
n
d
a
ti
o
n
b
y
3
.
8
a
n
d
2
.
1
in
t
h
e
c
a
se
o
f
th
e
p
re
c
isio
n
a
n
d
re
c
a
ll
m
e
a
su
re
s fo
r
p
e
rfo
rm
a
n
c
e
e
v
a
lu
a
ti
o
n
.
K
ey
w
o
r
d
s
:
B
E
R
T
Dim
en
s
i
o
n
al
r
ed
u
ctio
n
Scien
tific
ar
ticle
r
ec
o
m
m
en
d
atio
n
Su
r
v
ey
citatio
n
n
etwo
r
k
citatio
n
n
etwo
r
k
T
h
is i
s
a
n
o
p
e
n
a
c
c
e
ss
a
rticle
u
n
d
e
r th
e
CC B
Y
-
SA
li
c
e
n
se
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
Siv
asan
k
ar
i R
Dep
ar
tm
en
t o
f
C
o
m
p
u
ter
A
p
p
licatio
n
s
(
MCA)
SR
M
I
n
s
titu
te
o
f
Scien
ce
an
d
T
ec
h
n
o
lo
g
y
-
R
am
ap
u
r
am
C
am
p
u
s
C
h
en
n
ai,
T
am
il Na
d
u
,
I
n
d
ia
E
m
ail: m
an
ju
r
m
o
n
i@
g
m
ail.
co
m
1.
I
NT
RO
D
UCT
I
O
N
A
s
cien
tific
ar
ticle
r
ec
o
m
m
e
n
d
atio
n
s
y
s
tem
aim
s
to
r
ec
o
m
m
en
d
r
esear
ch
p
a
p
er
s
f
o
r
r
esear
ch
er
s
.
W
ith
a
r
ap
id
in
cr
ea
s
e
in
r
ese
ar
ch
ac
tiv
ities
,
it
tak
es
tim
e
f
o
r
r
esear
c
h
er
s
to
g
ath
er
th
e
d
ata
n
ee
d
ed
f
o
r
th
ei
r
r
esear
ch
.
T
h
e
d
ata
n
ee
d
ed
also
in
clu
d
es
p
r
ev
i
o
u
s
r
esear
ch
d
o
n
e
in
th
e
p
r
e
f
er
r
e
d
ar
ea
.
T
h
is
is
r
eq
u
ir
ed
f
o
r
t
h
e
r
esear
ch
er
s
to
s
tay
u
p
d
ated
an
d
to
f
o
cu
s
o
n
th
e
r
eq
u
ir
ed
ar
e
a
an
d
p
ar
t
o
f
th
e
r
esear
ch
.
No
t
o
n
ly
th
at,
b
u
t
with
th
e
h
elp
o
f
p
r
ev
i
o
u
s
ly
p
u
b
lis
h
ed
ar
ticles,
r
esear
ch
er
s
ca
n
ea
s
ily
id
en
tify
th
e
c
o
n
c
ep
t a
n
d
f
u
lf
il th
e
r
eq
u
ir
em
e
n
t
to
en
h
an
ce
th
e
id
ea
f
u
r
th
er
.
I
n
to
d
ay
'
s
d
ig
itally
co
n
n
ec
te
d
wo
r
ld
,
it
is
ef
f
o
r
tles
s
to
g
ath
er
n
ee
d
e
d
ar
ticles
th
r
o
u
g
h
a
d
ig
ital
s
u
p
p
o
r
t
s
y
s
tem
.
Yet
th
e
n
ee
d
f
o
r
th
e
r
ec
o
m
m
en
d
atio
n
s
y
s
tem
is
to
id
en
tif
y
th
e
m
o
s
t
r
elev
an
t
ar
tic
les
with
o
u
t
s
p
en
d
in
g
m
u
c
h
tim
e.
Scien
tific
r
ec
o
m
m
en
d
atio
n
s
y
s
tem
s
h
elp
r
esear
ch
er
s
b
y
id
en
tify
in
g
th
e
m
o
s
t
r
elev
an
t
ar
ticles
f
o
r
th
e
m
with
th
e
h
elp
o
f
a
s
ea
r
c
h
q
u
er
y
.
T
h
ese
r
ec
o
m
m
e
n
d
ati
o
n
s
y
s
tem
s
ca
n
b
e
ca
teg
o
r
ized
in
to
co
n
te
n
t
-
b
a
s
ed
s
y
s
tem
s
,
co
llab
o
r
ativ
e
s
y
s
tem
s
,
g
r
ap
h
-
b
ased
s
y
s
tem
s
,
an
d
h
y
b
r
i
d
r
ec
o
m
m
en
d
atio
n
s
y
s
tem
s
.
T
h
e
co
n
ten
t
-
b
ased
f
ilter
in
g
m
et
h
o
d
s
(
C
B
F)
is
a
class
ic
r
ec
o
m
m
en
d
atio
n
ap
p
r
o
ac
h
th
at
p
r
o
v
i
d
es
r
ec
o
m
m
e
n
d
atio
n
s
b
y
d
is
co
v
er
i
n
g
th
e
ass
o
ciatio
n
b
etwe
en
u
s
er
m
o
d
ellin
g
an
d
tex
tu
al
f
ea
tu
r
es
i
n
th
e
p
ap
er
.
T
h
e
a
d
v
an
tag
e
o
f
C
B
F
allo
ws
u
s
er
-
b
ased
p
er
s
o
n
aliza
tio
n
an
d
r
eq
u
ir
es
less
c
lass
if
icatio
n
wo
r
k
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
I
SS
N:
2252
-
8
7
7
6
S
r
vy
cite:
a
h
yb
r
id
s
cien
tifi
c
a
r
ticle
r
ec
o
mme
n
d
a
tio
n
s
ystem
(
S
iva
s
a
n
ka
r
i R
)
217
C
B
F
r
eq
u
ir
es
ad
d
r
ess
in
g
p
r
o
b
lem
s
lik
e
h
ig
h
co
m
p
u
tin
g
p
o
wer
,
lo
w
s
er
en
d
ip
ity
,
an
d
o
v
e
r
s
p
ec
ializatio
n
,
an
d
also
ig
n
o
r
es
th
e
q
u
ality
an
d
p
o
p
u
l
ar
ity
o
f
item
s
.
I
t
is
d
ep
en
d
en
t
o
n
ac
ce
s
s
to
th
e
item
'
s
f
ea
tu
r
es,
m
ak
in
g
it
u
n
s
u
itab
le
f
o
r
m
an
y
u
s
er
s
[
1
]
.
T
h
e
id
ea
o
f
co
llab
o
r
ativ
e
f
ilte
r
in
g
s
y
s
tem
s
(
C
F)
is
to
p
r
o
p
o
s
e
a
p
ap
er
to
u
s
er
s
wh
o
s
h
ar
e
co
m
m
o
n
in
ter
ests
.
I
t
im
p
lies
th
at
if
tw
o
u
s
er
s
A
an
d
B
h
a
v
e
s
im
ilar
in
ter
ests
,
th
e
ar
ticles
v
iewe
d
b
y
u
s
er
A
ca
n
b
e
r
ec
o
m
m
en
d
ed
to
u
s
er
B
.
C
F
is
co
n
ten
t
-
in
d
ep
en
d
e
n
t,
c
o
n
s
id
er
s
r
ea
l
-
wo
r
ld
q
u
ality
r
atin
g
s
,
an
d
m
ak
es
s
er
en
d
ip
ito
u
s
r
ec
o
m
m
en
d
atio
n
s
.
C
F
s
u
f
f
er
s
f
r
o
m
th
e
"c
o
ld
s
tar
t"
p
r
o
b
lem
,
wh
ich
o
cc
u
r
s
wh
en
th
e
u
s
er
o
r
th
e
item
is
n
ew
an
d
n
o
p
r
o
p
e
r
r
atin
g
in
f
o
r
m
atio
n
is
s
u
p
p
lied
b
y
th
e
u
s
er
s
[
2
],
[
3
]
.
I
n
g
r
a
p
h
-
b
ased
r
ec
o
m
m
en
d
atio
n
s
y
s
t
em
s
(
GB
)
th
e
f
ea
tu
r
es
o
f
a
n
ar
ticles
s
u
ch
as
tex
tu
al
co
n
ten
t,
b
ib
lio
g
r
ap
h
y
,
au
th
o
r
s
m
eta
d
ata
,
an
d
ar
ticle
m
eta
d
ata
ca
n
b
e
u
s
ed
to
co
n
s
tr
u
ct
th
e
g
r
ap
h
.
Af
ter
b
u
ild
in
g
a
g
r
ap
h
f
r
o
m
th
e
r
eq
u
i
r
ed
f
ea
tu
r
es
r
ec
o
m
m
en
d
atio
n
s
ar
e
co
m
p
u
te
d
b
ased
o
n
t
h
e
g
r
ap
h
alg
o
r
ith
m
s
.
GB
s
y
s
tem
s
ca
n
h
an
d
le
d
ata
s
p
ar
s
ity
b
u
t
r
e
q
u
ir
e
im
p
r
o
v
em
en
ts
t
o
h
an
d
le
c
o
m
p
lex
q
u
er
ies.
T
o
o
v
er
co
m
e
th
e
c
h
allen
g
es
o
f
t
h
e
ab
o
v
e
-
m
en
tio
n
ed
s
y
s
tem
,
two
o
r
m
o
r
e
ty
p
es
ca
n
b
e
co
m
b
in
e
d
to
p
r
o
d
u
ce
a
m
o
s
t
ef
f
ec
tiv
e
an
d
p
o
wer
f
u
l
r
ec
o
m
m
en
d
atio
n
.
Su
ch
s
y
s
tem
s
ar
e
ca
lled
h
y
b
r
id
r
ec
o
m
m
en
d
atio
n
s
y
s
tem
s
,
an
d
t
h
ey
h
a
v
e
th
e
ca
p
a
b
ilit
y
to
a
d
d
r
ess
two
o
r
m
o
r
e
is
s
u
es.
E
x
is
tin
g
r
ec
o
m
m
en
d
a
tio
n
alg
o
r
ith
m
s
illu
s
tr
ate
th
e
v
iab
ilit
y
o
f
r
ec
o
m
m
en
d
i
n
g
m
u
ltip
le
ty
p
es
o
f
f
ea
tu
r
es
an
d
r
elatio
n
s
h
ip
s
f
r
o
m
s
cien
tific
ar
ticles,
s
u
ch
as
h
is
to
r
ical
b
eh
av
io
u
r
an
d
r
ec
o
r
d
s
,
r
esear
ch
co
n
ten
t,
citatio
n
n
etwo
r
k
s
,
an
d
s
o
o
n
[
4
]
.
An
ef
f
icien
t
s
cien
tific
p
a
p
er
r
ec
o
m
m
en
d
e
r
s
y
s
tem
is
n
ee
d
ed
to
g
en
e
r
ate
h
ig
h
-
q
u
ality
r
ec
o
m
m
en
d
atio
n
s
f
o
r
n
o
v
ice
r
esear
ch
er
s
wh
o
lack
k
n
o
wled
g
e
in
th
e
r
esear
ch
ar
ea
[
2
]
,
[5
]
.
Sch
o
lar
s
h
av
e
co
n
d
u
cte
d
r
esear
ch
o
n
r
ec
o
m
m
en
d
atio
n
s
y
s
tem
s
,
b
u
t
th
er
e
is
a
lack
o
f
ex
p
lo
r
atio
n
o
f
th
e
r
esear
ch
s
tatu
s
an
d
f
u
tu
r
e
tr
en
d
s
[
6
].
Af
ter
ex
p
lo
r
in
g
s
ev
er
al
r
esear
ch
ar
ticles
[
7
]
-
[
9
]
,
it
is
co
n
clu
d
e
d
th
at
th
e
s
tate
-
of
-
th
e
-
ar
t
ar
ticles
o
r
r
ev
iew
ar
ticles
in
th
e
f
ield
o
f
s
tu
d
y
ar
e
co
m
p
letely
ig
n
o
r
ed
an
d
n
o
t
co
n
s
id
er
ed
as
im
p
o
r
tan
t
to
co
m
p
u
te
th
e
r
ec
o
m
m
en
d
atio
n
.
T
h
e
r
e
v
iew
ar
ticl
es
ar
e
p
r
im
ar
y
b
ec
au
s
e
th
ey
s
y
n
th
esis
e
p
ast
wo
r
k
s
,
s
u
m
m
ar
is
e
th
e
id
ea
o
f
th
e
wo
r
k
,
u
n
d
er
s
tan
d
well
th
e
p
r
o
b
le
m
th
at
h
as
b
ee
n
d
is
cu
s
s
ed
,
an
d
p
r
o
v
id
e
a
p
o
ten
tial
way
to
t
h
in
k
a
b
o
u
t
t
h
e
s
o
lu
tio
n
to
th
e
p
r
o
b
lem
u
n
d
er
d
is
cu
s
s
io
n
.
L
iter
atu
r
e
r
ev
i
ews
ar
e
in
clu
d
ed
i
n
all
r
esear
ch
p
r
o
jects,
r
eg
a
r
d
le
s
s
o
f
th
e
d
is
cip
lin
e.
A
liter
atu
r
e
r
ev
iew
is
a
r
esear
ch
a
r
ticle
wh
er
e
th
e
au
th
o
r
ty
p
ically
b
eg
in
s
b
y
s
u
m
m
ar
is
in
g
p
r
io
r
r
esear
c
h
to
m
ap
an
d
ev
alu
ate
th
e
r
esear
ch
ar
ea
,
in
s
p
ir
e
th
e
s
tu
d
y
'
s
p
u
r
p
o
s
e,
an
d
s
u
p
p
o
r
t
th
e
r
es
ea
r
ch
q
u
esti
o
n
an
d
h
y
p
o
th
es
es.
A
g
o
o
d
liter
atu
r
e
r
e
v
iew
wo
u
ld
d
is
cu
s
s
th
e
im
p
o
r
tan
t
wo
r
k
s
p
r
o
d
u
ce
d
o
v
er
tim
e
in
th
e
s
elec
ted
f
ield
o
f
s
tu
d
y
as
well
as
th
e
cu
r
r
en
t
tr
en
d
s
in
th
e
f
ield
o
f
r
esear
ch
.
T
h
is
r
esu
lts
in
a
lis
t
o
f
citatio
n
s
th
at
ar
e
r
eg
ar
d
e
d
as
s
ig
n
if
ica
n
t
an
d
h
ig
h
ly
p
e
r
tin
en
t
to
th
e
s
tu
d
y
.
T
h
ese
citatio
n
s
m
ay
b
e
u
s
ed
to
r
ec
o
m
m
e
n
d
a
r
elev
an
t
r
esear
ch
ar
ticle.
I
f
a
citatio
n
n
etwo
r
k
f
o
r
s
u
ch
a
s
u
r
v
ey
p
ap
er
is
b
u
ilt,
it
m
ay
r
esu
lt
in
a
n
etwo
r
k
th
at
co
m
p
r
is
es
s
ig
n
if
ican
t
ar
ticles
f
r
o
m
r
ec
en
t
tim
es
th
at
ca
n
b
e
u
s
ed
f
o
r
th
e
r
ec
o
m
m
en
d
atio
n
s
y
s
tem
s
.
T
h
is
wo
r
k
s
u
g
g
ests
a
Srv
y
cite,
a
h
y
b
r
id
s
cien
tific
ar
ticl
e
r
ec
o
m
m
en
d
atio
n
s
y
s
tem
th
at
b
u
ild
s
th
e
s
u
r
v
ey
citatio
n
n
etwo
r
k
(
SC
N)
alo
n
g
with
th
e
C
B
F
an
d
citat
io
n
n
etwo
r
k
s
(
C
N)
to
r
ec
o
m
m
en
d
r
elev
a
n
t a
r
ticles.
T
h
e
o
b
jectiv
e
o
f
th
is
s
tu
d
y
is
s
u
m
m
ar
ized
as f
o
llo
ws:
−
T
h
e
m
ain
o
b
jectiv
e
o
f
th
is
p
ap
er
is
to
im
p
r
o
v
e
th
e
ac
cu
r
ac
y
o
f
r
ec
o
m
m
en
d
atio
n
s
b
y
u
s
in
g
a
h
y
b
r
id
tech
n
iq
u
e
th
at
in
teg
r
ates
co
n
ten
t
-
b
ased
f
ilter
in
g
an
d
citatio
n
g
r
ap
h
s
.
T
o
p
r
o
v
e
th
is
o
b
je
ctiv
e,
we
h
av
e
in
tr
o
d
u
ce
d
a
n
o
v
el
citatio
n
g
r
ap
h
ca
lled
th
e
SC
N
g
r
ap
h
,
w
h
ich
is
co
n
s
tr
u
cted
u
s
in
g
all
t
y
p
es
o
f
s
u
r
v
ey
ar
ticles f
r
o
m
th
e
d
ataset
with
t
h
eir
citatio
n
s
.
−
Ap
ar
t f
r
o
m
th
e
SC
N,
we
also
em
p
lo
y
ed
th
e
C
N
g
r
ap
h
to
im
p
r
o
v
e
t
h
e
ac
cu
r
ac
y
o
f
o
u
r
r
esu
lts
.
−
Fo
r
clu
s
ter
in
g
,
we
u
s
ed
b
id
ir
ec
tio
n
al
en
co
d
e
r
r
ep
r
esen
tatio
n
s
f
r
o
m
tr
an
s
f
o
r
m
er
s
(
B
E
R
T
)
in
tex
t
em
b
ed
d
in
g
alo
n
g
c
o
s
in
e
s
im
ilar
ity
an
d
th
e
K
-
m
ea
n
s
alg
o
r
ith
m
.
−
Her
e,
th
e
d
ata
we
u
s
ed
is
h
ig
h
-
d
im
en
s
io
n
al
tex
t
d
ata
th
at
s
u
f
f
er
s
f
r
o
m
m
e
m
o
r
y
an
d
tim
e
co
m
p
lex
ity
.
T
o
ad
d
r
ess
th
is
p
r
o
b
lem
,
L
ASSO
o
r
L
1
r
eg
u
lar
izatio
n
is
u
s
ed
.
−
Fu
r
th
er
m
o
r
e
,
th
e
ex
p
er
im
e
n
ts
with
th
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
s
h
o
w
th
at
th
e
r
ec
o
m
m
en
d
ati
o
n
s
y
s
tem
n
o
w
wo
r
k
s
b
etter
an
d
g
iv
es m
o
r
e
a
cc
u
r
ate
r
esu
lts
co
m
p
ar
e
d
to
th
e
o
th
er
m
eth
o
d
s
p
r
esen
te
d
in
t
h
is
s
tu
d
y
.
T
o
s
u
m
m
ar
ize
th
e
co
n
tr
ib
u
tio
n
s
o
f
th
e
p
r
o
p
o
s
ed
s
y
s
tem
,
th
e
f
ir
s
t step
i
s
to
clu
s
ter
th
e
en
tir
e
d
ata
in
to
two
clu
s
ter
s
:
s
u
r
v
ey
ar
ticle
c
lu
s
ter
an
d
o
r
ig
in
al
ar
ticle
clu
s
ter
s
.
T
o
clu
s
ter
tex
t
d
ata,
th
is
s
y
s
tem
f
o
llo
ws
s
ev
er
al
s
tep
s
,
s
u
ch
as
tex
t
p
r
ep
r
o
ce
s
s
in
g
,
tex
t
em
b
ed
d
in
g
,
d
im
en
s
io
n
r
e
d
u
ctio
n
,
s
im
ilar
ity
ca
lcu
latio
n
,
an
d
clu
s
ter
in
g
.
Fo
llo
wed
b
y
clu
s
ter
in
g
is
g
r
ap
h
g
e
n
er
atio
n
,
wh
er
e
two
d
if
f
er
e
n
t
g
r
ap
h
s
(
SC
N
an
d
C
N)
ar
e
co
n
s
tr
u
cted
b
y
m
atc
h
in
g
th
e
s
ea
r
ch
q
u
er
y
o
f
th
e
u
s
er
.
Fin
ally
,
th
e
r
esu
lts
o
f
b
o
th
g
r
ap
h
s
ar
e
co
m
b
in
ed
to
g
en
er
ate
th
e
lis
t
o
f
f
in
al
ar
ticles
r
elev
an
t
to
th
e
s
ea
r
ch
q
u
er
y
.
Her
e,
th
e
p
r
o
p
o
s
ed
s
y
s
tem
is
h
y
b
r
id
,
wh
er
e
th
e
co
llab
o
r
atio
n
-
f
ilter
in
g
m
eth
o
d
is
u
s
ed
to
f
in
d
im
p
o
r
ta
n
t
n
o
d
es
in
th
e
n
etwo
r
k
an
d
c
o
n
ten
t
-
f
ilter
in
g
is
u
s
ed
in
th
e
later
p
r
o
ce
s
s
to
co
m
p
u
te
th
e
tex
t
s
im
ila
r
ity
s
co
r
e
alo
n
g
with
citatio
n
v
alu
es
co
m
p
u
ted
in
th
e
g
r
a
p
h
to
g
en
er
ate
th
e
f
in
al
r
ec
o
m
m
en
d
atio
n
s
.
Fig
u
r
e
1
ex
p
lain
s
th
e
ar
ch
itectu
r
e
o
f
th
e
p
r
o
p
o
s
ed
s
y
s
tem
.
T
h
e
r
est
o
f
th
e
p
ap
e
r
is
s
tr
u
ctu
r
ed
as
f
o
ll
o
ws:
s
ec
tio
n
2
d
is
cu
s
s
es
s
ev
er
al
r
elev
an
t
wo
r
k
s
th
at
a
r
e
r
e
lated
to
t
h
is
wo
r
k
.
Sectio
n
3
d
escr
ib
es
th
e
p
r
o
p
o
s
ed
m
eth
o
d
in
d
etail.
Sectio
n
4
p
r
esen
ts
th
e
o
u
tco
m
es
o
f
th
e
ex
p
er
im
en
ts
an
d
ev
alu
atio
n
s
.
Sectio
n
5
c
o
n
clu
d
es th
e
ar
ticle
an
d
ad
d
r
ess
es f
u
tu
r
e
wo
r
k
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
7
7
6
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
,
Vo
l.
1
4
,
No
.
1
,
A
p
r
il
20
2
5
:
216
-
22
8
218
Fig
u
r
e
1
.
Srv
y
cite
a
r
ch
itectu
r
e
2.
RE
L
AT
E
D
WO
RK
Sear
ch
en
g
in
es
f
o
r
ac
ad
em
ic
l
iter
atu
r
e
s
u
ch
as
Go
o
g
le
Sch
o
lar
,
Sem
an
tic
Sch
o
lar
,
an
d
Scien
ce
Gate
ar
e
u
s
ed
to
co
m
p
ile
th
e
ar
ti
cles
r
eq
u
ir
ed
f
o
r
th
e
liter
atu
r
e
r
ev
iew.
"Res
ea
r
ch
p
ap
er
r
ec
o
m
m
en
d
atio
n
"
,
"r
esear
ch
ar
ticle
r
ec
o
m
m
e
n
d
atio
n
"
,
"scien
tific
ar
ticle
r
ec
o
m
m
en
d
atio
n
"
,
an
d
"
s
cien
tific
p
ap
er
r
ec
o
m
m
en
d
atio
n
"
ar
e
u
s
ed
as
s
ea
r
ch
ter
m
s
.
Nea
r
ly
2
,
4
2
9
ite
m
s
o
v
er
th
e
p
r
ev
io
u
s
1
0
y
ea
r
s
wer
e
id
en
tifie
d
to
f
it
th
e
s
ea
r
ch
,
an
d
th
e
m
o
s
t
p
er
tin
en
t
an
d
im
p
o
r
ta
n
t
ar
ticles
f
r
o
m
th
is
g
r
o
u
p
wer
e
co
m
p
iled
an
d
u
s
ed
f
o
r
th
is
wo
r
k
.
Fig
u
r
e
2
a
r
ticles
p
u
b
lic
atio
n
co
u
n
t
p
er
y
ea
r
r
ep
r
esen
t
s
th
e
y
ea
r
ly
c
o
u
n
t
o
f
ar
ticles
p
u
b
lis
h
ed
f
r
o
m
2
0
1
4
to
2
0
2
3
.
A
h
y
b
r
id
r
ec
o
m
m
e
n
d
atio
n
s
y
s
tem
was
cr
ea
ted
in
th
e
wo
r
k
o
f
[
1
0
]
.
T
h
r
ee
s
ep
ar
ate
ap
p
r
o
ac
h
es
d
em
o
g
r
a
p
h
ic
-
b
ased
,
r
elatio
n
s
h
ip
n
etwo
r
k
-
b
ased
,
an
d
co
n
ten
t
-
b
ased
m
a
k
e
u
p
th
is
h
y
b
r
id
r
ec
o
m
m
en
d
atio
n
s
y
s
tem
.
I
n
th
is
s
tu
d
y
,
th
e
au
t
h
o
r
s
test
ed
th
e
r
ec
o
m
m
e
n
d
ati
o
n
s
y
s
tem
u
s
in
g
a
d
ata
s
et
f
r
o
m
R
esear
ch
Gate
.
L
aten
t
f
ac
to
r
s
wer
e
in
teg
r
ated
with
c
o
n
ten
t
-
b
ased
f
ilter
in
g
to
p
r
o
p
o
s
e
ar
ticles
to
r
esear
ch
er
s
in
t
h
e
r
ec
e
n
t
wo
r
k
b
y
[
1
1
]
.
T
h
e
f
ir
s
t
s
tag
e
in
c
r
ea
tin
g
a
n
a
p
p
r
o
p
r
iate
s
u
b
ject
f
o
r
th
e
in
q
u
ir
y
is
to
p
i
c
m
o
d
ellin
g
.
L
ate
n
t
d
ir
ich
let
allo
ca
tio
n
(
L
DA)
an
d
n
o
n
-
n
e
g
ativ
e
m
atr
ix
f
ac
to
r
i
za
tio
n
(
NM
F),
two
tech
n
iq
u
e
s
,
ar
e
em
p
lo
y
ed
in
to
p
ic
m
o
d
ellin
g
.
A
lis
t
o
f
ar
ti
cl
es
th
at
ar
e
s
im
ilar
is
th
en
s
u
g
g
ested
b
ased
o
n
th
e
s
em
an
ti
c.
Had
h
iatm
a
et
a
l.
[1
2
]
p
u
t
f
o
r
t
h
th
e
PP
R
_
T
C
f
r
am
ewo
r
k
,
wh
ich
u
s
es
a
m
o
d
i
f
ied
Pag
eRan
k
alg
o
r
ith
m
to
f
ac
ilit
ate
th
e
s
elec
tio
n
o
f
m
u
lti
-
to
p
ic
co
m
m
u
n
ities
.
T
h
e
in
itial
s
ec
tio
n
o
f
th
e
f
r
am
ewo
r
k
b
eg
in
s
with
th
e
ex
tr
ac
tio
n
o
f
s
ig
n
if
ican
t
ch
ar
ac
ter
is
tics
f
r
o
m
th
e
ac
a
d
em
ic
citatio
n
n
etwo
r
k
,
wh
ich
is
f
o
llo
wed
b
y
th
e
i
d
en
tific
a
tio
n
o
f
m
u
lti
-
to
p
ic
co
m
m
u
n
ities
th
at
ar
e
r
elev
an
t
to
th
e
q
u
er
y
.
Fin
ally
,
r
a
n
k
in
g
is
ca
lcu
lated
b
y
a
m
o
d
i
f
ied
P
ag
eRan
k
alg
o
r
ith
m
to
s
u
g
g
est
r
elev
an
t
ar
ticles
th
at
f
it
th
e
co
m
m
u
n
ities
.
A
d
ee
p
lear
n
in
g
-
b
ased
h
y
b
r
id
ar
ticl
e
r
ec
o
m
m
en
d
ati
o
n
s
y
s
tem
was
p
r
o
p
o
s
ed
b
y
th
e
a
ca
d
em
ics
[
1
3
]
.
T
h
is
ap
p
r
o
ac
h
g
r
o
u
p
s
p
ap
e
r
s
f
r
o
m
s
ev
er
al
d
i
s
cip
lin
es,
in
clu
d
in
g
co
m
p
u
ter
s
cien
ce
,
ec
o
n
o
m
ics,
an
d
m
ed
icin
e,
b
y
c
o
m
b
in
in
g
d
o
cu
m
e
n
t
s
im
ilar
ity
,
h
ier
ar
ch
i
ca
l
clu
s
ter
in
g
,
an
d
k
ey
wo
r
d
ex
tr
ac
tio
n
.
A
d
d
itio
n
ally
,
it
h
as
th
e
ab
ilit
y
to
p
r
o
v
id
e
co
n
s
u
m
er
s
with
item
s
th
at
ar
e
q
u
ite
s
im
ilar
in
ter
m
s
o
f
s
em
an
tics
.
T
h
e
r
esear
ch
er
s
em
p
l
o
y
ed
th
e
u
s
er
s
'
b
r
o
wsi
n
g
a
n
d
p
er
s
o
n
al
in
f
o
r
m
ati
o
n
in
th
e
s
tu
d
y
p
r
o
v
id
e
d
b
y
[
1
4
]
to
m
ak
e
a
r
ticle
r
ec
o
m
m
en
d
atio
n
s
.
T
h
e
d
e
n
s
ity
-
b
ased
s
p
atial
clu
s
ter
in
g
o
f
ap
p
licatio
n
s
with
n
o
is
e
(
DB
S
C
AN
)
clu
s
ter
in
g
m
eth
o
d
is
u
s
ed
to
g
r
o
u
p
p
er
ti
n
en
t
ar
ticles,
f
o
llo
wed
b
y
th
e
f
ir
ef
ly
alg
o
r
ith
m
f
o
r
d
ata
o
p
tim
is
atio
n
a
n
d
th
e
g
e
n
e
tic
alg
o
r
ith
m
f
o
r
d
ata
p
r
ed
icti
o
n
.
Fin
ally
,
u
s
er
s
ar
e
r
ec
o
m
m
en
d
ed
a
r
ticles
u
s
in
g
a
r
ec
o
m
m
en
d
ato
r
y
s
y
s
tem
b
ased
o
n
p
ar
ticip
ato
r
y
f
ilt
er
in
g
.
C
h
au
d
h
u
r
i
et
a
l.
[1
5
]
in
tr
o
d
u
ce
d
s
y
s
tem
atic
h
id
d
en
attr
ib
u
te
-
b
ased
r
ec
o
m
m
en
d
atio
n
en
g
in
e
(
SHAR
E
)
,
a
m
u
lti
-
cr
iter
ia
-
b
ased
p
er
s
o
n
a
lized
r
esear
ch
p
ap
er
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
I
SS
N:
2252
-
8
7
7
6
S
r
vy
cite:
a
h
yb
r
id
s
cien
tifi
c
a
r
ticle
r
ec
o
mme
n
d
a
tio
n
s
ystem
(
S
iva
s
a
n
ka
r
i R
)
219
r
ec
o
m
m
en
d
atio
n
s
y
s
tem
th
at
u
s
ed
m
an
y
asp
ec
ts
to
d
is
co
v
er
im
p
o
r
tan
t
in
s
ig
h
ts
to
u
n
iq
u
ely
r
ep
r
esen
t
pu
b
licatio
n
s
.
Au
th
o
r
s
also
ta
k
es
in
to
ac
co
u
n
t
th
e
d
esire
o
f
u
s
er
in
o
r
d
er
to
ca
p
t
u
r
e
th
e
d
y
n
am
ic
n
o
tio
n
o
f
u
s
er
s
.
SHAR
E
u
s
ed
h
y
b
r
id
r
an
k
in
g
m
eth
o
d
with
t
h
e
m
u
lti
-
cr
iter
ia
d
ec
is
io
n
a
n
aly
s
is
(
MCDA)
an
d
s
u
p
p
o
r
t
v
ec
to
r
m
ac
h
in
e
(
SVM
)
r
a
n
k
alg
o
r
ith
m
s
.
I
n
th
is
ca
s
e
,
MC
DA
is
u
s
ed
to
r
an
k
p
ap
er
s
b
y
an
aly
zin
g
m
u
ltip
le
ch
ar
ac
ter
is
tics
,
wh
er
ea
s
SVM
r
an
k
is
u
s
ed
to
r
ate
ar
ticles
d
ep
en
d
in
g
o
n
u
s
er
p
r
e
f
er
en
ce
s
.
SHAR
E
em
p
lo
y
ed
a
r
an
k
a
g
g
r
e
g
atio
n
m
eth
o
d
to
g
en
er
ate
cu
s
to
m
ized
lis
t o
f
r
an
k
in
g
s
.
Fig
u
r
e
2
.
Ar
ticles
p
u
b
licat
io
n
co
u
n
t p
e
r
y
ea
r
T
h
e
HNT
A
ap
p
r
o
ac
h
f
o
r
ac
a
d
em
ic
ar
ticle
r
ec
o
m
m
en
d
atio
n
was
p
u
t
o
u
t
b
y
[
1
6
]
.
T
h
is
tech
n
iq
u
e
s
u
g
g
ests
a
s
tr
ateg
y
b
y
f
u
s
in
g
tem
p
o
r
al
f
ea
tu
r
es
f
r
o
m
th
e
SC
HOL
AT
d
ataset
with
h
eter
o
g
en
eo
u
s
n
etwo
r
k
s
.
R
esear
ch
er
s
in
th
is
ca
s
e
co
m
b
in
ed
u
s
er
r
ese
ar
ch
i
n
ter
ests
wi
th
co
n
ten
t
in
f
o
r
m
atio
n
f
r
o
m
t
h
e
p
ap
er
s
.
I
n
o
r
d
er
to
d
eter
m
in
e
c
o
m
m
o
n
alities
,
th
e
u
s
er
s
'
r
esear
ch
in
ter
ests
ar
e
f
u
r
th
e
r
d
iv
id
e
d
in
to
th
o
s
e
th
at
ar
e
im
m
ed
iate
a
n
d
th
o
s
e
th
at
a
r
e
o
n
g
o
in
g
.
I
n
o
r
d
er
to
p
r
o
p
o
s
e
p
ap
er
s
to
s
ch
o
lar
s
,
all
th
e
esti
m
a
ted
v
ar
ia
b
les
ar
e
weig
h
ted
.
A
g
r
ap
h
ical
f
r
a
m
ewo
r
k
f
o
r
cit
atio
n
r
elatio
n
s
h
ip
s
was
cr
ea
ted
b
y
[
1
7
]
as
a
h
y
b
r
i
d
r
ec
o
m
m
en
d
atio
n
m
o
d
el
th
at
in
co
r
p
o
r
ates
citatio
n
an
d
co
n
t
en
t
-
b
ased
tech
n
iq
u
es.
W
eig
h
ts
ar
e
allo
ca
ted
f
o
r
th
e
citatio
n
o
f
th
e
p
ap
e
r
b
ased
o
n
th
e
s
ec
tio
n
in
wh
ich
it is
ap
p
lied
.
T
h
e
co
n
te
n
t
-
b
ased
s
im
ilar
ity
alg
o
r
ith
m
is
p
air
ed
with
th
is
citatio
n
weig
h
t
to
g
et
th
e
f
in
al
lis
t
o
f
r
ec
o
m
m
en
d
ed
p
ap
er
s
.
L
iu
et
a
l.
[1
8
]
b
u
ild
an
u
n
d
ir
ec
ted
p
ap
er
citatio
n
g
r
ap
h
f
r
o
m
u
s
er
s
’
q
u
er
ies.
T
h
ey
m
o
d
elle
d
th
e
p
r
o
b
lem
as
a
Stein
er
tr
e
e
p
r
o
b
lem
,
co
n
s
tr
u
cted
a
tr
ee
,
an
d
r
ec
o
m
m
e
n
d
ed
th
e
p
ap
er
s
.
T
h
e
a
u
th
o
r
s
c
o
n
s
tr
u
cted
th
e
g
r
a
p
h
b
y
s
ee
k
in
g
a
ll
th
e
k
ey
wo
r
d
s
f
r
o
m
th
e
u
s
er
s
’
q
u
er
y
,
an
d
th
en
ar
ticles
th
at
m
atch
ea
c
h
wo
r
d
ar
e
r
etr
iev
ed
.
Af
ter
r
etr
iev
i
n
g
an
in
d
i
v
id
u
al
lis
t
o
f
r
el
at
ed
p
ap
e
r
s
f
o
r
ea
ch
k
ey
wo
r
d
,
a
r
elatio
n
b
etwe
en
a
ll
th
e
k
ey
wo
r
d
s
ets
is
f
o
r
m
ed
as
a
s
u
b
g
r
ap
h
b
ased
o
n
th
e
s
im
ilar
ity
b
etwe
en
th
e
p
ap
er
s
.
B
ased
o
n
th
e
f
in
al
St
ein
er
tr
ee
g
e
n
er
ated
af
ter
m
er
g
in
g
s
elec
ted
tr
ee
s
,
r
ec
o
m
m
e
n
d
atio
n
s
a
r
e
m
a
d
e.
T
ao
et
a
l.
[
19
]
p
r
es
en
ted
a
n
a
r
ticle
to
o
f
f
er
a
lis
t
o
f
r
e
f
er
en
ce
s
f
o
r
f
u
r
th
er
s
tu
d
y
.
W
o
r
d
2
d
ec
is
also
u
s
ed
f
o
r
to
p
ic
v
ec
to
r
izatio
n
,
wh
ile
L
DA
is
u
s
ed
to
ex
tr
ac
t
k
ey
wo
r
d
s
f
r
o
m
s
u
b
jects
an
d
p
ap
er
s
.
Pa
p
er
v
ec
to
r
izatio
n
is
r
ep
r
esen
ted
b
y
Do
c2
V
ec
.
T
h
e
d
eg
r
ee
o
f
s
im
ilar
ity
b
etwe
en
to
p
ics
an
d
p
ap
e
r
s
is
d
eter
m
i
n
ed
u
s
in
g
th
ese
two
v
ec
to
r
s
.
A
lis
t
o
f
N
p
u
b
licatio
n
s
is
s
u
g
g
ested
to
th
e
u
s
er
as
a
r
eso
u
r
ce
b
ased
o
n
th
e
f
in
al
s
im
ilar
ity
v
alu
es.
B
y
co
m
p
ar
in
g
th
e
s
im
ilar
ity
o
f
t
wo
ar
ticles,
Du
et
a
l.
[2
0
]
cr
ea
ted
a
h
eter
o
g
e
n
eo
u
s
in
f
o
r
m
atio
n
n
etwo
r
k
.
Fo
r
n
etwo
r
k
co
n
s
tr
u
ctio
n
,
f
ac
to
r
s
s
u
ch
as
ar
ticle
co
n
ten
t,
citatio
n
s
,
p
ap
er
ar
ea
,
an
d
co
-
au
th
o
r
s
h
ip
ar
e
tak
en
in
to
ac
co
u
n
t.
T
h
e
au
th
o
r
s
cr
ea
te
w
o
r
d
s
eq
u
e
n
ce
s
u
s
in
g
n
atu
r
al
l
an
g
u
ag
e
m
o
d
els,
an
d
th
ey
em
p
lo
y
r
a
n
d
o
m
walk
s
to
m
o
v
e
o
v
er
t
h
e
n
etwo
r
k
.
Ma
et
a
l.
[2
1
]
,
b
u
ilt
a
h
eter
o
g
en
eo
u
s
b
ib
lio
g
r
a
p
h
ic
n
etw
o
r
k
.
T
h
e
s
u
g
g
ested
ap
p
r
o
ac
h
b
u
ild
s
th
e
n
etwo
r
k
b
y
em
p
lo
y
in
g
th
e
we
ig
h
ts
o
f
t
o
p
o
lo
g
ical
ch
a
r
ac
ter
i
s
tics
,
m
eta
-
p
ath
s
,
an
d
m
eta
-
g
r
ap
h
s
to
g
eth
e
r
with
th
em
.
T
h
e
W
-
r
an
k
m
eth
o
d
is
p
r
o
p
o
s
ed
b
y
th
e
r
esear
ch
er
s
in
[2
2
]
.
T
h
e
W
-
r
an
k
alg
o
r
ith
m
r
an
k
s
ar
ticles
b
y
weig
h
tin
g
th
e
lin
k
a
g
es
in
th
e
citatio
n
a
n
d
au
th
o
r
s
h
ip
n
et
wo
r
k
s
.
T
o
ca
lcu
late
th
e
citat
io
n
,
n
etwo
r
k
-
an
d
s
em
an
tic
-
b
ased
alg
o
r
ith
m
s
ar
e
co
u
p
led
t
o
p
r
o
p
o
s
e
th
e
m
o
s
t
ap
p
r
o
p
r
iate
citatio
n
.
Kan
ak
ia
et
a
l.
[2
3
]
cr
ea
te
d
a
h
y
b
r
id
r
ec
o
m
m
en
d
e
r
s
y
s
tem
th
at
co
m
b
in
es
b
o
th
co
n
ten
t
-
b
ased
(
C
B
)
an
d
co
-
citatio
n
-
b
ased
(
C
C
B
)
s
u
g
g
esti
o
n
s
to
g
en
er
ate
a
f
in
al
s
tatic
li
s
t o
f
r
ec
o
m
m
en
d
atio
n
s
f
o
r
ea
ch
ar
ticle
in
th
e
MA
G
.
T
h
is
m
eth
o
d
u
s
es a
weig
h
ted
m
ix
e
d
h
y
b
r
i
d
izatio
n
with
a
cu
s
to
m
izab
le
weig
h
tin
g
f
u
n
ctio
n
to
d
ir
e
ctly
co
n
tr
ast
ea
ch
r
ec
o
m
m
en
d
atio
n
p
air
f
r
o
m
e
ith
er
lis
t
an
d
m
er
g
e
b
o
th
lis
ts
to
g
et
a
f
in
al
p
ap
er
r
ec
o
m
m
en
d
atio
n
lis
t
f
o
r
p
r
ac
tically
ev
er
y
p
ap
er
in
th
e
g
r
ap
h
.
So
n
an
d
Kim
[2
4
]
p
r
esen
ted
an
ac
ad
e
m
ic
p
a
p
er
r
ec
o
m
m
en
d
atio
n
s
y
s
tem
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
7
7
6
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
,
Vo
l.
1
4
,
No
.
1
,
A
p
r
il
20
2
5
:
216
-
22
8
220
th
a
t
co
m
b
in
es
citatio
n
an
aly
s
is
an
d
n
etwo
r
k
an
aly
s
is
.
I
t
is
b
ased
o
n
m
u
ltil
ev
el
citatio
n
n
etwo
r
k
s
th
at
co
m
p
ar
e
all
in
d
ir
ec
tly
lin
k
ed
p
ap
e
r
s
to
th
e
p
ap
er
o
f
i
n
ter
est
to
in
s
p
ec
t
th
e
s
tr
u
ctu
r
al
an
d
s
em
an
tic
r
e
latio
n
s
h
ip
s
am
o
n
g
th
em
.
T
h
e
au
th
o
r
s
co
n
s
id
er
ed
th
e
m
u
tu
al
r
elatio
n
s
h
ip
s
am
o
n
g
th
e
p
ap
er
s
in
a
b
r
o
ad
n
etwo
r
k
b
ey
o
n
d
a
s
in
g
le
lev
el
an
d
e
v
alu
ated
th
e
s
ig
n
if
ican
ce
o
f
ea
ch
p
ap
er
th
r
o
u
g
h
ce
r
tain
ce
n
tr
ality
m
ea
s
u
r
es.
I
t
was
co
n
clu
d
e
d
to
o
u
tp
er
f
o
r
m
o
th
er
e
x
is
tin
g
m
o
d
els
b
ased
o
n
u
s
er
s
atis
f
ac
tio
n
d
ata.
Har
u
n
a
et
a
l.
[2
5
]
ex
p
lain
e
d
h
o
w
a
s
cien
tific
p
ap
er
d
ataset
is
co
n
v
er
ted
in
to
an
ac
ad
em
ic
citatio
n
n
etwo
r
k
b
y
co
n
v
er
tin
g
it
in
to
a
n
ac
a
d
em
ic
citatio
n
n
etwo
r
k
,
id
en
tif
y
in
g
co
m
m
u
n
ities
with
in
th
at
n
etwo
r
k
,
a
n
d
i
d
en
tify
in
g
m
u
lti
-
to
p
ic
co
m
m
u
n
ities
an
d
p
ap
er
s
b
y
u
s
in
g
th
e
L
DA
to
p
ic
m
o
d
el,
wh
ic
h
is
a
to
p
ic
m
o
d
el
th
at
is
b
ased
o
n
th
e
L
DA
co
n
ce
p
t.
T
h
e
y
em
p
lo
y
ed
th
is
tech
n
iq
u
e
to
s
elec
t
an
d
co
m
b
in
e
th
e
m
o
s
t
p
e
r
tin
en
t
n
-
to
p
ic
co
m
m
u
n
ities
f
r
o
m
a
s
et
o
f
m
u
lti
-
to
p
ic
co
m
m
u
n
ities
.
Mo
d
if
ied
Pag
eRan
k
,
PP
R
_
T
C
,
is
em
p
lo
y
ed
to
cr
ea
te
a
b
ias
p
r
o
b
ab
ilit
y
v
ec
to
r
a
n
d
tr
an
s
itio
n
p
r
o
b
a
b
ilit
y
m
atr
ix
f
o
r
to
p
ic
q
u
er
ies an
d
m
u
lti
-
to
p
ic
co
m
m
u
n
ities
.
I
n
th
e
wo
r
k
[
2
6
]
,
a
r
esear
ch
p
a
p
er
r
ec
o
m
m
en
d
atio
n
s
y
s
tem
th
at
co
m
b
in
es
a
cr
o
s
s
-
cr
awlin
g
-
b
a
s
ed
d
o
cu
m
en
t
g
ath
e
r
in
g
tech
n
iq
u
e
with
a
p
ap
er
co
n
n
ec
tio
n
an
al
y
ze
r
.
T
r
ad
itio
n
al
k
ey
wo
r
d
an
al
y
s
is
an
d
cit
atio
n
n
etwo
r
k
in
g
a
n
aly
s
is
ar
e
co
m
b
in
ed
in
th
e
s
u
g
g
ested
p
ap
er
r
elatio
n
s
h
ip
an
aly
ze
r
.
A
cr
o
s
s
-
cr
awlin
g
e
n
g
in
e
was
cr
ea
ted
to
co
llect
r
el
ev
an
t
ar
ticles
f
r
o
m
v
ar
io
u
s
d
ig
ital
lib
r
ar
ies,
wh
ile
a
p
ap
er
citatio
n
n
etwo
r
k
a
n
aly
ze
r
was
b
u
ilt
to
id
en
tify
t
h
e
r
elev
an
ce
o
f
p
ap
er
s
b
ased
o
n
tex
tu
al
a
n
d
p
ap
er
ci
tatio
n
n
etwo
r
k
an
al
y
s
is
.
Sh
ar
m
a
et
a
l.
[2
7
]
d
ev
elo
p
e
d
a
m
eth
o
d
f
o
r
en
co
d
in
g
ea
ch
d
o
cu
m
en
t
in
ter
m
s
o
f
its
co
n
ce
p
t
o
r
s
em
an
tics
.
T
h
ey
ac
co
m
p
lis
h
ed
th
is
b
y
b
u
i
ld
in
g
a
d
is
tr
ib
u
ted
r
ep
r
esen
tatio
n
in
h
ig
h
-
d
im
en
s
io
n
al
s
p
ac
e.
T
h
is
in
d
iv
id
u
al
r
ep
r
esen
tatio
n
is
u
s
ed
to
d
is
co
v
er
item
s
th
at
m
ee
t
th
e
n
ee
d
s
o
f
th
e
u
s
er
.
T
h
e
en
tire
r
ec
o
m
m
en
d
atio
n
p
r
o
ce
d
u
r
e
is
d
iv
id
ed
in
to
two
s
tag
es:
v
ec
to
r
cr
ea
tio
n
f
o
r
ca
n
d
id
ate
p
a
p
er
s
an
d
a
r
ec
o
m
m
en
d
atio
n
al
g
o
r
ith
m
.
Dh
an
d
a
an
d
Ver
m
a
[2
8
]
s
u
g
g
ested
a
p
er
s
o
n
alize
d
r
esear
ch
ar
ticle
r
ec
o
m
m
en
d
atio
n
s
y
s
te
m
th
at
u
s
es
an
in
cr
em
en
tal
r
esear
ch
p
ap
er
r
ep
o
s
ito
r
y
an
d
th
e
in
cr
em
e
n
tal
h
ig
h
-
u
tili
ty
item
s
et
m
in
in
g
ap
p
r
o
ac
h
ef
f
icien
t
in
cr
em
en
tal
h
ig
h
-
u
tili
ty
item
s
et
m
in
in
g
alg
o
r
ith
m
(
E
I
HI
)
.
T
h
e
tech
n
iq
u
e
p
er
f
o
r
m
s
th
e
s
u
g
g
esti
o
n
p
r
o
ce
s
s
in
two
s
tep
s
.
First,
ar
ticles
ar
e
s
e
lecte
d
f
o
r
th
e
r
esear
ch
e
r
'
s
ar
ea
o
f
in
ter
est
u
s
in
g
th
e
p
r
o
b
ab
ili
s
tic
laten
t
s
em
an
tic
an
aly
s
is
(
PLSA
)
alg
o
r
ith
m
,
wh
ich
ca
teg
o
r
izes
r
esear
ch
p
ap
e
r
s
in
to
'
k
'
c
lu
s
ter
s
,
an
d
th
en
v
a
lu
ab
le
p
u
b
licatio
n
s
ar
e
r
ec
o
m
m
en
d
ed
to
th
e
r
esear
ch
er
u
s
in
g
E
I
HI
to
m
in
e
d
y
n
a
m
ic
d
atasets
.
Xia
et
a
l.
[
29
]
ar
e
th
e
au
th
o
r
s
o
f
th
e
co
m
m
o
n
a
u
th
o
r
r
elatio
n
-
b
ased
r
ec
o
m
m
e
n
d
atio
n
tec
h
n
iq
u
e
(
C
AR
E
)
.
R
e
s
ea
r
ch
er
s
f
r
eq
u
en
tly
s
ee
k
o
u
t
wo
r
k
s
b
y
th
e
s
am
e
au
th
o
r
s
,
b
u
t
n
o
t
all
r
esear
ch
er
s
h
av
e
au
th
o
r
-
b
a
s
ed
s
ea
r
ch
h
ab
its
.
T
h
ese
two
f
ac
ts
f
o
r
m
t
h
e
b
asis
o
f
th
e
C
AR
E
tech
n
iq
u
e.
I
t
h
a
s
a
r
esear
ch
er
s
elec
tio
n
m
o
d
u
l
e
an
d
a
g
r
ap
h
-
b
ased
a
r
ticle
r
a
n
k
in
g
m
o
d
u
le.
T
h
e
f
ir
s
t
m
o
d
u
le
g
ath
er
s
d
ata
f
r
o
m
p
r
io
r
p
r
ef
e
r
en
ce
s
an
d
s
u
g
g
est
s
th
at
ac
ad
em
ics
wi
th
au
th
o
r
-
b
ased
s
ea
r
ch
h
ab
its
ar
e
its
tar
g
et
au
d
ien
ce
.
T
o
cr
ea
te
ar
ticle
r
atin
g
lis
ts
,
th
e
s
ec
o
n
d
m
o
d
u
le
u
s
es
a
g
r
ap
h
-
b
ased
r
an
d
o
m
walk
tech
n
iq
u
e.
Am
am
i
et
a
l.
[3
0
]
s
u
g
g
ested
an
L
DA
-
b
ased
ap
p
r
o
ac
h
f
o
r
r
ec
o
m
m
en
d
in
g
s
ch
o
lar
ly
ar
ticles.
Acc
o
r
d
in
g
to
th
is
s
tr
ateg
y
,
th
e
co
n
ten
t
th
at
th
e
r
esear
ch
er
s
elf
-
g
en
er
ated
s
h
o
u
ld
s
er
v
e
a
s
th
e
f
o
u
n
d
atio
n
f
o
r
th
e
cr
ea
tio
n
o
f
th
e
r
esear
ch
e
r
p
r
o
f
ile
b
ec
au
s
e
it
r
ev
ea
ls
th
e
t
o
p
ics
in
w
h
ich
t
h
e
r
esear
c
h
er
i
s
m
o
s
t
in
ter
ested
as
well
as
th
e
s
p
ec
ialized
la
n
g
u
ag
e
th
at
r
esear
ch
er
s
e
m
p
lo
y
wh
en
th
e
y
wr
ite
ar
ticles.
T
h
e
th
em
es
th
e
L
DA
alg
o
r
ith
m
e
x
tr
ac
ted
f
r
o
m
t
h
e
r
esear
ch
er
'
s
ea
r
lier
p
u
b
licati
o
n
s
ar
e
th
e
n
co
m
b
in
ed
to
cr
ea
te
th
e
r
esear
ch
er
p
r
o
f
ile.
An
o
f
f
icial
lan
g
u
ag
e
m
o
d
el
r
ep
r
esen
tatio
n
o
f
th
e
u
n
r
ea
d
ar
ticle
is
p
r
o
v
id
e
d
,
an
d
it
is
co
m
p
ar
ed
with
ea
ch
to
p
ic
d
ef
in
in
g
th
e
r
esear
ch
er
p
r
o
f
ile
in
o
r
d
er
to
d
eter
m
in
e
wh
eth
er
th
e
item
wo
u
ld
b
e
o
f
in
ter
est
to
th
e
r
esear
ch
er
.
C
h
ak
r
a
b
o
r
ty
et
a
l.
[3
1
]
d
ev
elo
p
ed
th
e
f
ac
ete
d
r
ec
o
m
m
e
n
d
atio
n
s
y
s
tem
f
o
r
s
cien
tific
ar
ticles
(
FeR
o
SA)
.
FeR
o
SA
i
s
d
esig
n
ed
to
p
r
o
p
er
l
y
ca
ter
to
t
h
e
en
d
u
s
er
'
s
ap
p
r
o
p
r
iate
k
n
o
wled
g
e
co
n
tex
t
b
y
n
o
t o
n
ly
r
ec
o
m
m
en
d
in
g
r
elev
an
t
s
cien
t
if
ic
p
ap
er
s
b
u
t
also
o
r
g
an
izin
g
th
e
r
ec
o
m
m
en
d
atio
n
s
in
to
f
ac
ets
g
iv
en
a
q
u
er
y
p
ap
er
.
T
h
is
ap
p
r
o
ac
h
aim
s
to
s
im
u
l
ate
th
e
tr
av
er
s
al
m
ec
h
an
is
m
o
f
a
u
s
er
s
tar
tin
g
f
r
o
m
a
k
n
o
wn
ar
ticle
u
s
in
g
a
p
r
in
cip
led
f
r
am
ewo
r
k
o
f
r
an
d
o
m
walk
s
with
r
estar
ts
.
T
o
s
y
s
tem
atica
lly
g
en
er
ate
th
e
m
o
s
t
r
elev
an
t
r
esu
lts
,
th
e
m
o
d
el
c
o
n
s
id
er
s
b
o
th
th
e
cit
atio
n
lin
k
s
a
n
d
t
h
e
co
n
ten
t
d
ata.
FeR
o
SA
d
iv
id
es
th
e
s
u
g
g
esti
o
n
s
in
to
f
o
u
r
n
atu
r
ally
o
cc
u
r
r
in
g
ca
teg
o
r
ies,
in
clu
d
in
g
b
ac
k
g
r
o
u
n
d
,
alter
n
a
tiv
e
ap
p
r
o
ac
h
es,
m
eth
o
d
s
,
an
d
co
m
p
ar
is
o
n
.
3.
M
E
T
H
O
D
3
.
1
.
P
r
o
blem
f
o
rm
ula
t
io
n
T
h
e
task
o
f
th
e
r
ec
o
m
m
en
d
atio
n
s
y
s
tem
is
to
p
r
ed
ict
th
e
lis
t o
f
ar
ticles
th
at
m
atch
th
e
u
s
er
q
u
er
y
an
d
to
r
ec
o
m
m
e
n
d
th
e
m
to
th
e
r
esear
ch
er
s
.
L
et
u
s
co
n
s
id
er
an
a
r
ticle
a
∈
A,
wh
er
e
A
is
th
e
s
et
o
f
all
ar
ticles
an
d
th
e
s
ea
r
ch
q
u
er
y
o
f
a
r
esear
ch
er
is
r
ep
r
esen
ted
as SQ.
SQr
=
{a
1
,
a2
,
…an
}
r
ep
r
esen
ts
th
e
s
et
o
f
all
ar
ticles
th
is
s
y
s
tem
r
ec
o
m
m
en
d
s
as
s
im
ilar
ar
ticles
b
y
r
elatin
g
th
e
SQ
o
f
a
r
esear
ch
er
r
.
T
o
r
ec
o
m
m
en
d
th
e
ar
ticle
s
in
SQr
,
th
is
s
y
s
tem
r
eq
u
ir
es
two
d
if
f
er
e
n
t
g
r
a
p
h
n
etwo
r
k
s
:
th
e
SC
N
an
d
th
e
C
N.
An
a
r
ticle
s
c
∈
SC
N,
wh
er
e
SC
N
i
s
th
e
s
et
o
f
s
u
r
v
ey
a
r
ticles,
an
d
cn
∈
C
N,
wh
er
e
C
N
is
th
e
s
et
o
f
n
o
r
m
al
ar
ticles
o
th
er
th
an
s
u
r
v
e
y
s
.
An
ar
ticle
s
c
is
s
a
id
to
b
e
an
elem
en
t o
f
SC
N
if
an
d
o
n
ly
if
s
c
∈
A;
s
im
ilar
ly
,
an
ar
ticle
cn
is
s
aid
to
b
e
an
elem
e
n
t
o
f
C
N
if
an
d
o
n
l
y
if
cn
∈
A.
SQr
is
co
m
p
u
ted
b
y
co
m
b
in
in
g
a
lis
t
o
f
th
e
to
p
r
ec
o
m
m
e
n
d
ed
a
r
ticles
f
r
o
m
SC
N
an
d
C
N
n
etwo
r
k
s
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
I
SS
N:
2252
-
8
7
7
6
S
r
vy
cite:
a
h
yb
r
id
s
cien
tifi
c
a
r
ticle
r
ec
o
mme
n
d
a
tio
n
s
ystem
(
S
iva
s
a
n
ka
r
i R
)
221
3
.
2
.
Clus
t
er
ing
I
n
o
r
d
er
to
g
r
o
u
p
tex
t
d
o
c
u
m
e
n
ts
in
t
o
m
e
an
in
g
f
u
l
cl
u
s
ter
s
b
ased
o
n
th
eir
co
n
te
n
t
an
d
s
im
i
lar
ity
,
tex
t
d
o
cu
m
e
n
t
clu
s
ter
in
g
is
a
u
s
ed
.
Du
e
to
its
ass
is
tan
ce
in
o
r
g
an
izin
g
,
s
u
m
m
ar
izi
n
g
,
a
n
d
g
ath
e
r
in
g
in
s
ig
h
ts
f
r
o
m
h
u
g
e
am
o
u
n
ts
o
f
tex
tu
al
d
ata,
it p
lay
s
a
v
ital r
o
le
in
m
an
y
n
atu
r
al
lan
g
u
ag
e
p
r
o
c
ess
in
g
(
NL
P)
an
d
in
f
o
r
m
atio
n
r
etr
iev
al
task
s
.
T
h
e
m
ain
o
b
jectiv
e
o
f
clu
s
ter
in
g
in
a
r
ec
o
m
m
en
d
atio
n
s
y
s
tem
is
to
p
u
t
s
im
ilar
d
o
cu
m
en
ts
to
g
eth
er
s
o
th
at
th
ey
ca
n
s
h
ar
e
co
m
m
o
n
th
em
es,
to
p
ics,
o
r
s
u
b
ject
m
atter
f
o
r
f
u
r
th
er
p
r
o
ce
s
s
in
g
o
f
co
n
ten
t.
T
h
e
en
tire
p
r
o
ce
s
s
o
f
clu
s
ter
in
g
ca
n
b
e
b
r
o
k
en
d
o
w
n
in
to
f
iv
e
s
tep
s
:
p
r
ep
r
o
ce
s
s
in
g
,
tex
t
r
ep
r
esen
tatio
n
o
r
em
b
ed
d
in
g
s
,
d
im
en
s
io
n
ality
r
ed
u
ctio
n
,
s
im
ilar
ity
m
ea
s
u
r
e
a
n
d
f
in
ally
cl
u
s
ter
in
g
.
3
.
2
.
1
.
T
ex
t
p
re
pro
ce
s
s
ing
T
ex
t
d
o
cu
m
en
ts
d
is
tr
ib
u
ted
v
i
a
d
ig
ital
m
e
d
ia
ar
e
u
n
p
r
o
ce
s
s
ed
an
d
ca
n
n
o
t
b
e
p
r
o
ce
s
s
ed
im
m
ed
iately
b
y
in
tellig
en
t
s
y
s
tem
s
.
T
h
e
te
x
t
d
o
c
u
m
en
t
co
n
tain
in
g
r
aw
te
x
t
d
ata
ca
n
tak
e
v
ar
i
o
u
s
f
o
r
m
s
,
s
u
ch
as
p
lain
te
x
t,
HT
ML
d
o
cu
m
en
ts
,
J
SON
f
iles
,
an
d
s
o
o
n
.
T
h
is
tex
t d
ata
als
o
in
clu
d
es
s
o
m
e
s
p
ec
ial
ch
ar
a
cter
s
,
wo
r
d
s
,
le
tter
s
,
an
d
n
u
m
b
e
r
s
th
at
a
r
e
ir
r
elev
a
n
t
to
th
e
c
o
n
tex
t
o
r
h
av
e
n
o
m
ea
n
in
g
.
T
ex
t
p
r
o
ce
s
s
in
g
is
r
eq
u
ir
ed
to
c
o
n
v
e
r
t
it
in
to
a
f
o
r
m
th
at
th
is
in
tellig
en
t
s
y
s
tem
ca
n
u
n
d
er
s
tan
d
.
T
e
x
t
p
r
o
ce
s
s
in
g
is
a
p
r
o
ce
s
s
wh
er
e
u
n
wan
ted
d
ata
f
r
o
m
r
aw
d
ata
is
r
em
o
v
e
d
to
p
r
o
ce
e
d
with
f
u
r
th
er
p
r
o
ce
s
s
in
g
.
T
ex
t
p
r
o
ce
s
s
in
g
ca
n
b
e
s
u
b
d
iv
id
ed
in
to
to
k
en
izatio
n
,
n
o
r
m
aliza
tio
n
,
s
to
p
wo
r
d
r
e
m
o
v
al
a
n
d
s
tem
m
in
g
.
T
o
k
e
n
izatio
n
a
n
d
n
o
r
m
aliza
tio
n
:
t
o
k
en
izatio
n
is
a
f
u
n
d
a
m
en
tal
tex
t
p
r
o
ce
s
s
in
g
tech
n
iq
u
e
th
a
t
en
tails
co
n
v
er
tin
g
r
aw
tex
t
f
r
o
m
tex
t
f
iles
o
r
d
ig
ital
m
ed
i
a
in
to
m
ea
n
in
g
f
u
l
u
n
its
k
n
o
wn
as
to
k
e
n
s
.
T
h
is
p
r
o
ce
s
s
d
iv
id
es
a
d
o
cu
m
en
t'
s
tex
t
in
to
d
is
cr
ete
u
n
its
b
y
in
s
er
tin
g
wh
ite
s
p
ac
es
o
r
d
is
tin
ct
m
ar
k
er
s
,
allo
win
g
ea
ch
u
n
it
to
co
r
r
esp
o
n
d
to
a
d
i
f
f
er
en
t
wo
r
d
in
th
e
tex
t.
Du
r
i
n
g
n
o
r
m
aliza
tio
n
,
tex
t
elem
en
ts
s
u
ch
as
co
d
e
tag
s
,
n
u
m
b
er
s
,
s
y
m
b
o
ls
,
s
p
ec
ial
ch
ar
ac
ter
tag
s
,
UR
L
s
,
an
d
ch
ar
ac
ter
s
th
at
d
elin
ea
te
s
ec
tio
n
s
with
in
en
tire
d
o
cu
m
e
n
ts
,
s
u
ch
as
u
n
d
er
s
c
o
r
e
(
_
)
,
at
(
@
)
,
p
er
ce
n
tag
e
(
%),
am
p
er
s
an
d
(
&
)
an
d
s
o
o
n
,
ar
e
r
em
o
v
ed
.
No
r
m
aliza
tio
n
also
elim
in
ates
s
y
m
b
o
ls
with
am
b
ig
u
o
u
s
m
e
an
in
g
s
,
n
o
n
-
E
n
g
lis
h
wo
r
d
s
,
a
n
d
p
u
n
ctu
atio
n
s
to
p
r
ep
r
o
ce
s
s
tex
t.
Sto
p
wo
r
d
r
e
m
o
v
al:
s
to
p
wo
r
d
s
ar
e
p
r
im
ar
ily
u
s
ed
to
p
r
o
v
id
e
g
r
am
m
ati
ca
l
s
tr
u
ctu
r
e
in
te
x
t
d
o
cu
m
e
n
ts
.
Sto
p
w
o
r
d
s
a
r
e
to
k
en
s
th
at
a
r
e
c
h
ar
ac
ter
ized
as
h
av
in
g
less
im
p
o
r
tan
ce
,
an
d
t
h
ey
ten
d
t
o
a
p
p
ea
r
f
r
eq
u
e
n
tly
in
th
e
d
o
c
u
m
en
t.
T
h
e
r
em
o
v
al
o
f
th
ese
s
to
p
wo
r
d
s
h
as
a
s
ig
n
if
ican
t
im
p
ac
t
o
n
t
h
e
s
y
s
tem
'
s
s
to
r
ag
e
an
d
p
r
o
ce
s
s
in
g
.
T
h
e
,
th
is
,
is
,
a
r
e,
h
e,
th
ese,
an
d
th
at
ar
e
s
o
m
e
ex
am
p
les
o
f
s
to
p
wo
r
d
s
in
E
n
g
lis
h
.
Stem
m
in
g
:
t
h
e
p
r
o
ce
s
s
o
f
r
em
o
v
in
g
all
p
r
ef
ix
es a
n
d
s
u
f
f
ix
es in
o
r
d
er
t
o
o
b
tain
r
o
o
t w
o
r
d
s
is
k
n
o
wn
as
s
tem
m
in
g
.
I
t is n
o
t
n
ec
ess
ar
y
to
ap
p
ly
s
tem
m
in
g
to
all
g
e
n
er
ated
t
o
k
en
s
.
On
ly
a
f
ew
t
o
k
en
s
,
s
u
ch
as
p
la
y
in
g
,
s
in
g
i
n
g
,
a
n
d
p
r
o
ce
s
s
in
g
,
r
eq
u
ir
e
s
tem
m
in
g
to
o
b
tain
th
e
co
r
r
esp
o
n
d
in
g
r
o
o
t w
o
r
d
s
p
lay
,
s
in
g
,
an
d
p
r
o
ce
s
s
.
3
.
2
.
2
.
T
ex
t
em
bedd
ing
s
Du
e
to
th
e
co
m
p
lex
n
atu
r
e
o
f
tex
t
to
k
en
s
,
it
is
s
till
im
p
o
s
s
ib
le
f
o
r
m
ac
h
i
n
e
lear
n
in
g
(
ML
)
alg
o
r
ith
m
s
to
u
n
d
er
s
tan
d
th
em
.
T
o
s
im
p
lify
th
is
p
r
o
b
le
m
,
tex
t
em
b
e
d
d
i
n
g
s
ar
e
u
s
ed
,
wh
ich
i
s
t
h
e
p
r
o
ce
s
s
o
f
r
ep
r
esen
tin
g
co
m
p
lex
to
k
en
s
in
to
a
f
o
r
m
th
at
is
s
im
p
le
to
u
n
d
er
s
tan
d
b
y
th
e
ML
alg
o
r
ith
m
s
.
T
h
e
co
n
v
er
ted
f
o
r
m
o
f
t
h
ese
co
m
p
lex
tex
t
to
k
en
s
in
to
a
s
im
p
lifie
d
r
ep
r
esen
tatio
n
is
ca
lled
a
v
ec
to
r
.
E
ac
h
wo
r
d
o
r
tex
t
to
k
en
is
ass
ig
n
ed
in
d
iv
id
u
al
n
u
m
e
r
ical
v
ec
t
o
r
s
.
T
r
ad
itio
n
al
m
eth
o
d
s
,
n
eu
r
a
l
n
etwo
r
k
s
with
wo
r
d
em
b
e
d
d
in
g
s
,
an
d
d
ee
p
lear
n
in
g
m
eth
o
d
s
ar
e
am
o
n
g
th
e
s
im
ilar
ity
m
ea
s
u
r
es
u
s
ed
to
co
m
p
u
te
th
e
v
ec
to
r
s
.
T
r
ad
i
tio
n
al
m
eth
o
d
s
ar
e
class
if
ied
as
eith
er
s
im
p
le
o
r
co
m
p
lex
.
On
e
-
h
o
t
e
n
co
d
in
g
,
b
ag
o
f
wo
r
d
s
(
B
o
W
)
,
ter
m
f
r
eq
u
en
cy
-
i
n
v
er
s
e
d
o
cu
m
e
n
t
f
r
eq
u
en
cy
(
TF
-
I
D
F),
an
d
wo
r
d
co
u
n
t
v
ec
to
r
s
ar
e
ex
am
p
les
o
f
s
im
p
le
tr
a
d
itio
n
al
em
b
ed
d
in
g
m
eth
o
d
s
.
Vec
to
r
s
ca
n
b
e
m
ea
s
u
r
ed
u
s
in
g
co
m
p
lex
m
eth
o
d
s
s
u
ch
as
laten
t sem
an
tic
an
aly
s
i
s
(
L
SA)
,
L
DA,
an
d
v
ec
to
r
s
p
ac
e
m
o
d
el
(
VSM)
.
W
o
r
d
2
Vec
,
Do
c2
Vec
,
an
d
Glo
v
e
ar
e
s
o
m
e
s
im
ilar
ity
m
ea
s
u
r
es
th
at
u
s
e
n
eu
r
a
l
n
etwo
r
k
s
with
wo
r
d
em
b
ed
d
i
n
g
s
.
Dee
p
lear
n
in
g
m
eth
o
d
s
s
u
ch
B
E
R
T
,
lo
n
g
-
s
h
o
r
t
ter
m
m
em
o
r
y
(
L
STM
)
,
b
id
ir
ec
tio
n
al
lo
n
g
s
h
o
r
t
-
te
r
m
m
em
o
r
y
(
B
i
-
L
STM
)
,
an
d
c
o
n
v
o
lu
ti
o
n
al
n
eu
r
al
n
etwo
r
k
(
C
NN)
ca
n
also
b
e
u
s
ed
.
W
e
h
av
e
s
elec
ted
B
E
R
T
f
o
r
o
u
r
p
r
o
p
o
s
ed
m
o
d
el
to
g
en
er
ate
tex
t
em
b
ed
d
i
n
g
.
T
h
is
m
eth
o
d
in
v
o
lv
es
u
s
in
g
co
m
p
u
ter
s
to
tr
ain
m
ac
h
in
e
lear
n
i
n
g
m
o
d
els
with
te
x
t
d
ata.
B
E
R
T
,
wh
ich
s
tan
d
s
f
o
r
B
E
R
T
,
g
ain
s
a
n
u
n
d
er
s
tan
d
i
n
g
o
f
wo
r
d
s
in
co
n
tex
t
d
u
r
i
n
g
tr
ain
i
n
g
.
Un
lik
e
s
o
m
e
m
o
d
er
n
lan
g
u
ag
e
r
e
p
r
e
s
en
tatio
n
th
eo
r
ies,
B
E
R
T
'
s
m
ain
o
b
jectiv
e
is
t
o
cr
ea
te
h
ig
h
ly
d
etailed
b
id
ir
ec
tio
n
al
r
ep
r
esen
tatio
n
s
f
r
o
m
u
n
lab
eled
tex
t.
I
t a
cc
o
m
p
lis
h
es th
is
b
y
co
n
s
id
er
in
g
b
o
th
lef
t a
n
d
r
ig
h
t
co
n
te
x
ts
th
r
o
u
g
h
o
u
t its
lay
er
s
.
3
.
2
.
3
.
Dim
ens
io
na
lity
re
du
ct
io
n
Dim
en
s
io
n
ality
r
ed
u
ctio
n
is
a
tech
n
iq
u
e
f
o
r
r
e
d
u
cin
g
th
e
n
u
m
b
er
o
f
f
ea
tu
r
es
o
r
v
ar
ia
b
les
i
n
a
d
ataset
wh
ile
r
etain
in
g
as
m
u
ch
u
s
ef
u
l
in
f
o
r
m
atio
n
as
p
o
s
s
ib
le.
T
h
is
m
eth
o
d
is
u
s
ed
in
th
is
ca
s
e
to
d
ea
l
with
h
ig
h
-
d
im
en
s
io
n
al
d
ata
in
a
r
ec
o
m
m
en
d
atio
n
s
y
s
tem
.
I
t
im
p
r
o
v
e
s
co
m
p
u
tatio
n
al
e
f
f
icien
cy
,
allev
iates
th
e
cu
r
s
e
o
f
d
im
en
s
io
n
ality
,
an
d
im
p
r
o
v
e
s
d
ata
in
ter
p
r
etab
ilit
y
.
T
h
e
m
o
s
t
co
m
m
o
n
l
y
u
s
ed
d
im
e
n
s
io
n
ality
r
ed
u
ctio
n
alg
o
r
ith
m
s
ar
e
p
r
in
cip
al
co
m
p
o
n
en
t
an
aly
s
is
(
PC
A)
,
L
DA1
,
T
-
d
is
tr
ib
u
ted
s
to
ch
asti
c
n
e
ig
h
b
o
r
em
b
e
d
d
in
g
(t
-
SNE)
,
least
ab
s
o
lu
te
s
h
r
in
k
ag
e
an
d
s
elec
tio
n
o
p
er
at
o
r
(
L
ASSO)
an
d
u
n
if
o
r
m
m
a
n
if
o
ld
ap
p
r
o
x
i
m
atio
n
an
d
p
r
o
jectio
n
(
UM
AP)
.
I
n
th
e
p
r
o
p
o
s
ed
m
eth
o
d
,
L
ASSO
is
u
s
ed
f
o
r
d
im
en
s
io
n
r
e
d
u
cti
o
n
.
T
h
e
r
ea
s
o
n
f
o
r
ch
o
o
s
in
g
L
ASSO
o
r
L
1
r
eg
u
l
ar
izatio
n
is
th
at
th
e
f
ea
tu
r
es
o
f
th
e
tex
t
d
ata
we
u
s
e
ar
e
h
ig
h
-
d
im
en
s
io
n
al.
W
h
en
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
7
7
6
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
,
Vo
l.
1
4
,
No
.
1
,
A
p
r
il
20
2
5
:
216
-
22
8
222
h
ig
h
-
d
im
e
n
s
io
n
al
d
ata
is
r
ed
u
ce
d
,
th
e
r
esu
ltan
t d
ataset
will b
ec
o
m
e
s
p
ar
s
e.
L
ASSO is
p
r
o
v
en
to
h
an
d
le
s
p
ar
s
e
d
ata
b
y
elim
in
atin
g
f
ea
tu
r
es th
at
ar
e
n
o
t im
p
o
r
tan
t
u
s
in
g
a
ze
r
o
co
ef
f
icien
t.
3
.
2
.
4
.
Sim
ila
rit
y
a
nd
dis
t
a
nc
e
m
ea
s
ures
T
h
e
s
im
ilar
ity
m
ea
s
u
r
e
ass
ess
es
th
e
s
im
ilar
ity
o
f
s
ev
er
al
ter
m
s
,
s
u
ch
as
wo
r
d
s
,
p
h
r
ases
,
d
o
cu
m
en
ts
,
o
r
co
n
ce
p
ts
.
T
h
e
g
o
al
o
f
ass
es
s
in
g
th
e
s
im
ilar
ity
b
etwe
en
tw
o
p
h
r
ases
is
to
d
eter
m
in
e
th
e
d
eg
r
ee
to
wh
ich
two
co
n
ce
p
ts
ar
e
co
n
ce
p
tu
ally
r
elate
d
.
I
t
will
n
o
t
c
o
m
p
ar
e
wo
r
d
s
th
at
ar
e
lex
ico
g
r
ap
h
ically
s
im
ilar
.
T
h
e
s
h
o
r
test
d
is
tan
ce
b
etwe
en
two
item
s
will
b
e
d
eter
m
in
ed
in
th
e
ca
s
e
o
f
th
e
d
is
tan
ce
m
ea
s
u
r
e
.
T
h
e
p
r
im
ar
y
g
o
al
o
f
t
h
ese
m
etr
ics
is
to
g
r
o
u
p
p
ap
er
s
th
at
ar
e
s
im
ilar
to
g
eth
er
.
C
o
s
in
e
s
im
ilar
ity
,
J
ac
ca
r
d
co
e
f
f
icien
t,
an
d
in
n
er
p
r
o
d
u
cts
ar
e
th
e
m
o
s
t
o
f
ten
u
s
ed
s
im
ilar
ity
m
etr
ics,
wh
er
ea
s
d
is
tan
ce
m
ea
s
u
r
em
en
ts
in
clu
d
e
E
u
clid
ea
n
d
is
tan
ce
,
Ma
n
h
attan
d
is
tan
ce
(
L
1
d
i
s
tan
ce
)
,
J
en
s
en
-
Sh
an
n
o
n
d
i
v
er
g
e
n
ce
,
an
d
h
am
m
in
g
d
is
tan
ce
.
W
e
em
p
lo
y
co
s
in
e
s
im
ilar
ity
to
m
ea
s
u
r
e
th
e
s
im
ilar
ity
b
etwe
en
th
e
v
ec
to
r
s
b
ec
au
s
e
we
h
av
e
u
s
ed
th
e
K
-
m
ea
n
s
alg
o
r
ith
m
f
o
r
cl
u
s
ter
in
g
.
T
h
e
co
s
in
e
s
im
ilar
ity
m
ea
s
u
r
e
is
v
er
y
v
e
r
s
atile
wh
en
co
m
b
in
ed
wi
th
K
-
m
ea
n
s
.
I
t a
ls
o
co
n
cl
u
d
es th
at
two
v
ec
to
r
s
ar
e
clo
s
e
if
th
e
d
is
tan
ce
ca
lcu
lated
b
etwe
en
th
em
is
s
m
all.
3
.
2
.
5
.
Clus
t
er
ing
a
lg
o
rit
hm
s
Do
cu
m
en
ts
ar
e
d
iv
i
d
ed
i
n
to
g
r
o
u
p
s
u
s
in
g
cl
u
s
ter
in
g
tec
h
n
iq
u
es.
T
h
e
g
o
al
o
f
clu
s
ter
in
g
alg
o
r
ith
m
s
is
to
p
r
o
d
u
ce
s
ep
ar
a
te
cl
u
s
ter
b
ased
o
n
th
e
d
o
cu
m
en
t
s
im
ilar
ity
.
I
n
o
th
er
w
o
r
d
s
,
p
a
p
er
s
with
in
o
n
e
clu
s
ter
s
h
o
u
ld
b
e
as
s
im
ilar
as
p
o
s
s
ib
le,
an
d
d
o
cu
m
en
ts
in
s
id
e
s
ep
a
r
ate
clu
s
ter
s
s
h
o
u
ld
b
e
as
d
is
s
im
ilar
as
p
o
s
s
ib
le.
C
lu
s
ter
in
g
alg
o
r
ith
m
s
a
r
e
b
asi
ca
lly
class
if
ied
in
to
th
r
ee
t
y
p
e
s
p
ar
titi
o
n
ed
-
b
ased
clu
s
ter
in
g
,
h
ier
ar
c
h
ical
-
b
ased
clu
s
ter
in
g
an
d
d
e
n
s
ity
-
b
ased
c
lu
s
ter
in
g
.
Fo
r
o
u
r
p
r
o
p
o
s
ed
a
p
p
r
o
ac
h
,
th
e
K
-
m
ea
n
s
alg
o
r
it
h
m
is
ch
o
s
en
f
o
r
its
s
im
p
licity
an
d
e
f
f
icien
cy
in
h
an
d
lin
g
h
ig
h
-
d
im
e
n
s
io
n
al
d
ata.
Als
o
,
th
is
is
th
e
r
o
o
t
o
f
all
o
th
er
clu
s
ter
in
g
alg
o
r
ith
m
s
.
T
h
e
p
r
im
ar
y
r
ea
s
o
n
f
o
r
s
elec
tin
g
th
is
alg
o
r
ith
m
is
th
at
th
e
d
ataset
we
u
s
ed
is
h
ig
h
-
d
im
en
s
io
n
al
an
d
th
e
n
u
m
b
er
o
f
clu
s
ter
s
is
clea
r
ly
k
n
o
w
n
.
So
,
to
r
ed
u
ce
t
h
e
co
m
p
lex
ity
o
f
o
u
r
a
p
p
r
o
ac
h
an
d
to
p
r
o
v
id
e
a
n
ac
cu
r
ate
r
esu
lt,
th
is
clu
s
ter
in
g
was c
h
o
s
en
f
o
r
th
e
s
tu
d
y
.
3
.
3
.
Cit
a
t
io
n net
wo
r
k
s
T
h
e
p
r
im
ar
y
o
b
jectiv
e
o
f
th
is
s
tu
d
y
is
to
p
r
o
p
o
s
e
ar
ticle
r
ec
o
m
m
en
d
atio
n
s
to
r
esear
ch
er
s
b
y
u
s
in
g
a
citatio
n
n
etwo
r
k
f
o
r
m
ed
with
all
s
u
r
v
ey
ar
ticles
f
o
r
th
e
s
ea
r
ch
to
p
ic.
T
o
h
a
v
e
a
b
etter
u
n
d
er
s
tan
d
in
g
o
f
th
is
wo
r
k
,
it
is
m
an
d
ato
r
y
to
k
n
o
w
ab
o
u
t
th
e
ty
p
es
o
f
r
esear
ch
.
A
r
esear
ch
ar
ticle
p
u
b
lis
h
ed
wo
u
ld
b
e:
o
r
ig
in
al
r
esear
ch
,
r
etr
o
s
p
ec
tiv
e
s
tu
d
y
,
ca
s
e
s
tu
d
y
,
m
eth
o
d
o
lo
g
ical
s
tu
d
y
,
o
p
in
i
o
n
s
tu
d
y
,
s
h
o
r
t
c
o
m
m
u
n
icatio
n
,
an
d
r
ev
iew
ar
ticles.
I
n
o
u
r
s
y
s
tem
,
we
ca
teg
o
r
ize
all
ty
p
es
o
f
r
esear
ch
ar
ticles
ex
ce
p
t
r
ev
iew
a
r
ticles
as
o
n
e
class
an
d
r
e
v
iew
ar
ticles
as
a
s
ep
ar
ate
class
.
T
h
e
r
ev
iew
ar
ticle
its
elf
m
ay
b
e
a
n
ar
r
ativ
e,
s
y
s
tem
atic,
o
r
m
eta
-
an
aly
s
is
r
ev
iew
th
at
s
u
m
m
ar
iz
es
ex
is
tin
g
tr
en
d
s
an
d
g
a
p
s
i
n
th
e
f
ield
o
f
r
esear
ch
.
B
y
ac
co
u
n
tin
g
f
o
r
a
s
in
g
le
s
u
r
v
ey
ar
ticle,
it
is
p
o
s
s
ib
le
to
f
in
d
th
e
im
p
er
ativ
e
r
esear
c
h
d
is
co
v
er
ies
in
th
e
f
ield
.
W
ith
th
ese
ar
ticles,
it
i
s
p
o
s
s
ib
le
to
o
b
tain
citatio
n
s
f
o
r
th
ese
s
ig
n
if
ican
t
r
esear
ch
ar
t
icles
in
a
f
ield
o
v
er
a
p
er
io
d
o
f
s
tu
d
y
th
at
ca
n
b
e
u
s
ed
to
r
ec
o
m
m
en
d
r
elev
an
t
a
r
ticles.
Fo
r
th
is
p
u
r
p
o
s
e,
th
is
p
r
o
p
o
s
ed
wo
r
k
co
n
s
tr
u
cts
two
d
if
f
er
en
t
n
etwo
r
k
s
:
SC
N
an
d
C
N.
3
.
3
.
1
.
Su
rv
ey
cit
a
t
i
o
n net
wo
r
k
SC
N
is
a
d
ir
ec
ted
g
r
ap
h
co
n
s
tr
u
cted
with
a
lis
t
o
f
s
u
r
v
ey
ar
ticles
id
en
tifie
d
as
s
u
r
v
ey
s
in
th
e
d
atab
ase.
T
h
e
s
et
UA=
{a
1
,
a
2
,
…,
an
}
co
n
s
is
t
o
f
all
ar
ticles
f
o
r
m
ed
b
y
co
m
b
i
n
in
g
b
o
th
g
r
ap
h
SC
N
an
d
C
N.
L
et
u
s
co
n
s
id
er
th
e
UA=
{a
1
,
a2
,
…,
an
}
an
d
SA={
s
a1
,
s
a
2
,
….
,
s
an
}
as
th
e
s
et
o
f
n
o
d
es
f
o
r
m
ed
with
all
ar
ticles
f
r
o
m
s
u
r
v
ey
cl
u
s
ter
s
alo
n
g
with
t
h
eir
citatio
n
s
,
a
n
d
a
r
elatio
n
b
etwe
en
th
e
n
o
d
es
f
r
o
m
s
a1
t
o
s
a2
ca
n
b
e
r
ef
e
r
r
ed
to
as
cited
b
y
.
C
o
n
s
id
er
an
y
ar
ticle
f
r
o
m
SA
is
an
elem
en
t
o
f
s
et
A,
s
u
ch
t
h
at
s
ai
∈
A
is
alwa
y
s
tr
u
e.
E
ac
h
n
o
d
e
in
th
is
g
r
ap
h
SC
N
i
s
as
s
ig
n
ed
th
r
ee
v
alu
es:
ar
ticl
e
I
D
(
Aid
)
,
cited
b
y
v
alu
e
(
C
B
v
)
,
an
d
citin
g
v
alu
e
(
C
v
)
.
Ar
ticle
I
D
is
th
e
p
r
im
ar
y
u
n
i
q
u
e
i
d
en
tifie
r
th
at
is
u
s
ed
t
o
id
e
n
tify
ea
c
h
a
r
ticle
s
ep
ar
ately
i
n
th
is
n
etwo
r
k
.
C
B
v
is
th
e
t
o
tal
co
u
n
t
f
o
r
th
e
n
u
m
b
er
o
f
tim
es
th
e
a
r
ticle
h
as
b
ee
n
“c
ited
b
y
”
o
th
er
ar
ti
cles.
Similar
ly
,
C
v
is
th
e
to
tal
n
u
m
b
er
o
f
ar
ticles
th
at
t
h
e
c
u
r
r
en
t
n
o
d
e
is
citin
g
.
Fig
u
r
e
3
illu
s
tr
ates
a
clea
r
ex
p
lan
atio
n
o
f
t
h
e
SC
N
g
r
ap
h
with
s
ev
en
n
o
d
es:
A,
B
,
C
,
X,
Y,
W
,
an
d
Z
.
B
y
co
n
s
id
er
in
g
th
e
b
elo
w
ex
am
p
le,
th
e
en
tire
r
ec
o
m
m
en
d
atio
n
s
y
s
tem
ca
n
b
e
ex
p
lain
e
d
as
f
o
llo
ws:
UA=
{A,
B
,
C
,
Y
,
W
,
X,
Y,
Z
},
an
d
SA={
A,
B
,
C
}.
Her
e
A,
B
,
a
n
d
C
ar
e
s
u
r
v
ey
ar
ticles
id
en
tifie
d
as
r
elev
an
t
to
th
e
u
s
er
’
s
s
ea
r
ch
q
u
er
y
.
an
d
th
en
,
with
th
ese
n
o
d
es,
th
e
cit
atio
n
s
o
f
th
e
ar
ticles
in
A,
B
,
an
d
C
ar
e
r
ep
r
esen
ted
in
F
ig
u
r
e
3
B
)
.
Fig
u
r
e
3
B
)
r
ep
r
esen
ts
th
e
citatio
n
s
o
f
ea
ch
n
o
d
e;
f
o
r
ex
am
p
le,
ar
ticle
A
h
as
cited
ar
ticle
s
X
an
d
Y
in
th
e
f
ig
u
r
e.
T
h
e
co
r
r
esp
o
n
d
in
g
g
r
ap
h
o
f
th
e
n
o
d
es
in
F
ig
u
r
e
3
B
)
is
co
n
s
tr
u
cted
in
to
th
e
g
r
ap
h
as
s
h
o
wn
in
Fig
u
r
e
3
A)
SC
N
g
r
ap
h
.
Fig
u
r
e
3
C
)
ex
p
lai
n
s
th
e
s
tr
u
ctu
r
e
o
f
n
o
d
e
A
a
n
d
its
v
alu
es,
wh
ich
ar
e
u
s
ed
in
th
e
g
r
ap
h
t
o
f
in
d
th
e
m
o
s
t in
f
lu
en
tial n
o
d
e
in
th
e
g
r
ap
h
SC
N.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
I
SS
N:
2252
-
8
7
7
6
S
r
vy
cite:
a
h
yb
r
id
s
cien
tifi
c
a
r
ticle
r
ec
o
mme
n
d
a
tio
n
s
ystem
(
S
iva
s
a
n
ka
r
i R
)
223
3
.
3
.
2
.
Cit
a
t
io
n
n
et
wo
r
k
C
N
is
co
n
s
tr
u
cted
with
a
lis
t
o
f
ar
ticles
o
th
er
th
an
th
e
o
n
es
th
at
ar
e
id
en
tifie
d
as
s
u
r
v
ey
a
r
ticles.
T
h
is
n
etwo
r
k
is
f
o
r
m
e
d
with
a
s
et
o
f
n
o
d
es
OA=
{o
a1
,
o
a2
,
…,
o
an
},
wh
er
e
OA
r
ep
r
esen
ts
th
e
s
et
o
f
n
o
d
es
f
r
o
m
th
e
o
r
i
g
in
al
clu
s
ter
alo
n
g
with
th
eir
citatio
n
s
.
T
h
e
r
elatio
n
s
h
ip
o
f
th
is
g
r
ap
h
,
C
N,
ca
n
b
e
r
ef
er
r
ed
to
as
“c
ited
b
y
”
b
etwe
en
two
n
o
d
es
f
r
o
m
o
a1
to
o
a2
.
Similar
to
SC
N,
th
e
n
o
d
es
in
C
N
h
o
ld
th
r
ee
v
al
u
es:
C
B
v
,
Aid
,
an
d
C
v
.
C
o
n
s
id
er
an
y
ar
ticle
f
r
o
m
OA
is
an
elem
en
t
o
f
s
et
A,
s
u
ch
th
at
o
ai
∈
UA
is
alwa
y
s
tr
u
e
an
d
UA=
{a
1
,
a2
,
…,
a
n
}.
T
h
e
g
r
a
p
h
r
ep
r
esen
tatio
n
o
f
th
e
g
r
a
p
h
C
N
is
s
im
ilar
th
at
o
f
g
r
ap
h
S
C
N.
T
h
e
v
al
u
es
in
th
e
n
o
d
es
o
f
th
ese
g
r
ap
h
s
(
SC
N
an
d
C
N)
ar
e
u
s
ed
to
f
in
d
a
lis
t
o
f
p
r
o
m
i
n
en
t
ar
ticles,
wh
i
ch
ar
e
u
s
ed
b
y
th
e
r
ec
o
m
m
en
d
atio
n
s
y
s
tem
to
f
in
d
r
elev
an
t
ar
ticles
b
ased
o
n
th
e
tex
t
s
im
ilar
ity
in
d
ex
o
f
ea
ch
r
ec
o
m
m
e
n
d
e
d
ar
ticle.
Fig
u
r
e
3
.
SC
N
g
r
ap
h
an
d
n
o
d
e
s
tr
u
ctu
r
e
4.
E
XP
E
R
I
M
E
N
T
A
L
SE
T
UP
T
h
e
en
tire
d
ataset
is
d
iv
id
ed
in
to
two
g
r
o
u
p
s
:
g
r
o
u
p
s
o
f
s
u
r
v
ey
a
r
ticles
an
d
g
r
o
u
p
s
o
f
o
r
i
g
in
al
r
esear
ch
ar
ticles.
R
ev
iew,
m
et
a
-
an
aly
s
is
r
ev
iew
an
d
s
y
s
tem
a
tic
r
ev
iew
ar
ticles
m
ak
e
u
p
th
e
ar
ticles
in
s
u
r
v
e
y
ar
ticles
clu
s
ter
.
All
o
th
er
v
ar
ieties
o
f
r
esear
ch
ar
ticles,
in
clu
d
in
g
ca
s
e
s
tu
d
ies,
m
eth
o
d
o
lo
g
ical
r
esear
ch
ar
ticles,
s
h
o
r
t
co
m
m
u
n
icatio
n
ar
ticles,
an
d
o
th
er
s
,
ar
e
in
clu
d
ed
in
th
e
o
r
ig
in
al
r
esear
ch
clu
s
ter
.
T
h
e
p
r
o
ce
d
u
r
e
u
s
ed
to
d
iv
id
e
th
e
d
ataset
in
t
o
two
d
is
tin
ct
g
r
o
u
p
s
b
eg
in
s
with
clu
s
ter
in
g
.
C
lu
s
ter
in
g
is
th
e
in
itial
p
h
ase
o
f
th
e
r
ec
o
m
m
en
d
atio
n
s
y
s
tem
.
Usi
n
g
th
is
clu
s
ter
in
g
,
t
h
e
d
ata
s
et
i
s
d
iv
id
ed
in
to
s
u
r
v
e
y
cl
u
s
ter
a
n
d
o
r
ig
in
al
ar
ticle
clu
s
ter
s
.
C
lu
s
ter
in
g
tex
t
d
ata
i
s
a
tim
e
-
co
n
s
u
m
in
g
p
r
o
ce
d
u
r
e
th
at
co
m
p
r
is
es
p
r
ep
r
o
ce
s
s
in
g
,
tex
t
em
b
ed
d
in
g
s
,
d
im
en
s
io
n
ality
r
ed
u
ctio
n
,
s
im
ilar
ity
m
ea
s
u
r
em
en
ts
,
an
d
clu
s
ter
in
g
.
As
a
r
esu
lt
o
f
th
e
clu
s
ter
in
g
,
two
d
is
tin
ct
clu
s
ter
s
ar
e
g
en
er
ated
.
T
h
e
n
ex
t
p
h
ase
o
f
th
e
r
ec
o
m
m
en
d
a
tio
n
s
y
s
tem
is
th
e
co
n
s
tr
u
ctio
n
o
f
SC
N
an
d
C
N
u
s
in
g
th
ese
clu
s
ter
s
.
Fu
r
th
er
p
r
o
ce
s
s
in
g
is
co
m
p
u
ted
b
y
u
tili
zin
g
th
ese
two
g
r
ap
h
s
,
SC
N
an
d
C
N.
T
h
e
s
im
ilar
ity
s
co
r
e
o
f
ea
ch
c
o
n
n
ec
tio
n
is
o
b
tain
e
d
f
o
llo
win
g
t
h
e
b
u
ild
i
n
g
o
f
t
h
e
SC
N.
T
h
e
s
am
e
p
r
o
ce
d
u
r
e
is
d
o
n
e
f
o
r
th
e
a
r
ticles
in
th
e
n
etwo
r
k
C
N,
an
d
t
h
e
y
ar
e
o
n
ce
a
g
ain
g
r
o
u
p
ed
b
ased
o
n
tex
t
s
i
m
ilar
ity
s
co
r
e
alo
n
g
with
co
m
p
u
ted
citatio
n
v
al
u
e
.
T
h
e
f
in
d
in
g
s
o
f
th
ese
two
clu
s
ter
s
ar
e
th
en
jo
in
t
to
ca
lcu
late
th
e
weig
h
ted
av
er
ag
e
o
f
th
e
s
im
ilar
ity
s
c
o
r
e.
T
h
e
r
ec
o
m
m
e
n
d
atio
n
al
g
o
r
ith
m
m
ak
es
r
ec
o
m
m
e
n
d
at
io
n
s
b
a
s
ed
o
n
th
e
ass
es
s
ed
s
im
ilar
ity
s
co
r
e.
Fo
r
o
u
r
s
tu
d
y
,
we
h
av
e
c
o
llected
ar
ticles
f
r
o
m
s
o
u
r
ce
s
lik
e
Go
o
g
le
Sch
o
lar
,
DB
L
P,
an
d
I
E
E
E
Xp
l
o
r
e,
wh
ich
c
o
n
tain
o
v
er
6
0
7
,
3
3
1
ar
ticles
f
r
o
m
co
m
p
u
ter
s
cien
ce
m
ajo
r
s
.
Fro
m
th
e
co
llected
d
ata,
we
h
av
e
r
em
o
v
e
d
o
v
er
2
4
,
4
2
1
ar
ticles,
as
th
ey
h
av
e
a
f
ew
m
is
s
in
g
attr
ib
u
tes
t
h
at
ar
e
co
n
s
id
er
ed
im
p
o
r
tan
t
f
o
r
o
u
r
s
tu
d
y
.
T
h
e
a
r
ticles
in
th
is
d
ataset
a
r
e
id
e
n
t
if
ied
as
ar
ticles
f
r
o
m
m
ac
h
in
e
lear
n
in
g
,
a
r
tific
ial
in
tellig
en
ce
,
im
ag
e
p
r
o
ce
s
s
in
g
,
n
atu
r
al
lan
g
u
ag
e
p
r
o
ce
s
s
in
g
,
co
m
p
iler
d
es
ig
n
,
in
ter
n
e
t
o
f
th
in
g
s
(
I
o
T
)
,
s
em
an
tic
web
,
d
ata
m
in
in
g
,
c
o
m
p
u
ter
v
is
io
n
,
r
ea
l
-
tim
e
co
m
p
u
tin
g
,
c
y
b
er
s
ec
u
r
ity
,
c
o
m
p
u
ter
n
etwo
r
k
s
,
an
d
alg
o
r
ith
m
s
.
Fo
r
th
e
s
tu
d
y
,
t
h
e
m
ajo
r
attr
ib
u
tes
o
r
f
ea
tu
r
es
we
h
av
e
co
n
ce
n
tr
ated
o
n
a
r
e
tex
t
attr
ib
u
tes:
th
e
titl
e
o
f
th
e
ar
ticle
a
n
d
th
e
a
b
s
tr
ac
t
o
f
th
e
a
r
ticle;
an
d
m
eta
-
attr
ib
u
tes:
cited
b
y
ar
ticles
an
d
citin
g
ar
ticles.
T
h
e
NL
T
K
tex
t
d
ata
is
u
s
ed
to
p
r
ep
r
o
ce
s
s
th
is
d
ataset.
Fo
r
e
m
b
ed
d
in
g
t
h
is
p
r
e
-
p
r
o
ce
s
s
ed
tex
t
d
ata,
we
h
av
e
u
tili
ze
d
th
e
B
E
R
T
m
o
d
el.
T
h
e
v
ar
iab
les
u
s
ed
f
o
r
th
is
m
o
d
el
d
u
r
in
g
th
e
tr
ain
in
g
ar
e
lis
ted
in
T
ab
le
1
.
T
h
en
,
f
o
r
d
im
en
s
io
n
r
ed
u
ctio
n
,
L
1
r
eg
u
lar
izatio
n
is
u
s
ed
with
an
a
lp
h
a
v
al
u
e
o
f
0
.
2
.
Fin
ally
,
th
e
clu
s
ter
s
ar
e
f
o
r
m
ed
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
7
7
6
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
,
Vo
l.
1
4
,
No
.
1
,
A
p
r
il
20
2
5
:
216
-
22
8
224
u
s
in
g
K
-
m
ea
n
s
clu
s
ter
in
g
.
Af
ter
clu
s
ter
in
g
,
two
clu
s
ter
s
f
o
r
m
ed
,
wh
er
e
th
e
f
ir
s
t
cl
u
s
ter
co
n
s
is
ted
o
f
3
0
,
4
5
4
n
u
m
b
er
ar
ticles
o
f
t
y
p
e
s
u
r
v
e
y
o
r
r
ev
iew
ar
ticles,
an
d
th
e
s
ec
o
n
d
cl
u
s
ter
co
n
s
is
ted
o
f
5
2
,
4
5
6
n
u
m
b
er
ar
ticles
clu
s
ter
ed
as
o
r
ig
in
al
r
esear
ch
ar
ticles.
T
h
en
t
h
e
ar
ticles
th
at
m
atch
th
e
u
s
er
q
u
er
y
f
r
o
m
b
o
th
clu
s
ter
s
ar
e
s
elec
ted
s
ep
ar
ately
to
g
en
er
ate
th
e
SC
N
an
d
C
N
g
r
ap
h
s
.
T
h
en
th
e
C
B
v
an
d
C
v
n
o
d
es
f
o
r
ea
ch
n
o
d
e
ar
e
ca
lcu
lated
.
W
ith
co
m
p
u
ted
C
B
v
an
d
C
v
,
im
p
o
r
tan
t
n
o
d
es
f
r
o
m
th
e
n
etwo
r
k
s
SC
N
an
d
C
N
ar
e
id
en
tifie
d
s
ep
ar
ately
.
T
h
en
th
e
W
eig
h
ted
T
ex
t
Similar
ity
(
W
T
S)
s
co
r
es
o
f
th
ese
s
el
ec
ted
n
o
d
es
ar
e
co
m
p
u
ted
u
s
in
g
(
1
)
.
T
h
en
a
f
i
n
al
r
ec
o
m
m
en
d
atio
n
lis
t is g
en
er
ated
an
d
p
r
esen
ted
as a
lis
t o
f
r
ec
o
m
m
en
d
ed
ar
tic
les.
(
)
=
(
)
∗
(
(
)
)
+
(
)
∗
(
(
)
)
2
(
1
)
W
h
er
e:
WTS
(
ai
)
:
weig
h
ted
tex
t similar
ity
s
co
r
e
f
o
r
ar
ticle
ai
.
TS
(
ai
)
:
tex
t similar
ity
s
co
r
e
o
f
ar
ticle
ai
co
m
p
u
ted
b
y
co
m
p
a
r
in
g
th
e
r
esear
ch
er
'
s
q
u
er
y
.
S
C
N
(
C
B
v
(
ai
)
)
:
to
tal
cited
b
y
c
o
u
n
t
o
f
ar
ticle
ai
in
t
h
e
SC
N.
CN
(
C
B
v
(
ai
)
)
:
to
tal
cited
b
y
co
u
n
t
o
f
a
r
ticle
ai
in
th
e
C
N.
I
n
th
e
(
1
)
,
th
e
C
v
v
alu
es
(
r
ep
r
esen
tin
g
th
e
n
u
m
b
er
o
f
citin
g
p
ap
er
s
)
h
av
e
b
ee
n
ef
f
ec
tiv
ely
in
co
r
p
o
r
ated
alo
n
g
s
id
e
th
e
cit
ed
b
y
c
o
u
n
ts
(
C
B
v
)
an
d
tex
t
s
im
ilar
ity
(
T
S)
v
alu
es.
T
h
e
d
i
v
is
io
n
b
y
2
in
th
e
f
o
r
m
u
la
en
s
u
r
es
t
h
at
th
e
weig
h
ted
av
e
r
ag
e
o
f
th
e
te
x
t
s
im
ilar
ity
s
co
r
es,
c
o
n
s
id
er
in
g
b
o
th
t
h
e
SC
N
an
d
C
N,
is
co
m
p
u
ted
ac
cu
r
ately
.
T
ab
le
1.
P
ar
am
eter
s
ettin
g
s
f
o
r
th
e
p
r
o
p
o
s
ed
m
o
d
el
S
S
.
N
o
P
r
o
c
e
ss
A
l
g
o
r
i
t
h
ms
V
a
l
u
e
1.
Te
x
t
p
r
e
p
r
o
c
e
ss
i
n
g
N
LTK
-
2.
Te
x
t
e
mb
e
d
d
i
n
g
B
ER
T
b
e
r
t
-
l
a
r
g
e
-
u
n
c
a
se
d
3.
D
i
me
n
si
o
n
r
e
d
u
c
t
i
o
n
LA
S
S
O
A
l
p
h
a
v
a
l
u
e
,
a
0
.
2
4.
K
-
me
a
n
s
c
l
u
s
t
e
r
i
n
g
N
u
mb
e
r
o
f
c
l
u
s
t
e
r
s,
K
2
5.
G
r
a
p
h
c
o
n
st
r
u
c
t
i
o
n
Th
r
e
s
h
o
l
d
f
o
r
t
e
x
t
si
mi
l
a
r
i
t
y
,
TS
0
.
5
5.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
NS
T
h
e
p
r
o
p
o
s
ed
S
r
v
y
cite
m
et
h
o
d
is
co
m
p
ar
ed
with
th
e
b
el
o
w
-
b
aselin
e
m
eth
o
d
s
.
T
h
e
ac
c
u
r
ac
y
o
f
o
u
r
p
r
o
p
o
s
ed
m
eth
o
d
is
ev
alu
ate
d
u
s
in
g
th
e
m
etr
ics
p
r
ec
is
io
n
a
n
d
r
ec
all.
Pre
cisi
o
n
is
a
m
ea
s
u
r
e
th
at
is
u
s
ed
to
id
en
tify
th
e
p
r
o
p
o
s
itio
n
o
f
th
e
co
r
r
ec
t
p
r
ed
ictio
n
p
r
ed
icted
b
y
th
e
m
eth
o
d
s
.
R
ec
all
is
a
m
ea
s
u
r
e
th
at
id
en
tifie
s
th
e
p
o
r
tio
n
o
f
an
ac
cu
r
ate
r
es
u
lt
f
r
o
m
t
h
e
p
r
ed
icate
r
esu
lts
.
Z
h
u
et
a
l
.
[
3
]
r
ec
o
m
m
en
d
u
s
in
g
a
s
ea
r
ch
q
u
e
r
y
b
ased
o
n
th
e
b
e
h
av
io
u
r
o
f
t
h
e
u
s
er
alo
n
g
with
th
e
cu
r
r
en
t
d
r
af
t
o
f
th
e
a
r
ticle.
T
h
is
m
eth
o
d
ass
u
m
es
u
s
er
s
’
p
r
e
v
io
u
s
p
u
b
licatio
n
s
an
d
cited
ar
ticles
wil
l
r
em
ain
th
e
s
am
e
an
d
r
ec
o
m
m
en
d
s
ar
ticles.
N
o
t
o
n
ly
th
at,
b
u
t
th
is
m
eth
o
d
d
o
es
n
o
t
e
v
en
c
o
n
s
id
er
th
e
co
n
te
n
t
o
f
t
h
e
p
u
b
lis
h
ed
ar
ticles.
Fo
r
tex
t
f
ea
tu
r
es,
it
o
n
ly
c
o
n
s
id
er
s
th
e
to
p
ics
o
f
p
r
e
v
io
u
s
ly
p
u
b
lis
h
ed
ar
ticle
s
an
d
th
e
d
r
af
t
o
f
th
e
c
u
r
r
en
t
ar
ticle.
I
t
is
a
c
o
llab
o
r
a
tiv
e
m
eth
o
d
;
h
e
n
ce
it
s
u
f
f
er
s
f
r
o
m
th
e
co
ld
s
tar
t
p
r
o
b
lem
if
a
n
ew
u
s
er
en
ter
s
th
e
s
y
s
tem
.
C
h
au
d
h
u
r
i
et
a
l
.
[1
5
]
p
r
o
p
o
s
ed
a
wo
r
k
b
y
d
ev
elo
p
in
g
a
r
ec
o
m
m
en
d
at
io
n
s
y
s
tem
b
y
u
s
in
g
a
d
ir
ec
t
tex
t
f
ea
tu
r
e
a
lo
n
g
with
f
o
u
r
n
e
w
in
d
ir
ec
t
f
ea
tu
r
es
to
ca
lcu
late
th
e
d
iv
e
r
s
ity
in
k
e
y
wo
r
d
s
,
c
o
m
p
lex
ity
in
s
en
ten
c
es,
citatio
n
an
aly
s
is
o
v
er
tim
e,
an
d
q
u
ality
o
f
th
e
ar
ticles.
Ma
b
u
d
e
et
a
l.
[
8
]
p
r
o
p
o
s
ed
an
in
teg
r
ated
ap
p
r
o
ac
h
to
r
ec
o
m
m
en
d
r
esear
ch
ar
ti
cles
an
d
ex
p
er
tis
e
r
ec
o
m
m
en
d
atio
n
s
y
s
tem
s
,
wh
ich
ar
e
b
asically
two
d
if
f
er
e
n
t
r
ec
o
m
m
en
d
atio
n
s
y
s
tem
s
.
B
u
t
b
o
th
o
f
th
ese
s
y
s
tem
s
u
s
e
th
e
s
am
e
s
et
o
f
d
ata
s
ets
to
r
ec
o
m
m
en
d
two
d
i
f
f
er
en
t
o
u
t
p
u
ts
;
th
e
r
ef
o
r
e,
th
ese
two
r
ec
o
m
m
en
d
atio
n
s
y
s
tem
s
ar
e
in
teg
r
ated
in
to
a
s
in
g
le
s
y
s
tem
.
I
n
th
e
wo
r
k
p
r
o
p
o
s
ed
b
y
Z
h
an
g
et
a
l.
[2
2
]
,
cr
ea
ted
a
m
u
ltid
im
en
s
io
n
al
m
o
d
el
b
ased
o
n
th
e
u
s
er
’
s
ac
ad
em
ic
p
o
r
tr
ait.
Her
e,
th
e
ar
ticles
ar
e
r
ec
o
m
m
en
d
ed
b
ased
o
n
b
asic
in
f
o
r
m
atio
n
a
b
o
u
t
th
e
u
s
er
,
s
u
ch
as
ag
e,
g
e
n
d
er
,
ed
u
ca
tio
n
,
m
ajo
r
,
id
en
ti
ty
,
an
d
d
ir
ec
tio
n
o
f
th
e
r
esear
ch
.
T
h
e
n
,
b
ased
-
o
n
attr
ib
u
tes
lik
e
ac
ad
e
m
ic
m
o
tiv
atio
n
,
co
g
n
itiv
e
s
ty
le,
an
d
d
o
m
ain
k
n
o
wled
g
e
o
f
ac
ad
em
ic
p
er
s
o
n
ality
ar
e
an
aly
s
ed
,
an
d
f
in
ally
,
b
e
h
av
io
u
r
ch
a
r
ac
ter
is
tics
ar
e
co
m
p
u
ted
u
s
in
g
th
e
b
asic
in
f
o
r
m
atio
n
alr
ea
d
y
o
b
tain
ed
.
L
i
et
a
l.
[1
6
]
p
r
o
p
o
s
ed
a
g
r
ap
h
-
b
ased
r
ec
o
m
m
e
n
d
atio
n
s
y
s
tem
,
MA
R
ec
,
wh
er
e
th
e
au
th
o
r
s
u
s
ed
an
attr
ib
u
ted
h
eter
o
g
en
e
o
u
s
in
f
o
r
m
atio
n
n
etwo
r
k
(
aH
I
N
)
n
etwo
r
k
f
o
r
in
f
o
r
m
atio
n
e
x
tr
ac
tio
n
,
GAT
f
o
r
g
e
n
er
atio
n
o
f
n
etw
o
r
k
em
b
ed
d
i
n
g
to
id
en
tify
t
h
e
u
s
er
m
eta
p
ath
,
a
n
d
B
i
-
L
STM
in
f
u
s
ed
with
atten
tio
n
-
awa
r
e
m
ec
h
a
n
is
m
s
to
ca
p
tu
r
e
r
ec
en
t
ar
ticles
an
d
r
ec
o
m
m
e
n
d
th
em
to
u
s
er
s
.
Sm
ail
et
a
l.
[1
1
]
co
m
p
ar
ed
two
to
p
to
p
ic
m
o
d
ellin
g
tech
n
iq
u
es,
L
DA
an
d
NM
F.
T
h
e
L
DA
is
im
p
lem
en
ted
u
s
in
g
Scik
it,
an
d
th
e
NM
F
is
im
p
lem
en
ted
u
s
in
g
Gen
s
im
an
d
Ma
llet
p
ac
k
a
g
es.
T
h
e
au
th
o
r
s
co
n
clu
d
e
d
t
h
at
u
s
in
g
th
e
NM
F
with
th
e
Ma
llet
p
ac
k
ag
e
f
o
r
to
p
ic
m
o
d
ellin
g
p
r
o
d
u
ce
d
b
ett
er
p
r
ec
is
io
n
f
o
r
th
e
r
ec
o
m
m
e
n
d
atio
n
o
f
r
esear
ch
ar
ticles
in
th
e
d
atase
t
u
s
ed
.
T
a
b
le
2
s
u
m
m
ar
izes
th
e
co
m
p
a
r
i
s
o
n
o
f
th
e
p
r
o
p
o
s
ed
s
r
v
y
cite
m
o
d
el
with
s
elec
ted
b
aselin
e
m
o
d
els.
I
n
T
a
b
le
2
,
t
h
e
co
m
p
ar
is
o
n
s
tu
d
y
c
o
m
p
ar
e
s
th
e
m
eth
o
d
o
lo
g
y
,
f
ea
tu
r
es
u
s
ed
b
y
t
h
e
m
eth
o
d
s
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
I
SS
N:
2252
-
8
7
7
6
S
r
vy
cite:
a
h
yb
r
id
s
cien
tifi
c
a
r
ticle
r
ec
o
mme
n
d
a
tio
n
s
ystem
(
S
iva
s
a
n
ka
r
i R
)
225
to
p
er
f
o
r
m
r
ec
o
m
m
en
d
atio
n
s
,
th
e
d
ataset
u
s
ed
,
an
d
e
v
alu
at
io
n
m
et
r
ics
u
s
ed
to
ca
lc
u
late
th
e
ac
cu
r
ac
y
o
f
th
e
d
is
cu
s
s
ed
m
o
d
els
.
Fig
u
r
e
4
p
r
esen
ts
th
e
p
r
ec
is
io
n
an
d
r
ec
all
r
esu
lts
f
r
o
m
th
e
Srv
y
cit
e
m
o
d
el.
W
e
h
av
e
ca
lcu
lated
th
e
r
esu
lts
f
o
r
p
r
ec
is
io
n
(
Fig
u
r
e
4
(
a)
)
an
d
r
ec
all
(
Fig
u
r
e
4
(
b
)
)
@
K
v
alu
es
f
o
r
5
,
1
0
,
2
0
,
3
0
,
4
0
,
an
d
5
0
.
Fro
m
th
e
r
esu
lt,
it
is
clea
r
th
at
th
e
r
ec
all
m
ea
s
u
r
e
o
f
th
e
p
r
o
p
o
s
ed
m
o
d
el
s
h
o
ws
b
etter
im
p
r
o
v
em
e
n
t
f
o
r
@
K
v
alu
es
ab
o
v
e
1
0
.
Similar
ly
,
f
o
r
p
r
ec
is
io
n
,
th
e
r
e
is
a
c
o
n
s
id
er
ab
le
in
cr
ea
s
e
i
n
th
e
r
es
u
lt
with
@
K
v
alu
es
gr
ea
ter
th
an
2
0
.
W
ith
o
u
r
d
ata
s
et,
it
ca
n
b
e
co
n
clu
d
ed
th
at
t
h
er
e
is
a
co
n
s
tan
t
in
cr
ea
s
e
in
th
e
ac
cu
r
ac
y
o
f
th
e
r
esu
lts
with
h
ig
h
er
v
alu
es
o
f
K
u
s
in
g
th
e
p
r
o
p
o
s
ed
Srv
y
cite
m
eth
o
d
.
B
ased
o
n
o
u
r
d
ataset,
it
ca
n
b
e
in
f
er
r
e
d
th
at
th
er
e
ex
is
ts
a
co
n
s
i
s
ten
t
au
g
m
en
tatio
n
in
th
e
ac
cu
r
ac
y
o
f
th
e
r
esu
lts
as
th
e
@
K
v
al
u
es
in
cr
ea
s
e
wh
en
em
p
lo
y
in
g
th
e
p
r
o
p
o
s
ed
Srv
y
cite
m
eth
o
d
.
T
h
is
o
b
s
er
v
atio
n
s
u
g
g
ests
th
at
th
e
Srv
y
cit
e
m
o
d
el
ef
f
ec
tiv
el
y
en
h
an
ce
s
th
e
p
r
ec
is
io
n
an
d
r
ec
all
o
f
r
ec
o
m
m
e
n
d
ed
ar
ticles,
p
ar
ticu
lar
ly
as
th
e
s
ize
o
f
th
e
r
e
co
m
m
en
d
atio
n
lis
t
ex
p
an
d
s
.
T
ab
le
2
.
C
o
m
p
a
r
is
o
n
o
f
p
r
o
p
o
s
ed
m
o
d
els with
th
e
b
aselin
e
m
o
d
els
S.
N
o
W
o
r
k
M
e
t
h
o
d
o
l
o
g
y
F
e
a
t
u
r
e
s
o
f
p
a
p
e
r
s
u
s
e
d
D
a
t
a
s
e
t
Ev
a
l
u
a
t
i
o
n
me
t
r
i
c
e
s
1.
P
r
o
p
o
se
d
w
o
r
k
K
-
me
a
n
s
+
s
u
r
v
e
y
c
i
t
a
t
i
o
n
+
c
i
t
a
t
i
o
n
n
e
t
w
o
r
k
A
b
st
r
a
c
t
+
t
i
t
l
e
+
c
i
t
a
t
i
o
n
n
e
t
w
o
r
k
+
su
r
v
e
y
p
a
p
e
r
n
e
t
w
o
r
k
D
B
LP,
G
o
o
g
l
e
S
c
h
o
l
a
r
,
I
EEE
P
r
e
c
i
s
i
o
n
r
e
c
a
l
l
a
c
c
u
r
a
c
y
2.
Zh
u
e
t
a
l
.
[
3
]
B
i
d
i
r
e
c
t
i
o
n
a
l
r
e
c
u
r
r
e
n
t
n
e
u
r
a
l
n
e
t
w
o
r
k
(
R
N
N
)
w
i
t
h
a
t
t
e
n
t
i
o
n
mec
h
a
n
i
sm
(
t
e
x
t
f
e
a
t
u
r
e
e
mb
e
d
d
i
n
g
:
P
V
-
D
M
)
Ti
t
l
e
+
a
u
t
h
o
r
p
u
b
l
i
c
a
t
i
o
n
a
n
d
c
i
t
a
t
i
o
n
i
n
f
o
r
ma
t
i
o
n
D
B
LP
P
r
e
c
i
s
i
o
n
@
2
0
R
e
c
a
l
l
@
2
0
M
R
R
N
D
C
G
@
2
0
3.
C
h
a
u
d
h
u
r
i
e
t
a
l
.
[1
5
]
e
n
s
e
mb
l
e
mo
d
e
l
s
-
A
d
a
B
o
o
st
,
b
a
g
g
i
n
g
,
v
o
t
i
n
g
,
a
n
d
s
t
a
c
k
i
n
g
1
5
a
r
t
i
c
l
e
f
e
a
t
u
r
e
s
+
p
a
p
e
r
met
a
d
a
t
a
+
k
e
y
w
o
r
d
d
i
v
e
r
si
t
y
mea
s
u
r
e
me
n
t
+
se
n
t
e
n
c
e
c
o
m
p
l
e
x
i
t
y
a
n
a
l
y
s
i
s +
c
i
t
a
t
i
o
n
a
n
a
l
y
si
s
o
v
e
r
t
i
m
e
+
sc
i
e
n
t
i
f
i
c
q
u
a
l
i
t
y
m
e
a
s
u
r
e
m
e
n
t
S
c
o
p
u
s
P
r
e
c
i
s
i
o
n
R
e
c
a
l
l
F
1
sc
o
r
e
A
c
c
u
r
a
c
y
R
M
S
E
4.
S
mai
l
e
t
a
l
.
[1
1
]
H
I
N
+
G
A
T
+
R
N
N
To
p
i
c
+
v
e
n
u
e
+
p
a
p
e
r
+
u
s
e
r
D
B
LP,
S
p
o
t
i
f
y
,
a
n
d
Y
e
l
p
F1
-
sc
o
r
e
@
5
,
A
c
c
u
r
a
c
y
a
n
d
N
D
C
G
@
5
.
5.
M
a
b
u
d
e
e
t
a
l
.
[
8
]
M
a
t
r
i
x
f
a
c
t
o
r
i
z
a
t
i
o
n
+
K
M
e
a
n
s
P
u
b
l
i
c
a
t
i
o
n
c
o
u
n
t
,
c
i
t
a
t
i
o
n
s
,
u
ser
r
a
t
i
n
g
s,
a
n
d
a
r
t
i
c
l
e
k
e
y
p
h
r
a
se
S
c
o
p
u
s
P
r
e
c
i
s
i
o
n
r
e
c
a
l
l
F
1
sc
o
r
e
a
c
c
u
r
a
c
y
R
M
S
E
(
a)
(
b
)
Fig
u
r
e
4
.
Srv
y
cite
@
K
v
alu
e
(
a)
p
r
ec
is
io
n
an
d
(
b
)
r
ec
all
T
h
e
p
r
ec
is
io
n
an
d
r
ec
all
r
es
u
lts
o
b
tain
ed
f
r
o
m
th
e
Srv
y
c
ite
m
o
d
el
wer
e
ca
lcu
lated
f
o
r
d
if
f
er
en
t
v
alu
es
o
f
K
(
5
,
1
0
,
2
0
,
3
0
,
4
0
,
an
d
5
0
)
.
Pre
cisi
o
n
m
ea
s
u
r
es
t
h
e
p
r
o
p
o
r
tio
n
o
f
r
elev
a
n
t
ar
tic
les
am
o
n
g
th
e
to
tal
r
ec
o
m
m
en
d
ed
ar
ticles,
wh
ile
r
ec
all
as
s
ess
e
s
th
e
p
r
o
p
o
r
tio
n
o
f
r
elev
a
n
t
ar
ticles
r
etr
iev
ed
b
y
th
e
r
ec
o
m
m
en
d
atio
n
s
y
s
tem
o
u
t
o
f
all
r
elev
an
t
ar
ticles
av
aila
b
le
in
th
e
d
ataset.
W
e
o
b
s
er
v
ed
th
at
th
e
r
ec
all
m
ea
s
u
r
e
o
f
t
h
e
p
r
o
p
o
s
ed
m
o
d
el
ex
h
ib
ited
b
et
ter
im
p
r
o
v
e
m
en
t
f
o
r
K
v
alu
es
ab
o
v
e
1
0
.
Similar
ly
,
p
r
ec
is
io
n
s
h
o
wed
a
co
n
s
id
er
a
b
le
in
cr
ea
s
e
f
o
r
K
v
alu
es
g
r
ea
ter
th
an
2
0
.
T
h
ese
f
in
d
in
g
s
in
d
icate
a
co
n
s
is
ten
t
in
cr
ea
s
e
in
th
e
ac
cu
r
ac
y
o
f
th
e
r
esu
lts
with
h
ig
h
er
v
alu
es
o
f
K
u
s
in
g
th
e
p
r
o
p
o
s
ed
Srv
y
cite
m
eth
o
d
.
Fu
r
th
er
m
o
r
e,
we
co
m
p
ar
ed
th
e
p
er
f
o
r
m
a
n
ce
o
f
o
u
r
p
r
o
p
o
s
ed
Srv
y
cite
m
eth
o
d
with
b
aselin
e
m
eth
o
d
s
b
y
e
v
alu
atin
g
p
r
ec
is
io
n
an
d
r
ec
all.
T
h
e
p
er
ce
n
tag
e
i
m
p
r
o
v
e
m
en
t
m
etr
ic
was
em
p
lo
y
ed
to
q
u
an
tify
th
e
ex
te
n
t
to
wh
ich
o
u
r
m
eth
o
d
o
u
tp
er
f
o
r
m
ed
th
e
b
aselin
e
m
eth
o
d
s
.
W
e
o
b
s
er
v
ed
a
s
ig
n
if
ican
t
im
p
r
o
v
em
en
t
o
f
1
5
–
2
0
p
er
ce
n
t
in
b
o
th
Evaluation Warning : The document was created with Spire.PDF for Python.