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.
9
,
No
.
4
,
A
u
g
u
s
t
201
9
,
p
p
.
3
1
6
7
~
3
1
7
4
I
SS
N:
2088
-
8708
,
DOI
: 1
0
.
1
1
5
9
1
/
i
j
ec
e
.
v
9
i
4
.
p
p
3
1
6
7
-
3174
3167
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ia
e
s
co
r
e
.
co
m/
jo
u
r
n
a
ls
/in
d
ex
.
p
h
p
/
I
JE
C
E
A
ro
bust au
thors
hip attri
bution o
n
big perio
d
M
ub
in Sh
o
u
k
a
t
T
a
m
bo
li
1
,
R
a
j
esh
P
ra
s
a
d
2
1
M
a
to
sh
ri
C
o
ll
e
g
e
o
f
En
g
in
e
e
rin
g
a
n
d
Re
se
a
rc
h
Ce
n
tre
,
I
n
d
ia
2
S
in
h
g
a
d
I
n
stit
u
te o
f
T
e
c
h
n
o
lo
g
y
a
n
d
S
c
ien
c
e
Na
rh
e
,
I
n
d
ia
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Au
g
26
,
2
0
1
8
R
ev
i
s
ed
Mar
13
,
2
0
1
9
A
cc
ep
ted
A
p
r
4
,
2
0
1
9
A
u
th
o
rsh
ip
a
tt
ri
b
u
t
io
n
is
a
tas
k
t
o
id
e
n
ti
f
y
th
e
w
rit
e
r
o
f
u
n
k
n
o
wn
tex
t
a
n
d
c
a
teg
o
rize
it
to
k
n
o
w
n
w
rit
e
r.
W
r
it
in
g
sty
le
o
f
e
a
c
h
a
u
th
o
r
is
d
ist
in
c
t
a
n
d
c
a
n
b
e
u
se
d
f
o
r
th
e
d
isc
rim
in
a
ti
o
n
.
T
h
e
re
a
re
d
iff
e
re
n
t
p
a
ra
m
e
t
e
rs
re
sp
o
n
sib
le
f
o
r
re
c
ti
fy
in
g
su
c
h
c
h
a
n
g
e
s.
W
h
e
n
th
e
w
rit
in
g
s
a
m
p
les
c
o
ll
e
c
ted
f
o
r
a
n
a
u
th
o
r
w
h
e
n
it
b
e
l
o
n
g
s
t
o
sm
a
ll
p
e
rio
d
,
it
c
a
n
p
a
rti
c
i
p
a
te
e
f
f
i
c
ien
tl
y
f
o
r
id
e
n
ti
f
ica
ti
o
n
o
f
u
n
k
n
o
w
n
sa
m
p
le
.
In
th
is
p
a
p
e
r
a
u
t
h
o
r
i
d
e
n
ti
f
ica
ti
o
n
p
ro
b
lem
c
o
n
sid
e
re
d
w
h
e
re
w
rit
in
g
sa
m
p
le
is
n
o
t
a
v
a
il
a
b
le
o
n
t
h
e
sa
m
e
ti
m
e
p
e
rio
d
.
S
u
c
h
e
v
id
e
n
c
e
s
c
o
ll
e
c
ted
o
v
e
r
l
o
n
g
p
e
rio
d
o
f
ti
m
e
.
A
n
d
c
h
a
ra
c
t
e
r
n
-
g
ra
m
,
w
o
rd
n
-
g
ra
m
a
n
d
p
o
s
n
-
g
ra
m
f
e
a
tu
re
s
u
se
d
to
b
u
i
ld
th
e
m
o
d
e
l.
As
th
e
y
a
r
e
c
o
n
tri
b
u
ti
n
g
to
w
a
rd
s
st
y
l
e
o
f
w
r
it
e
r
in
ter
m
s
o
f
c
o
n
ten
t
a
s
we
ll
a
s
sta
ti
stic
c
h
a
ra
c
teristic
o
f
w
rit
in
g
st
y
le.
We
a
p
p
li
e
d
su
p
p
o
rt
v
e
c
to
r
m
a
c
h
in
e
a
lg
o
rit
h
m
f
o
r
c
las
si
f
ica
ti
o
n
.
Ef
f
e
c
ti
v
e
re
su
lt
s
a
n
d
o
u
tco
m
e
c
a
m
e
o
u
t
f
ro
m
th
e
e
x
p
e
ri
m
e
n
ts.
W
h
il
e
d
isc
ri
m
in
a
ti
n
g
a
m
o
n
g
m
u
lt
ip
le
a
u
th
o
rs,
c
o
r
p
u
s
se
lec
ti
o
n
a
n
d
c
o
n
str
u
c
ti
o
n
w
e
re
th
e
m
o
st
ted
io
u
s
tas
k
w
h
ich
w
a
s
i
m
p
lem
e
n
ted
e
ffe
c
ti
v
e
l
y
.
It
is
o
b
se
rv
e
d
th
a
t
a
c
c
u
ra
c
y
v
a
ried
o
n
f
e
a
tu
re
ty
p
e
.
W
o
rd
a
n
d
c
h
a
ra
c
ter n
-
g
ra
m
h
a
v
e
sh
o
w
n
g
o
o
d
a
c
c
u
ra
c
y
th
a
n
P
o
S n
-
g
ra
m
.
K
ey
w
o
r
d
s
:
Au
t
h
o
r
id
en
tific
atio
n
A
ttrib
u
tio
n
Featu
r
e
ex
tr
ac
tio
n
St
y
lo
m
etr
y
SVM
Co
p
y
rig
h
t
©
2
0
1
9
In
stit
u
te o
f
A
d
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
:
Mu
b
in
S
h
o
u
k
at
T
a
m
b
o
li,
Dep
ar
t
m
en
t o
f
C
o
m
p
u
ter
E
n
g
i
n
ee
r
in
g
,
Ma
to
s
h
r
i Co
lle
g
e
o
f
E
n
g
i
n
ee
r
i
n
g
a
n
d
R
esear
c
h
C
e
n
tr
e,
Nash
i
k
,
Ma
h
ar
ash
tr
a,
I
n
d
ia
.
E
m
ail:
m
u
b
i
n
.
ta
m
b
o
li
@
g
m
ail.
co
m
1.
I
NT
RO
D
UCT
I
O
N
Au
t
h
o
r
s
h
ip
id
en
tific
atio
n
i
s
d
is
ti
n
g
u
is
h
i
n
g
tas
k
o
f
r
ec
o
g
n
iz
in
g
w
r
i
ter
o
f
a
g
i
v
en
co
n
ten
t
f
r
o
m
h
is
w
r
iti
n
g
s
t
y
le.
R
esear
c
h
o
f
au
t
h
o
r
id
en
tif
ica
tio
n
h
a
s
g
o
t
ex
p
o
n
en
tial
g
r
o
w
th
i
n
r
ec
en
t
y
ea
r
s
d
u
e
to
its
v
alu
ab
le
co
n
tr
ib
u
tio
n
in
f
o
r
e
n
s
ic,
li
n
g
u
is
t
ic
r
esear
ch
,
s
o
cial
p
s
y
c
h
o
lo
g
y
,
l
iter
ar
y
s
ci
en
ce
,
s
o
cial
m
ed
ia
a
n
al
y
s
is
an
d
e
-
co
m
m
er
ce
ac
tiv
i
ties
.
Du
e
to
th
e
b
lo
o
m
o
f
I
n
ter
n
et
co
m
m
u
n
icatio
n
h
as
b
ec
o
m
e
ea
s
ier
an
d
co
m
m
o
n
tu
r
n
w
a
s
d
iv
er
ted
in
to
v
en
o
m
o
u
s
m
o
v
e
m
en
ts
.
W
e
id
en
t
if
y
s
u
c
h
s
u
s
p
i
cio
u
s
e
n
titi
e
s
o
v
er
t
h
e
n
et
w
o
r
k
.
A
ct
u
all
y
,
th
i
s
i
s
a
ted
io
u
s
tas
k
b
u
t
it
ca
n
b
e
s
i
m
p
lif
ied
u
s
in
g
au
th
o
r
s
h
ip
attr
ib
u
tio
n
.
Ge
n
er
all
y
,
m
es
s
ag
e
s
o
n
w
eb
ar
e
n
a
m
e
less
.
Ma
n
y
au
t
h
o
r
s
i
n
t
h
eir
w
r
iti
n
g
,
th
e
y
d
o
n
’
t
g
iv
e
th
eir
g
e
n
u
in
e
ch
ar
ac
ter
d
ata.
Fo
r
ex
a
m
p
le,
n
a
m
e,
a
g
e,
s
ex
a
n
d
ad
d
r
ess
.
I
n
n
u
m
er
o
u
s
ab
u
s
e
s
o
r
w
r
o
n
g
d
o
in
g
i
n
s
ta
n
ce
s
o
f
o
n
lin
e
m
es
s
ag
e
s
,
it
is
r
eq
u
i
r
ed
to
f
in
d
th
e
r
ea
l
id
en
tit
y
o
f
a
u
th
o
r
s
.
Alo
n
g
t
h
e
s
e
li
n
es,
th
e
o
b
s
cu
r
it
y
o
f
o
n
li
n
e
m
es
s
ag
e
s
f
o
r
ce
s
s
o
m
e
k
i
n
d
o
f
d
if
f
ic
u
lt
ies
to
id
en
ti
f
y
a
u
t
h
o
r
o
f
co
n
te
n
t
s
a
v
ailab
le
o
n
I
n
ter
n
et.
I
n
t
h
e
n
atu
r
e
o
f
attr
ib
u
tio
n
,
ca
te
g
o
r
izatio
n
is
m
ad
e
f
o
r
u
n
k
n
o
w
n
tex
t
d
o
cu
m
en
t
to
o
n
e,
f
r
o
m
a
li
m
ited
s
et
o
f
ca
n
d
i
d
a
te
au
t
h
o
r
w
h
o
s
e
d
ata
-
s
et
is
in
ter
m
s
o
f
w
r
iti
n
g
s
a
m
p
le
[
1
]
.
W
r
itin
g
o
f
ea
ch
a
u
t
h
o
r
is
u
n
iq
u
el
y
id
en
ti
f
ied
w
it
h
w
r
iti
n
g
s
t
y
le.
So
,
w
r
i
tin
g
s
t
y
le
w
o
u
ld
b
ec
o
m
e
o
n
e
m
etr
ic
o
f
d
is
cr
i
m
in
a
tin
g
au
th
o
r
s
.
T
h
er
e
ar
e
f
ac
to
r
s
af
f
ec
t
t
h
e
p
er
f
o
r
m
an
ce
o
f
attr
ib
u
tio
n
s
y
s
t
e
m
:
n
u
m
b
er
o
f
au
th
o
r
s
,
w
r
iti
n
g
s
a
m
p
le,
s
ize
o
f
w
r
it
in
g
s
a
m
p
le,
p
er
io
d
o
f
k
n
o
w
n
an
d
u
n
k
n
o
w
n
w
r
iti
n
g
s
a
m
p
les
a
v
ailab
le.
W
r
itin
g
s
t
y
le
o
f
a
u
t
h
o
r
ch
a
n
g
es
o
v
er
ti
m
e.
T
o
o
b
s
er
v
e
s
u
c
h
c
h
an
g
e
an
d
i
ts
e
f
f
ec
t
o
n
a
u
th
o
r
id
en
ti
f
ica
tio
n
,
d
ataset
s
h
o
u
ld
co
n
tain
s
a
m
p
l
e
o
v
er
a
b
ig
tim
e
p
er
io
d
,
s
o
th
at
p
er
f
o
r
m
an
ce
o
f
s
y
s
te
m
ca
n
b
e
ev
alu
ated
co
r
r
ec
tly
.
T
h
er
e
ar
e
s
ev
er
al
p
ar
a
m
eter
s
w
h
ich
a
f
f
ec
ts
w
r
i
tin
g
s
t
y
le
o
f
i
n
d
iv
id
u
al
w
h
e
n
b
ig
p
er
io
d
c
o
n
s
id
er
ed
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
9
,
No
.
4
,
A
u
g
u
s
t
201
9
:
3
1
6
7
-
3174
3168
T
h
ese
f
ac
to
r
s
ef
f
ec
ts
o
n
s
t
y
le
o
f
w
r
iti
n
g
in
c
lu
d
es
ed
u
ca
tio
n
,
n
atio
n
alit
y
,
g
e
n
r
e,
ag
e,
to
p
ic,
f
o
r
m
al
co
n
te
n
t,
w
r
itte
n
in
t
h
e
s
a
m
e
p
er
io
d
.
T
h
ese
s
a
m
p
les
ar
e
w
r
itte
n
o
v
er
d
if
f
er
e
n
t
ti
m
e
p
er
io
d
,
co
g
n
i
tiv
e
d
is
to
r
tio
n
,
th
i
n
k
i
n
g
,
e
m
o
tio
n
s
,
p
s
y
c
h
o
lo
g
y
etc.
T
h
ese
ch
alle
n
g
es
h
av
e
o
n
ce
in
a
w
h
ile
b
ee
n
te
n
d
ed
to
b
y
th
e
r
esear
c
h
g
r
o
u
p
.
Ma
n
y
t
ec
h
n
iq
u
e
s
h
a
v
e
s
h
o
w
n
r
e
m
ar
k
ab
le
ad
eq
u
ac
y
in
d
is
t
in
g
u
is
h
i
n
g
t
h
e
g
e
n
u
i
n
e
w
r
iter
s
,
a
n
d
in
r
ev
ie
w
s
e
v
er
al
r
esear
ch
e
s
m
a
d
e
b
y
co
n
s
id
er
in
g
s
a
m
p
le
i
n
s
a
m
e
ti
m
e
p
er
io
d
h
en
ce
d
is
to
r
t
io
n
in
w
r
i
tin
g
s
t
y
l
e
co
n
s
id
er
ed
n
e
g
li
g
ib
le
[
2
]
.
A
r
t
if
icial
n
e
u
r
al
n
et
w
o
r
k
ca
n
b
e
u
s
ed
b
y
tex
t
p
r
o
ce
s
s
i
n
g
[
3
]
.
T
h
er
e
is
r
e
m
ar
k
ab
le
ch
an
g
e
i
n
w
r
it
in
g
s
t
y
le
at
d
if
f
er
en
t
a
g
e
at
co
m
p
o
s
itio
n
n
o
v
el,
co
n
cl
u
d
e
t
h
at
v
o
ca
b
u
lar
y
s
ize
li
m
ited
o
v
e
r
ti
m
e
[
4
]
.
T
em
p
o
r
al
ch
an
g
e
s
o
cc
u
r
r
ed
in
w
r
iti
n
g
s
t
y
le
a
n
d
v
o
c
ab
u
lar
y
u
s
a
g
e
o
n
s
h
o
r
t te
x
t o
v
er
ti
m
e.
A
ut
h
o
r
d
o
h
av
e
c
h
a
n
g
e
i
n
h
is
w
r
i
tin
g
s
t
y
le
b
u
t
d
if
f
er
en
t
a
u
t
h
o
r
s
h
a
v
e
d
if
f
er
en
t
s
t
y
le
[
5
]
.
T
im
e
a
n
d
to
p
ics
ar
e
r
esp
o
n
s
ib
le
f
o
r
ch
an
g
e
i
n
w
r
iti
n
g
s
t
y
le
s
an
d
also
af
f
ec
t t
h
e
ac
cu
r
ac
y
o
f
at
tr
ib
u
tio
n
tas
k
[
6
-
7
]
.
I
n
tr
ad
itio
n
al
ap
p
r
o
ac
h
es,
t
h
e
r
e
ar
e
s
o
m
e
d
is
ad
v
a
n
ta
g
es.
F
ir
s
t
i
s
t
h
at
w
r
iti
n
g
s
a
m
p
le
u
s
ed
in
th
e
attr
ib
u
tio
n
ta
s
k
ar
e
ir
r
esp
ec
ti
v
e
o
f
ti
m
e
p
er
io
d
.
P
r
o
b
ab
ly
all
s
a
m
p
le
s
b
elo
n
g
to
n
ea
r
p
er
io
d
s
.
So
,
th
e
ti
m
e
ef
f
ec
t
o
n
e
v
er
y
s
a
m
p
le
is
n
u
l
lif
ied
.
Seco
n
d
is
t
h
at
r
esear
c
h
li
m
i
ted
to
f
i
x
ed
n
u
m
b
er
o
f
f
ea
tu
r
es
w
o
r
k
ed
o
n
li
m
ited
to
s
i
m
ilar
it
y
-
b
a
s
ed
ap
p
r
o
ac
h
f
o
r
co
n
ten
t
t
y
p
es
o
f
f
ea
t
u
r
es
a
n
d
d
o
cu
m
e
n
t
s
ta
tis
tic
f
ea
tu
r
e
s
u
s
e
s
m
ac
h
in
e
lear
n
i
n
g
ap
p
r
o
ac
h
es.
A
za
r
b
o
n
y
ad
H
.
[
8
]
f
o
cu
s
ed
o
n
au
th
o
r
id
e
n
ti
f
icatio
n
w
i
th
t
h
e
co
n
s
id
er
atio
n
o
f
ti
m
e
p
er
io
d
b
u
t a
p
p
r
o
ac
h
tim
e
p
er
io
d
lim
ited
to
f
o
u
r
y
ea
r
s
a
n
d
u
s
ed
f
ea
tu
r
e
t
y
p
e
i
s
ch
ar
ac
t
er
4
g
r
am
.
I
n
th
i
s
p
ap
er
,
w
e
f
o
u
n
d
ac
c
u
r
ac
y
af
f
ec
ted
o
n
id
e
n
ti
f
icatio
n
,
f
o
r
d
o
cu
m
e
n
t
s
a
m
p
le
co
llecte
d
o
v
er
b
ig
ti
m
e.
W
e
ac
cu
m
u
late
d
ataset
f
r
o
m
n
e
w
s
ar
ticle
s
an
d
co
llec
ted
letter
s
f
r
o
m
f
a
m
o
u
s
p
er
s
o
n
alitie
s
.
F
o
r
m
u
late
n
e
w
ap
p
r
o
ac
h
f
o
r
au
t
h
o
r
id
e
n
ti
f
icatio
n
f
r
o
m
th
e
i
n
s
p
ir
ati
o
n
o
f
ti
m
e
-
ba
s
ed
lan
g
u
a
g
e
m
o
d
el
s
[
5
,
9
,
1
0
]
.
Data
s
et
co
n
s
is
t
o
f
w
r
it
in
g
s
a
m
p
le
w
r
itte
n
b
y
a
u
t
h
o
r
o
v
er
a
n
d
av
er
a
g
e
th
ir
t
y
y
ea
r
s
o
f
ti
m
e
p
er
io
d
.
Featu
r
es,
a
s
elec
ti
v
e
s
et
o
f
at
tr
ib
u
tes
o
n
w
h
ic
h
m
ac
h
i
n
e
lear
n
in
g
ap
p
r
o
ac
h
is
f
o
llo
w
ed
.
O
u
r
m
et
h
o
d
o
lo
g
y
w
o
r
k
ed
o
n
f
ea
t
u
r
es
as
c
h
ar
ac
ter
s
eq
u
en
c
in
g
,
w
o
r
d
s
eq
u
e
n
ci
n
g
,
p
ar
t
o
f
s
p
ee
ch
s
eq
u
e
n
cin
g
a
n
d
co
m
b
in
at
io
n
o
f
t
h
e
m
.
T
h
is
is
f
o
llo
w
ed
b
y
m
ac
h
i
n
e
lear
n
i
n
g
ap
p
r
o
ac
h
f
o
r
cla
s
s
i
f
icatio
n
.
SVM
s
u
p
er
v
is
ed
m
ac
h
i
n
e
lear
n
in
g
ap
p
r
o
ac
h
u
s
ed
f
o
r
d
is
tin
g
u
i
s
h
in
g
t
h
e
u
n
k
n
o
w
n
s
a
m
p
le
to
m
ak
e
it
k
n
o
w
n
.
I
n
d
escr
ib
ed
ap
p
r
o
ac
h
,
a
B
ag
o
f
W
o
r
d
(
B
o
W
)
r
ep
r
esen
tatio
n
s
tr
ateg
y
ap
p
lied
.
Si
m
p
li
f
ied
co
r
p
u
s
s
tr
u
ct
u
r
e
is
u
s
ed
,
a
r
a
w
tex
t
s
a
m
p
les
d
is
tin
g
u
is
h
ed
o
v
er
ti
m
e
s
p
ac
e
w
it
h
t
h
e
id
en
ti
f
icatio
n
o
f
m
o
n
t
h
an
d
y
ea
r
i
n
w
h
ich
d
o
cu
m
en
t
w
r
itte
n
.
R
e
m
a
in
i
n
g
p
ap
er
is
o
r
g
an
ized
as
b
elo
w
.
Sectio
n
2
d
escr
ib
es
th
e
r
elate
d
w
o
r
k
d
o
n
e
in
th
e
ar
ea
.
Me
th
o
d
o
lo
g
y
d
escr
ib
ed
in
s
ec
tio
n
3
.
Secti
o
n
4
elab
o
r
ates
ex
p
er
i
m
e
n
t
r
esu
lt
s
a
n
d
d
is
cu
s
s
io
n
a
n
d
co
n
clu
s
io
n
b
e
th
e
la
s
t
s
ec
tio
n
5
o
f
th
is
p
ap
er
.
2.
RE
L
AT
E
D
S
T
UDY
Au
t
h
o
r
s
h
ip
attr
ib
u
t
io
n
w
o
r
k
s
in
th
r
ee
m
ain
d
o
m
ai
n
s
-
a
u
th
o
r
id
en
tif
icatio
n
,
s
i
m
ilar
it
y
d
et
ec
tio
n
an
d
ch
ar
ac
ter
izatio
n
.
I
n
id
e
n
ti
f
icat
io
n
tas
k
,
t
h
e
h
i
s
to
r
y
o
f
a
u
t
h
o
r
s
h
ip
s
t
y
le
i
s
k
n
o
w
n
i
n
ad
v
an
c
e
an
d
li
k
eli
h
o
o
d
o
f
w
r
iti
n
g
s
a
m
p
le
w
it
h
av
ailab
l
e
in
f
o
r
m
atio
n
.
Au
t
h
o
r
ch
ar
ac
ter
izatio
n
o
u
tli
n
es
t
h
e
attr
ib
u
tes
o
f
a
w
r
i
ter
an
d
p
r
o
d
u
ce
s
th
e
w
r
iter
p
r
o
f
ile
i
n
lig
h
t
o
f
h
i
s
o
r
h
er
w
o
r
k
.
So
m
e
o
f
th
e
s
e
q
u
ali
ties
i
n
cl
u
d
e
s
ex
,
i
n
s
tr
u
cti
v
e
a
n
d
s
o
cial
f
o
u
n
d
atio
n
an
d
d
ialec
t
co
m
m
o
n
alit
y
.
I
n
s
i
m
ilar
it
y
d
etec
tio
n
,
w
o
r
k
o
f
d
i
f
f
er
en
t
au
th
o
r
s
is
co
m
p
ar
ed
w
it
h
s
i
n
g
le
au
t
h
o
r
to
f
in
d
it
its
clo
s
en
ess
[
1
1
]
.
Deta
iled
r
ev
iew
[
7
]
d
escr
ib
es
th
e
m
e
th
o
d
o
lo
g
ies
f
o
r
au
th
o
r
s
h
ip
attr
ib
u
tio
n
s
tar
ti
n
g
f
r
o
m
s
t
y
lo
m
etr
y
f
ea
tu
r
es,
w
h
ich
co
n
s
is
t
o
f
le
x
ical
f
ea
tu
r
es,
ch
ar
ac
ter
f
ea
t
u
r
es,
s
y
n
tact
ic
f
ea
t
u
r
e,
s
e
m
a
n
tic
f
ea
t
u
r
es,
a
p
p
licatio
n
s
p
ec
i
f
ic
f
ea
t
u
r
es.
L
e
x
ical
h
i
g
h
lig
h
t
s
ar
e
w
o
r
d
s
o
r
ch
ar
ac
ter
-
b
ase
d
f
ac
t
u
al
m
ea
s
u
r
es
o
f
a
le
x
ica
l
v
ar
iet
y
[
1
2
]
.
C
h
ar
ac
ter
b
as
ed
f
ea
t
u
r
es
co
n
s
i
s
t
o
f
s
eq
u
e
n
ce
o
f
c
h
ar
ac
ter
s
,
m
ea
s
u
r
ed
a
t
ch
ar
ac
ter
lev
e
l.
S
y
n
tactic
f
ea
t
u
r
e
in
v
o
lv
e
s
s
y
n
tax
o
f
la
n
g
u
a
g
e
u
s
ed
f
o
r
w
r
itin
g
an
d
s
e
m
a
n
ti
c
co
r
r
esp
o
n
d
s
to
m
ea
n
i
n
g
o
f
s
en
ten
ce
f
o
r
w
h
ich
N
L
P
to
o
ls
c
an
b
e
u
s
ed
,
ap
p
licatio
n
ty
p
e
in
v
o
l
v
es
s
p
ec
i
f
icall
y
t
y
p
e
o
f
d
o
cu
m
e
n
t,
d
o
m
ai
n
o
f
d
is
cu
s
s
io
n
in
co
r
p
u
s
[
1
1
]
.
Featu
r
es
ar
e
n
o
th
i
n
g
b
u
t
s
t
y
le
m
ar
k
er
s
.
P
ap
er
[
1
3
]
u
s
e
s
a
s
i
m
ilar
it
y
-
b
a
s
ed
ap
p
r
o
ac
h
w
it
h
r
an
d
o
m
f
ea
t
u
r
es,
w
h
ich
ca
p
tu
r
es
in
f
o
r
m
at
io
n
ab
o
u
t
to
p
ic
an
d
w
r
it
in
g
s
t
y
le
a
n
d
ap
p
lied
n
o
v
el
alg
o
r
ith
m
to
p
r
o
d
u
ce
r
esu
lt.
R
es
u
lt
b
y
t
h
e
m
w
as
o
b
s
er
v
ed
o
n
ab
o
u
t
1
0
0
0
au
t
h
o
r
w
it
h
9
3
.
2
%
p
r
ec
is
io
n
.
Fo
cu
s
w
as
g
iv
e
n
o
n
l
y
f
o
r
ch
ar
ac
ter
4
-
g
r
a
m
,
a
s
in
g
le
f
ea
t
u
r
e
w
i
th
ab
o
u
t
2
0
0
0
0
attr
ib
u
tes.
T
h
en
th
is
r
es
u
lt
w
a
s
co
m
p
ar
ed
w
it
h
d
is
tr
ib
u
tin
g
d
ataset
i
n
to
t
r
ain
in
g
a
n
d
test
i
n
g
,
r
ep
ea
ted
l
y
.
R
esear
ch
i
n
v
is
u
alize
d
ev
e
n
t
d
r
iv
en
ap
p
r
o
ac
h
[
1
4
]
ca
n
b
e
v
is
u
al
ized
an
d
in
ter
p
r
eted
.
A
p
p
r
o
ac
h
w
a
s
b
ased
o
n
v
is
u
aliza
t
io
n
o
f
f
i
n
g
er
p
r
in
t
co
m
p
ar
is
o
n
.
Feat
u
r
e
s
et
is
b
ased
o
n
t
w
o
t
y
p
e
s
,
s
et
o
n
e
is
u
n
i
f
ied
a
n
d
an
o
th
er
is
clas
s
s
p
ec
i
f
ic.
A
g
r
o
u
p
o
f
w
r
iti
n
g
s
t
y
le
f
ea
t
u
r
e
g
ath
er
ed
as
ev
id
e
n
ce
u
n
it
alo
n
g
w
i
t
h
it
s
s
co
r
in
g
v
ec
to
r
.
An
al
y
s
i
s
w
as
m
ad
e
o
n
s
u
c
h
ev
e
n
t
s
co
r
e
to
f
i
n
d
tar
g
e
ted
cr
ea
to
r
.
Su
itab
ilit
y
o
f
m
et
h
o
d
is
li
m
ited
to
2
0
au
th
o
r
s
.
W
o
r
d
n
-
g
r
a
m
,
ch
ar
ac
ter
n
-
g
r
a
m
,
P
o
S
n
-
g
r
a
m
f
ea
tu
r
es
u
s
ed
to
co
n
s
tr
u
ct
ev
en
t.
T
h
e
ap
p
r
o
ac
h
ex
p
lain
ed
i
n
[
1
5
]
,
b
ased
o
n
th
eo
r
etica
l
s
tu
d
y
o
f
f
u
n
ctio
n
w
o
r
d
s
in
au
t
h
o
r
s
h
ip
attr
ib
u
t
io
n
.
Fu
n
ctio
n
w
o
r
d
s
ar
e
n
o
th
i
n
g
b
u
t
e
x
tr
an
eo
u
s
w
o
r
d
s
u
s
ed
w
i
th
o
u
t
a
f
f
ec
ti
n
g
m
ea
n
i
n
g
o
f
s
e
n
te
n
ce
.
Sta
m
ato
s
[
1
6
]
elab
o
r
ated
f
ast
te
s
t
ca
teg
o
r
izatio
n
m
e
th
o
d
s
.
I
n
th
e
w
o
r
k
,
u
s
ed
N
L
P
to
o
l
to
er
a
d
icate
th
e
s
t
y
l
is
tic
f
ac
ts
.
Mu
l
t
ip
le
r
eg
r
ess
io
n
an
d
d
is
cr
i
m
i
n
an
t
an
al
y
s
i
s
clas
s
i
f
i
ca
tio
n
m
o
d
els
w
er
e
u
s
ed
to
ca
teg
o
r
ize
cr
ea
t
o
r
’
s
f
ac
t
s
.
O
n
e
o
f
th
e
r
esear
c
h
p
ap
er
[
1
7
]
d
ea
ls
w
it
h
s
o
u
r
ce
co
d
e
w
r
itte
n
b
y
d
if
f
er
en
t
p
r
o
g
r
a
m
m
er
w
h
ic
h
w
as
id
en
ti
f
i
ed
o
n
th
e
b
asis
o
n
n
-
g
r
a
m
a
u
t
h
o
r
p
r
o
f
ile.
T
h
is
m
eth
o
d
is
b
ased
o
n
b
y
te
lev
el
n
-
g
r
a
m
f
ea
tu
r
es
o
n
d
if
f
er
en
t
s
o
u
r
ce
co
d
e
w
r
itte
n
b
y
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8708
A
r
o
b
u
s
t a
u
th
o
r
s
h
ip
a
ttr
ib
u
tio
n
o
n
b
ig
p
erio
d
(
Mu
b
in
S
h
o
u
k
a
t Ta
mb
o
li
)
3169
d
if
f
er
e
n
t
a
u
t
h
o
r
in
j
av
a
o
r
C
++
lan
g
u
a
g
e.
I
n
t
h
is
ap
p
r
o
ac
h
,
f
ix
ed
n
u
m
b
er
o
f
m
o
s
t
f
r
e
q
u
en
t
n
-
g
r
a
m
f
r
o
m
s
o
u
r
ce
co
d
e
f
ile
w
as c
o
n
s
id
er
ed
.
A
s
i
m
ilar
it
y
m
ea
s
u
r
e
d
ef
i
n
ed
b
y
Kesej
i
[
1
8
]
w
as
u
s
ed
to
f
in
d
r
ela
ti
v
e
d
is
ta
n
ce
a
m
o
n
g
t
w
o
s
t
y
le
s
.
On
t
h
e
a
v
ailab
le
d
ataset,
t
h
e
ap
p
r
o
ac
h
s
h
o
w
s
8
8
%
ac
c
u
r
ac
y
.
Sp
ati
u
m
-
L
1
,
a
n
e
f
f
ec
ti
v
e
au
th
o
r
v
er
i
f
icatio
n
m
o
d
el
b
ased
a
s
tr
u
c
tu
r
e
o
f
u
n
s
u
p
er
v
i
s
ed
lear
n
i
n
g
al
g
o
r
ith
m
[
1
9
]
.
I
t
m
ak
e
s
u
s
e
o
f
2
0
0
m
o
s
tl
y
o
cc
u
r
r
ed
ter
m
s
o
f
u
n
k
n
o
w
n
tex
t.
T
h
e
m
et
h
o
d
d
escr
ib
ed
in
th
e
p
ap
er
w
a
s
d
is
tan
ce
b
ased
.
Au
t
h
o
r
h
as
r
ep
r
esen
ted
h
is
w
o
r
k
i
n
P
A
N
C
L
E
F
2
0
1
4
an
d
g
iv
e
s
g
o
o
d
co
m
p
ar
ativ
e
r
es
u
lt
a
g
ain
s
t
e
x
i
s
tin
g
al
g
o
r
ith
m
s
.
U
s
ed
f
ea
tu
r
es
i
n
t
h
e
m
et
h
o
d
s
w
er
e
w
o
r
d
t
y
p
e
an
d
p
u
n
ct
u
atio
n
s
y
m
b
o
l
s
.
A
ca
s
e
s
t
u
d
y
o
n
v
o
ca
b
u
lar
y
c
h
an
g
es
[
4
]
o
n
le
x
ical,
s
y
n
tactic
a
n
d
d
is
co
u
r
s
e
w
a
s
o
b
s
er
v
ed
.
C
o
r
p
u
s
f
r
o
m
n
o
v
el
w
a
s
u
s
ed
f
o
r
ex
p
er
i
m
e
n
ts
a
n
d
o
n
l
y
f
ir
s
t
5
0
0
0
0
w
o
r
d
s
u
s
ed
.
Ob
s
er
v
ed
f
ac
ts
wer
e
co
u
n
t
o
f
v
o
ca
b
u
lar
y
s
ize,
r
ich
n
e
s
s
a
n
d
u
n
iq
u
e
w
o
r
d
s
as
a
o
n
e
m
ea
s
u
r
e
an
d
w
o
r
d
n
-
g
r
a
m
,
w
o
r
d
len
g
th
was
o
th
er
an
d
th
ir
d
t
y
p
e
as
o
cc
u
r
r
en
ce
s
o
f
v
ag
u
e
an
d
in
d
ef
i
n
ite
w
o
r
d
s
w
er
e
ass
u
m
ed
a
s
f
ac
ts
.
A
t
d
if
f
er
en
t
ag
e
o
f
co
m
p
o
s
itio
n
f
o
u
n
d
v
a
r
iatio
n
s
o
n
co
n
s
id
er
ed
f
ac
ts
.
An
o
th
er
c
h
alle
n
g
in
g
task
i
n
a
u
t
h
o
r
s
h
ip
attr
ib
u
tio
n
i
s
to
h
a
n
d
le
m
e
s
s
a
g
es
w
i
th
v
ar
y
in
g
le
n
g
th
.
Dig
i
tal
co
m
m
u
n
i
ca
tio
n
o
v
er
i
n
ter
n
et
is
al
w
a
y
s
i
n
f
o
r
m
o
f
s
h
o
r
t
m
es
s
ag
e
s
in
ter
m
s
o
f
e
m
ail,
c
h
at
m
es
s
ag
e
s
,
t
w
ee
ts
etc.
Z
h
en
g
[
1
1
]
co
m
es
w
i
th
o
n
e
o
f
th
e
id
ea
to
id
en
ti
f
y
au
t
h
o
r
f
o
r
o
n
lin
e
m
e
s
s
a
g
es.
E
n
g
li
s
h
a
n
d
C
h
in
e
s
e
lan
g
u
a
g
e
c
h
o
s
en
f
o
r
ex
p
er
i
m
en
ti
n
g
.
Featu
r
e
s
w
er
e
co
n
s
id
er
ed
in
g
r
o
u
p
s
a
s
le
x
ical
f
e
atu
r
es,
w
o
r
d
-
b
ased
f
ea
t
u
r
es,
s
y
n
tact
ic
f
ea
t
u
r
es,
co
n
ten
t
-
s
p
ec
i
f
ic
f
ea
t
u
r
es,
s
tr
u
ctu
r
al
f
ea
t
u
r
es,
C
4
.
5
,
NN,
SVM
class
if
icatio
n
m
o
d
el
s
u
s
ed
f
o
r
id
en
ti
f
icatio
n
.
R
es
u
lt
s
b
ased
o
n
f
ea
t
u
r
e
t
y
p
e
s
an
d
tec
h
n
iq
u
es
li
k
e
SV
M
an
d
NN
p
r
o
d
u
ce
d
ch
alle
n
g
i
n
g
r
es
u
lt
s
.
C
o
g
n
iti
v
e
er
r
o
r
is
an
o
t
h
er
a
s
p
ec
t
f
o
r
attr
ib
u
tio
n
.
A
w
r
iter
m
a
k
es
b
asic
m
is
tak
e
s
w
i
th
r
eg
ar
d
to
f
e
w
is
s
u
e
s
s
p
ec
i
f
icall
y
:
C
au
s
al
P
r
e
m
is
e,
P
r
o
b
ab
ilit
y
J
u
d
g
m
en
t
a
n
d
C
o
n
d
itio
n
al
R
elia
n
ce
.
T
h
e
r
esear
ch
er
’
s
b
lu
n
d
er
s
ar
e
i
m
p
er
ativ
e
s
i
n
ce
th
e
y
h
a
v
e
a
lar
g
e
ef
f
ec
t
o
n
h
i
s
co
n
clu
s
io
n
s
an
d
s
i
n
ce
co
m
p
ar
ati
v
e
m
i
s
tak
e
s
f
r
eq
u
e
n
tl
y
h
ap
p
en
w
h
e
n
in
d
i
v
id
u
al
s
,
b
o
th
s
p
ec
iali
s
ts
an
d
a
m
ate
u
r
s
ar
e
co
n
f
r
o
n
ted
w
it
h
th
e
ch
al
len
g
es
o
f
B
a
y
esia
n
in
f
er
en
ce
[
2
0
]
.
C
h
en
g
[
2
1
]
r
ec
o
g
n
izes
t
h
at
t
h
e
is
s
u
e
o
f
s
ex
u
al
o
r
ien
tatio
n
r
ec
o
g
n
izab
le
p
r
o
o
f
f
r
o
m
te
x
t
i
s
an
in
ter
ac
tio
n
b
et
w
ee
n
p
s
y
c
h
o
-
li
n
g
u
is
tic
s
,
n
o
n
s
p
ec
if
ic
w
r
i
tin
g
s
t
y
les
o
f
m
e
n
an
d
lad
ies.
I
n
th
eir
s
tu
d
ie
s
th
e
y
h
av
e
u
s
ed
th
r
ee
alg
o
r
it
h
m
s
v
iz
SVM,
B
a
y
esia
n
lo
g
i
s
tic
r
eg
r
ess
io
n
a
n
d
A
d
ab
o
o
s
t
d
ec
is
io
n
tr
ee
.
A
cc
u
r
ac
y
ca
p
t
u
r
ed
is
ar
o
u
n
d
8
5
%.
C
o
n
tr
ib
u
ted
f
ea
tu
r
e
s
i
n
d
is
cr
i
m
i
n
atio
n
w
er
e
f
u
n
ctio
n
w
o
r
d
s
,
w
o
r
d
-
b
ased
f
ea
t
u
r
es,
s
tr
u
ct
u
r
al
f
ea
tu
r
es.
T
h
e
m
o
d
el
ap
p
lied
o
n
C
o
r
p
u
s
f
r
o
m
r
eu
ter
s
an
d
en
r
o
n
e
m
a
il
d
ataset.
T
ex
t
is
r
ep
r
esen
ted
in
t
h
e
f
o
r
m
o
f
v
ec
to
r
s
.
Ma
n
y
t
y
p
es
o
f
w
i
n
d
o
w
a
lg
o
r
ith
m
s
ar
e
ap
p
lied
to
d
is
cr
i
m
i
n
ate
a
m
o
n
g
s
e
v
er
al
au
t
h
o
r
s
[
2
2
]
to
p
r
o
d
u
ce
co
m
p
r
o
m
i
s
in
g
ac
c
u
r
ac
y
.
W
r
itep
r
in
t,
a
n
e
w
tech
n
iq
u
e
in
tr
o
d
u
ce
d
[
1
2
]
in
w
h
ic
h
s
lid
in
g
w
in
d
o
w
f
ea
t
u
r
es
w
er
e
co
n
s
id
er
ed
f
o
r
th
e
ap
p
licatio
n
o
f
la
n
g
u
a
g
e
m
o
d
el.
Ma
n
y
t
y
p
e
s
o
f
f
ea
t
u
r
es
w
er
e
ac
cu
m
u
lated
an
d
ap
p
lied
o
n
t
h
is
n
e
w
m
o
d
el
w
h
ich
p
r
o
d
u
ce
s
ac
c
u
r
ac
y
ar
o
u
n
d
g
r
ea
ter
th
an
9
0
%.
A
ll
t
h
ese
r
ev
ie
w
ed
r
esear
ch
es
f
o
cu
s
e
d
o
n
th
e
m
et
h
o
d
s
f
o
r
au
th
o
r
attr
ib
u
tio
n
an
d
id
en
ti
f
icatio
n
,
ti
m
e
w
h
en
t
h
e
d
o
cu
m
e
n
t g
e
n
er
ated
w
as
n
o
t c
o
n
s
id
er
ed
.
A
za
r
b
o
n
y
ad
[
8
]
s
t
u
d
ied
attr
ib
u
tio
n
w
h
er
e
w
r
iti
n
g
n
at
u
r
e
o
f
au
th
o
r
c
h
a
n
g
e
s
.
T
h
ese
te
m
p
o
r
al
ch
a
n
g
es
ar
e
o
b
s
er
v
ed
w
it
h
r
esp
ec
t
to
w
o
r
d
d
is
tr
ib
u
tio
n
in
w
r
iti
n
g
s
a
m
p
les.
I
n
th
e
ex
p
er
i
m
en
t
d
o
n
e,
t
w
ee
t
a
n
d
en
r
o
n
e
m
ail
d
ata
s
et
w
as
co
n
s
id
er
ed
o
v
er
th
e
p
er
io
d
o
f
5
y
ea
r
s
.
C
h
ar
ac
ter
4
-
g
r
a
m
w
as
t
h
e
o
b
s
er
v
ed
f
ea
tu
r
es
.
T
em
p
o
r
al
ch
an
g
es
w
er
e
ca
p
t
u
r
ed
w
it
h
al
g
o
r
ith
m
d
ef
i
n
ed
i
n
p
ap
er
[
9
,
1
0
]
,
tim
e
-
b
ased
l
an
g
u
a
g
e
m
o
d
el
an
d
ca
lcu
lated
f
r
o
m
li
n
ea
r
r
eg
r
ess
io
n
tech
n
iq
u
es.
R
e
s
ea
r
ch
w
o
r
k
[
4
]
elab
o
r
ates
ch
an
g
e
i
n
v
o
ca
b
u
lar
y
u
s
ag
e
b
y
a
w
r
i
ter
an
d
p
r
o
v
ed
t
h
at
s
ize
o
f
v
o
ca
b
u
lar
y
g
o
es
o
n
d
ec
r
ea
s
in
g
o
v
er
t
i
m
e.
T
i
m
e
f
r
a
m
e
o
v
er
th
e
w
o
r
k
w
as
b
ig
ab
o
u
t
u
p
to
3
5
y
ea
r
s
.
Au
th
o
r
[
6
]
also
in
v
esti
g
ated
an
d
co
n
clu
d
ed
w
r
iti
n
g
s
t
y
le
o
f
au
t
h
o
r
ch
an
g
es
o
v
er
ti
m
e
an
d
au
t
h
o
r
s
h
ip
v
er
if
icatio
n
ac
cu
r
ac
y
i
n
cr
ea
s
es
w
h
e
n
r
ec
o
r
d
is
co
m
p
o
s
ed
in
a
b
r
ief
p
er
io
d
o
f
ti
m
e.
R
ev
ie
w
o
f
d
if
f
er
e
n
t
m
et
h
o
d
s
o
f
clas
s
i
f
ica
tio
n
a
n
d
th
e
ir
r
es
u
lts
is
e
lab
o
r
ated
in
p
ap
er
[
2
]
.
An
d
[
2
3
]
u
s
es
m
ac
h
i
n
e
lear
n
i
n
g
ap
p
r
o
ac
h
f
o
r
co
n
t
en
t
t
y
p
es
o
f
f
ea
t
u
r
e
w
h
er
e
w
r
iti
n
g
s
a
m
p
les
ar
e
s
h
o
r
ts
a
n
d
ir
r
esp
ec
tiv
e
o
f
ti
m
e.
I
n
[
2
4
]
s
h
o
w
th
e
v
ar
iatio
n
o
f
f
ea
tu
r
es
o
v
er
t
i
m
e.
Var
iat
io
n
w
er
e
n
o
t
s
tat
io
n
ar
y
in
t
h
e
w
o
r
k
.
B
u
t
in
d
icate
s
t
h
er
e
is
c
h
an
g
e
i
n
w
r
iti
n
g
s
t
y
le
o
f
au
th
o
r
.
I
n
th
e
s
u
r
v
e
y
[
2
5
]
,
d
escr
ib
es
d
if
f
e
r
en
t
t
y
p
es
o
f
f
ea
t
u
r
es
an
d
th
e
ir
co
m
b
i
n
atio
n
a
n
d
ap
p
lied
o
n
ar
ab
ic
tex
t f
o
r
to
th
e
au
th
o
r
s
h
ip
attr
ib
u
tio
n
tas
k
.
2
.
1
.
F
ea
t
ure
s
elec
t
io
n
Au
t
h
o
r
s
h
ip
a
ttrib
u
tio
n
i
s
a
n
i
n
f
o
r
m
atio
n
r
etr
ie
v
al
ta
s
k
,
w
h
er
e
f
ea
t
u
r
e
u
s
ed
i
n
th
e
w
o
r
k
ca
n
af
f
ec
t
th
e
o
u
tco
m
e
s
p
r
o
d
u
ce
d
.
Fro
m
ab
o
v
e
liter
atu
r
e,
f
ea
t
u
r
es
ar
e
b
r
o
a
d
ly
ca
te
g
o
r
ized
in
t
y
p
es
s
u
c
h
as
w
r
iti
n
g
s
t
y
le
an
d
co
n
ten
t
s
p
ec
i
f
ic.
I
n
w
r
i
tin
g
s
t
y
le,
w
r
iti
n
g
s
t
y
le
o
f
a
u
t
h
o
r
is
ca
p
tu
r
ed
in
d
i
f
f
er
en
t
m
ea
n
s
as
s
t
y
le
o
f
a
u
th
o
r
.
I
n
co
n
ten
t
s
p
ec
if
ic,
i
m
p
o
r
ta
n
ce
is
g
iv
e
n
t
o
w
r
i
tin
g
co
n
te
n
t
an
d
its
m
ea
n
i
n
g
.
I
n
co
n
te
n
t
s
p
ec
if
ic
f
ea
t
u
r
es,
w
e
ar
e
m
o
s
tl
y
co
n
ce
r
n
ed
ab
o
u
t
w
o
r
d
n
-
g
r
a
m
,
ch
ar
ac
ter
n
-
g
r
a
m
,
f
r
eq
u
e
n
c
y
o
f
w
o
r
d
u
s
a
g
e,
co
g
n
iti
v
e
er
r
o
r
s
m
ad
e,
co
n
ten
t
s
p
ec
i
f
ic
f
ea
t
u
r
e
s
etc.
T
h
ese
f
ea
tu
r
es
ar
e
co
n
s
i
d
er
ed
as
g
r
o
u
p
b
ec
au
s
e
ea
ch
ty
p
e
o
f
t
h
e
f
ea
t
u
r
e
is
co
n
s
is
tin
g
o
f
m
a
n
y
attr
ib
u
tes.
I
t is al
w
a
y
s
ev
al
u
ated
i
n
g
r
o
u
p
s
.
So
m
eti
m
e
er
r
o
r
o
cc
u
r
r
ed
in
d
o
cu
m
en
t c
an
al
s
o
b
e
tr
ea
ted
as
o
n
e
o
f
t
h
e
e
f
f
ec
tiv
e
f
ea
t
u
r
es.
T
h
er
e
ar
e
f
ea
t
u
r
es
w
h
ich
ca
p
t
u
r
es
b
o
th
a
s
p
ec
t
co
n
ten
t
an
d
s
t
y
le,
ch
ar
ac
ter
n
-
g
r
a
m
is
o
n
e
o
f
th
e
m
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
9
,
No
.
4
,
A
u
g
u
s
t
201
9
:
3
1
6
7
-
3174
3170
2
.
2
.
Cha
ra
ct
er
n
-
g
ra
m
I
n
lin
g
u
is
tic
ar
it
h
m
etic,
ch
ar
a
cter
n
-
g
r
a
m
i
s
co
n
s
i
s
ti
n
g
o
f
a
co
n
tin
u
al
s
eq
u
en
ce
o
f
n
ter
m
s
in
a
g
iv
e
n
s
a
m
p
le.
I
t
ca
n
b
e
p
h
o
n
e
m
e
wh
ich
is
a
g
e
s
t
u
r
e
o
f
s
o
u
n
d
w
h
i
ch
d
if
f
er
en
t
iates
o
n
e
f
r
o
m
a
n
o
th
er
,
s
y
llab
le
w
h
ic
h
d
is
tin
g
u
is
h
es
s
eq
u
e
n
ce
o
f
s
p
ee
ch
s
o
u
n
d
in
w
o
r
d
s
e.
g
.
m
ath
e
m
atic
s
co
m
p
o
s
ed
o
f
t
w
o
s
y
ll
ab
les
v
iz
m
at
h
e
a
n
d
m
atic
s
,
letter
s
,
w
o
r
d
s
etc.
I
n
co
n
tin
u
al
s
eq
u
e
n
ce
,
it
i
s
n
ee
d
ed
to
d
ef
in
e
w
h
et
h
er
it
i
s
i
n
w
o
r
d
o
r
co
m
p
lete
d
o
cu
m
en
t.
I
t
m
ea
n
s
w
h
eth
er
it
co
n
s
id
er
s
th
e
s
p
ac
e,
o
p
er
ato
r
,
an
d
p
u
n
ct
u
atio
n
i
n
w
o
r
d
s
o
r
n
o
t.
2
.
3
.
Wo
rd
n
-
g
ra
m
I
n
th
is
co
n
tex
t,
te
x
t
is
v
ie
w
ed
as
s
eq
u
e
n
ce
o
f
w
o
r
d
s
.
I
n
w
o
r
d
n
-
g
r
a
m
w
o
r
d
s
ar
e
co
llected
,
w
h
ic
h
h
as
n
co
n
t
ig
u
o
u
s
w
o
r
d
s
.
I
t
ca
p
t
u
r
es
co
n
ten
t
s
p
ec
i
f
ic
i
n
f
o
r
m
atio
n
r
ath
er
th
a
n
s
t
y
lis
tic
i
n
f
o
r
m
atio
n
f
r
o
m
t
h
e
co
r
p
u
s
[
7
]
.
A
cc
o
r
d
in
g
to
s
u
r
v
e
y
m
ad
e
b
y
Sta
m
atato
s
,
w
o
r
d
n
-
g
r
a
m
i
s
u
s
ed
f
o
r
au
th
o
r
id
en
ti
f
icatio
n
.
I
t d
o
esn
’
t
al
w
a
y
s
g
i
v
e
p
r
o
m
i
s
i
n
g
r
es
u
l
ts
r
ath
er
th
a
n
o
th
er
s
.
Ma
n
y
ti
m
es,
s
u
c
h
co
n
tig
u
o
u
s
w
o
r
d
s
ar
e
n
o
t
alw
a
y
s
o
cc
u
r
r
in
g
i
n
s
y
s
te
m
,
s
p
ec
if
i
c
all
y
s
h
o
r
t
tex
t.
I
t
m
a
y
n
o
t
g
i
v
e
co
r
r
ec
t
in
f
o
r
m
atio
n
a
ll
t
h
e
ti
m
e
b
ec
au
s
e
it
i
s
in
ca
p
ab
le
w
h
en
w
r
it
in
g
er
r
o
r
is
in
tr
o
d
u
ce
d
in
s
a
m
p
le.
I
t
ca
p
tu
r
es
h
u
m
a
n
b
eh
a
v
io
r
b
u
t
it
is
p
o
s
s
ib
le
th
at
t
h
e
b
eh
av
io
r
m
a
y
c
h
an
g
e
o
v
er
ti
m
e.
Fo
r
s
h
o
r
t
tex
t,
t
h
er
e
is
le
s
s
p
o
s
s
i
b
ili
t
y
o
f
ca
p
tu
r
i
n
g
s
u
c
h
r
ep
ea
tin
g
b
eh
a
v
io
r
.
I
n
p
ap
er
[
2
3
]
w
o
r
d
n
-
g
r
a
m
f
ea
tu
r
e
is
u
s
ed
to
g
at
h
er
s
e
m
a
n
ticall
y
m
ea
n
in
g
f
u
l
i
n
f
o
r
m
ati
o
n
f
r
o
m
s
a
m
p
le
o
f
s
h
o
r
t te
x
t.
2
.
4
.
P
a
rt
o
f
s
peec
h
n
-
g
ra
m
T
h
is
f
ea
t
u
r
e
ca
p
tu
r
es
s
t
y
lis
tic
in
f
o
r
m
atio
n
f
r
o
m
g
i
v
e
n
s
a
m
p
l
e
o
f
tex
t.
T
o
g
en
er
ate
t
h
i
s
f
ea
t
u
r
e,
f
ir
s
t
i
t
is
r
eq
u
ir
ed
to
tag
th
e
tex
t
s
a
m
p
le.
I
t
is
th
e
w
a
y
o
f
i
n
cr
ea
s
in
g
a
w
o
r
d
in
co
n
te
n
t
(
co
r
p
u
s
)
as
co
m
p
ar
in
g
to
a
s
p
ec
if
ic
g
r
a
m
m
at
ical
f
o
r
m
,
in
lig
h
t
o
f
t
h
e
t
w
o
.
I
ts
d
ef
i
n
it
io
n
an
d
r
elatio
n
s
h
ip
w
it
h
n
ea
r
b
y
w
o
r
d
,
s
en
ten
ce
a
n
d
p
ar
ag
r
ap
h
.
I
t is f
ir
m
l
y
f
i
x
i
n
g
to
co
r
p
u
s
et
y
m
o
lo
g
ical.
Un
iv
er
s
al
ta
g
s
et
co
n
s
is
t o
f
f
o
llo
w
i
n
g
tag
s
.
2
.
5
.
F
un
ct
io
n
w
o
rds
Fu
n
ctio
n
w
o
r
d
s
ar
e
w
o
r
d
in
g
w
h
o
s
e
d
esig
n
is
to
ad
d
th
e
lin
g
u
is
tic
s
tr
u
c
tu
r
e
as
o
p
p
o
s
ed
to
th
e
s
ig
n
i
f
ica
n
ce
o
f
a
s
en
te
n
ce
.
I
t
is
o
p
en
class
w
o
r
d
,
w
h
ic
h
in
cl
u
d
es
ad
j
ec
tiv
e,
n
o
u
n
,
v
er
b
etc.
E
x
am
p
le
o
f
f
u
n
ctio
n
w
o
r
d
s
ar
e
‘
o
f
’
,
‘
at
’
,
‘
i
n
’
,
‘
t
h
at
’
,
‘
d
o
’
etc.
F
u
n
ctio
n
w
o
r
d
s
ca
n
b
e
co
n
s
id
er
ed
as
a
b
ase
f
o
r
te
x
t
u
al
co
m
p
ar
is
o
n
.
I
n
g
e
n
er
al
co
u
n
t
o
f
f
u
n
ctio
n
w
o
r
d
s
i
n
a
n
y
s
a
m
p
le
is
lar
g
e
w
h
ic
h
ca
n
a
ct
a
s
d
is
ti
n
g
u
is
h
ed
f
ea
t
u
r
e
f
o
r
id
en
ti
f
y
i
n
g
au
t
h
o
r
.
I
t
i
s
n
o
t
f
o
cu
s
i
n
g
to
w
ar
d
s
t
h
e
co
n
t
en
t
o
f
te
x
t;
it
f
o
cu
s
es
o
n
s
t
y
lis
tic
f
ea
t
u
r
e
o
f
a
n
au
th
o
r
.
I
t
ca
n
n
o
t
b
e
d
ir
ec
tl
y
a
p
p
lied
to
d
is
cr
im
i
n
ate
b
et
w
ee
n
au
th
o
r
s
.
I
n
[
2
6
]
m
et
h
o
d
o
lo
g
y
h
a
v
e
m
o
r
e
t
h
a
n
1
7
5
f
u
n
ctio
n
w
o
r
d
s
.
2
.
6
.
Cla
s
s
if
ica
t
io
n
Au
t
h
o
r
s
h
ip
v
er
i
f
icatio
n
p
r
o
b
le
m
ca
n
b
e
s
o
lv
ed
b
y
s
i
m
i
lar
it
y
-
b
a
s
ed
ap
p
r
o
ac
h
an
d
m
ac
h
in
e
lear
n
i
n
g
ap
p
r
o
ac
h
.
Ma
ch
in
e
lear
n
in
g
class
i
f
icatio
n
m
o
d
el
f
o
llo
w
s
i
n
s
ta
n
ce
-
b
ased
ap
p
r
o
ac
h
.
I
n
i
n
s
ta
n
ce
-
b
ased
ap
p
r
o
ac
h
ea
ch
tr
ain
in
g
s
a
m
p
l
e
is
id
en
tif
ied
u
n
iq
u
e
l
y
i
n
attr
ib
u
tio
n
m
o
d
el.
C
la
s
s
i
f
icatio
n
m
o
d
el
u
s
es
v
ar
io
u
s
s
u
p
er
v
i
s
ed
an
d
u
n
s
u
p
er
v
i
s
ed
m
et
h
o
d
f
o
r
th
e
attr
ib
u
tio
n
.
I
n
th
e
d
escr
ip
tio
n
w
e
o
b
e
y
SV
M
m
ac
h
in
e
lear
n
i
n
g
m
et
h
o
d
as
i
n
s
ta
n
ce
-
b
ased
ap
p
r
o
ac
h
f
o
r
class
i
f
ica
tio
n
[
7
]
,
[
2
3
]
,
[
1
9
]
.
I
n
m
a
c
h
i
n
e
lear
n
i
n
g
b
ased
ap
p
r
o
ac
h
,
w
r
iti
n
g
s
t
y
le
o
f
ea
ch
k
n
o
w
n
a
u
th
o
r
i
s
id
e
n
ti
f
ied
as
tr
ain
in
g
s
a
m
p
le,
w
h
ic
h
is
u
s
ed
to
b
u
ild
class
i
f
ier
,
w
h
ic
h
o
n
n
ex
t
u
s
ed
to
clas
s
i
f
y
u
n
k
n
o
w
n
s
a
m
p
le.
Her
e
is
n
ee
d
o
f
e
n
o
u
g
h
s
a
m
p
le
to
tr
ai
n
clas
s
i
f
ier
s
o
it
ca
n
b
e
u
s
ed
o
n
f
u
r
t
h
er
[
1
3
]
.
Dec
is
io
n
tr
ee
a
lg
o
r
ith
m
[
2
7
]
u
s
ed
in
au
t
h
o
r
id
en
ti
f
icatio
n
f
o
r
Ma
r
at
h
i
la
n
g
u
a
g
e.
Si
m
ilar
it
y
b
ase
d
ap
p
r
o
ac
h
is
an
o
th
er
o
n
e,
h
er
e
d
is
tan
ce
m
etr
ic
is
u
s
ed
to
d
is
cr
i
m
i
n
ate
b
et
w
ee
n
t
w
o
s
a
m
p
l
e,
if
o
n
e
s
a
m
p
le
is
m
o
s
t
s
i
m
ilar
w
ith
a
n
o
th
er
t
h
e
n
co
n
cl
u
d
e
th
at
i
s
w
r
itte
n
w
it
h
s
a
m
e
au
t
h
o
r
.
T
h
is
is
d
ir
ec
t
ap
p
r
o
ac
h
h
en
ce
n
o
t
co
n
s
id
er
ed
in
an
i
n
s
ta
n
ce
-
b
ase
d
ap
p
r
o
ac
h
.
Su
p
p
o
r
t
Vec
to
r
Ma
ch
i
n
e
ac
t
as
d
is
cr
i
m
i
n
ati
n
g
clas
s
i
f
ier
w
h
ic
h
s
ep
ar
ates
d
ata
s
a
m
p
le
th
r
o
u
g
h
a
h
y
p
er
p
la
n
e.
A
n
u
m
b
er
o
f
h
y
p
er
p
lan
es se
p
ar
ate
o
u
t d
ata
s
a
m
p
le
i
n
to
n
u
m
b
er
o
f
clas
s
es.
I
t is
s
u
itab
le
to
w
o
r
k
w
it
h
h
i
g
h
d
i
m
e
n
s
io
n
al
d
ata,
h
en
ce
s
u
i
tab
le
f
o
r
o
u
r
ap
p
r
o
ac
h
.
I
t
co
n
s
i
s
ts
o
f
s
et
o
f
tr
ain
in
g
p
o
in
t
ac
t
as
v
ec
to
r
w
h
ic
h
is
r
ep
r
esen
ted
in
b
ag
o
f
w
o
r
d
s
[
2
8
]
f
o
r
m
.
Var
io
u
s
k
er
n
el
f
u
n
c
tio
n
s
c
a
n
b
e
u
s
ed
,
it
is
a
k
in
d
o
f
alg
o
r
ith
m
u
s
ed
f
o
r
p
atter
n
an
a
l
y
s
i
s
.
A
k
er
n
el
i
s
u
s
ed
to
m
ak
e
lin
ea
r
m
o
d
el
to
n
o
n
li
n
ea
r
m
o
d
el.
SVM
s
u
p
p
o
r
ts
f
o
r
d
if
f
er
e
n
t t
y
p
e
o
f
k
er
n
e
l [
2
9
]
.
3.
M
E
T
H
O
DO
L
O
G
Y
Ou
r
ap
p
r
o
ac
h
is
b
ased
o
n
r
e
p
r
esen
tatio
n
o
f
s
a
m
p
led
tex
t
to
k
en
an
d
t
h
ei
r
g
r
o
u
p
i
n
g
s
.
M
o
s
t
s
i
m
ilar
au
th
o
r
f
o
r
th
e
id
en
t
if
ica
tio
n
b
ased
th
e
cla
s
s
i
f
icatio
n
m
ad
e
b
y
o
u
r
s
y
s
te
m
.
W
e
f
o
llo
w
e
d
m
ac
h
in
e
lear
n
i
n
g
ap
p
r
o
ac
h
to
b
u
ild
m
o
d
el.
Fea
tu
r
es
r
ep
r
esen
tatio
n
in
ter
m
s
o
f
v
ec
to
r
an
d
u
s
ed
in
s
tan
ce
b
ase
ap
p
r
o
ac
h
an
d
f
o
llo
w
ed
b
y
SVM
a
lg
o
r
i
t
h
m
f
o
r
co
n
s
tr
u
c
tin
g
m
o
d
el.
Dif
f
er
en
t
f
e
at
u
r
e
s
e
ts
u
s
e
d
to
s
etu
p
m
o
d
el,
f
o
r
v
er
i
f
icatio
n
o
f
th
e
m
o
d
el
w
e
d
e
f
i
n
e
o
n
e
s
et
o
f
s
a
m
p
le
s
a
s
tr
ain
in
g
a
n
d
o
n
e
s
et
a
s
te
s
ti
n
g
.
An
d
f
r
o
m
t
h
is
w
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8708
A
r
o
b
u
s
t a
u
th
o
r
s
h
ip
a
ttr
ib
u
tio
n
o
n
b
ig
p
erio
d
(
Mu
b
in
S
h
o
u
k
a
t Ta
mb
o
li
)
3171
d
ec
id
e
th
e
ef
f
ec
ti
v
e
n
ess
o
f
o
u
r
alg
o
r
ith
m
.
P
r
o
p
o
s
ed
m
et
h
o
d
o
lo
g
y
b
u
i
ld
s
w
it
h
f
ea
t
u
r
e
r
ep
r
esen
tat
io
n
,
f
ea
tu
r
e
s
elec
tio
n
,
ap
p
licatio
n
o
f
cla
s
s
if
icatio
n
alg
o
r
it
h
m
to
b
u
i
ld
m
o
d
el
a
n
d
th
e
n
v
er
i
f
y
in
g
r
es
u
lts
.
I
n
t
h
e
f
ea
tu
r
e
s
elec
tio
n
,
i
n
itial
l
y
w
e
r
e
m
o
v
e
o
u
t
all
t
h
e
s
p
ec
ial
s
y
m
b
o
l
s
,
UR
L
s
,
f
u
n
ctio
n
w
o
r
d
s
f
r
o
m
th
e
te
x
t.
F
u
n
ct
io
n
w
o
r
d
s
r
e
m
o
v
ed
as
i
t
is
co
m
m
o
n
l
y
u
s
ed
r
ep
ea
ti
n
g
ter
m
s
,
an
d
i
n
t
h
e
p
r
o
ce
d
u
r
e
ex
tr
ac
tio
n
o
f
s
t
y
le
i
s
t
h
e
p
r
im
ar
y
o
b
j
ec
tiv
e.
W
h
ile
ex
tr
ac
tin
g
ch
ar
ac
ter
,
w
o
r
d
an
d
p
o
s
s
eq
u
en
ce
ar
e
r
esp
ec
tiv
e
to
ea
ch
s
en
te
n
ce
.
A
ll
th
e
ex
tr
ac
ted
f
ea
t
u
r
es
ar
e
r
ep
r
es
en
ted
in
th
e
b
a
g
o
f
w
o
r
d
s
.
R
est
o
f
th
e
p
r
o
ce
d
u
r
e
is
lis
t
ed
in
th
e
a
lg
o
r
it
h
m
g
iv
e
n
b
elo
w
.
G
iv
en:
n
n
u
m
b
er
o
f
s
a
m
p
le
f
r
o
m
w
h
i
ch
k
s
a
m
p
le
ac
t a
s
tr
ain
i
n
g
s
a
m
p
le
an
d
n
-
k
ac
t a
s
test
i
n
g
s
a
m
p
les.
Alg
o
rit
h
m
:
V
=
f
ea
tu
r
e
Vec
to
r
in
ter
m
s
o
f
B
ag
o
f
W
o
r
d
s
r
ep
r
esen
tatio
n
(
v
k
U
v
n
-
k
)
t
= {
ch
a
r
-
n
-
g
r
a
m,
w
o
r
d
-
n
-
g
r
a
m,
p
o
s
-
n
-
g
r
a
m
}
R
ep
ea
t f
o
r
ea
ch
s
a
m
p
le
S
i
Text
=
P
r
ep
r
o
ce
s
s
in
g
(
s
a
mp
le
)
V
=
B
u
ild
F
ea
tu
r
e
(
f
ea
tu
r
e
t
y
p
e
t
i
,
Text
)
Fo
r
ea
ch
s
a
m
p
le
f
ea
tu
r
e
f
r
o
m
V
V
ne
w
,
k
, V
ne
w
,
n
-
k
=
R
ed
u
ce
Feat
u
r
eDi
m
e
n
s
io
n
(
V
)
R
=
B
u
ild
C
lass
if
ier
SVM(
P
o
ly
n
o
m
ial
Ker
n
el(
p
o
w
2
)
,
V
k
,
V
n
-
k
)
Fo
r
ea
ch
s
a
m
p
le
s
f
r
o
m
s
et
o
f
test
i
n
g
s
a
m
p
le
b
est
m
atc
h
f
o
r
s
is
d
er
iv
ed
f
r
o
m
R
O
utput
:
E
ac
h
s
a
m
p
le
f
r
o
m
te
s
ti
n
g
s
et
i
s
ass
i
g
n
ed
to
b
est
m
atch
cla
s
s
,
n
u
m
b
er
o
f
clas
s
es a
r
e
s
a
m
e
as
au
th
o
r
s
et.
T
h
e
m
ain
id
ea
w
it
h
ab
o
v
e
m
eth
o
d
o
lo
g
y
i
s
t
h
at
it
w
o
r
k
s
i
n
g
o
o
d
w
a
y
f
o
r
te
x
t
s
a
m
p
le
w
h
ic
h
ar
e
co
llected
f
r
o
m
u
s
er
o
v
er
lo
n
g
p
er
io
d
,
it
p
r
o
v
es
t
h
e
al
g
o
r
ith
m
p
r
o
d
u
ce
s
e
f
f
ec
ti
v
e
r
es
u
lt
s
.
I
n
o
u
r
B
a
g
o
f
W
o
r
d
s
a
m
p
le
is
a
d
ictio
n
ar
y
-
b
ased
r
ep
r
esen
tatio
n
o
f
d
ata.
I
n
Featu
r
e
ex
tr
ac
tio
n
m
et
h
o
d
o
lo
g
y
,
w
e
u
s
ed
NL
T
K
to
o
ls
to
ex
tr
ac
t
c
h
ar
ac
ter
w
o
r
d
,
P
o
S
n
-
g
r
a
m
s
.
W
e
u
s
ed
s
k
lear
n
lib
r
ar
y
to
r
ep
r
esen
t
o
u
r
d
a
ta
in
b
ag
o
f
w
o
r
d
s
f
o
r
m
atted
v
ec
to
r
.
A
ll
t
h
ese
f
e
atu
r
es
co
m
b
i
n
ed
to
g
et
h
er
w
it
h
co
n
ca
te
n
atio
n
o
p
er
atio
n
to
f
r
o
m
alo
n
e
f
ea
t
u
r
e
v
ec
to
r
.
I
n
th
e
s
ec
t
io
n
o
f
n
o
r
m
a
lizat
io
n
,
w
e
n
o
r
m
alize
d
th
e
s
a
m
p
le
a
m
o
n
g
t
h
e
w
h
o
le
f
ea
t
u
r
e
v
ec
to
r
.
4.
E
XP
E
R
I
M
E
NT
R
E
SU
L
T
S
AND
DIS
CUSS
I
O
N
W
e
h
av
e
co
llected
n
u
m
er
o
u
s
E
n
g
li
s
h
la
n
g
u
a
g
e
d
ata
f
r
o
m
r
ea
l
w
o
r
ld
en
tit
y
.
W
e
h
a
v
e
co
llected
d
ig
ital
as
w
e
ll
as
h
an
d
w
r
itte
n
d
o
cs
w
h
o
s
e
d
i
g
ital
co
p
y
i
s
av
ailab
le
o
n
li
n
e.
W
e
h
a
v
e
c
o
llected
d
ata
o
f
1
1
au
th
o
r
s
o
f
v
ar
io
u
s
ti
m
e
s
p
a
n
.
T
h
e
p
er
io
d
in
w
h
ic
h
d
ata
i
s
av
ailab
le
f
r
o
m
4
y
ea
r
s
to
m
o
r
e
th
an
3
0
y
ea
r
s
.
Mix
d
atasets
w
er
e
co
n
s
id
er
ed
.
I
n
o
u
r
d
ataset,
m
i
x
ed
ti
m
e
p
er
io
d
c
o
r
p
u
s
is
u
s
ed
.
Fro
m
av
ailab
le
r
eso
u
r
ce
s
,
w
e
ac
c
u
m
u
late
h
a
n
d
w
r
itte
n
le
tter
s
o
f
f
e
w
a
u
t
h
o
r
s
w
h
ich
co
n
v
er
ted
to
d
ig
itized
f
o
r
m
.
An
d
o
n
f
u
r
th
er
u
s
ed
i
n
id
en
ti
f
icatio
n
p
r
o
ce
s
s
.
O
u
t
o
f
elev
en
a
u
th
o
r
s
,
s
i
x
a
u
t
h
o
r
s
co
r
p
u
s
w
a
s
f
r
o
m
ti
m
e
p
er
io
d
5
t
o
8
an
d
f
i
v
e
a
u
t
h
o
r
h
av
i
n
g
co
r
p
u
s
f
r
o
m
ti
m
e
p
er
io
d
2
5
to
3
8
.
A
n
ac
cu
m
u
lat
ed
co
r
p
u
s
s
ize
ab
o
u
t
m
o
r
e
t
h
an
3
0
0
w
o
r
d
s
i
n
a
d
o
cu
m
en
t.
Fro
m
te
x
t
s
a
m
p
le
w
e
f
ir
s
t
r
e
m
o
v
ed
s
p
ec
ial
s
y
m
b
o
ls
,
a
n
d
r
ep
ea
ted
w
o
r
d
s
as
i
n
f
o
r
m
at
o
f
letter
w
er
e
r
e
m
o
v
ed
.
A
S
C
I
I
ch
ar
ac
ter
s
k
e
p
t
as
it
i
s
i
n
s
a
m
p
les.
All
te
x
t
s
a
m
p
le
n
o
t
f
r
o
m
s
a
m
e
ti
m
e
p
er
io
d
s
o
o
u
r
s
y
s
te
m
b
ec
o
m
e
s
r
o
b
u
s
t.
W
e
u
s
e
b
a
g
o
f
w
o
r
d
s
f
o
r
m
at
f
o
r
r
ep
r
esen
t
atio
n
o
f
o
u
r
f
ea
t
u
r
e
v
ec
to
r
,
as
u
s
ed
f
ea
tu
r
es
w
er
e
ch
ar
ac
ter
n
g
r
a
m
,
w
o
r
d
n
g
r
a
m
an
d
p
o
s
n
g
r
a
m
.
W
e
u
tili
ze
d
all
s
u
c
h
f
ea
tu
r
e
u
n
i
q
u
el
y
.
W
e
ch
ec
k
ed
co
m
p
ar
ati
v
e
r
es
u
lt
f
o
r
th
e
s
a
m
e.
W
h
en
w
e
co
n
s
tr
u
cted
f
ea
tu
r
e
v
ec
to
r
,
w
e
s
elec
t
k
b
est
f
ea
t
u
r
e
f
r
o
m
t
h
e
m
a
n
d
ap
p
lied
class
if
icat
io
n
m
o
d
el
t
o
g
en
er
ate
r
es
u
lt.
Selec
ted
f
e
atu
r
e
ev
al
u
ated
o
n
t
h
e
b
asi
s
o
f
h
o
w
co
r
r
ec
tl
y
th
e
id
en
ti
f
icatio
n
o
f
au
t
h
o
r
m
ad
e.
Fo
llo
w
i
n
g
T
ab
le
1
illu
s
tr
ate
h
o
w
co
r
r
ec
tl
y
th
e
a
u
t
h
o
r
id
en
ti
f
ied
.
T
ab
le
1
.
T
h
e
p
er
f
o
r
m
an
ce
o
f
au
th
o
r
s
h
ip
attr
ib
u
t
io
n
F
e
a
t
u
r
e
T
i
me
sp
a
n
A
c
c
u
r
a
c
y
K
a
p
p
a
S
t
a
t
i
s
t
i
c
s
P
O
S
2
g
r
a
m
A
v
e
r
a
g
e
3
0
y
e
a
r
s
7
8
.
6
5
0
.
7
6
P
O
S
4
g
r
a
m
7
0
.
1
7
0
.
6
7
W
o
r
d
2
g
r
a
m
8
4
.
2
1
0
.
8
2
W
o
r
d
3
g
r
a
m
7
0
.
4
6
0
.
6
7
W
o
r
d
4
g
r
a
m
5
0
.
8
7
0
.
4
6
C
h
a
r
4
g
r
a
m
8
6
.
9
8
0
.
8
6
C
h
a
r
5
g
r
a
m
8
8
.
5
9
0
.
8
7
P
O
S
2
W
o
r
d
d
2
8
3
.
3
3
0
.
8
1
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
9
,
No
.
4
,
A
u
g
u
s
t
201
9
:
3
1
6
7
-
3174
3172
On
u
s
e
o
f
s
elec
ted
f
ea
t
u
r
e
t
y
p
e
in
d
icate
d
i
n
T
ab
le
1
,
th
e
r
es
u
lt
v
ar
ies
f
r
o
m
5
0
%
to
8
8
%.
W
h
en
ta
l
k
ab
o
u
t
P
OS
as
f
ea
t
u
r
es,
r
es
u
lt
i
s
g
o
o
d
b
u
t
n
o
t
e
n
o
u
g
h
,
it
s
h
o
w
s
f
o
r
p
o
s
2
g
r
a
m
7
8
%
an
d
p
o
s
4
g
r
a
m
7
0
%,
m
ea
n
s
as
n
in
cr
ea
s
es
f
o
r
P
OS
-
n
-
g
r
a
m
p
er
f
o
r
m
a
n
ce
o
f
clas
s
if
ier
d
eg
r
ad
ed
.
I
n
w
o
r
d
-
n
-
g
r
a
m
w
o
r
d
2
g
r
a
m
g
i
v
es
g
o
o
d
a
cc
u
r
ac
y
as
t
w
o
co
n
ti
n
u
o
u
s
w
o
r
d
o
cc
u
r
r
en
ce
is
g
o
o
d
ch
ar
ac
ter
is
tic
to
id
en
ti
f
y
a
u
t
h
o
r
h
ab
it
th
a
n
a
n
y
o
t
h
er
f
ea
t
u
r
e.
I
t
m
a
k
es
s
e
n
s
e
f
o
r
s
el
ec
tin
g
f
ea
t
u
r
e
as
co
m
p
ar
ed
to
o
th
er
s
.
Fo
r
ch
ar
ac
ter
4
an
d
5
g
r
a
m
in
d
icate
s
clo
s
e
r
esu
lt,
m
o
s
t
o
f
th
e
r
esear
ch
wo
r
k
s
u
g
g
e
s
t
th
e
ch
ar
ac
ter
4
g
r
a
m
ar
e
g
o
o
d
f
ea
tu
r
e
to
ca
p
tu
r
e
u
s
er
s
w
r
iti
n
g
s
t
y
le
as
w
ell
co
n
te
n
t
u
s
ed
in
co
r
p
u
s
.
B
u
t
it
m
a
k
es
n
o
s
e
n
s
e,
h
o
w
i
t
g
ets
ca
p
tu
r
e
b
o
th
,
if
u
s
ed
,
w
e
id
en
tif
ied
h
ab
it
u
al
m
is
tak
e,
r
ep
etitio
n
o
f
w
r
i
tin
g
,
as
w
o
r
d
s
b
r
o
k
en
i
n
to
n
ch
a
r
ac
ter
,
it
d
o
esn
’
t
h
a
v
e
a
n
y
m
ea
n
.
T
h
o
u
g
h
s
u
c
h
p
r
o
b
lem
s
ar
is
e
i
n
ch
ar
ac
ter
n
g
r
a
m
,
it
g
i
v
es p
r
o
m
is
i
n
g
r
esu
lt
th
an
a
n
y
o
t
h
er
f
ea
t
u
r
es.
W
e
also
tr
ied
f
o
r
co
m
b
in
ed
f
e
atu
r
e
t
y
p
e
as
p
o
s
-
2
-
g
r
a
m
a
n
d
w
o
r
d
-
2
-
g
r
a
m
co
m
b
i
n
ed
l
y
,
b
u
t
w
it
h
th
is
,
f
ea
t
u
r
e
s
ize
i
n
cr
ea
s
es
a
n
d
h
e
n
ce
ex
ec
u
tio
n
ti
m
e
f
o
r
th
e
s
y
s
te
m
i
n
cr
ea
s
e
s
.
B
u
t
r
es
u
lt
s
ar
e
ef
f
ec
t
iv
e
b
u
t
as
co
m
p
ar
ed
to
w
o
r
d
-
2
-
g
r
a
m
i
t
i
s
les
s
.
T
ab
le
1
s
h
o
w
s
t
h
e
ac
c
u
r
ac
y
a
n
d
k
ap
p
a
s
tati
s
tic
f
o
r
v
ar
io
u
s
f
ea
tu
r
e
t
y
p
e.
Kap
p
a
s
tatis
tic
u
s
ed
to
m
ea
s
u
r
e
in
ter
r
ater
r
eliab
ilit
y
.
W
h
e
n
s
tatis
t
ic
is
g
r
ea
ter
th
a
n
0
.
8
th
en
ag
r
ee
m
en
t
lev
el
s
tr
o
n
g
a
n
d
i
f
les
s
t
h
an
0
.
6
th
en
it
is
w
ee
k
.
I
n
T
ab
le
1
,
w
o
r
d
-
2
-
g
r
a
m
,
c
h
ar
-
4
-
g
r
a
m
,
ch
ar
-
5
-
g
r
a
m
a
n
d
p
o
s
2
w
o
r
d
2
g
r
am
s
h
o
w
s
s
tr
o
n
g
lev
el
o
f
ag
r
ee
m
e
n
t.
Fi
g
u
r
e
1
s
h
o
w
s
th
e
A
cc
u
r
ac
y
p
lo
t
wh
en
co
m
p
ar
i
n
g
w
i
th
d
if
f
er
e
n
t
f
ea
t
u
r
es
w
it
h
S
VM
.
W
e
also
co
m
p
ar
ed
o
u
r
r
es
u
lt
s
w
h
en
SVM
clas
s
i
f
ier
u
s
ed
w
it
h
o
th
er
s
y
s
te
m
s
w
h
e
n
n
ai
v
e
b
a
y
es a
n
d
r
an
d
o
m
f
o
r
est
m
et
h
o
d
s
ap
p
lied
as sh
o
w
n
i
n
T
ab
le
2
.
Fig
u
r
e
1
.
A
cc
u
r
ac
y
p
lo
t
w
h
e
n
co
m
p
ar
i
n
g
w
i
th
d
i
f
f
er
en
t
f
ea
t
u
r
es
w
i
th
S
VM
T
ab
le
2
.
C
o
m
p
ar
ativ
e
r
es
u
lt
wh
en
u
s
i
n
g
d
if
f
er
en
t c
la
s
s
i
f
ica
ti
o
n
m
et
h
o
d
s
F
e
a
t
u
r
e
Ty
p
e
A
c
c
u
r
a
c
y
S
V
M
N
a
i
v
e
B
a
y
e
s
R
a
n
d
o
m F
o
r
e
st
P
O
S
2
g
r
a
m
7
8
.
6
5
6
6
.
0
8
7
5
.
4
3
P
O
S
4
g
r
a
m
7
0
.
1
7
6
6
.
9
5
6
4
.
3
2
W
o
r
d
2
g
r
a
m
8
4
.
2
1
7
8
.
3
6
7
4
.
1
2
W
o
r
d
3
g
r
a
m
7
0
.
4
6
7
2
.
2
2
7
0
.
3
2
W
o
r
d
4
g
r
a
m
5
0
.
8
7
5
9
.
5
0
6
4
.
0
3
C
h
a
r
4
g
r
a
m
8
6
.
9
8
8
0
.
5
5
8
2
.
1
6
C
h
a
r
5
g
r
a
m
8
8
.
5
9
8
0
.
5
5
8
0
.
1
7
A
cr
o
s
s
all
t
h
e
f
ea
t
u
r
es
t
y
p
e
S
VM
d
o
es
b
est.
On
ly
i
n
ca
s
e
o
f
w
o
r
d
4
g
r
a
m
,
SVM
s
h
o
w
s
p
o
o
r
r
esu
lt
w
h
ile
n
ai
v
e
b
a
y
e
s
an
d
r
an
d
o
m
f
o
r
est
d
id
g
o
o
d
.
W
h
en
w
i
th
n
aiv
e
b
a
y
es
a
n
d
r
an
d
o
m
f
o
r
est
class
i
f
icatio
n
m
et
h
o
d
s
co
m
p
ar
ed
o
n
b
i
g
p
er
io
d
d
ataset
th
e
n
r
a
n
d
o
m
f
o
r
es
t
m
et
h
o
d
d
id
b
est
f
o
r
P
OS
2
-
g
r
a
m
an
d
c
h
ar
ac
ter
4
-
g
r
a
m
a
n
d
f
o
r
all
o
th
er
s
n
ai
v
e
b
ay
e
s
d
id
b
est.
W
h
ile,
th
e
ac
c
u
r
ac
y
co
m
p
ar
ed
w
it
h
t
h
e
m
et
h
o
d
s
in
[
1
3
]
an
d
[
1
7
]
,
s
h
o
w
n
in
T
ab
le
3
.
T
h
ese
m
et
h
o
d
s
ar
e
s
i
m
ilar
it
y
b
ased
,
an
d
SV
M
is
a
m
ac
h
i
n
e
lear
n
i
n
g
alg
o
r
ith
m
.
SVM
o
u
tp
er
f
o
r
m
s
t
h
e
o
th
er
t
w
o
.
I
n
o
u
r
d
ataset
p
er
f
o
r
m
an
ce
o
f
SC
A
P
is
w
o
r
s
t a
s
co
m
p
ar
ed
to
Featu
r
e
s
a
m
p
li
n
g
[
1
3
]
an
d
SVM.
T
ab
le
3
.
C
o
m
p
ar
in
g
t
h
e
r
esu
l
t
s
w
ith
t
h
e
attr
ib
u
tio
n
m
et
h
o
d
f
o
r
ch
ar
ac
ter
4
-
g
r
a
m
M
e
t
h
o
d
s
S
C
A
P
[
1
7
]
F
e
a
t
u
r
e
S
a
m
p
l
i
n
g
[
1
3
]
S
V
M
A
c
c
u
r
a
c
y
5
7
.
4
4
8
4
.
8
9
8
6
.
9
8
P
r
e
c
i
si
o
n
5
8
.
1
3
8
6
.
1
2
8
8
.
2
5
R
e
c
a
l
l
4
8
.
6
9
8
2
.
3
3
8
3
.
4
6
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8708
A
r
o
b
u
s
t a
u
th
o
r
s
h
ip
a
ttr
ib
u
tio
n
o
n
b
ig
p
erio
d
(
Mu
b
in
S
h
o
u
k
a
t Ta
mb
o
li
)
3173
5.
CO
NCLU
SI
O
N
I
n
th
is
w
o
r
k
,
w
e
s
tu
d
ied
a
n
d
i
m
p
le
m
e
n
ted
a
s
y
s
te
m
w
h
ic
h
i
s
ca
p
ab
le
to
h
a
n
d
le
au
th
o
r
s
h
ip
attr
ib
u
tio
n
p
r
o
b
lem
f
o
r
te
x
t
w
r
it
ten
b
y
a
u
th
o
r
s
at
d
i
f
f
er
en
t
ti
m
e
f
r
a
m
e.
E
x
p
er
i
m
e
n
t
w
as
m
ad
e
o
n
E
n
g
l
is
h
te
x
t
co
r
p
u
s
w
r
itte
n
b
y
a
u
t
h
o
r
at
d
if
f
er
e
n
t
ti
m
e
-
f
r
a
m
e
.
T
ex
t
co
r
p
u
s
f
r
o
m
th
e
n
e
w
s
g
r
o
u
p
co
llectio
n
,
l
etter
co
llectio
n
s
ar
e
co
n
s
id
er
ed
.
W
h
ile
p
er
f
o
r
m
i
n
g
ex
p
er
i
m
e
n
ts
f
o
cu
s
is
g
i
v
e
n
o
n
l
y
o
n
f
ea
tu
r
e
s
et,
h
o
w
f
ea
tu
r
es
e
x
tr
ac
ted
an
d
r
ep
r
esen
ted
an
d
th
e
s
elec
tio
n
p
r
o
ce
d
u
r
e
o
f
f
ea
tu
r
e
s
d
u
r
in
g
m
o
d
el
b
u
ild
in
g
.
I
n
th
e
e
x
p
er
i
m
en
ts
,
th
e
SVM
m
et
h
o
d
g
iv
e
s
v
er
y
g
o
o
d
r
esu
lt
as
co
m
p
ar
ed
to
n
av
ie
b
a
y
e
s
an
d
r
an
d
o
m
f
o
r
est
a
lg
o
r
it
h
m
f
o
r
ch
ar
ac
ter
n
-
g
r
a
m
,
w
o
r
d
n
-
g
r
a
m
an
d
p
o
s
n
-
g
r
a
m
.
Hig
h
est
ac
h
ie
v
ed
r
es
u
lt
w
as
8
8
.
5
9
%
w
it
h
c
h
ar
ac
ter
5
-
g
r
a
m
f
ea
tu
r
es.
C
o
m
b
i
n
ed
f
ea
t
u
r
es
o
f
p
o
s
2
-
g
r
a
m
a
n
d
wo
r
d
2
-
g
r
a
m
p
r
o
d
u
ce
s
r
es
u
lt
s
u
p
to
8
3
.
3
3
%
w
h
er
e
in
d
i
v
id
u
all
y
th
e
y
p
r
o
d
u
ce
d
7
8
.
6
5
%
an
d
8
4
.
2
1
%
r
esp
ec
tiv
el
y
.
Data
s
et
co
llected
o
v
er
l
o
n
g
ti
m
e
p
er
io
d
av
er
ag
e
3
0
y
ea
r
s
o
f
ti
m
e
s
p
a
n
.
Ou
r
ex
p
er
i
m
en
t
is
b
ased
o
n
f
ea
tu
r
e
s
elec
t
io
n
cr
iter
ia.
W
e
ca
n
b
r
o
ad
en
o
u
r
v
ie
w
to
m
ak
e
o
u
r
s
y
s
te
m
m
o
r
e
r
o
b
u
s
t.
I
n
f
u
t
u
r
e
w
o
r
k
ca
n
b
e
ex
ten
d
ed
b
y
ca
te
g
o
r
izin
g
d
i
f
f
e
r
en
t c
o
r
p
u
s
t
y
p
e.
I
m
p
r
o
v
in
g
s
y
s
te
m
b
y
i
n
cr
ea
s
i
n
g
ac
cu
r
ac
y
o
f
s
y
s
te
m
.
RE
F
E
R
E
NC
E
S
[1
]
M
in
Ya
n
g
,
Ka
m
-
P
u
i
Ch
o
w
,
"
A
u
th
o
rsh
ip
a
tt
ri
b
u
t
io
n
f
o
r
f
o
re
n
sic
in
v
e
stig
a
ti
o
n
w
it
h
th
o
u
sa
n
d
s
o
f
a
u
th
o
rs,
"
IFI
P
In
ter
n
a
t
io
n
a
l
I
n
f
o
rm
a
ti
o
n
S
e
c
u
rit
y
Co
n
fer
e
n
c
e
,
S
p
rin
g
e
r
,
Be
rli
n
,
H
e
id
e
lb
e
rg
,
p
p
.
3
3
9
-
3
5
0
,
2
0
1
4
.
[2
]
T
a
m
b
o
li
M
.
S
.
a
n
d
P
ra
sa
d
,
R.
S
.
,
"
A
u
th
o
rsh
i
p
a
n
a
ly
sis
a
n
d
id
e
n
ti
f
ica
ti
o
n
tec
h
n
iq
u
e
s:
A
re
v
i
e
w
,
"
In
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
Co
m
p
u
ter
A
p
p
l
ica
ti
o
n
s
,
v
o
l.
77
,
n
o
.
16
,
2
0
1
3
.
[3
]
Ra
jes
h
P
ra
sa
d
,
U.
V.
Ku
lk
a
rn
i,
a
n
d
Ja
y
a
sh
re
e
R.
P
ra
sa
d
,
"
A
n
o
v
e
l
e
v
o
lu
ti
o
n
a
ry
c
o
n
n
e
c
ti
o
n
ist
te
x
t
su
m
m
a
rize
r
(ECT
S
),
"
An
ti
-
c
o
u
n
ter
feiti
n
g
,
S
e
c
u
rity
a
n
d
Id
e
n
ti
fi
c
a
ti
o
n
in
C
o
mm
u
n
ic
a
ti
o
n
,
A
S
ID
2
0
0
9
.
3
rd
In
ter
n
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
o
n
.
IEE
E
,
2
0
0
9
.
[4
]
L
a
n
c
a
sh
ire
I.
a
n
d
Hirst
G
.
,
"
V
o
c
a
b
u
lary
c
h
a
n
g
e
s
in
Ag
a
th
a
Ch
risti
e
’s
m
y
ste
rie
s
a
s
a
n
in
d
ica
ti
o
n
o
f
d
e
m
e
n
ti
a
:
A
c
a
se
stu
d
y
,
"
1
9
th
An
n
u
a
l
R
o
tma
n
Res
e
a
rc
h
In
sti
tu
te
Co
n
fer
e
n
c
e
,
Co
g
n
it
ive
Ag
i
n
g
:
Res
e
a
rc
h
a
n
d
Pr
a
c
ti
c
e
,
p
p
.
8
-
10
,
2
0
0
9
.
[5
]
Ca
n
F
.
a
n
d
P
a
tt
o
n
J.
M
.
,
"
Ch
a
n
g
e
o
f
w
rit
in
g
st
y
le
w
it
h
ti
m
e
,
"
Co
mp
u
ter
s
a
n
d
th
e
Hu
ma
n
it
ies
,
v
o
l.
38
,
n
o
.
1,
pp.
61
-
82
,
2
0
0
4
.
[6
]
V
a
n
Da
m
M
.
a
n
d
Ha
u
f
f
C.
,
"
Lar
g
e
-
sc
a
le
a
u
th
o
r
v
e
rif
ica
ti
o
n
:
tem
p
o
ra
l
a
n
d
t
o
p
ica
l
i
n
f
lu
e
n
c
e
s,
"
Pro
c
e
e
d
in
g
s
o
f
t
h
e
3
7
t
h
i
n
ter
n
a
t
io
n
a
l
ACM
S
IGIR
c
o
n
fer
e
n
c
e
o
n
Res
e
a
rc
h
&
d
e
v
e
lo
p
me
n
t
in
in
f
o
rm
a
ti
o
n
re
triev
a
l
,
A
CM
2
0
1
4
,
p
p
.
1
0
3
9
-
1
0
4
2
,
2
0
1
4
.
[7
]
S
tam
a
t
a
to
s
E.
,
"
A
su
rv
e
y
o
f
m
o
d
e
rn
a
u
th
o
rsh
i
p
a
tt
rib
u
ti
o
n
m
e
th
o
d
s,
"
J
o
u
rn
a
l
o
f
th
e
Asso
c
ia
ti
o
n
fo
r
In
fo
rm
a
t
i
o
n
S
c
ien
c
e
a
n
d
T
e
c
h
n
o
l
o
g
y
,
v
o
l.
60
,
n
o
.
3,
5
3
8
-
5
5
6
,
2
0
0
9
.
[8
]
A
z
a
rb
o
n
y
a
d
H.,
De
h
g
h
a
n
i
M
.
,
M
a
rx
M
.
a
n
d
Ka
m
p
s
J.,
"
T
i
m
e
-
a
w
a
re
a
u
th
o
rsh
ip
a
tt
ri
b
u
ti
o
n
f
o
r
sh
o
rt
tex
t
stre
a
m
s,
"
In
Pro
c
e
e
d
in
g
s
o
f
t
h
e
3
8
t
h
In
ter
n
a
ti
o
n
a
l
ACM
S
IGIR
Co
n
fer
e
n
c
e
o
n
Res
e
a
rc
h
a
n
d
De
v
e
lo
p
me
n
t
in
In
fo
rm
a
t
io
n
Retrie
v
a
l
,
p
p
.
7
2
7
-
7
3
0
,
2
0
1
5
.
[9
]
Zh
a
i
C.
a
n
d
L
a
ff
e
rt
y
J.,
"
A
stu
d
y
o
f
s
m
o
o
th
in
g
m
e
th
o
d
s
f
o
r
la
n
g
u
a
g
e
m
o
d
e
ls
a
p
p
li
e
d
t
o
a
d
h
o
c
in
f
o
rm
a
ti
o
n
re
tri
e
v
a
l,
"
In
ACM
S
IGIR
Fo
r
u
m
,
ACM
2
0
1
7
,
v
o
l.
5
1
,
n
o
.
2
,
p
p
.
2
6
8
-
2
7
6
,
2
0
1
7
.
[1
0
]
Ke
ik
h
a
M
.
,
G
e
ra
n
i
S
.
a
n
d
Cre
sta
n
i
F
.
,
"
T
i
m
e
-
b
a
s
e
d
re
lev
a
n
c
e
m
o
d
e
ls,
"
In
Pro
c
e
e
d
in
g
s
o
f
th
e
3
4
t
h
in
ter
n
a
ti
o
n
a
l
ACM
S
IGIR
c
o
n
fer
e
n
c
e
o
n
Res
e
a
rc
h
a
n
d
d
e
v
e
lo
p
me
n
t
i
n
I
n
fo
rm
a
ti
o
n
Retrie
v
a
l
,
ACM
2
0
1
1
,
p
p
.
1
0
8
7
-
1
0
8
8
,
2
0
1
1
.
[1
1
]
Zh
e
n
g
R.
,
L
i
J.,
Ch
e
n
H.
a
n
d
H
u
a
n
g
Z.
,
"
A
f
r
a
m
e
w
o
rk
f
o
r
a
u
th
o
r
sh
ip
i
d
e
n
t
if
ica
ti
o
n
o
f
o
n
li
n
e
m
e
ss
a
g
e
s:
W
rit
in
g
-
st
y
le
f
e
a
tu
re
s
a
n
d
c
las
si
f
ica
ti
o
n
tec
h
n
iq
u
e
s,
"
J
o
u
rn
a
l
o
f
t
h
e
Asso
c
ia
ti
o
n
fo
r
In
fo
rm
a
ti
o
n
S
c
ien
c
e
a
n
d
T
e
c
h
n
o
l
o
g
y
,
v
o
l.
57
,
n
o
.
3,
p
p
.
3
7
8
-
3
9
3
,
2
0
0
6
.
[1
2
]
A
b
b
a
si
A
.
a
n
d
Ch
e
n
H.,
"
W
rit
e
p
rin
ts:
A
sty
lo
m
e
tri
c
a
p
p
ro
a
c
h
to
id
e
n
ti
ty
-
lev
e
l
id
e
n
ti
f
ica
ti
o
n
a
n
d
si
m
il
a
rit
y
d
e
tec
ti
o
n
in
c
y
b
e
rsp
a
c
e
,
"
ACM
T
r
a
n
sa
c
ti
o
n
s
o
n
In
f
o
rm
a
ti
o
n
S
y
ste
ms
(
T
OIS
)
,
v
o
l.
26
,
n
o
.
2,
p
p
.
7
,
2
0
0
8
.
[1
3
]
Ko
p
p
e
l
M
.
,
S
c
h
ler
J.
a
n
d
A
r
g
a
m
o
n
S
.
,
"
A
u
th
o
rsh
i
p
a
tt
rib
u
ti
o
n
in
th
e
w
il
d
,
"
L
a
n
g
u
a
g
e
Res
o
u
rc
e
s
a
n
d
Eva
l
u
a
ti
o
n
,
v
o
l.
45
,
n
o
.
1
,
p
p
.
83
-
9
4
,
2
0
1
1
.
[1
4
]
D
in
g
S
.
H.,
F
u
n
g
B.
a
n
d
De
b
b
a
b
i
M
.
,
"
A
v
isu
a
li
z
a
b
le
e
v
id
e
n
c
e
-
d
riv
e
n
a
p
p
ro
a
c
h
f
o
r
a
u
th
o
rsh
ip
a
tt
r
ib
u
ti
o
n
,
"
ACM
T
ra
n
sa
c
ti
o
n
s
o
n
In
f
o
rm
a
ti
o
n
a
n
d
S
y
ste
m S
e
c
u
rity (
T
IS
S
EC)
,
v
o
l.
17
,
n
o
.
3
,
p
p
.
1
2
,
2
0
1
5
.
[1
5
]
Ke
ste
m
o
n
t
M
.
,
"
F
u
n
c
ti
o
n
W
o
rd
s
in
A
u
th
o
rsh
i
p
A
tt
rib
u
ti
o
n
.
F
r
o
m
Blac
k
M
a
g
ic
to
T
h
e
o
r
y
?
,"
In
Pro
c
e
e
d
in
g
s
o
f
th
e
3
rd
W
o
rk
sh
o
p
o
n
C
o
mp
u
ta
t
io
n
a
l
L
in
g
u
isti
c
s f
o
r L
it
e
ra
tu
re
(
CL
FL
)
,
p
p
.
5
9
-
66
,
2
0
1
4
.
[1
6
]
S
tam
a
t
a
to
s
E.
,
F
a
k
o
tak
is
N.
a
n
d
Ko
k
k
in
a
k
is
G
.
,
"
A
u
to
m
a
ti
c
tex
t
c
a
teg
o
riza
ti
o
n
i
n
term
s
o
f
g
e
n
re
a
n
d
a
u
th
o
r,
"
Co
mp
u
t
a
ti
o
n
a
l
li
n
g
u
isti
c
s
,
v
o
l
.
26
,
n
o
.
4
,
p
p
.
4
7
1
-
4
9
5
,
2
0
0
0
.
[1
7
]
F
ra
n
tze
sk
o
u
G
.
,
S
tam
a
tato
s E
.
,
Gritza
li
s S
.
,
Ch
a
sk
i
C.
E.
a
n
d
Ho
w
a
ld
B.
S
.
,
"
Id
e
n
ti
fy
in
g
a
u
th
o
rsh
i
p
b
y
b
y
te
-
le
v
e
l
n
-
g
ra
m
s:
T
h
e
so
u
rc
e
c
o
d
e
a
u
t
h
o
r
p
ro
f
il
e
(S
CA
P
)
m
e
th
o
d
,
"
I
n
ter
n
a
t
i
o
n
a
l
J
o
u
rn
a
l
o
f
Di
g
it
a
l
Evid
e
n
c
e
,
v
o
l.
6
,
n
o
.
1
,
p
p
.
1
-
1
8
,
2
0
0
7
.
[1
8
]
Ke
še
lj
V
.
,
P
e
n
g
F
.
,
Ce
rc
o
n
e
N
.
a
n
d
T
h
o
m
a
s
C.
,
"
N
-
g
ra
m
-
b
a
s
e
d
a
u
th
o
r
p
r
o
f
il
e
s
f
o
r
a
u
th
o
rsh
i
p
a
tt
rib
u
t
io
n
,
"
In
Pro
c
e
e
d
in
g
s
o
f
th
e
c
o
n
fer
e
n
c
e
p
a
c
if
ic
a
ss
o
c
ia
ti
o
n
f
o
r
c
o
mp
u
ta
ti
o
n
a
l
li
n
g
u
isti
c
s
,
P
A
CL
IN
G
2
0
0
3
,
v
o
l
.
3
,
p
p
.
2
5
5
-
264
,
2
0
1
3
.
[1
9
]
Ko
c
h
e
r
M
.
a
n
d
S
a
v
o
y
J.,
"
A
si
m
p
le
a
n
d
e
f
f
icie
n
t
a
lg
o
rit
h
m
f
o
r
a
u
th
o
rsh
ip
v
e
rif
ica
ti
o
n
,
"
J
o
u
r
n
a
l
o
f
th
e
Asso
c
ia
t
i
o
n
fo
r In
f
o
rm
a
ti
o
n
S
c
ien
c
e
a
n
d
T
e
c
h
n
o
l
o
g
y
,
v
o
l.
68
,
n
o
.
1,
p
p
.
2
5
9
-
2
6
9
,
2
0
1
7
.
[2
0
]
Bu
rn
s
K.
,
"
Ba
y
e
sia
n
in
f
e
re
n
c
e
i
n
d
isp
u
ted
a
u
t
h
o
rs
h
ip
:
A
c
a
se
s
tu
d
y
o
f
c
o
g
n
it
iv
e
e
rro
rs
a
n
d
a
n
e
w
s
y
st
e
m
f
o
r
d
e
c
isio
n
s
u
p
p
o
rt
,
"
In
fo
rm
a
ti
o
n
S
c
ien
c
e
s
,
v
o
l.
1
7
6
,
n
o
.
1
1
,
p
p
.
1
5
7
0
-
1
5
8
9
,
2
0
0
6
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
9
,
No
.
4
,
A
u
g
u
s
t
201
9
:
3
1
6
7
-
3174
3174
[2
1
]
Ch
e
n
g
N.,
C
h
a
n
d
ra
m
o
u
li
R.
a
n
d
S
u
b
b
a
lak
sh
m
i
K.
P
.
,
"
A
u
th
o
r
g
e
n
d
e
r
id
e
n
ti
f
ica
ti
o
n
f
ro
m
tex
t,
"
Dig
it
a
l
In
v
e
stig
a
ti
o
n
,
v
o
l.
8
,
n
o
.
1
,
p
p
.
78
-
8
8
,
2
0
1
1
.
[2
2
]
A
r
g
a
m
o
n
S
.
,
Š
a
ric
M
.
a
n
d
S
tein
S
.
S
.
,
"
S
ty
le
m
in
in
g
o
f
e
lec
tro
n
ic
m
e
ss
a
g
e
s
f
o
r
m
u
lt
ip
le
a
u
t
h
o
rsh
ip
d
isc
rim
in
a
ti
o
n
:
f
irst
re
su
lt
s,”
In
Pro
c
e
e
d
in
g
s
o
f
th
e
n
in
th
ACM
S
IGKD
D
in
ter
n
a
ti
o
n
a
l
c
o
n
fer
e
n
c
e
o
n
Kn
o
wled
g
e
d
isc
o
v
e
ry
a
n
d
d
a
t
a
min
in
g
,
ACM
2
0
0
3
,
p
p
.
4
7
5
-
480
,
2
0
0
3
.
[2
3
]
Ro
c
h
a
A
.
,
S
c
h
e
irer
W
.
J.,
F
o
rsta
ll
C.
W
.
,
Ca
v
a
lca
n
te
T
.
,
T
h
e
o
p
h
i
lo
A
.
,
S
h
e
n
B.
a
n
d
S
tam
a
tato
s
E
.
,
"
A
u
th
o
rsh
i
p
a
tt
rib
u
t
io
n
f
o
r
so
c
ial
m
e
d
ia
f
o
re
n
sic
s,
"
IEE
E
T
ra
n
sa
c
ti
o
n
s
o
n
In
f
o
rm
a
ti
o
n
Fo
re
n
sic
s
a
n
d
S
e
c
u
rity
,
v
o
l.
12
,
n
o
.
1
,
p
p
.
5
-
3
3
,
2
0
1
7
.
[2
4
]
T
a
m
b
o
li
,
M
u
b
in
S
h
o
u
k
a
t
a
n
d
Ra
jes
h
S
.
P
ra
sa
d
.
,
"
F
e
a
tu
re
S
e
l
e
c
ti
o
n
i
n
T
im
e
Aw
a
re
A
u
th
o
rs
h
ip
A
tt
rib
u
ti
o
n
"
In
ter
n
a
t
io
n
a
l
C
o
n
fer
e
n
c
e
on
A
d
v
a
n
c
e
s in
C
o
mm
u
n
ica
ti
o
n
a
n
d
C
o
mp
u
ti
n
g
T
e
c
h
n
o
lo
g
y
(
ICACCT
)
.
I
EE
E
,
2
0
1
8
.
[2
5
]
M
o
h
a
m
m
e
d
AL
-
S
a
re
m
,
A
b
d
e
l
-
Ha
m
id
E
m
a
ra
,
"
T
h
e
e
ffe
c
t
o
f
train
in
g
se
t
si
z
e
in
a
u
th
o
rsh
ip
a
tt
ri
b
u
ti
o
n
:
a
p
p
li
c
a
ti
o
n
o
n
sh
o
rt
A
ra
b
ic
tex
ts,
"
In
ter
n
a
ti
o
n
a
l
J
o
u
r
n
a
l
o
f
El
e
c
trica
l
a
n
d
Co
mp
u
ter
E
n
g
i
n
e
e
rin
g
(
IJ
ECE
)
,
v
o
l.
9
,
n
o
.
1
,
p
p
.
6
5
2
-
6
5
9
,
2
0
1
9
[2
6
]
Ke
ste
m
o
n
t
M
.
,
"
F
u
n
c
ti
o
n
W
o
rd
s
in
A
u
th
o
rsh
i
p
A
tt
rib
u
ti
o
n
.
F
r
o
m
Blac
k
M
a
g
ic
to
T
h
e
o
r
y
?
,
"
In
Pro
c
e
e
d
in
g
s
o
f
th
e
3
rd
W
o
rk
sh
o
p
o
n
C
o
mp
u
ta
t
io
n
a
l
L
in
g
u
isti
c
s f
o
r L
it
e
ra
tu
re
(
CL
FL
)
,
p
p
.
5
9
-
66
,
2
0
1
4
.
[2
7
]
S
u
n
i
l
Ka
le
a
n
d
Ra
jes
h
S
.
P
ra
sa
d
,
"
A
u
th
o
r
Id
e
n
ti
f
ica
ti
o
n
u
sin
g
S
e
q
u
e
n
t
ial
M
in
im
a
l
Op
ti
m
iza
ti
o
n
w
it
h
ru
le
-
b
a
se
d
De
c
isio
n
T
re
e
o
n
In
d
ian
L
it
e
ra
tu
r
e
in
M
a
ra
th
i
,
"
Pro
c
e
d
ia
Co
mp
u
te
r S
c
ien
c
e
,
v
o
l.
1
3
2
,
p
p
.
1
0
8
6
-
1
1
0
1
,
2
0
1
8
.
[2
8
]
M
.
Rizz
o
Irf
a
n
,
M
.
A
li
F
a
u
z
i,
T
ib
y
a
n
i,
Nu
ru
l
Dy
a
h
M
e
n
tari,
"
Tw
it
ter
S
e
n
ti
m
e
n
t
A
n
a
l
y
sis
o
n
2
0
1
3
C
u
rricu
l
u
m
Us
in
g
En
se
m
b
le
F
e
a
tu
re
s
a
n
d
K
-
Ne
a
re
st
Ne
i
g
h
b
o
r
,"
In
t
e
rn
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
En
g
in
e
e
rin
g
(
IJ
ECE
)
,
v
o
l.
8
,
n
o
.
6
,
p
p
.
5
4
0
9
-
1
4
,
2
0
1
8
[2
9
]
Die
d
e
rich
Jo
a
c
h
im
,
e
t
a
l.
"
A
u
th
o
rsh
ip
a
tt
ri
b
u
ti
o
n
w
it
h
su
p
p
o
rt
v
e
c
to
r
m
a
c
h
in
e
s,
"
Ap
p
li
e
d
i
n
tell
ig
e
n
c
e
,
v
o
l.
19
,
n
o
.
1
-
2
,
p
p
.
1
0
9
-
1
2
3
,
2
0
0
3
.
Evaluation Warning : The document was created with Spire.PDF for Python.