I
nte
rna
t
io
na
l J
o
urna
l o
f
E
lect
rica
l a
nd
Co
m
pu
t
er
E
ng
ineering
(
I
J
E
CE
)
Vo
l.
15
,
No
.
3
,
J
u
n
e
20
25
,
p
p
.
3139
~
3
1
4
8
I
SS
N:
2088
-
8
7
0
8
,
DOI
: 1
0
.
1
1
5
9
1
/ijece.
v
15
i
3
.
pp
3
1
3
9
-
3
1
4
8
3139
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ec
e.
ia
esco
r
e.
co
m
Enha
ncing
cy
ber
bully
ing
det
ec
tion
wit
h
adv
a
nced t
e
x
t
prepro
cess
ing
an
d ma
chine l
ea
rni
ng
Ra
k
esh
B
a
pu
Dhum
a
le
1
,
Aj
a
y
K
um
a
r
D
a
s
s
2
,
Am
it
Um
br
a
j
ka
a
r
3
,
P
r
a
deep
M
a
ne
1
1
D
e
p
a
r
t
me
n
t
o
f
El
e
c
t
r
o
n
i
c
s a
n
d
T
e
l
e
c
o
mm
u
n
i
c
a
t
i
o
n
E
n
g
i
n
e
e
r
i
n
g
,
A
I
S
S
M
S
I
n
st
i
t
u
t
e
o
f
I
n
f
o
r
mat
i
o
n
Te
c
h
n
o
l
o
g
y
,
P
u
n
e
,
I
n
d
i
a
2
D
e
p
a
r
t
me
n
t
o
f
El
e
c
t
r
o
n
i
c
s a
n
d
T
e
l
e
c
o
mm
u
n
i
c
a
t
i
o
n
E
n
g
i
n
e
e
r
i
n
g
,
S
i
n
h
g
a
d
C
o
l
l
e
g
e
o
f
E
n
g
i
n
e
e
r
i
n
g
,
P
u
n
e
,
I
n
d
i
a
3
S
c
h
o
o
l
o
f
M
e
c
h
a
n
i
c
a
l
En
g
i
n
e
e
r
i
n
g
,
D
.
Y
.
P
a
t
i
l
I
n
t
e
r
n
a
t
i
o
n
a
l
U
n
i
v
e
r
si
t
y
,
P
u
n
e
,
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
Au
g
2
4
,
2
0
2
4
R
ev
is
ed
Feb
7
,
2
0
2
5
Acc
ep
ted
Ma
r
4
,
2
0
2
5
Th
e
u
se
o
f
so
c
ial
m
e
d
ia
a
n
d
th
e
in
tern
e
t
h
a
s
b
e
e
n
in
c
re
a
sin
g
d
ra
m
a
ti
c
a
ll
y
in
re
c
e
n
t
y
e
a
rs.
Cy
b
e
r
-
b
u
l
ly
i
n
g
is
t
h
e
term
u
se
d
t
o
d
e
sc
rib
e
th
e
m
isu
s
e
o
f
so
c
ial
m
e
d
ia
b
y
s
o
m
e
p
e
o
p
le
wh
o
m
a
k
e
th
re
a
ten
in
g
c
o
m
m
e
n
ts.
T
h
is
h
a
s
a
d
e
v
a
sta
ti
n
g
in
fl
u
e
n
c
e
o
n
p
e
o
p
le'
s
li
v
e
s,
e
sp
e
c
ially
th
o
se
o
f
c
h
il
d
re
n
a
n
d
tee
n
a
g
e
rs,
a
n
d
c
a
n
lea
d
to
fe
e
li
n
g
s
o
f
d
e
p
re
ss
io
n
a
n
d
s
u
icid
a
l
t
h
o
u
g
h
ts.
T
h
e
m
e
th
o
d
o
lo
g
y
p
ro
p
o
se
d
i
n
th
is
p
a
p
e
r
in
c
lu
d
e
s
fo
u
r
ste
p
s
fo
r
i
d
e
n
ti
f
y
in
g
c
y
b
e
rb
u
ll
y
i
n
g
:
p
re
p
ro
c
e
ss
in
g
,
fe
a
tu
re
e
x
trac
ti
o
n
,
c
las
sifica
ti
o
n
,
a
n
d
e
v
a
lu
a
ti
o
n
.
T
h
e
first
ste
p
is
to
c
re
a
te
a
lab
e
led
,
v
a
ried
d
a
tas
e
t.
Wo
rd
2
V
ec
a
n
d
term
fre
q
u
e
n
c
y
-
i
n
v
e
rse
d
o
c
u
m
e
n
t
fre
q
u
e
n
c
y
a
re
u
se
d
in
fe
a
tu
re
e
x
trac
ti
o
n
to
tran
sfo
rm
te
x
t
i
n
to
h
ig
h
-
d
ime
n
si
o
n
a
l
v
e
c
to
rs.
Wo
rd
2
V
ec
c
re
a
tes
wo
rd
e
m
b
e
d
d
in
g
s
u
sin
g
t
h
e
sk
ip
-
g
ra
m
a
n
d
c
o
n
ti
n
u
o
u
s
b
a
g
-
of
-
wo
rd
s
m
o
d
e
ls,
wh
il
e
term
fre
q
u
e
n
c
y
-
i
n
v
e
rse
d
o
c
u
m
e
n
t
fre
q
u
e
n
c
y
a
ss
e
ss
e
s
th
e
tex
t'
s
term
re
lev
a
n
c
y
.
S
u
p
p
o
rt
v
e
c
to
r
m
a
c
h
in
e
c
las
sifiers
a
re
u
s
e
d
in
th
e
m
o
d
e
l,
a
n
d
th
e
ir
e
ffe
c
ti
v
e
n
e
ss
is
c
o
m
p
a
re
d
to
th
a
t
o
f
o
t
h
e
r
tec
h
n
i
q
u
e
s
li
k
e
lo
g
isti
c
re
g
re
ss
io
n
a
n
d
n
a
ï
v
e
Ba
y
e
s
.
Th
e
c
las
sifiers
su
p
p
o
rt
v
e
c
to
r
m
a
c
h
in
e
,
n
a
ïv
e
Ba
y
e
s
,
a
n
d
lo
g
isti
c
re
g
re
ss
i
o
n
we
re
a
ss
e
ss
e
d
.
Th
e
m
a
x
imu
m
a
c
c
u
ra
c
y
wa
s
9
5
%
fo
r
th
e
su
p
p
o
rt
v
e
c
to
r
c
las
sifier
with
sk
ip
-
g
ra
m
a
n
d
9
3
%
fo
r
c
o
n
ti
n
u
o
u
s
b
a
g
-
of
-
wo
rd
s
.
F
o
r
se
n
ti
m
e
n
t
c
a
teg
o
ries
,
F
1
-
sc
o
re
s,
r
e
c
a
ll
,
a
n
d
p
re
c
isio
n
we
re
c
o
m
p
u
ted
.
T
h
e
a
v
e
ra
g
e
p
re
c
isio
n
a
n
d
re
c
a
ll
we
re
0
.
7
7
a
n
d
0
.
7
9
,
re
sp
e
c
ti
v
e
ly
.
K
ey
w
o
r
d
s
:
C
y
b
er
b
u
lly
in
g
Dete
ctio
n
On
lin
e
th
r
ea
ts
So
cial
m
ed
ia
So
cial
m
ed
ia
m
is
u
s
e
Su
p
p
o
r
t
v
ec
to
r
m
ac
h
in
es
T
ex
t
class
if
icatio
n
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
:
R
ak
esh
B
ap
u
Dh
u
m
ale
Dep
ar
tm
en
t o
f
E
lectr
o
n
ics an
d
T
elec
o
m
m
u
n
icatio
n
E
n
g
in
ee
r
in
g
,
AI
SS
MS
I
n
s
titu
te
o
f
I
n
f
o
r
m
atio
n
T
ec
h
n
o
lo
g
y
Pu
n
e,
I
n
d
ia
E
m
ail: r
b
d
.
s
co
e@
g
m
ail.
co
m
1.
I
NT
RO
D
UCT
I
O
N
So
cial
m
ed
ia
ad
d
ictio
n
h
as
in
cr
ea
s
ed
g
lo
b
ally
in
d
ir
ec
t
p
r
o
p
o
r
tio
n
to
th
e
p
h
e
n
o
m
en
al
g
r
o
wth
in
d
ata
s
er
v
ice
av
ailab
ilit
y
.
Similar
t
o
o
th
e
r
n
atio
n
s
,
c
y
b
er
b
u
lly
in
g
h
as
s
h
ar
p
ly
i
n
cr
ea
s
ed
[
1
]
.
W
ith
ev
er
y
o
n
e
liv
in
g
o
n
d
ig
ital
an
d
o
n
lin
e
p
latf
o
r
m
s
in
th
is
W
eb
4
.
0
er
a,
it
i
s
ex
tr
em
ely
ch
allen
g
in
g
to
s
af
eg
u
ar
d
s
o
ciety
ag
ain
s
t
th
e
s
tar
tlin
g
in
cr
ea
s
e
in
c
y
b
er
cr
im
e
[
2
]
.
I
t
h
as
b
ee
n
d
is
co
v
er
ed
th
at
teen
ag
e
r
s
ar
e
t
h
e
m
ain
v
ictim
s
o
f
cy
b
er
b
u
lly
in
g
[
3
]
.
C
y
b
er
b
u
lly
in
g
is
a
d
a
n
g
er
o
u
s
an
d
d
am
a
g
in
g
b
eh
av
io
r
th
at
ca
n
m
a
k
e
v
ictim
s
tr
y
s
u
icid
e
an
d
s
u
f
f
e
r
last
in
g
co
n
s
eq
u
en
c
es
[
4
]
.
C
y
b
er
b
u
lly
in
g
d
etec
tio
n
ca
n
b
e
s
ee
n
as
a
class
if
icatio
n
p
r
o
b
lem
in
wh
ic
h
p
o
s
ts
m
ad
e
o
n
lin
e
ar
e
class
if
ied
as
eith
er
b
u
lly
in
g
o
r
n
o
n
-
b
u
lly
in
g
o
n
es
[
5
]
.
A
s
y
s
tem
th
at
en
h
a
n
ce
s
th
e
p
er
f
o
r
m
an
ce
o
f
d
etec
tin
g
cy
b
er
b
u
lly
in
g
will
b
e
cr
ea
ted
v
ia
a
v
ar
iety
o
f
co
m
p
u
te
r
v
is
io
n
(
C
V)
,
m
ac
h
in
e
lear
n
in
g
(
ML
)
an
d
n
atu
r
al
la
n
g
u
ag
e
p
r
o
ce
s
s
in
g
(
NL
P)
tech
n
iq
u
es
[
6
]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
15
,
No
.
3
,
J
u
n
e
20
25
:
3
1
3
9
-
3
1
4
8
3140
So
cial
m
ed
ia'
s
r
is
e
h
as
led
to
an
in
cr
ea
s
e
in
cy
b
e
r
b
u
lly
i
n
g
,
wh
ic
h
ca
lls
f
o
r
ef
f
ec
tiv
e
d
etec
tio
n
tech
n
iq
u
es.
I
n
[
7
]
,
a
h
y
b
r
id
r
a
n
d
o
m
f
o
r
est
-
b
ased
C
NN
m
o
d
el
f
o
r
tex
t
class
if
icatio
n
is
p
r
e
s
en
ted
.
I
t
ac
h
iev
es
g
o
o
d
ac
cu
r
ac
y
an
d
f
aster
r
esu
lts
f
o
r
tim
ely
in
ter
v
e
n
tio
n
th
a
n
p
r
ev
i
o
u
s
ML
an
d
d
ee
p
l
ea
r
n
in
g
(
DL
)
m
eth
o
d
s
.
T
h
e
wo
r
k
[
8
]
,
im
p
r
o
v
e
d
a
s
y
s
tem
th
at
u
s
es
ter
m
f
r
eq
u
en
c
y
-
in
v
er
s
e
d
o
cu
m
en
t
f
r
eq
u
en
c
y
(
T
F
-
I
DF)
(
b
ig
r
am
)
f
ea
tu
r
e
ex
tr
ac
tio
n
a
n
d
m
u
lti
-
class
if
icatio
n
alg
o
r
ith
m
s
to
d
etec
t
s
ix
d
if
f
er
en
t
ty
p
es
o
f
c
y
b
er
b
u
lly
in
g
twee
ts
with
h
ig
h
ac
cu
r
ac
y
.
T
r
a
d
itio
n
al
class
if
ier
s
wer
e
o
u
tp
er
f
o
r
m
ed
b
y
e
n
s
em
b
le
ap
p
r
o
ac
h
es
i
n
teg
r
atin
g
d
ec
is
io
n
tr
ee
s
,
r
an
d
o
m
f
o
r
est
,
a
n
d
XGBo
o
s
t; th
e
s
tack
in
g
class
if
ier
ac
h
iev
ed
an
ac
cu
r
ac
y
o
f
9
0
.
7
1
%.
E
f
f
ec
tiv
e
d
etec
tio
n
tech
n
i
q
u
e
s
ar
e
r
eq
u
i
r
ed
b
ec
au
s
e
o
f
th
e
s
u
r
g
e
in
c
y
b
er
b
u
lly
in
g
th
at
h
as
r
esu
lted
f
r
o
m
th
e
in
cr
ea
s
ed
u
s
ag
e
o
f
s
o
cial
m
ed
ia
p
latf
o
r
m
s
.
I
n
o
r
d
er
to
ca
teg
o
r
ize
ag
g
r
ess
iv
e
twee
ts
,
Mu
n
ee
r
et
a
l.
[
9
]
p
r
o
p
o
s
es
an
au
to
m
ated
m
eth
o
d
t
h
at
u
s
es
a
s
tack
in
g
en
s
em
b
le
o
f
f
iv
e
ML
alg
o
r
it
h
m
s
.
I
t
o
u
tp
er
f
o
r
m
s
co
n
v
en
tio
n
al
m
o
d
els
an
d
p
r
i
o
r
s
tu
d
ies,
with
an
ac
cu
r
ac
y
r
ate
o
f
9
4
.
0
0
%.
T
h
e
s
tu
d
y
[
1
0
]
ad
d
r
ess
es
th
e
ch
allen
g
e
o
f
id
e
n
tify
in
g
cy
b
e
r
b
u
lly
in
g
c
o
m
m
en
ts
with
s
u
b
tle
ex
p
r
ess
io
n
s
b
y
p
r
o
p
o
s
in
g
a
n
o
v
el
ap
p
r
o
ac
h
u
s
in
g
th
e
b
i
d
ir
ec
tio
n
al
en
c
o
d
er
r
ep
r
esen
tatio
n
s
f
r
o
m
tr
a
n
s
f
o
r
m
er
s
(
B
E
R
T
)
p
r
e
-
t
r
ain
i
n
g
m
o
d
el
an
d
th
e
B
iS
R
U+
+
m
o
d
el
with
atten
tio
n
m
ec
h
an
is
m
s
,
im
p
r
o
v
in
g
cla
s
s
if
icatio
n
ac
cu
r
ac
y
an
d
g
en
er
aliza
tio
n
th
r
o
u
g
h
a
m
u
lti
-
task
lear
n
in
g
f
r
am
ewo
r
k
.
I
n
o
r
d
er
t
o
ad
d
r
ess
th
e
p
r
o
b
l
em
o
f
c
y
b
er
b
u
lly
in
g
o
n
s
o
cia
l
m
ed
ia,
Gs
et
a
l.
[
1
1
]
s
u
g
g
e
s
ts
a
ML
-
b
ased
d
etec
tio
n
m
eth
o
d
th
at
m
ak
es
u
s
e
o
f
a
v
ar
iety
o
f
d
ata
s
ets
an
d
s
o
p
h
is
ticated
NL
P
te
ch
n
iq
u
es.
I
t
ac
h
iev
es
h
ig
h
ac
cu
r
ac
y
b
y
u
s
in
g
s
u
p
p
o
r
t
v
ec
to
r
m
ac
h
in
e
(
SVM)
as
th
e
m
o
s
t
ef
f
icien
t
alg
o
r
ith
m
a
n
d
em
p
h
asizes
th
e
s
ig
n
if
ican
ce
o
f
eth
ical
co
n
s
id
er
atio
n
s
f
o
r
im
p
r
o
v
in
g
o
n
lin
e
s
af
ety
.
T
h
e
s
tu
d
y
im
p
r
o
v
es
th
e
id
en
tific
atio
n
o
f
cy
b
er
b
u
lly
in
g
i
n
C
h
in
a
b
y
p
u
ttin
g
f
o
r
th
a
h
y
b
r
i
d
m
o
d
el
t
h
at
co
m
b
in
es
d
ee
p
b
id
ir
ec
tio
n
al
lo
n
g
s
h
o
r
t
-
ter
m
m
em
o
r
y
(
Bi
-
L
STM
)
with
XL
Net
[
1
2
]
.
T
h
is
m
o
d
el
b
ea
ts
b
o
th
s
tan
d
ar
d
an
d
d
ee
p
l
ea
r
n
in
g
b
aselin
es,
im
p
r
o
v
in
g
ac
cu
r
ac
y
b
y
1
.
1
3
%
o
v
er
B
E
R
T
,
4
.
2
9
% o
v
er
SVM,
an
d
1
.
4
9
%
o
v
er
g
ate
d
r
ec
u
r
r
en
t u
n
it
.
T
h
e
s
tu
d
y
r
e
p
u
r
p
o
s
es
th
e
Sen
tim
en
t1
4
0
d
ataset
to
id
en
tif
y
cy
b
er
b
u
lly
in
g
o
n
T
witter
u
s
in
g
ar
tific
ial
in
tellig
en
ce
.
Hig
h
p
r
ec
is
io
n
i
s
d
em
o
n
s
tr
ated
b
y
n
ai
v
e
B
ay
es
(
NB
)
an
d
B
i
-
L
STM
m
o
d
els;
y
et,
th
e
s
tu
d
y
em
p
h
asizes
th
e
n
ee
d
f
o
r
im
p
r
o
v
ed
m
o
d
els
a
n
d
a
wid
e
r
r
a
n
g
e
o
f
d
ata
[
1
3
]
.
T
h
e
p
r
o
b
lem
o
f
c
y
b
er
b
u
lly
in
g
is
g
ettin
g
wo
r
s
e
as
s
o
cial
m
ed
i
a
ch
an
g
es.
T
h
e
wo
r
k
p
r
esen
ts
co
m
b
in
atio
n
al
n
etwo
r
k
f
o
r
b
u
lly
in
g
d
etec
tio
n
(
C
NB
D)
,
a
d
ee
p
lear
n
in
g
m
o
d
el
th
at
im
p
r
o
v
es
ac
cu
r
ac
y
,
p
r
e
cisi
o
n
,
an
d
r
ec
all
in
id
en
tif
y
in
g
cy
b
er
b
u
lly
in
g
i
n
p
h
o
to
g
r
ap
h
s
.
I
t
c
o
m
b
in
es
b
in
a
r
y
e
n
co
d
er
im
ag
e
tr
an
s
f
o
r
m
er
(
B
E
iT)
an
d
m
u
lti
-
lay
e
r
p
er
ce
p
tr
o
n
(
ML
P)
an
d
is
au
g
m
en
ted
b
y
OC
R
an
d
im
ag
e
ca
p
tio
n
in
g
[
1
4
]
.
T
h
e
s
tu
d
y
cr
ea
tes
B
u
lly
Fil
ter
NeT
,
a
cy
b
er
b
u
lly
in
g
d
etec
tio
n
s
y
s
tem
f
o
r
B
en
g
ali
s
o
c
ial
m
ed
ia
wr
itin
g
s
u
s
in
g
tr
an
s
f
o
r
m
er
-
b
a
s
ed
an
d
n
o
n
-
co
n
te
x
tu
al
em
b
e
d
d
in
g
.
T
h
e
B
an
g
laB
E
R
T
-
b
ased
B
u
lly
Fil
ter
NeT
s
u
cc
ess
f
u
lly
h
an
d
les
out
-
of
-
v
o
ca
b
u
lar
y
is
s
u
es
with
an
ac
cu
r
a
cy
o
f
8
8
.
0
4
%
[
1
5
]
.
T
ap
s
o
b
a
et
a
l.
[
1
6
]
s
u
g
g
est
a
ML
s
tr
ateg
y
th
at
co
m
b
in
es
lo
n
g
s
h
o
r
t
ter
m
m
em
o
r
y
(
L
STM
)
,
r
an
d
o
m
f
o
r
est
(
R
F)
an
d
SVM
to
im
p
r
o
v
e
cy
b
er
th
r
ea
t
id
en
tific
atio
n
in
B
u
r
k
i
n
a
Fas
o
.
W
ith
R
F,
th
e
ap
p
r
o
ac
h
ac
h
iev
es
b
etter
ac
cu
r
ac
y
in
cy
b
e
r
b
u
lly
i
n
g
,
p
h
is
h
in
g
a
n
d
o
n
lin
e
f
r
au
d
s
.
T
h
e
p
ap
e
r
[
1
7
]
u
s
es
laten
t
s
em
an
tic
an
aly
s
is
f
o
r
s
en
tim
en
t
a
n
aly
s
is
to
class
if
y
cy
b
er
b
u
lly
in
g
twee
ts
,
ac
h
ie
v
i
n
g
h
i
g
h
er
ac
cu
r
ac
y
th
an
ex
is
tin
g
m
eth
o
d
s
,
th
u
s
aid
in
g
in
r
e
d
u
cin
g
o
n
lin
e
ab
u
s
e
an
d
s
u
p
p
o
r
tin
g
s
af
e
s
o
cial
m
ed
ia
u
s
ag
e.
T
h
e
s
tu
d
y
p
r
o
p
o
s
es
a
ML
ap
p
r
o
ac
h
u
s
in
g
a
m
u
ltin
o
m
ial
NB
class
if
ier
to
d
etec
t
c
y
b
er
b
u
ll
y
in
g
o
n
s
o
cial
n
etwo
r
k
s
,
ac
h
iev
in
g
8
8
.
7
6
%
ac
cu
r
ac
y
in
c
lass
if
y
in
g
s
h
am
in
g
,
s
ex
u
al
h
ar
ass
m
en
t,
an
d
r
ac
is
m
,
with
f
u
zz
y
r
u
le
s
ets f
o
r
b
u
lly
in
g
s
tr
en
g
th
[
1
8
]
.
I
n
o
r
d
er
to
d
etec
t
cy
b
e
r
b
u
lly
in
g
in
Ar
ab
ic,
t
h
e
s
tu
d
y
in
v
e
s
tig
ates
ML
m
eth
o
d
s
.
B
ec
au
s
e
o
f
t
h
eir
h
ig
h
ac
cu
r
ac
y
,
Al
Har
b
i
et
a
l.
[
1
9
]
s
u
g
g
est
r
id
g
e
a
n
d
lo
g
is
tic
r
eg
r
ess
io
n
(
L
R
)
.
T
h
e
g
o
al
is
to
u
s
e
th
e
s
e
m
eth
o
d
s
to
id
en
tif
y
c
y
b
er
b
u
lly
in
g
in
Ar
ab
ic.
W
ith
ML
,
cy
b
er
b
u
lly
in
g
ca
n
b
e
au
to
m
a
tically
d
etec
ted
b
y
an
aly
zin
g
lin
g
u
is
tic
p
atter
n
s
.
Han
i
et
a
l.
[
2
0
]
s
u
g
g
ests
a
cla
s
s
if
ier
-
b
ased
s
u
p
er
v
is
ed
s
tr
ate
g
y
.
Ne
u
r
al
n
etwo
r
k
o
u
tp
er
f
o
r
m
s
o
th
e
r
m
o
d
els
with
an
ac
cu
r
ac
y
o
f
9
2
.
8
%,
ac
co
r
d
in
g
to
ev
alu
atio
n
.
C
y
b
er
b
u
ll
y
in
g
v
ia
W
ik
ip
ed
ia
an
d
T
witter
is
d
etec
ted
b
y
N
L
P
an
d
ML
with
o
v
e
r
9
0
%
a
n
d
8
0
%
ac
cu
r
ac
y
,
r
esp
ec
tiv
el
y
[
2
1
]
.
T
h
e
g
o
al
o
f
[
2
2
]
is
to
co
m
b
in
e
ML
a
n
d
NL
P
to
p
r
o
v
id
e
a
n
ef
f
icien
t
m
eth
o
d
f
o
r
id
en
tify
i
n
g
a
b
u
s
iv
e
an
d
h
a
r
ass
in
g
m
ess
ag
es
s
en
t
o
n
lin
e.
T
h
e
s
tu
d
y
u
s
es
T
F
-
I
DF
an
d
b
a
g
-
of
-
wo
r
d
s
(
B
o
W
)
v
ar
ia
b
les
to
ex
am
in
e
th
e
ac
cu
r
ac
y
o
f
f
o
u
r
ML
alg
o
r
ith
m
s
.
T
h
e
g
o
al
o
f
[
2
3
]
is
to
co
m
b
in
e
M
L
an
d
NL
P
to
p
r
o
v
id
e
an
e
f
f
icien
t
m
eth
o
d
f
o
r
id
en
tify
in
g
ab
u
s
iv
e
an
d
h
ar
a
s
s
in
g
m
ess
ag
es
s
en
t
o
n
lin
e.
T
h
e
s
tu
d
y
u
s
es
T
F
-
I
DF
an
d
B
o
W
v
ar
iab
les
to
ex
am
in
e
th
e
ac
cu
r
ac
y
o
f
f
o
u
r
ML
alg
o
r
ith
m
s
.
Dev
elo
p
i
n
g
r
e
liab
le
n
u
m
er
ical
r
ep
r
esen
tatio
n
s
o
f
tex
t
m
ess
ag
es
is
a
m
ajo
r
co
n
ce
r
n
i
n
th
is
s
tu
d
y
.
I
n
o
r
d
er
to
im
p
r
o
v
e
r
ep
r
ese
n
tatio
n
lear
n
in
g
,
th
e
w
o
r
k
p
r
e
s
en
ts
th
e
s
em
an
tic
-
en
h
an
ce
d
m
ar
g
in
alize
d
d
en
o
i
s
in
g
au
to
-
e
n
co
d
e
r
(
SMSDA)
,
an
en
h
an
ce
d
d
ee
p
lear
n
i
n
g
m
o
d
el
th
at
au
g
m
e
n
ts
s
tack
ed
d
en
o
is
in
g
a
u
to
-
en
c
o
d
e
r
s
with
s
em
an
tic
d
r
o
p
o
u
t n
o
is
e
an
d
s
p
ar
s
ity
co
n
s
tr
ain
ts
[
2
4
]
.
T
h
is
r
esear
ch
p
r
o
v
id
es
a
n
o
v
el
m
eth
o
d
th
at
ac
h
iev
es
9
5
%
ac
cu
r
ac
y
in
cy
b
er
th
r
ea
t
d
et
ec
tio
n
b
y
in
teg
r
atin
g
W
o
r
d
2
Vec
with
SVM.
Pre
p
r
o
ce
s
s
in
g
,
f
ea
tu
r
e
e
x
tr
ac
tio
n
,
an
d
class
if
icatio
n
u
s
in
g
W
o
r
d
2
Vec
an
d
TF
-
I
DF
ar
e
all
p
ar
t
o
f
th
e
m
et
h
o
d
o
lo
g
y
;
SVM
p
er
f
o
r
m
s
b
etter
th
an
o
th
er
m
eth
o
d
s
lik
e
lo
g
is
tic
r
eg
r
ess
io
n
an
d
n
aïv
e
B
ay
es.
T
h
e
em
p
l
o
y
m
en
t
o
f
b
o
th
s
k
ip
-
g
r
am
an
d
co
n
ti
n
u
o
u
s
b
a
g
-
of
-
wo
r
d
s
m
o
d
els
,
wh
ich
im
p
r
o
v
e
th
e
f
ea
tu
r
e
ex
tr
ac
tio
n
p
r
o
ce
s
s
,
is
a
s
ig
n
if
ican
t
ad
v
an
ce
in
W
o
r
d
2
Vec
.
T
h
e
s
u
g
g
ested
ap
p
r
o
a
ch
p
er
f
o
r
m
s
b
etter
th
an
p
r
ev
io
u
s
s
tu
d
ies
an
d
h
a
s
an
av
er
ag
e
p
r
ec
is
io
n
a
n
d
r
ec
all
o
f
0
.
7
7
an
d
0
.
7
9
,
r
esp
e
ctiv
ely
.
T
h
e
s
tu
d
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
-
8
7
0
8
E
n
h
a
n
cin
g
cy
b
erb
u
llyin
g
d
etec
tio
n
w
ith
a
d
va
n
ce
d
text
p
r
ep
r
o
ce
s
s
in
g
…
(
R
a
ke
s
h
B
a
p
u
Dh
u
ma
le
)
3141
em
p
h
asizes
th
e
s
ig
n
if
ican
ce
o
f
tack
lin
g
th
e
r
ap
id
ly
ex
p
an
d
in
g
is
s
u
e
o
f
cy
b
er
b
u
lly
in
g
,
p
ar
ticu
lar
ly
its
d
etr
im
en
tal
ef
f
ec
ts
o
n
teen
a
g
er
s
,
an
d
th
e
u
s
ef
u
l
n
ess
o
f
th
is
r
esear
ch
f
o
r
p
r
o
ac
tiv
e
m
ea
s
u
r
es
lik
e
r
ea
l
-
tim
e
s
u
r
v
eillan
ce
.
T
h
is
wo
r
k
is
n
o
t
o
n
ly
e
f
f
ec
tiv
e
b
u
t
also
s
ca
lab
le,
p
av
in
g
th
e
way
f
o
r
f
u
tu
r
e
r
esear
ch
to
im
p
r
o
v
e
s
ea
r
ch
ac
cu
r
ac
y
a
n
d
c
o
m
p
atib
ilit
y
ac
r
o
s
s
d
if
f
er
en
t p
latf
o
r
m
s
an
d
lan
g
u
ag
es.
2.
M
AT
E
R
I
AL
S AN
D
M
E
T
H
O
DS
T
h
er
e
ar
e
f
o
u
r
s
tag
es
in
th
e
s
u
g
g
ested
m
eth
o
d
o
lo
g
y
f
o
r
d
etec
tin
g
cy
b
er
b
u
lly
in
g
,
p
r
e
p
r
o
ce
s
s
in
g
,
f
ea
tu
r
e
e
x
tr
ac
tio
n
,
class
if
icati
o
n
,
a
n
d
e
v
alu
atio
n
.
T
h
is
s
ec
ti
o
n
g
o
es
o
v
er
ea
c
h
s
tep
in
d
e
p
th
.
A
lab
ele
d
d
ataset
is
co
m
p
iled
with
ca
s
es
o
f
b
o
t
h
n
o
n
-
cy
b
er
b
u
lly
in
g
a
n
d
c
y
b
e
r
b
u
lly
in
g
.
T
h
e
tar
g
et
p
o
p
u
latio
n
s
h
o
u
ld
b
e
well
-
r
ep
r
esen
ted
i
n
th
is
d
ataset,
w
h
ich
s
h
o
u
ld
b
e
d
i
v
er
s
e.
On
e
o
f
th
e
m
o
s
t
im
p
o
r
tan
t
s
tep
s
in
th
e
d
etec
tio
n
o
f
cy
b
er
b
u
lly
in
g
is
p
r
ep
r
o
ce
s
s
in
g
.
I
t
en
tails
r
em
o
v
in
g
s
p
a
m
co
n
ten
t
an
d
clea
n
in
g
tex
ts
s
u
ch
as
r
em
o
v
in
g
p
u
n
ctu
atio
n
an
d
s
to
p
wo
r
d
s
.
Pre
p
r
o
ce
s
s
in
g
h
as
b
ee
n
u
tili
ze
d
in
th
e
p
r
o
p
o
s
ed
m
o
d
el
to
r
em
o
v
e
an
d
s
o
r
t
-
o
u
t
u
n
wan
ted
n
o
is
e
i
n
tex
t
d
etec
ti
o
n
lik
e
r
e
p
etitiv
e
wo
r
d
s
,
s
p
ec
i
al
ch
ar
ac
ter
s
a
n
d
s
to
p
w
o
r
d
s
ar
e
r
em
o
v
ed
.
Af
ter
th
at,
s
tem
m
in
g
was
u
s
ed
to
t
h
e
r
em
ain
in
g
wo
r
d
s
to
r
esto
r
e
th
em
to
th
eir
o
r
ig
in
al
f
o
r
m
s
.
A
clea
n
twitter
d
ataset
is
p
r
o
d
u
ce
d
b
y
th
is
p
r
ep
r
o
ce
s
s
in
g
,
wh
ich
ca
n
b
e
u
s
ed
to
r
u
n
t
h
e
r
ec
o
m
m
en
d
e
d
m
o
d
el
an
d
p
r
o
d
u
ce
p
r
ed
ictio
n
s
.
A
cr
u
cial
s
tag
e
in
tex
t
class
if
icatio
n
f
o
r
cy
b
e
r
b
u
lly
i
n
g
d
etec
tio
n
is
f
ea
tu
r
e
e
x
tr
ac
tio
n
.
W
o
r
d
2
Vec
an
d
TF
-
I
DF a
p
p
r
o
ac
h
es h
av
e
b
ee
n
em
p
lo
y
ed
f
o
r
f
ea
tu
r
e
e
x
tr
ac
tio
n
in
th
e
p
r
o
p
o
s
ed
m
o
d
el
wh
ich
is
b
ased
o
n
wo
r
d
s
tatis
t
ics.
T
h
is
m
o
d
el
s
o
lely
t
ak
es
in
to
ac
c
o
u
n
t
wo
r
d
ex
p
r
e
s
s
io
n
s
th
at
ar
e
c
o
n
s
is
ten
t
ac
r
o
s
s
all
tex
ts
.
As
a
r
esu
lt,
o
n
e
o
f
th
e
m
o
s
t
o
f
ten
u
tili
ze
d
f
ea
tu
r
e
ex
tr
ac
tio
n
m
eth
o
d
s
f
o
r
tex
t
d
etec
tio
n
is
T
F
-
I
DF.
W
o
r
d
2
Vec
is
a
n
eu
r
al
n
etwo
r
k
with
two
lay
er
s
th
at
p
r
o
ce
s
s
es tex
t b
y
"v
ec
to
r
izin
g
"
wo
r
d
s
.
C
o
r
p
u
s
es o
f
tex
t ser
v
es a
s
its
in
p
u
t
an
d
a
s
et
o
f
v
ec
to
r
s
attr
ib
u
te
v
ec
to
r
s
th
at
r
ef
lect
th
e
wo
r
d
s
in
th
at
s
tr
u
ctu
r
e
s
er
v
e
as
its
o
u
tp
u
t.
T
h
e
W
o
r
d
2
Vec
tech
n
iq
u
e
cr
ea
tes
a
h
ig
h
-
d
im
e
n
s
io
n
al
v
ec
to
r
f
o
r
ev
er
y
wo
r
d
b
y
u
tili
zin
g
two
h
id
d
en
lay
er
s
o
f
s
h
allo
w
n
eu
r
al
n
etwo
r
k
s
: th
e
s
k
ip
-
g
r
am
m
o
d
e
l (
SGM)
an
d
th
e
c
o
n
tin
u
o
u
s
b
ag
-
of
-
w
o
r
d
s
(
C
B
OW
)
.
T
h
e
co
n
tin
u
o
u
s
b
a
g
-
of
-
wo
r
d
s
m
o
d
el
(
B
W
M)
u
s
es
th
e
s
u
r
r
o
u
n
d
in
g
co
n
tex
t
wo
r
d
s
to
f
o
r
ec
ast
th
e
m
id
d
le
wo
r
d
.
T
h
e
wo
r
d
s
th
at
co
m
e
b
ef
o
r
e
an
d
a
f
ter
th
e
cu
r
r
en
t
(
m
id
d
le)
wo
r
d
m
ak
e
u
p
th
e
co
n
tex
t.
B
ec
au
s
e
it
d
o
esn
'
t
m
atter
wh
at
wo
r
d
s
a
r
e
u
s
ed
in
wh
at
s
eq
u
en
ce
in
t
h
e
co
n
tex
t,
th
is
ar
ch
itectu
r
e
is
k
n
o
wn
as
a
b
ag
-
of
-
wo
r
d
s
m
o
d
el.
T
h
e
c
o
n
tin
u
o
u
s
SKM
f
o
r
ec
asts
wo
r
d
s
th
at
f
all
with
in
a
s
p
ec
if
ic
r
an
g
e
b
o
th
b
ef
o
r
e
an
d
af
ter
th
e
cu
r
r
en
t
wo
r
d
in
a
p
h
r
ase.
T
h
e
ex
am
p
le
is
g
iv
en
in
Fig
u
r
e
1
.
A
SGM
p
r
ed
icts
th
e
co
n
tex
t
(
o
r
n
eig
h
b
o
r
s
)
o
f
a
wo
r
d
g
iv
e
n
th
e
w
o
r
d
its
elf
,
wh
er
ea
s
a
B
W
M
p
r
ed
icts
a
wo
r
d
b
ased
o
n
t
h
e
s
u
r
r
o
u
n
d
in
g
co
n
tex
t.
SKM,
o
r
n
-
SKM
th
at
p
er
m
it
to
k
en
s
to
b
e
s
k
ip
p
ed
as
g
iv
en
in
Fig
u
r
e
1
,
ar
e
u
s
ed
to
tr
ain
th
e
m
o
d
el
.
A
s
et
o
f
s
k
ip
-
g
r
am
p
air
in
g
s
o
f
(
ta
r
g
et_
w
o
r
d
,
co
n
text_
w
o
r
d
)
wh
er
e
co
n
text_
w
o
r
d
ap
p
ea
r
s
in
th
e
n
ea
r
b
y
c
o
n
t
ex
t
o
f
ta
r
g
et_
w
o
r
d
ca
n
b
e
u
s
ed
to
in
d
icate
th
e
co
n
tex
t
o
f
a
wo
r
d
.
A
win
d
o
w
s
ize
d
ef
in
es
th
e
co
n
tex
t
wo
r
d
s
f
o
r
ea
ch
o
f
th
e
wo
r
d
s
in
th
is
p
h
r
ase.
T
h
e
s
p
an
o
f
wo
r
d
s
th
at
ca
n
b
e
r
eg
ar
d
ed
as c
o
n
tex
t w
o
r
d
s
o
n
ea
ch
s
id
e
o
f
a
tar
g
et
wo
r
d
d
e
p
en
d
s
o
n
th
e
win
d
o
w
s
ize.
T
h
in
k
o
f
a
g
r
o
u
p
o
f
w
o
r
d
s
Ϣ
.
I
f
th
e
s
lid
in
g
win
d
o
w
s
ize
(
Ḉ
)
is
2
,
a
n
d
Ϣ
is
th
e
in
p
u
t
(
ce
n
ter
wo
r
d
)
,
th
e
n
th
e
co
n
tex
t
wo
r
d
s
ar
e
Ϣ
−
2
,
Ϣ
−
1
,
Ϣ
+
1
,
an
d
Ϣ
+
2
.
T
h
e
SGM
u
s
in
g
th
e
n
e
u
r
al
n
etwo
r
k
'
s
b
ac
k
p
r
o
p
ag
atio
n
eq
u
atio
n
s
,
w
h
ich
is
im
p
lem
en
ted
i
n
th
is
wo
r
k
is
s
h
o
wn
in
Fig
u
r
e
1
.
Op
tim
izin
g
th
e
lik
elih
o
o
d
o
f
p
r
ed
ictin
g
c
o
n
tex
t
wo
r
d
s
b
a
s
ed
o
n
th
e
ta
r
g
et
wo
r
d
is
th
e
SGM’
s
tr
ain
in
g
g
o
al.
T
h
e
aim
f
o
r
a
wo
r
d
s
eq
u
en
ce
co
n
s
is
tin
g
o
f
Ϣ
−
2
,
Ϣ
−
1
,
Ϣ
+
1
,
an
d
Ϣ
+
2
ca
n
b
e
ex
p
r
ess
ed
a
s
(
1
)
f
o
r
th
e
a
v
er
ag
e
l
o
g
p
r
o
b
a
b
i
lity
.
l
og
(
(
)
)
=
1
∑
∑
l
og
(
(
Ϣ
+
|
Ϣ
)
)
−
Ḉ
≤
k
≤
Ḉ
=
1
(
1
)
T
h
e
b
asic SKM
d
ef
in
es th
is
p
r
o
b
ab
ilit
y
u
s
in
g
t
h
e
s
o
f
t
-
m
a
x
f
u
n
ctio
n
as g
iv
e
n
in
(
2
)
.
(
c
on
te
xt
_
wo
r
d
ta
r
ge
t
_
wo
r
d
⁄
)
=
(
′
c
o
nt
ex
t
_
wo
r
d
t
arg
et
_
wo
r
d
)
∑
(
′
t
arg
et
_
wo
r
d
)
=
1
(
2
)
W
h
er
e,
th
e
v
o
ca
b
u
lar
y
s
ize
is
V
.
t
ar
g
et
_
w
o
r
d
is
th
e
tar
g
et
wo
r
d
's
v
ec
to
r
r
ep
r
esen
tatio
n
.
′
co
n
t
ex
t
_
w
o
r
d
is
th
e
co
n
tex
t
wo
r
d
'
s
v
ec
to
r
r
ep
r
esen
tatio
n
.
T
h
e
ex
p
o
n
en
tiated
d
o
t
p
r
o
d
u
ct
b
etwe
en
th
e
tar
g
et
wo
r
d
v
ec
to
r
an
d
th
e
co
n
tex
t
wo
r
d
v
ec
to
r
is
r
ep
r
esen
ted
b
y
th
e
n
u
m
e
r
ato
r
.
T
h
e
to
tal
o
f
th
ese
ex
p
o
n
en
tiated
d
o
t
p
r
o
d
u
cts
f
o
r
ea
ch
wo
r
d
in
th
e
lex
ico
n
is
th
e
d
en
o
m
in
ato
r
.
I
n
o
r
d
er
to
ca
lcu
late
t
h
e
d
e
n
o
m
in
ato
r
o
f
th
is
f
o
r
m
u
latio
n
,
a
th
o
r
o
u
g
h
s
o
f
tm
ax
o
v
e
r
th
e
en
tire
v
o
ca
b
u
lar
y
m
u
s
t
b
e
p
er
f
o
r
m
e
d
,
wh
ich
f
r
eq
u
e
n
tly
co
n
s
is
ts
o
f
a
s
ig
n
if
ican
t
n
u
m
b
er
o
f
p
h
r
a
s
es
(
1
0
5
-
10
7
).
An
ef
f
ec
tiv
e
ap
p
r
o
x
im
atio
n
f
o
r
a
co
m
p
lete
s
o
f
tm
ax
is
th
e
n
o
is
e
co
n
tr
asti
v
e
esti
m
atio
n
(
NC
E
)
lo
s
s
f
u
n
ctio
n
.
I
t
is
p
o
s
s
ib
le
to
s
im
p
lify
th
e
NC
E
lo
s
s
b
y
u
s
in
g
n
eg
ativ
e
s
am
p
lin
g
with
t
h
e
g
o
al
o
f
lear
n
in
g
wo
r
d
em
b
e
d
d
in
g
s
r
ath
er
th
an
m
o
d
elin
g
th
e
wo
r
d
d
is
tr
ib
u
tio
n
.
T
h
e
g
o
al
o
f
s
i
m
p
lifie
d
n
eg
ativ
e
s
am
p
lin
g
f
o
r
a
tar
g
et
wo
r
d
is
to
is
o
late
th
e
co
n
tex
t w
o
r
d
f
r
o
m
n
eg
ativ
e
s
am
p
les th
at
ar
e
s
ele
cted
f
r
o
m
a
wo
r
d
n
o
is
e
d
is
tr
ib
u
tio
n
(
(
)
)
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
15
,
No
.
3
,
J
u
n
e
20
25
:
3
1
3
9
-
3
1
4
8
3142
Fig
u
r
e
1
.
Sk
im
g
r
am
a
r
ch
itectu
r
e
u
s
in
g
b
ac
k
p
r
o
p
a
g
atio
n
Mo
r
e
ex
ac
tly
,
co
n
s
id
er
in
g
th
e
lo
s
s
f
o
r
a
ta
r
g
et_
w
o
r
d
as
a
clas
s
if
icatio
n
p
r
o
b
lem
b
e
twee
n
th
e
co
n
text_
w
o
r
d
an
d
n
eg
ativ
e
s
a
m
p
les
is
an
ef
f
icien
t
ap
p
r
o
x
im
atio
n
o
f
f
u
ll
s
o
f
tm
a
x
o
v
er
t
h
e
v
o
ca
b
u
lar
y
f
o
r
a
s
k
ip
-
g
r
am
p
air
.
A
ta
r
g
et_
w
o
r
d
an
d
c
o
n
text_
w
o
r
d
p
air
is
co
n
s
id
er
ed
n
eg
ativ
e
if
an
d
o
n
ly
if
th
e
co
n
text_
w
o
r
d
d
o
es
n
o
t
o
cc
u
r
in
t
h
e
w
in
d
o
w
_
s
iz
e
n
eig
h
b
o
r
h
o
o
d
o
f
th
e
ta
r
g
et_
w
o
r
d
.
(
b
etter
,
b
ac
k
)
i
s
p
air
o
f
n
e
g
ativ
e
s
am
p
les
o
f
ex
am
p
le
“Y
o
u
b
et
ter
watc
h
y
o
u
r
b
ac
k
.
”
SKG
u
s
in
g
b
ac
k
p
r
o
p
a
g
atio
n
n
eu
r
al
n
etwo
r
k
(
B
PNN)
is
s
h
o
wn
in
Fig
u
r
e
1
.
T
h
e
v
ar
iab
les ar
e
d
ef
in
ed
as b
el
o
w.
V:
C
o
u
n
t o
f
d
is
tin
ct
ter
m
s
in
o
u
r
tex
t c
o
r
p
u
s
(
v
o
ca
b
u
lar
y
)
X:
I
n
p
u
t la
y
e
r
(
s
in
g
le
h
o
t e
n
co
d
in
g
o
f
o
u
r
in
p
u
t w
o
r
d
)
N:
T
h
e
q
u
an
tity
o
f
n
eu
r
o
n
s
with
in
th
e
n
e
u
r
al
n
etwo
r
k
'
s
h
id
d
en
lay
er
W
: Weig
h
ts
s
ep
ar
atin
g
th
e
in
p
u
t la
y
er
f
r
o
m
th
e
h
id
d
e
n
lay
er
W
'
: Weig
h
ts
s
ep
ar
atin
g
th
e
o
u
tp
u
t la
y
er
f
r
o
m
th
e
h
id
d
e
n
lay
er
Y:
A
s
o
f
t
-
m
ax
o
u
t
p
u
t la
y
er
with
p
r
o
b
ab
ilit
ies f
o
r
ea
ch
w
o
r
d
in
o
u
r
le
x
ico
n
TF
-
I
DF
is
th
e
p
r
o
ce
d
u
r
e
f
o
r
f
ig
u
r
in
g
o
u
t
h
o
w
r
elev
an
t
a
wo
r
d
is
to
a
tex
t
in
a
co
r
p
u
s
o
r
s
er
ies.
A
wo
r
d
'
s
m
ea
n
in
g
in
c
r
ea
s
es
in
d
ir
ec
t
p
r
o
p
o
r
tio
n
to
h
o
w
o
f
ten
it
ap
p
ea
r
s
in
th
e
tex
t
;
b
u
t,
th
e
co
r
p
u
s
'
s
(
d
ataset's)
wo
r
d
f
r
eq
u
en
cy
b
a
lan
ce
s
th
is
o
u
t.
T
h
e
T
F
-
I
DF
weig
h
t
o
f
a
ter
m
in
a
d
o
cu
m
en
t
is
r
ep
r
esen
ted
m
ath
em
atica
lly
as (
3
)
.
W
(
ḏ
,
ṯ
)
=
(
ḏ
,
ṯ
)
∗
l
og
(
Ṉ
ḏ
(
ṯ
)
)
(
3
)
T
h
e
n
u
m
b
e
r
Ṉ
in
th
is
in
s
tan
ce
in
d
icate
s
th
e
to
tal
n
u
m
b
er
o
f
d
o
cu
m
e
n
ts
,
an
d
th
e
n
u
m
b
er
o
f
d
o
cu
m
e
n
ts
in
th
e
co
r
p
u
s
th
at
in
clu
d
e
th
e
ter
m
Ṯ
is
d
en
o
ted
b
y
Ḏ
(
Ṯ
)
.
T
h
e
f
ir
s
t
co
m
p
o
n
e
n
t
in
(
3
)
im
p
r
o
v
es
m
em
o
r
y
,
an
d
t
h
e
s
ec
o
n
d
ter
m
im
p
r
o
v
es wo
r
d
em
b
e
d
d
in
g
ac
cu
r
ac
y
.
I
n
th
is
wo
r
k
,
SVM
class
if
ier
s
h
av
e
b
ee
n
u
s
ed
to
d
ete
r
m
in
e
if
a
twee
t
q
u
alif
ies
as
cy
b
er
b
u
lly
in
g
o
r
n
o
t.
SVM
class
if
ier
p
er
f
o
r
m
a
n
ce
is
co
n
tr
asted
with
th
at
o
f
L
ig
h
tGB
M,
SGD,
R
F,
Ad
aBo
o
s
t,
an
d
NB
.
SVM
o
p
tim
izes
th
e
d
is
tan
ce
(
m
ar
g
i
n
)
b
etwe
en
two
ca
teg
o
r
ies
b
y
tr
an
s
f
o
r
m
in
g
th
e
o
r
ig
i
n
al
f
ea
tu
r
e
s
p
ac
e
in
to
a
u
s
er
-
d
ef
in
e
d
h
ig
h
e
r
-
d
im
e
n
s
io
n
al
s
p
ac
e
b
ased
o
n
a
k
er
n
e
l.
SVM
f
ir
s
t
m
ak
es
an
ap
p
r
o
x
im
atio
n
o
f
th
e
h
y
p
er
p
lan
e
d
iv
i
d
in
g
th
e
two
c
ateg
o
r
ies.
3.
I
M
P
L
E
M
E
NT
A
T
I
O
N
T
h
e
f
o
llo
win
g
s
en
ten
c
e
is
an
ex
am
p
le
tak
e
n
in
to
co
n
s
id
er
atio
n
.
"Yo
u
b
etter
watc
h
y
o
u
r
b
ac
k
.
"
T
o
k
en
izin
g
t
h
e
s
en
ten
ce
,
o
r
d
iv
id
in
g
it
in
to
d
is
cr
ete
wo
r
d
s
o
r
to
k
e
n
s
,
is
th
e
f
ir
s
t
s
tep
in
th
e
p
r
o
ce
s
s
.
T
o
k
en
izatio
n
f
ac
ilit
ates
th
e
co
n
v
er
s
io
n
o
f
th
e
s
en
ten
ce
i
n
to
a
f
o
r
m
at
th
at
is
ea
s
ily
p
r
o
ce
s
s
ed
f
o
r
wo
r
d
em
b
ed
d
in
g
s
an
d
o
th
er
NL
P ta
s
k
s
.
T
h
e
d
etail
s
tep
s
ar
e
b
elo
w
:
a.
First s
tep
is
to
cr
ea
te
a
v
o
ca
b
u
lar
y
to
s
av
e
m
a
p
p
in
g
s
f
r
o
m
to
k
en
s
to
in
teg
er
in
d
ices.
{
0
∶
<
>
,
1
∶
<
>
,
2
∶
<
ℎ
>
,
3
∶
<
>
,
4
∶
<
>
}
b.
I
n
o
r
d
er
to
s
to
r
e
m
ap
p
i
n
g
s
b
et
wee
n
in
teg
er
in
d
ices a
n
d
to
k
e
n
s
,
cr
ea
te
an
in
v
er
ted
v
o
ca
b
u
l
ar
y
.
{
<
>
:
0
,
<
>
:
1
,
<
ℎ
>
:
2
,
<
>
:
3
,
<
>
:
4
}
c.
Af
ter
th
at,
th
e
s
en
ten
ce
m
u
s
t b
e
v
ec
to
r
ized
.
[
0
,
1
,
2
,
3
,
4
]
.
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
-
8
7
0
8
E
n
h
a
n
cin
g
cy
b
erb
u
llyin
g
d
etec
tio
n
w
ith
a
d
va
n
ce
d
text
p
r
ep
r
o
ce
s
s
in
g
…
(
R
a
ke
s
h
B
a
p
u
Dh
u
ma
le
)
3143
d.
Gen
er
ate
s
k
ip
-
g
r
a
m
p
air
s
f
r
o
m
th
e
s
en
ten
ce
.
Few
ex
am
p
le
s
o
f
p
o
s
itiv
e
(
t
a
r
g
et_
w
o
r
d
,
co
n
text_
w
o
r
d
)
p
air
ar
e
s
h
o
wn
in
Fig
u
r
e
2.
e.
Neg
ativ
e
s
am
p
lin
g
f
o
r
o
n
e
s
k
ip
-
g
r
am
:
T
h
e
s
k
ip
-
g
r
am
s
f
u
n
c
tio
n
s
lid
es
ac
r
o
s
s
a
s
p
ec
if
ied
win
d
o
w
s
p
an
t
o
r
etu
r
n
all
a
f
f
ir
m
ativ
e
s
k
ip
-
g
r
a
m
p
air
s
.
W
o
r
d
s
f
r
o
m
th
e
v
o
ca
b
u
lar
y
m
u
s
t
b
e
r
a
n
d
o
m
ly
s
am
p
led
in
o
r
d
er
to
g
en
er
ate
m
o
r
e
s
k
ip
-
g
r
am
p
air
in
g
s
th
at
will
b
e
u
s
ed
as
tr
ain
in
g
n
eg
ativ
e
ex
am
p
les.
T
h
e
p
r
o
ce
s
s
in
v
o
lv
es
ap
p
ly
in
g
t
h
e
ap
p
r
o
ac
h
t
o
th
e
t
ar
g
et
wo
r
d
o
f
a
s
in
g
le
s
k
ip
-
g
r
am
an
d
p
ass
in
g
th
e
c
o
n
tex
t
w
o
r
d
as
th
e
tr
u
e
class
to
p
r
ev
en
t it
f
r
o
m
b
ein
g
s
am
p
led
.
f.
C
o
n
s
tr
u
ct
o
n
e
tr
ain
in
g
ex
am
p
le:
Neg
ativ
e
s
am
p
led
co
n
tex
t
wo
r
d
s
f
o
r
a
p
ar
ticu
lar
p
o
s
itiv
e
s
k
ip
-
g
r
am
th
at
d
o
n
o
t
o
cc
u
r
in
th
e
win
d
o
w
s
ize
n
eig
h
b
o
r
h
o
o
d
o
f
th
e
ta
r
g
et
_
w
o
r
d
.
C
o
m
b
in
e
th
e
o
n
e
co
n
t
ex
t
wo
r
d
th
at
is
p
o
s
itiv
e
an
d
th
e
o
n
e
th
at
is
n
e
g
ativ
e.
Fo
r
ev
er
y
tar
g
et
wo
r
d
,
th
is
y
ield
s
a
s
et
o
f
p
o
s
itiv
e
s
k
ip
-
g
r
a
m
s
with
lab
el
as 1
an
d
n
eg
ativ
e
s
am
p
le
s
with
lab
el
as 0
.
T
h
e
tr
ain
in
g
s
am
p
les ar
e
s
h
o
wn
in
T
ab
le
1
.
g.
T
h
e
p
r
o
ce
s
s
f
o
r
f
ig
u
r
in
g
o
u
t
a
wo
r
d
'
s
r
elev
an
ce
to
a
te
x
t
in
a
co
r
p
u
s
o
r
s
er
ies
is
ca
lled
T
F
-
I
DF.
Ma
th
em
atica
lly
,
a
ter
m
'
s
T
F
-
I
DF
weig
h
t
in
a
d
o
c
u
m
en
t
is
r
ep
r
esen
ted
b
y
(
2
)
.
L
et
u
s
c
o
n
s
id
er
f
o
llo
win
g
s
en
ten
ce
s
.
̶
Sen
ten
ce
1
: Y
o
u
b
etter
watc
h
y
o
u
r
b
ac
k
̶
Sen
ten
ce
2
: Watch
y
o
u
r
b
ac
k
̶
Sen
ten
ce
3
: Be c
ar
ef
u
l w
ith
y
o
u
r
b
ac
k
.
Fig
u
r
e
2
.
Sk
ip
-
g
r
am
m
o
d
el
T
ab
le
1
.
T
h
e
tr
ain
in
g
s
am
p
les
W
o
r
d
C
o
n
t
e
x
t
W
o
r
d
La
b
e
l
s
0
{1
,
2
,
3
,
4
}
{1
,
1
,
0
,
0
}
1
{0
,
2
,
3
,
4
}
{1
,
1
,
1
,
0
}
2
{0
,
1
,
3
,
4
}
{1
,
1
,
1
,
1
}
3
{0
,
1
,
2
,
4
}
{0
,
1
,
1
,
1
}
4
{0
,
1
,
2
,
3
}
{0
,
0
,
1
,
1
}
↓
↓
↓
A
f
e
w
w
o
r
d
s
a
r
e
a
b
se
n
t
f
r
o
m
t
h
e
se
n
t
e
n
c
e
p
r
o
v
i
d
e
d
.
Ju
s
t
l
i
k
e
se
v
e
r
a
l
o
t
h
e
r
i
n
d
i
c
e
s
t
h
a
t
w
e
r
e
u
t
i
l
i
z
e
d
i
n
t
h
e
t
r
a
i
n
i
n
g
samp
l
e
s
,
t
h
e
y
a
r
e
p
a
r
t
o
f
t
h
e
v
o
c
a
b
u
l
a
r
y
.
V
{3
4
,
5
,
2
,
9
1
,
7
}
{0
,
1
,
1
,
1
}
T
h
e
wo
r
d
’
s
r
elev
a
n
ce
to
a
tex
t
in
co
r
p
u
s
is
ca
lcu
lated
u
s
in
g
(
5
)
an
d
(
6
)
.
T
h
e
ca
lc
u
latio
n
s
ar
e
g
iv
en
in
T
ab
le
s
2
,
3
a
n
d
4
.
=
(
5
)
=
l
og
(
)
(
6
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
15
,
No
.
3
,
J
u
n
e
20
25
:
3
1
3
9
-
3
1
4
8
3144
T
ab
le
2
.
T
er
m
f
r
e
q
u
en
c
y
W
o
r
d
s
S
e
n
t
1
S
e
n
t
2
S
e
n
t
3
y
o
u
1
/
5
0
/
3
0
/
5
b
e
t
t
e
r
1
/
5
0
/
3
0
/
5
w
a
t
c
h
1
/
5
1
/
3
0
/
5
y
o
u
r
1
/
5
1
/
3
1
/
5
b
a
c
k
1
/
5
1
/
3
1
/
5
be
0
/
5
0
/
3
1
/
5
c
a
r
e
f
u
l
0
/
5
0
/
3
1
/
5
w
i
t
h
0
/
5
0
/
3
1
/
5
T
ab
le
3
.
I
n
v
er
s
e
d
o
cu
m
e
n
t f
r
e
q
u
en
cy
W
o
r
d
s
I
D
F
y
o
u
l
o
g
(
3
/
1
)
=
0
.
4
7
b
e
t
t
e
r
l
o
g
(
3
/
1
)
=
0
.
4
7
w
a
t
c
h
l
o
g
(
3
/
2
)
=
0
.
1
7
y
o
u
r
l
o
g
(
3
/
3
)
=
0
.
0
0
b
a
c
k
l
o
g
(
3
/
3
)
=
0
.
0
0
be
l
o
g
(
3
/
1
)
=
0
.
4
7
c
a
r
e
f
u
l
l
o
g
(
3
/
1
)
=
0
.
4
7
w
i
t
h
l
o
g
(
3
/
1
)
=
0
.
4
7
T
ab
le
4
.
T
er
m
f
r
e
q
u
en
c
y
in
v
e
r
s
e
d
o
cu
m
en
t
f
r
eq
u
e
n
cy
(
T
F X
I
DF)
W
o
r
d
s
S
e
n
t
1
S
e
n
t
2
S
e
n
t
3
y
o
u
0
.
0
9
4
0
.
0
0
0
0
.
0
0
0
b
e
t
t
e
r
0
.
0
9
4
0
.
0
0
0
0
.
0
0
0
w
a
t
c
h
0
.
0
3
4
0
.
5
6
6
0
.
0
0
0
y
o
u
r
0
.
0
0
0
0
.
0
0
0
0
.
0
0
0
b
a
c
k
0
.
0
0
0
0
.
0
0
0
0
.
0
0
0
be
0
.
0
0
0
0
.
0
0
0
0
.
0
9
4
c
a
r
e
f
u
l
0
.
0
0
0
0
.
0
0
0
0
.
0
9
4
w
i
t
h
0
.
0
0
0
0
.
0
0
0
0
.
0
9
4
Fin
d
in
g
th
e
o
p
tim
al
h
y
p
e
r
p
la
n
e
to
d
iv
id
e
t
h
e
n
e
g
ativ
e
an
d
p
o
s
itiv
e
class
es
is
th
e
id
ea
b
e
h
in
d
SVM
class
if
icatio
n
.
B
y
ap
p
ly
in
g
t
h
e
k
er
n
el
m
eth
o
d
,
SVM
m
ay
also
o
p
er
ate
o
n
h
ig
h
-
d
im
e
n
s
io
n
al
d
atasets
.
T
h
e
SVM
k
er
n
el
p
er
f
o
r
m
s
a
v
ar
iety
o
f
task
s
,
in
clu
d
in
g
a
d
d
itiv
e
m
u
lti
-
q
u
ad
r
atic
in
v
er
s
e,
s
ig
m
o
id
,
lin
ea
r
,
p
o
ly
n
o
m
ial
an
d
Gau
s
s
ian
R
B
F
f
u
n
ctio
n
s
.
SVM
Po
ly
n
o
m
ial
is
th
e
k
er
n
el
f
u
n
ctio
n
u
s
ed
in
th
is
s
tu
d
y
.
W
h
en
a
h
y
p
er
p
lan
e
ca
n
s
ep
ar
ate
th
e
d
ata,
a
lin
ea
r
SVM
is
u
s
ed
;
wh
en
o
n
ly
c
u
r
v
e
d
lin
es
ca
n
d
o
s
o
,
a
n
o
n
-
lin
ea
r
SVM
is
u
tili
ze
d
.
An
SVM
k
er
n
el
th
at
u
s
es
a
p
o
ly
n
o
m
ial
f
u
n
cti
o
n
t
o
tr
an
s
lat
e
d
ata
in
t
o
a
h
ig
h
er
-
d
im
e
n
s
io
n
al
s
p
ac
e
is
ca
lled
a
p
o
ly
n
o
m
ial
k
er
n
el.
T
o
d
o
th
is
,
it
tak
es
t
h
e
d
o
t
p
r
o
d
u
ct
o
f
t
h
e
p
o
ly
n
o
m
ial
f
u
n
cti
o
n
in
th
e
n
ew
s
p
ac
e
an
d
th
e
d
ata
p
o
in
ts
in
th
e
o
r
ig
in
al
s
p
ac
e.
W
h
en
th
e
d
ata
ca
n
n
o
t
b
e
s
ep
a
r
ated
lin
ea
r
l
y
,
th
is
k
er
n
el
is
f
r
e
q
u
en
tly
em
p
lo
y
ed
in
SVM
class
if
icati
o
n
task
s
.
Occ
asio
n
ally
,
a
h
y
p
er
p
lan
e
th
at
d
iv
id
es
th
e
class
e
s
ca
n
b
e
f
o
u
n
d
v
i
a
th
e
p
o
ly
n
o
m
ial
k
er
n
el
b
y
tr
an
s
f
er
r
in
g
th
e
d
ata
in
to
a
h
ig
h
er
-
d
im
en
s
io
n
al
s
p
ac
e.
A
n
u
m
b
er
o
f
p
ar
am
eter
s
,
s
u
ch
as
th
e
p
o
ly
n
o
m
ial'
s
d
eg
r
ee
an
d
co
ef
f
icie
n
t,
ca
n
b
e
ad
j
u
s
ted
to
en
h
a
n
ce
th
e
p
er
f
o
r
m
an
ce
o
f
th
e
p
o
ly
n
o
m
ial
k
er
n
el.
T
h
e
p
o
l
y
n
o
m
ial
k
er
n
el
f
o
r
p
o
ly
n
o
m
ials
o
f
d
e
g
r
ee
d
is
d
ef
in
ed
as (
7
).
(
⃗
⃗
⃗
,
⃗
⃗
⃗
)
=
(
+
1
)
(
7
)
w
h
er
e
K
(
X
i
⃗
⃗
⃗
,
X
j
⃗
⃗
⃗
)
is
k
er
n
el
f
u
n
ctio
n
,
X
i
⃗
⃗
⃗
,
X
j
⃗
⃗
⃗
ar
e
in
p
u
ts
a
n
d
d
is
a
n
o
r
d
er
.
B
y
ex
p
r
ess
in
g
it
as
a
QP
p
r
o
b
lem
an
d
u
tili
zin
g
a
co
m
m
o
n
ly
ac
ce
s
s
ib
le
n
u
m
er
ical
an
aly
s
is
lib
r
ar
y
to
s
o
lv
e
it,
th
e
h
y
p
er
p
lan
e
in
th
e
id
ea
l
SVM
i
s
ac
h
iev
ed
.
T
h
e
s
eq
u
en
tial te
ch
n
iq
u
e
is
an
ad
d
itio
n
al,
v
e
r
y
s
tr
aig
h
tf
o
r
war
d
o
p
tio
n
.
4.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
T
h
e
ex
p
er
im
en
t'
s
s
ettin
g
s
ar
e
s
h
o
wn
in
th
is
s
ec
t
io
n
alo
n
g
with
an
ex
p
lan
atio
n
o
f
th
eir
s
ig
n
if
ican
ce
.
Py
th
o
n
3
.
8
a
n
d
Go
o
g
le
C
o
lab
GPUs
ar
e
u
s
ed
.
T
h
e
s
u
g
g
este
d
cy
b
er
b
u
lly
in
g
m
o
d
el
an
d
o
t
h
er
b
aselin
e
m
o
d
els
wer
e
im
p
lem
en
te
d
u
s
in
g
th
e
T
en
s
o
r
Flo
w
lib
r
ar
y
,
w
h
ich
is
u
s
ed
f
o
r
a
p
p
licatio
n
s
lik
e
C
V
an
d
NL
P.
W
e
u
s
ed
a
p
u
b
licly
av
ailab
le
d
ataset,
as
cited
in
r
ef
er
en
ce
[
2
5
]
o
f
o
u
r
m
an
u
s
cr
ip
t.
T
o
h
elp
t
h
e
c
y
b
er
b
u
lly
in
g
m
o
d
el
g
r
asp
th
e
co
n
tex
t
an
d
an
aly
z
e
th
e
im
p
o
r
tan
t
in
f
o
r
m
atio
n
in
th
e
te
x
t,
an
in
p
u
t
m
atr
i
x
o
f
3
5
,
8
7
3
twee
ts
was
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
-
8
7
0
8
E
n
h
a
n
cin
g
cy
b
erb
u
llyin
g
d
etec
tio
n
w
ith
a
d
va
n
ce
d
text
p
r
ep
r
o
ce
s
s
in
g
…
(
R
a
ke
s
h
B
a
p
u
Dh
u
ma
le
)
3145
cr
ea
ted
an
d
to
k
e
n
ized
u
s
in
g
th
e
T
en
s
o
r
Flo
w
lib
r
ar
y
'
s
wo
r
d
s
eq
u
e
n
ce
an
al
y
s
is
.
Prio
r
to
to
k
e
n
izatio
n
,
a
p
r
ep
r
o
ce
s
s
in
g
p
h
ase
was
u
s
ed
to
elim
in
ate
m
is
s
in
g
an
d
d
u
p
l
icate
d
o
cu
m
en
ts
,
er
r
o
n
eo
u
s
tex
t
f
o
r
m
ats,
an
d
tex
t
co
n
ten
t
lo
s
s
.
I
n
o
r
d
e
r
to
r
ed
u
ce
n
o
is
e
an
d
th
e
s
ize
o
f
th
e
f
ea
tu
r
e
s
et,
wo
r
d
s
th
at
p
r
o
v
id
e
d
n
o
s
ig
n
if
ican
ce
to
th
e
p
h
r
ase
wer
e
elim
in
ated
f
r
o
m
th
e
tex
t.
T
h
e
y
also
wo
u
ld
n
'
t
af
f
ec
t
tex
t
p
r
o
ce
s
s
in
g
f
o
r
t
h
e
in
ten
d
ed
p
u
r
p
o
s
e.
SVM
clas
s
if
ier
s
wer
e
b
u
ilt
an
d
ev
alu
ated
with
NB
an
d
L
R
f
o
r
th
e
p
u
r
p
o
s
e
o
f
class
if
y
in
g
c
y
b
er
b
u
lly
in
g
tex
ts
.
E
ac
h
class
if
ier
was
in
itialize
d
an
d
th
en
tr
ai
n
ed
u
s
in
g
th
e
s
ca
led
o
v
er
s
am
p
le
d
tr
ain
in
g
d
ata
u
s
in
g
t
h
e
f
it
m
eth
o
d
.
Af
te
r
th
e
test
d
ataset
was
ca
teg
o
r
ized
,
th
e
ac
cu
r
ac
y
s
co
r
e
tech
n
i
q
u
e
was
u
s
ed
t
o
ev
alu
ate
th
e
test
's
ef
f
ec
tiv
en
ess
.
Fig
u
r
e
3
d
em
o
n
s
tr
ates
th
e
b
est
p
er
f
o
r
m
in
g
cla
s
s
if
ier
s
,
wh
ich
wer
e
b
ased
o
n
SVM,
NB
an
d
L
R
.
T
h
e
ac
cu
r
ac
y
p
er
ce
n
ta
g
es
o
f
th
r
ee
d
is
tin
ct
class
if
ier
s
,
n
am
ely
NB
,
LR
,
an
d
s
u
p
p
o
r
t
v
ec
to
r
class
if
ier
,
ar
e
g
iv
en
in
th
e
Fig
u
r
e
3
a
f
ter
th
e
y
h
av
e
b
ee
n
tr
ain
e
d
with
two
d
is
tin
ct
tr
ain
in
g
m
o
d
els,
C
B
OW
an
d
s
k
ip
-
g
r
am
.
T
h
e
h
ig
h
est
ac
cu
r
ac
y
in
th
is
ca
teg
o
r
y
was
9
3
%
f
o
r
th
e
s
u
p
p
o
r
t
v
ec
to
r
class
if
ier
with
C
B
OW
,
f
o
llo
wed
b
y
8
9
%
f
o
r
NB
an
d
9
1
%
f
o
r
LR
.
NB
in
cr
ea
s
ed
to
9
2
%,
LR
to
9
3
%,
an
d
s
u
p
p
o
r
t
v
ec
to
r
class
if
ier
o
n
ce
m
o
r
e
h
a
d
th
e
h
ig
h
est
ac
cu
r
ac
y
at
9
5
%
f
o
r
th
e
s
k
ip
-
g
r
am
tr
ai
n
in
g
m
o
d
el.
Fo
r
b
o
th
tr
ain
in
g
m
o
d
els,
th
e
s
u
p
p
o
r
t
v
ec
to
r
class
if
ier
s
h
o
wed
th
e
b
est
o
v
e
r
all
ac
cu
r
ac
y
o
f
all
th
e
class
if
ier
s
,
s
co
r
in
g
9
3
%
f
o
r
C
B
OW
an
d
9
5
%
f
o
r
s
k
ip
-
g
r
am
.
Desp
ite
b
ein
g
less
ac
cu
r
ate
th
an
th
e
o
th
er
two
class
if
ier
s
,
NB
m
an
ag
ed
to
ac
h
iev
e
im
p
r
ess
iv
e
r
esu
lts
,
s
co
r
in
g
8
9
%
f
o
r
C
B
OW
an
d
9
2
%
f
o
r
s
k
ip
-
g
r
am
.
I
n
co
m
p
a
r
is
o
n
to
th
e
C
B
OW
m
o
d
el,
th
e
s
k
ip
-
g
r
am
tr
ain
in
g
ap
p
r
o
ac
h
ty
p
ically
y
ield
e
d
g
r
ea
ter
ac
cu
r
ac
y
p
e
r
ce
n
tag
es
ac
r
o
s
s
all
clas
s
if
ier
s
.
Acc
o
r
d
in
g
to
th
is
an
aly
s
is
,
th
e
co
m
b
in
atio
n
th
at
wo
r
k
s
b
est
a
m
o
n
g
t
h
o
s
e
ex
am
i
n
ed
is
th
e
s
u
p
p
o
r
t
v
ec
to
r
class
if
ier
co
m
b
i
n
ed
with
s
k
ip
-
g
r
am
,
wh
ich
y
ield
s
th
e
m
ax
im
u
m
ac
cu
r
ac
y
.
T
h
e
p
r
ec
is
io
n
,
r
ec
all,
an
d
F1
-
s
co
r
e
f
o
r
th
e
th
r
ee
s
en
tim
en
t
ca
teg
o
r
ies
lik
e
p
o
s
itiv
e,
n
e
u
tr
al,
an
d
n
eg
ativ
e
ar
e
g
iv
e
n
in
th
e
T
a
b
le
5
.
T
h
e
class
if
ier
p
r
o
d
u
ce
d
a
n
F1
-
s
co
r
e
o
f
0
.
8
6
,
r
ec
all
o
f
0
.
8
1
,
a
n
d
p
r
ec
is
io
n
o
f
0
.
9
2
f
o
r
n
e
g
ativ
e
s
en
tim
en
t.
T
h
e
lev
els
f
o
r
n
eu
t
r
al
s
en
tim
en
t
wer
e
0
.
5
7
,
0
.
6
9
,
an
d
0
.
6
3
,
in
th
at
o
r
d
er
.
An
F1
-
s
co
r
e
o
f
0
.
6
9
,
a
r
ec
all
o
f
0
.
7
5
,
an
d
a
p
r
ec
is
io
n
o
f
0
.
6
4
wer
e
o
b
tain
ed
f
r
o
m
p
o
s
itiv
e
s
en
tim
en
t
ca
teg
o
r
izatio
n
.
T
h
e
o
v
er
all
a
v
er
ag
e
p
r
ec
is
io
n
,
r
ec
all,
an
d
F1
-
s
co
r
e
wer
e
0
.
7
9
,
0
.
7
7
,
a
n
d
0
.
7
8
,
r
esp
ec
tiv
el
y
.
T
en
s
o
r
Flo
w
b
aselin
e
m
o
d
els
an
d
a
cy
b
er
b
u
lly
in
g
m
o
d
el
w
er
e
im
p
lem
en
ted
in
th
e
ex
p
er
i
m
en
t
u
s
in
g
Go
o
g
le
C
o
lab
GPUs
an
d
Py
th
o
n
3
.
8
.
T
o
aid
th
e
m
o
d
el
in
c
o
m
p
r
e
h
en
d
in
g
co
n
tex
t
an
d
an
aly
zin
g
im
p
o
r
tan
t
d
ata,
a
n
in
p
u
t
m
atr
ix
co
n
s
is
tin
g
o
f
3
5
,
8
7
3
twee
ts
was
g
en
er
ated
an
d
to
k
en
ized
.
Pre
p
r
o
ce
s
s
in
g
m
eth
o
d
s
r
ed
u
ce
d
n
o
is
e
an
d
f
ea
tu
r
e
s
et
s
ize
b
y
r
em
o
v
in
g
d
u
p
licate
an
d
m
is
s
in
g
d
o
cu
m
en
ts
,
in
co
r
r
ec
t
tex
t
f
o
r
m
a
ts
,
an
d
u
n
n
ec
ess
ar
y
wo
r
d
s
.
T
o
ca
teg
o
r
ize
tex
ts
i
n
clu
d
in
g
cy
b
er
b
u
lly
i
n
g
,
NB
an
d
L
R
wer
e
ev
alu
ated
with
SVM
class
if
ier
s
.
Acc
u
r
ac
y
s
co
r
es
wer
e
u
s
ed
to
ev
alu
ate
ea
ch
class
if
ier
'
s
p
er
f
o
r
m
a
n
ce
af
ter
it
h
a
d
b
ee
n
tr
ain
ed
u
s
in
g
s
ca
led
o
v
er
s
am
p
led
tr
ai
n
in
g
d
ata.
C
las
s
if
ier
s
Fig
u
r
e
3
.
Acc
u
r
ac
y
p
er
ce
n
tag
e
b
y
class
if
ier
an
d
tr
ain
i
n
g
m
o
d
el
ac
cu
r
ac
y
(
%)
T
ab
le
5
.
C
lass
if
icatio
n
r
ep
o
r
t
o
f
s
u
p
p
o
r
t v
ec
to
r
class
if
ier
P
r
e
c
i
s
i
o
n
R
e
c
a
l
l
F1
-
sc
o
r
e
N
e
g
a
t
i
v
e
0
.
9
2
0
.
8
1
0
.
8
6
N
e
u
t
r
a
l
0
.
5
7
0
.
6
9
0
.
6
3
P
o
si
t
i
v
e
0
.
6
4
0
.
7
5
0
.
6
9
A
v
e
r
a
g
e
/
T
o
t
a
l
0
.
7
9
0
.
7
7
0
.
7
8
T
ain
tin
g
m
o
d
el
ac
cu
r
ac
y
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
15
,
No
.
3
,
J
u
n
e
20
25
:
3
1
3
9
-
3
1
4
8
3146
5.
CO
NCLU
SI
O
N
C
y
b
er
b
u
lly
in
g
,
in
wh
ich
p
eo
p
le
u
s
e
s
o
cial
m
ed
ia
to
p
o
s
t
th
r
ea
ten
in
g
r
e
m
ar
k
s
,
is
a
m
ajo
r
i
s
s
u
e.
T
h
is
ca
n
h
av
e
a
s
er
io
u
s
n
eg
ativ
e
e
f
f
ec
t
o
n
p
eo
p
le'
s
liv
es,
e
s
p
ec
ial
ly
th
at
o
f
k
id
s
an
d
teen
ag
er
s
,
r
esu
ltin
g
in
s
u
icid
al
th
o
u
g
h
ts
an
d
d
ep
r
ess
io
n
.
T
h
e
s
tu
d
y
s
u
g
g
ests
a
f
o
u
r
-
s
tep
p
r
o
ce
d
u
r
e
th
at
in
v
o
lv
es
p
r
e
p
r
o
ce
s
s
in
g
,
f
ea
tu
r
e
ex
tr
ac
tio
n
,
class
if
icatio
n
,
an
d
ev
alu
atio
n
to
id
en
tify
cy
b
e
r
b
u
lly
in
g
.
Ma
k
in
g
a
v
ar
ied
la
b
eled
d
ataset
is
th
e
f
ir
s
t
s
tep
.
W
o
r
d
2
Vec
an
d
T
F
-
I
DF
ar
e
u
s
ed
in
f
ea
tu
r
e
ex
tr
ac
tio
n
to
tr
an
s
f
o
r
m
tex
t
in
to
v
ec
to
r
s
.
SVM
cla
s
s
if
ier
s
ar
e
u
s
ed
in
th
e
m
o
d
el,
an
d
th
eir
e
f
f
icac
y
is
co
m
p
ar
ed
to
th
at
o
f
lo
g
is
tic
r
eg
r
ess
io
n
an
d
NB
.
T
h
e
h
ig
h
est
ac
cu
r
ac
y
was
9
3
%
with
C
B
OW
an
d
9
5
%
with
th
e
s
u
p
p
o
r
t
v
ec
to
r
class
if
ier
em
p
lo
y
in
g
s
k
ip
-
g
r
am
.
Fo
r
s
en
tim
en
t
ca
teg
o
r
ies,
th
e
av
er
a
g
e
p
r
ec
is
io
n
an
d
r
ec
all
wer
e
0
.
7
7
a
n
d
0
.
7
9
,
r
esp
ec
tiv
ely
.
F
UNDING
I
NF
O
R
M
A
T
I
O
N
Au
th
o
r
s
s
tate
n
o
f
u
n
d
in
g
in
v
o
lv
ed
.
AUTHO
R
CO
NT
RI
B
UT
I
O
NS ST
A
T
E
M
E
N
T
T
h
is
jo
u
r
n
al
u
s
es
th
e
C
o
n
tr
ib
u
to
r
R
o
les
T
ax
o
n
o
m
y
(
C
R
ed
iT)
to
r
ec
o
g
n
ize
in
d
iv
id
u
al
au
th
o
r
co
n
tr
ib
u
tio
n
s
,
r
ed
u
ce
au
th
o
r
s
h
ip
d
is
p
u
tes,
an
d
f
ac
ilit
ate
co
llab
o
r
atio
n
.
Na
m
e
o
f
Aut
ho
r
C
M
So
Va
Fo
I
R
D
O
E
Vi
Su
P
Fu
R
ak
esh
B
ap
u
Dh
u
m
ale
✓
✓
✓
✓
✓
✓
✓
✓
✓
Ajay
Ku
m
ar
Dass
✓
✓
✓
✓
✓
Am
it Um
b
r
ajk
aa
r
✓
✓
✓
✓
✓
Pra
d
ee
p
Ma
n
e
✓
✓
✓
✓
✓
C
:
C
o
n
c
e
p
t
u
a
l
i
z
a
t
i
o
n
M
:
M
e
t
h
o
d
o
l
o
g
y
So
:
So
f
t
w
a
r
e
Va
:
Va
l
i
d
a
t
i
o
n
Fo
:
Fo
r
mal
a
n
a
l
y
s
i
s
I
:
I
n
v
e
s
t
i
g
a
t
i
o
n
R
:
R
e
so
u
r
c
e
s
D
:
D
a
t
a
C
u
r
a
t
i
o
n
O
:
W
r
i
t
i
n
g
-
O
r
i
g
i
n
a
l
D
r
a
f
t
E
:
W
r
i
t
i
n
g
-
R
e
v
i
e
w
&
E
d
i
t
i
n
g
Vi
:
Vi
su
a
l
i
z
a
t
i
o
n
Su
:
Su
p
e
r
v
i
s
i
o
n
P
:
P
r
o
j
e
c
t
a
d
mi
n
i
st
r
a
t
i
o
n
Fu
:
Fu
n
d
i
n
g
a
c
q
u
i
si
t
i
o
n
CO
NF
L
I
C
T
O
F
I
N
T
E
R
E
S
T
ST
A
T
E
M
E
NT
Au
th
o
r
s
s
tate
n
o
co
n
f
lict o
f
in
t
er
est.
DATA AV
AI
L
AB
I
L
I
T
Y
T
h
e
d
ata
th
at
s
u
p
p
o
r
t
th
e
f
in
d
in
g
s
o
f
th
is
s
tu
d
y
ar
e
av
aila
b
le
o
n
r
eq
u
est
f
r
o
m
th
e
co
r
r
esp
o
n
d
in
g
au
th
o
r
,
R
B
D.
RE
F
E
R
E
NC
E
S
[
1
]
C
.
C
h
e
n
g
,
Y
.
c
h
i
n
g
L
a
u
,
L.
C
h
a
n
,
a
n
d
J
.
W
.
L
u
k
,
“
P
r
e
v
a
l
e
n
c
e
o
f
s
o
c
i
a
l
med
i
a
a
d
d
i
c
t
i
o
n
a
c
r
o
ss
3
2
n
a
t
i
o
n
s:
M
e
t
a
-
a
n
a
l
y
s
i
s
w
i
t
h
su
b
g
r
o
u
p
a
n
a
l
y
si
s
o
f
c
l
a
ss
i
f
i
c
a
t
i
o
n
sc
h
e
me
s
a
n
d
c
u
l
t
u
r
a
l
v
a
l
u
e
s
,
”
A
d
d
i
c
t
i
v
e
Be
h
a
v
i
o
rs
,
v
o
l
.
1
1
7
,
p
.
1
0
6
8
4
5
,
2
0
2
1
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
a
d
d
b
e
h
.
2
0
2
1
.
1
0
6
8
4
5
.
[
2
]
R
.
S
.
D
e
o
r
a
a
n
d
D
.
M
.
C
h
u
d
a
sam
a
,
“
B
r
i
e
f
st
u
d
y
o
f
c
y
b
e
r
c
r
i
me
o
n
a
n
i
n
t
e
r
n
e
t
,
”
J
o
u
r
n
a
l
o
f
C
o
m
m
u
n
i
c
a
t
i
o
n
E
n
g
i
n
e
e
r
i
n
g
&
S
y
s
t
e
m
s
,
v
o
l
.
1
1
,
n
o
.
1
,
p
p
.
1
–
6
,
2
0
2
1
,
d
o
i
:
1
0
.
3
7
5
9
1
/
J
o
C
ES.
[
3
]
S
.
A
l
i
m
,
“
C
y
b
e
r
b
u
l
l
y
i
n
g
i
n
t
h
e
w
o
r
l
d
o
f
t
e
e
n
a
g
e
r
s
a
n
d
s
o
c
i
a
l
me
d
i
a
:
A
l
i
t
e
r
a
t
u
r
e
r
e
v
i
e
w
,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
rn
a
l
o
f
C
y
b
e
r
Be
h
a
v
i
o
r,
Psy
c
h
o
l
o
g
y
a
n
d
L
e
a
rn
i
n
g
,
v
o
l
.
6
,
n
o
.
2
,
p
p
.
6
8
–
9
5
,
2
0
1
6
,
d
o
i
:
1
0
.
4
0
1
8
/
I
JC
B
P
L
.
2
0
1
6
0
4
0
1
0
5
.
[
4
]
R
.
D
e
n
n
e
h
y
,
S
.
M
e
a
n
e
y
,
M
.
C
r
o
n
i
n
,
a
n
d
E
.
A
r
e
n
sm
a
n
,
“
Th
e
p
s
y
c
h
o
s
o
c
i
a
l
i
m
p
a
c
t
s
o
f
c
y
b
e
r
v
i
c
t
i
mi
sat
i
o
n
a
n
d
b
a
r
r
i
e
r
s
t
o
se
e
k
i
n
g
so
c
i
a
l
s
u
p
p
o
r
t
:
Y
o
u
n
g
p
e
o
p
l
e
’
s p
e
r
s
p
e
c
t
i
v
e
s,
”
C
h
i
l
d
r
e
n
a
n
d
Y
o
u
t
h
S
e
rv
i
c
e
s
Re
v
i
e
w
,
v
o
l
.
1
1
1
,
n
o
.
F
e
b
r
u
a
r
y
2
0
2
0
,
p
.
1
0
4
8
7
2
,
2
0
2
0
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
c
h
i
l
d
y
o
u
t
h
.
2
0
2
0
.
1
0
4
8
7
2
.
[
5
]
V
.
S
a
h
a
n
a
,
K
.
M
.
A
.
K
u
m
a
r
,
a
n
d
A
.
A
.
D
a
r
e
m,
“
A
c
o
m
p
a
r
a
t
i
v
e
a
n
a
l
y
si
s
o
f
mac
h
i
n
e
l
e
a
r
n
i
n
g
t
e
c
h
n
i
q
u
e
s
f
o
r
c
y
b
e
r
b
u
l
l
y
i
n
g
d
e
t
e
c
t
i
o
n
o
n
F
o
r
mS
p
r
i
n
g
i
n
t
e
x
t
u
a
l
mo
d
a
l
i
t
y
,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
C
o
m
p
u
t
e
r
N
e
t
w
o
rk
a
n
d
I
n
f
o
rm
a
t
i
o
n
S
e
c
u
r
i
t
y
,
v
o
l
.
1
5
,
n
o
.
4
,
p
p
.
3
6
–
4
7
,
2
0
2
3
,
d
o
i
:
1
0
.
5
8
1
5
/
i
j
c
n
i
s.
2
0
2
3
.
0
4
.
0
4
.
[
6
]
C
.
R
a
j
,
A
.
A
g
a
r
w
a
l
,
G
.
B
h
a
r
a
t
h
y
,
B
.
N
a
r
a
y
a
n
,
a
n
d
M
.
P
r
a
sa
d
,
“
C
y
b
e
r
b
u
l
l
y
i
n
g
d
e
t
e
c
t
i
o
n
:
H
y
b
r
i
d
m
o
d
e
l
s
b
a
se
d
o
n
ma
c
h
i
n
e
l
e
a
r
n
i
n
g
a
n
d
n
a
t
u
r
a
l
l
a
n
g
u
a
g
e
p
r
o
c
e
ssi
n
g
t
e
c
h
n
i
q
u
e
s,”
E
l
e
c
t
r
o
n
i
c
s
(
S
w
i
t
zer
l
a
n
d
)
,
v
o
l
.
1
0
,
n
o
.
2
2
,
p
p
.
1
–
2
0
,
2
0
2
1
,
d
o
i
:
1
0
.
3
3
9
0
/
e
l
e
c
t
r
o
n
i
c
s
1
0
2
2
2
8
1
0
.
[
7
]
A
.
F
.
A
l
q
a
h
t
a
n
i
a
n
d
M
.
I
l
y
a
s,
“
A
n
e
n
sem
b
l
e
-
b
a
se
d
m
u
l
t
i
-
c
l
a
ss
i
f
i
c
a
t
i
o
n
m
a
c
h
i
n
e
l
e
a
r
n
i
n
g
c
l
a
ss
i
f
i
e
r
s
a
p
p
r
o
a
c
h
t
o
d
e
t
e
c
t
mu
l
t
i
p
l
e
c
l
a
ss
e
s
o
f
c
y
b
e
r
b
u
l
l
y
i
n
g
,
”
M
a
c
h
i
n
e
L
e
a
rn
i
n
g
a
n
d
K
n
o
w
l
e
d
g
e
E
x
t
r
a
c
t
i
o
n
,
v
o
l
.
6
,
n
o
.
1
,
p
p
.
1
5
6
–
1
7
0
,
2
0
2
4
,
d
o
i
:
1
0
.
3
3
9
0
/
ma
k
e
6
0
1
0
0
0
9
.
[
8
]
T.
N
.
H
a
r
sh
i
t
h
a
e
t
a
l
.
,
“
P
r
o
Te
c
t
:
a
h
y
b
r
i
d
d
e
e
p
l
e
a
r
n
i
n
g
mo
d
e
l
f
o
r
p
r
o
a
c
t
i
v
e
d
e
t
e
c
t
i
o
n
o
f
c
y
b
e
r
b
u
l
l
y
i
n
g
o
n
s
o
c
i
a
l
med
i
a
,
”
Fr
o
n
t
i
e
rs
i
n
Art
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
,
v
o
l
.
7
,
2
0
2
4
,
d
o
i
:
1
0
.
3
3
8
9
/
f
r
a
i
.
2
0
2
4
.
1
2
6
9
3
6
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
-
8
7
0
8
E
n
h
a
n
cin
g
cy
b
erb
u
llyin
g
d
etec
tio
n
w
ith
a
d
va
n
ce
d
text
p
r
ep
r
o
ce
s
s
in
g
…
(
R
a
ke
s
h
B
a
p
u
Dh
u
ma
le
)
3147
[
9
]
A
.
M
u
n
e
e
r
,
A
.
A
l
w
a
d
a
i
n
,
M
.
G
.
R
a
g
a
b
,
a
n
d
A
.
A
l
q
u
s
h
a
i
b
i
,
“
C
y
b
e
r
b
u
l
l
y
i
n
g
d
e
t
e
c
t
i
o
n
o
n
so
c
i
a
l
m
e
d
i
a
u
si
n
g
s
t
a
c
k
i
n
g
e
n
sem
b
l
e
l
e
a
r
n
i
n
g
a
n
d
e
n
h
a
n
c
e
d
B
E
R
T
,
”
I
n
f
o
r
m
a
t
i
o
n
(
S
w
i
t
z
e
rl
a
n
d
)
,
v
o
l
.
1
4
,
n
o
.
8
,
2
0
2
3
,
d
o
i
:
1
0
.
3
3
9
0
/
i
n
f
o
1
4
0
8
0
4
6
7
.
[
1
0
]
G
.
X
i
n
g
y
i
a
n
d
H
.
M
.
A
d
n
a
n
,
“
P
o
t
e
n
t
i
a
l
c
y
b
e
r
b
u
l
l
y
i
n
g
d
e
t
e
c
t
i
o
n
i
n
s
o
c
i
a
l
me
d
i
a
p
l
a
t
f
o
r
ms
b
a
se
d
o
n
a
m
u
l
t
i
-
t
a
sk
l
e
a
r
n
i
n
g
f
r
a
mew
o
r
k
,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
D
a
t
a
a
n
d
N
e
t
w
o
rk
S
c
i
e
n
c
e
,
v
o
l
.
8
,
n
o
.
1
,
p
p
.
2
5
–
3
4
,
2
0
2
4
,
d
o
i
:
1
0
.
5
2
6
7
/
j
.
i
j
d
n
s.
2
0
2
3
.
1
0
.
0
2
1
.
[
1
1
]
N
.
G
s,
A
.
S
h
e
n
o
y
y
,
K
.
C
h
a
t
u
r
y
a
,
L.
Jc,
a
n
d
J.
S
h
r
e
e
,
“
D
e
t
e
c
t
i
o
n
o
f
c
y
b
e
r
b
u
l
l
y
i
n
g
u
si
n
g
N
LP
a
n
d
ma
c
h
i
n
e
l
e
a
r
n
i
n
g
i
n
s
o
c
i
a
l
n
e
t
w
o
r
k
s
f
o
r
b
i
-
l
a
n
g
u
a
g
e
,
”
I
n
t
e
rn
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
S
c
i
e
n
t
i
f
i
c
Re
s
e
a
r
c
h
&
En
g
i
n
e
e
r
i
n
g
T
r
e
n
d
s
,
v
o
l
.
1
0
,
n
o
.
1
,
p
p
.
2
3
9
5
–
2
5
6
6
,
2
0
2
4
.
[
1
2
]
M
.
S
.
I
sl
a
m
,
A
.
N
.
O
r
n
o
,
a
n
d
M
.
A
r
i
f
u
z
z
a
ma
n
,
“
A
p
p
r
o
a
c
h
t
o
so
c
i
a
l
med
i
a
c
y
b
e
r
b
u
l
l
y
i
n
g
a
n
d
h
a
r
a
ssm
e
n
t
d
e
t
e
c
t
i
o
n
u
si
n
g
a
d
v
a
n
c
e
d
ma
c
h
i
n
e
l
e
a
r
n
i
n
g
,
”
Av
a
i
l
a
b
l
e
a
t
S
S
RN
4
7
0
5
2
6
1
.
M
a
r
.
1
2
,
2
0
2
4
,
d
o
i
:
1
0
.
2
1
2
0
3
/
r
s
.
3
.
r
s
-
4
0
3
1
5
5
4
/
v
1
.
[
1
3
]
A
.
O
r
e
l
a
j
a
e
t
a
l
.
,
“
A
t
t
r
i
b
u
t
e
-
s
p
e
c
i
f
i
c
c
y
b
e
r
b
u
l
l
y
i
n
g
d
e
t
e
c
t
i
o
n
u
s
i
n
g
a
r
t
i
f
i
c
i
a
l
i
n
t
e
l
l
i
g
e
n
c
e
,
”
J
o
u
rn
a
l
o
f
E
l
e
c
t
r
o
n
i
c
&
I
n
f
o
rm
a
t
i
o
n
S
y
s
t
e
m
s
,
v
o
l
.
6
,
n
o
.
1
,
p
p
.
1
0
–
2
1
,
2
0
2
4
,
d
o
i
:
1
0
.
3
0
5
6
4
/
j
e
i
s.
v
6
i
1
.
6
2
0
6
.
[
1
4
]
S
.
P
e
r
i
c
h
e
r
l
a
a
n
d
E
.
I
l
a
v
a
r
a
s
a
n
,
“
O
v
e
r
c
o
mi
n
g
t
h
e
c
h
a
l
l
e
n
g
e
o
f
c
y
b
e
r
b
u
l
l
y
i
n
g
d
e
t
e
c
t
i
o
n
i
n
i
m
a
g
e
s:
A
d
e
e
p
l
e
a
r
n
i
n
g
a
p
p
r
o
a
c
h
w
i
t
h
i
ma
g
e
c
a
p
t
i
o
n
i
n
g
a
n
d
O
C
R
i
n
t
e
g
r
a
t
i
o
n
,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
C
o
m
p
u
t
i
n
g
a
n
d
D
i
g
i
t
a
l
S
y
st
e
m
s
,
v
o
l
.
1
5
,
n
o
.
1
,
p
p
.
3
9
3
–
4
0
1
,
2
0
2
4
,
d
o
i
:
1
0
.
1
2
7
8
5
/
i
j
c
d
s
/
1
5
0
1
3
0
.
[
1
5
]
K
.
S
a
i
f
u
l
l
a
h
,
M
.
I
.
K
h
a
n
,
S
.
J
a
ma
l
,
a
n
d
I
.
H
.
S
a
r
k
e
r
,
“
C
y
b
e
r
b
u
l
l
y
i
n
g
t
e
x
t
i
d
e
n
t
i
f
i
c
a
t
i
o
n
:
A
d
e
e
p
l
e
a
r
n
i
n
g
a
n
d
t
r
a
n
sf
o
r
mer
-
b
a
se
d
l
a
n
g
u
a
g
e
mo
d
e
l
i
n
g
a
p
p
r
o
a
c
h
,
”
EAI
En
d
o
rs
e
d
T
ra
n
s
a
c
t
i
o
n
s
o
n
I
n
d
u
st
ri
a
l
N
e
t
w
o
r
k
s
a
n
d
I
n
t
e
l
l
i
g
e
n
t
S
y
s
t
e
m
s
,
v
o
l
.
1
1
,
n
o
.
1
,
p
p
.
1
–
1
2
,
2
0
2
4
,
d
o
i
:
1
0
.
4
1
0
8
/
E
ETI
N
I
S
.
V
1
1
I
1
.
4
7
0
3
.
[
1
6
]
W
.
C
.
T
a
p
s
o
b
a
e
t
a
l
.
,
“
C
y
b
e
r
t
h
r
e
a
t
’
s
d
e
t
e
c
t
i
o
n
u
si
n
g
ma
c
h
i
n
e
l
e
a
r
n
i
n
g
a
l
g
o
r
i
t
h
ms,”
p
p
.
0
–
9
,
2
0
2
4
.
[
1
7
]
D
.
C
.
J
o
y
W
i
n
n
i
e
W
i
s
e
,
S
.
A
m
b
a
r
e
e
sh
,
B
.
P
.
R
a
m
e
s
h
,
D
.
S
u
g
u
m
a
r
,
J.
P
.
B
h
i
ma
v
a
r
a
p
u
,
a
n
d
A
.
S
.
K
u
m
a
r
,
“
La
t
e
n
t
sem
a
n
t
i
c
a
n
a
l
y
si
s b
a
s
e
d
se
n
t
i
me
n
t
a
l
a
n
a
l
y
si
s o
f
t
w
e
e
t
s
i
n
s
o
c
i
a
l
me
d
i
a
f
o
r
t
h
e
c
l
a
ssi
f
i
c
a
t
i
o
n
o
f
c
y
b
e
r
b
u
l
l
y
i
n
g
t
e
x
t
,
”
I
n
t
e
rn
a
t
i
o
n
a
l
J
o
u
rn
a
l
o
f
I
n
t
e
l
l
i
g
e
n
t
S
y
s
t
e
m
s
a
n
d
A
p
p
l
i
c
a
t
i
o
n
s
i
n
E
n
g
i
n
e
e
ri
n
g
,
v
o
l
.
1
2
,
n
o
.
7
s
,
p
p
.
2
6
–
3
5
,
2
0
2
4
.
[
1
8
]
A
.
A
k
h
t
e
r
,
K
.
A
.
U
z
z
a
l
,
a
n
d
M
.
M
.
A
.
P
o
l
a
s
h
,
“
C
y
b
e
r
b
u
l
l
y
i
n
g
d
e
t
e
c
t
i
o
n
a
n
d
c
l
a
ss
i
f
i
c
a
t
i
o
n
u
s
i
n
g
mu
l
t
i
n
o
m
i
a
l
n
a
ï
v
e
B
a
y
e
s
a
n
d
f
u
z
z
y
l
o
g
i
c
,
”
I
n
t
e
rn
a
t
i
o
n
a
l
J
o
u
rn
a
l
o
f
M
a
t
h
e
m
a
t
i
c
a
l
S
c
i
e
n
c
e
s
a
n
d
C
o
m
p
u
t
i
n
g
,
v
o
l
.
5
,
n
o
.
4
,
p
p
.
1
–
1
2
,
2
0
1
9
,
d
o
i
:
1
0
.
5
8
1
5
/
i
j
msc
.
2
0
1
9
.
0
4
.
0
1
.
[
1
9
]
B
.
Y
.
A
l
H
a
r
b
i
,
M
.
S
.
A
l
H
a
r
b
i
,
N
.
J
.
A
l
Z
a
h
r
a
n
i
,
M
.
M
.
A
l
s
h
e
a
i
l
,
a
n
d
D
.
M
.
I
b
r
a
h
i
m,
“
U
si
n
g
m
a
c
h
i
n
e
l
e
a
r
n
i
n
g
a
l
g
o
r
i
t
h
ms
f
o
r
a
u
t
o
m
a
t
i
c
c
y
b
e
r
b
u
l
l
y
i
n
g
d
e
t
e
c
t
i
o
n
i
n
A
r
a
b
i
c
so
c
i
a
l
me
d
i
a
,
”
J
o
u
r
n
a
l
o
f
I
n
f
o
rm
a
t
i
o
n
T
e
c
h
n
o
l
o
g
y
M
a
n
a
g
e
m
e
n
t
,
v
o
l
.
1
2
,
n
o
.
2
,
p
p
.
1
2
3
–
1
3
0
,
2
0
2
0
,
d
o
i
:
1
0
.
2
2
0
5
9
/
JI
TM
.
2
0
2
0
.
7
5
7
9
6
.
[
2
0
]
J.
H
a
n
i
,
M
.
N
a
s
h
a
a
t
,
M
.
A
h
m
e
d
,
Z.
Emad
,
E.
A
mer
,
a
n
d
A
.
M
o
h
a
m
me
d
,
“
S
o
c
i
a
l
m
e
d
i
a
c
y
b
e
r
b
u
l
l
y
i
n
g
d
e
t
e
c
t
i
o
n
u
si
n
g
mac
h
i
n
e
l
e
a
r
n
i
n
g
,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
Ad
v
a
n
c
e
d
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
a
n
d
A
p
p
l
i
c
a
t
i
o
n
s
,
v
o
l
.
1
0
,
n
o
.
5
,
2
0
1
9
,
d
o
i
:
1
0
.
1
4
5
6
9
/
I
JA
C
S
A
.
2
0
1
9
.
0
1
0
0
5
8
7
.
[
2
1
]
V
.
Ja
i
n
,
V
.
K
u
mar,
V
.
P
a
l
,
a
n
d
D
.
K
.
V
i
s
h
w
a
k
a
r
m
a
,
“
D
e
t
e
c
t
i
o
n
o
f
c
y
b
e
r
b
u
l
l
y
i
n
g
o
n
s
o
c
i
a
l
m
e
d
i
a
u
si
n
g
m
a
c
h
i
n
e
l
e
a
r
n
i
n
g
,
”
i
n
Pro
c
e
e
d
i
n
g
s
-
5
t
h
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
C
o
m
p
u
t
i
n
g
M
e
t
h
o
d
o
l
o
g
i
e
s
a
n
d
C
o
m
m
u
n
i
c
a
t
i
o
n
,
I
C
C
MC
2
0
2
1
,
2
0
2
1
,
p
p
.
1
0
9
1
–
1
0
9
6
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
C
M
C
5
1
0
1
9
.
2
0
2
1
.
9
4
1
8
2
5
4
.
[
2
2
]
M
.
M
.
I
sl
a
m,
M
.
A
.
U
d
d
i
n
,
L.
I
sl
a
m
,
A
.
A
k
t
e
r
,
S
.
S
h
a
r
mi
n
,
a
n
d
U
.
K
.
A
c
h
a
r
j
e
e
,
“
C
y
b
e
r
b
u
l
l
y
i
n
g
d
e
t
e
c
t
i
o
n
o
n
so
c
i
a
l
n
e
t
w
o
r
k
s
u
si
n
g
mac
h
i
n
e
l
e
a
r
n
i
n
g
a
p
p
r
o
a
c
h
e
s
,
”
i
n
2
0
2
0
I
EEE
Asi
a
-
P
a
c
i
f
i
c
C
o
n
f
e
re
n
c
e
o
n
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
a
n
d
D
a
t
a
E
n
g
i
n
e
e
ri
n
g
,
C
S
D
E
2
0
2
0
,
2
0
2
0
,
p
p
.
5
–
1
0
,
d
o
i
:
1
0
.
1
1
0
9
/
C
S
D
E5
0
8
7
4
.
2
0
2
0
.
9
4
1
1
6
0
1
.
[
2
3
]
T.
H
.
Te
n
g
a
n
d
K
.
D
.
V
a
r
a
t
h
a
n
,
“
C
y
b
e
r
b
u
l
l
y
i
n
g
d
e
t
e
c
t
i
o
n
i
n
s
o
c
i
a
l
n
e
t
w
o
r
k
s
:
A
c
o
m
p
a
r
i
so
n
b
e
t
w
e
e
n
ma
c
h
i
n
e
l
e
a
r
n
i
n
g
a
n
d
t
r
a
n
sf
e
r
l
e
a
r
n
i
n
g
a
p
p
r
o
a
c
h
e
s,
”
I
EEE
A
c
c
e
ss
,
v
o
l
.
1
1
,
p
p
.
5
5
5
3
3
–
5
5
5
6
0
,
2
0
2
3
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
3
.
3
2
7
5
1
3
0
.
[
2
4
]
K
.
S
i
d
d
h
a
r
t
h
a
,
K
.
R
.
K
u
m
a
r
,
K
.
J
.
V
a
r
ma,
M
.
A
mo
g
h
,
a
n
d
M
.
S
a
ms
o
n
,
“
C
y
b
e
r
b
u
l
l
y
i
n
g
d
e
t
e
c
t
i
o
n
u
si
n
g
mac
h
i
n
e
l
e
a
r
n
i
n
g
,
”
2
0
2
2
2
n
d
Asi
a
n
C
o
n
f
e
re
n
c
e
o
n
I
n
n
o
v
a
t
i
o
n
i
n
T
e
c
h
n
o
l
o
g
y
,
AS
I
A
N
C
O
N
2
0
2
2
,
p
p
.
1
–
4
,
2
0
2
2
,
d
o
i
:
1
0
.
1
1
0
9
/
A
S
I
A
N
C
O
N
5
5
3
1
4
.
2
0
2
2
.
9
9
0
9
2
0
1
.
[
2
5
]
H
.
H
o
ss
e
i
n
mar
d
i
,
R
.
I
.
R
a
f
i
q
,
R
.
H
a
n
,
Q
.
L
v
,
a
n
d
S
.
M
i
s
h
r
a
,
“
P
r
e
d
i
c
t
i
o
n
o
f
c
y
b
e
r
b
u
l
l
y
i
n
g
i
n
c
i
d
e
n
t
s
i
n
a
m
e
d
i
a
-
b
a
se
d
s
o
c
i
a
l
n
e
t
w
o
r
k
,
”
Pro
c
e
e
d
i
n
g
s
o
f
t
h
e
2
0
1
6
I
EEE/
AC
M
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
Ad
v
a
n
c
e
s
i
n
S
o
c
i
a
l
N
e
t
w
o
r
k
s
A
n
a
l
y
si
s
a
n
d
Mi
n
i
n
g
,
AS
O
N
AM
2
0
1
6
,
n
o
.
A
u
g
u
st
,
p
p
.
1
8
6
–
1
9
2
,
2
0
1
6
,
d
o
i
:
1
0
.
1
1
0
9
/
A
S
O
N
A
M
.
2
0
1
6
.
7
7
5
2
2
3
3
.
B
I
O
G
RAP
H
I
E
S O
F
AUTH
O
RS
Ra
k
e
sh
Ba
p
u
Dhu
m
a
le
is
a
n
a
ss
o
c
iate
p
ro
fe
ss
o
r
in
th
e
De
p
a
rtme
n
t
o
f
El
e
c
tro
n
ics
a
n
d
Tele
c
o
m
m
u
n
ica
ti
o
n
En
g
in
e
e
rin
g
a
t
AISS
M
S
In
stit
u
te
o
f
In
f
o
rm
a
ti
o
n
Tec
h
n
o
l
o
g
y
,
P
u
n
e
,
with
o
v
e
r
1
6
y
e
a
rs
o
f
a
c
a
d
e
m
ic
e
x
p
e
rien
c
e
.
He
h
o
l
d
s
a
P
h
.
D.
i
n
e
lec
tro
n
ics
a
n
d
tele
c
o
m
m
u
n
ica
ti
o
n
e
n
g
in
e
e
r
in
g
a
n
d
h
a
s
p
u
b
li
sh
e
d
9
7
in
ter
n
a
ti
o
n
a
l
jo
u
rn
a
l
p
a
p
e
rs,
2
2
c
o
n
fe
re
n
c
e
p
a
p
e
rs,
8
Co
p
y
ri
g
h
ts
a
n
d
6
p
a
ten
ts.
He
sp
e
c
ialize
s
in
a
rti
ficia
l
in
telli
g
e
n
c
e
a
n
d
m
a
c
h
in
e
lea
rn
in
g
.
He
h
a
s
se
c
u
re
d
re
se
a
rc
h
f
u
n
d
in
g
e
x
c
e
e
d
in
g
₹
2
4
.
8
lak
h
a
n
d
c
o
n
tri
b
u
ted
a
s
tec
h
n
ica
l
p
r
o
g
ra
m
c
h
a
i
r
fo
r
I
EE
E
E
S
CI
c
o
n
fe
re
n
c
e
s.
He
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
rb
d
.
sc
o
e
@g
m
a
il
.
c
o
m
.
Aja
y
K
u
m
a
r
D
a
ss
h
a
s
b
e
e
n
i
n
d
e
fe
n
se
f
o
r
3
6
y
e
a
rs
a
fter
g
ra
d
u
a
ti
n
g
i
n
e
lec
tri
c
a
l
e
n
g
in
e
e
r
in
g
.
P
a
ra
ll
e
ll
y
,
h
e
h
a
s
b
e
e
n
c
lo
se
ly
a
ss
o
c
iate
d
wi
th
a
c
a
d
e
m
ic
field
wit
h
a
b
o
u
t
ten
p
a
p
e
rs
t
o
h
is
c
re
d
it
i
n
las
t
th
re
e
y
e
a
rs,
p
u
b
li
s
h
e
d
in
in
tern
a
ti
o
n
a
l
j
o
u
r
n
a
ls.
He
h
a
s
1
8
y
e
a
rs
o
f
e
x
p
e
rien
c
e
in
d
e
fe
n
se
p
r
o
d
u
c
ti
o
n
a
s
QA
e
x
p
e
rt.
He
is
M
.
Tec
h
.
in
p
o
we
r
e
lec
tro
n
ics
a
n
d
in
d
u
strial
d
riv
e
s
f
o
ll
o
we
d
b
y
a
d
o
c
to
ra
te
in
e
lec
tro
n
ics
e
n
g
i
n
e
e
rin
g
with
h
is
re
se
a
rc
h
wo
rk
in
AI
field
.
C
u
rre
n
tl
y
h
e
is
wo
r
k
in
g
a
s
v
ice
p
re
si
d
e
n
t
in
a
p
o
we
r
g
e
n
e
ra
ti
o
n
c
o
m
p
a
n
y
.
He
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
a
jay
k
d
a
ss
d
e
f@g
m
a
il
.
c
o
m
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
15
,
No
.
3
,
J
u
n
e
20
25
:
3
1
3
9
-
3
1
4
8
3148
Am
it
Um
b
r
a
jk
a
a
r
is
a
s
so
c
iate
p
ro
fe
ss
o
r
in
De
p
a
rtme
n
t
o
f
M
e
c
h
a
n
ica
l
En
g
i
n
e
e
rin
g
a
t
S
c
h
o
o
l
o
f
En
g
in
e
e
rin
g
,
D
Y
P
a
ti
l
I
n
tern
a
ti
o
n
a
l
Un
iv
e
rsity
,
P
u
n
e
with
o
v
e
r
2
7
y
e
a
rs
o
f
a
c
a
d
e
m
ic
e
x
p
e
rien
c
e
.
He
h
o
ld
s
P
h
.
D
.
in
m
e
c
h
a
n
ica
l
e
n
g
in
e
e
rin
g
a
n
d
h
a
s p
u
b
li
sh
e
d
1
9
p
a
p
e
rs
in
i
n
tern
a
ti
o
n
a
l
j
o
u
r
n
a
ls
a
n
d
i
n
tern
a
ti
o
n
a
l
c
o
n
fe
re
n
c
e
.
He
h
a
s
p
u
b
li
s
h
e
d
0
6
p
a
ten
ts
a
n
d
5
c
o
p
y
ri
g
h
ts.
His
c
o
re
a
re
a
o
f
re
se
a
rc
h
wo
rk
a
re
v
iz.
s
y
ste
m
d
e
sig
n
,
v
i
b
ra
ti
o
n
a
n
a
ly
sis
,
imp
lem
e
n
tatio
n
o
f
AN
N
a
n
d
m
a
c
h
in
e
lea
rn
i
n
g
i
n
fa
u
l
t
d
iag
n
o
sis
d
o
m
a
in
.
He
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
a
m
e
e
th
u
m
b
ra
jk
a
a
r@g
m
a
il
.
c
o
m
.
Pra
d
e
e
p
M
a
n
e
BE,
M
E
(E&T
C),
P
h
.
D
.
,
is
th
e
p
ri
n
c
ip
a
l
a
t
AIS
S
M
S
IOIT,
P
u
n
e
.
Wi
th
8
6
j
o
u
r
n
a
l
p
a
p
e
rs,
6
6
c
o
n
fe
r
e
n
c
e
p
u
b
l
ica
ti
o
n
s,
a
n
d
1
1
c
o
-
a
u
t
h
o
re
d
e
n
g
in
e
e
rin
g
b
o
o
k
s,
h
is
c
o
n
tri
b
u
ti
o
n
s
a
re
n
o
tab
le.
He
h
a
s
g
u
i
d
e
d
7
P
h
.
D
.
sc
h
o
lars
a
s
a
re
c
o
g
n
ize
d
g
u
id
e
a
t
S
P
P
U
a
n
d
Bh
a
ra
ti
Vid
y
a
p
e
e
th
.
A
fe
ll
o
w
o
f
IEI
,
IET
E,
a
n
d
a
m
e
m
b
e
r
o
f
IEE
E,
IS
A,
a
n
d
IS
TE
,
h
e
h
a
s
re
c
e
iv
e
d
a
wa
rd
s
su
c
h
a
s
b
e
st
p
ri
n
c
ip
a
l
(S
P
P
U,
2
0
2
1
)
a
n
d
n
a
ti
o
n
a
l
a
wa
rd
s
fro
m
IS
T
E
a
n
d
C
S
I.
He
wa
s
CO
-
P
I
fo
r
th
e
IS
RO
-
UO
P
re
se
a
rc
h
g
ra
n
t
a
n
d
h
a
s
se
rv
e
d
o
n
th
e
BOS
fo
r
El
e
c
tr
o
n
ics
F
a
c
u
l
t
y
a
t
S
P
P
U
a
n
d
B
h
a
r
a
t
i
V
i
d
y
a
p
e
e
t
h
.
H
e
c
a
n
b
e
c
o
n
t
a
c
t
e
d
a
t
:
p
b
m
a
n
e
6
8
2
9
@
g
m
a
i
l
.
c
o
m
.
Evaluation Warning : The document was created with Spire.PDF for Python.