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.
11
,
No
.
4
,
A
u
g
u
s
t
2021
,
p
p
.
3
6
1
7
~
3
6
2
8
I
SS
N:
2088
-
8708
,
DOI
: 1
0
.
1
1
5
9
1
/
i
j
ec
e
.
v
11
i
4
.
pp
3
6
1
7
-
3
6
2
8
3617
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ec
e.
ia
esco
r
e.
co
m
Tex
t
cla
ss
ificatio
n
m
o
d
el f
o
r
m
eth
a
m
phe
ta
m
i
ne
-
r
el
a
ted
tw
eets
in Sou
theas
t
Asia
using
dual da
ta p
reproces
sing
t
e
ch
niques
Na
ro
ng
s
a
k
Cha
y
a
ng
k
o
o
n,
Ano
ng
na
rt
Sriv
iho
k
De
p
a
rtme
n
t
o
f
Co
m
p
u
ter S
c
ien
c
e
,
F
a
c
u
lt
y
o
f
S
c
ien
c
e
,
Ka
se
tsa
rt
Un
iv
e
rsit
y
,
T
h
a
il
a
n
d
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Au
g
2
8
,
2
0
2
0
R
ev
i
s
ed
J
an
4
,
2
0
2
1
A
cc
ep
ted
J
an
1
8
,
2
0
2
1
M
e
th
a
m
p
h
e
ta
m
in
e
a
d
d
ictio
n
is
a
p
ro
m
in
e
n
t
p
r
o
b
lem
in
S
o
u
t
h
e
a
st
A
sia
.
Dru
g
a
d
d
icts
o
f
ten
d
isc
u
ss
il
leg
a
l
a
c
ti
v
it
ies
o
n
p
o
p
u
lar
so
c
ial
n
e
tw
o
rk
in
g
se
r
v
ice
s.
T
h
e
se
in
d
iv
id
u
a
ls
s
p
re
a
d
m
e
ss
a
g
e
s
o
n
so
c
ial
m
e
d
ia
a
s
a
m
e
a
n
s
o
f
b
o
t
h
b
u
y
in
g
a
n
d
se
ll
in
g
d
ru
g
s
o
n
li
n
e
.
T
h
is
p
a
p
e
r
p
ro
p
o
se
s
a
m
o
d
e
l,
th
e
“
tex
t
c
las
si
f
ica
ti
o
n
m
o
d
e
l
o
f
m
e
th
a
m
p
h
e
ta
m
in
e
t
w
e
e
ts
in
S
o
u
th
e
a
st
As
ia”
(T
M
TA
),
to
id
e
n
ti
fy
w
h
e
th
e
r
a
twe
e
t
f
ro
m
S
o
u
th
e
a
st
A
sia
is
re
late
d
to
m
e
th
a
m
p
h
e
tam
in
e
a
b
u
se
.
T
h
e
r
e
se
a
rc
h
a
d
d
re
ss
e
s
th
e
we
a
k
n
e
ss
o
f
b
a
g
o
f
w
o
rd
s
(Bo
W
)
b
y
in
tro
d
u
c
in
g
Bo
W
a
n
d
W
o
rd
2
V
e
c
f
e
a
tu
re
se
lec
ti
o
n
(BW
F
)
tec
h
n
iq
u
e
s.
A
d
o
m
a
in
-
b
a
se
d
f
e
a
t
u
re
se
lec
ti
o
n
m
e
th
o
d
w
a
s
p
e
r
f
o
rm
e
d
u
sin
g
th
e
Bo
W
d
a
tas
e
t
a
n
d
W
o
rd
2
Ve
c
.
T
h
e
B
W
F
d
a
tas
e
t
p
ro
v
id
e
d
a
s
m
a
ll
e
r
n
u
m
b
e
r
o
f
f
e
a
tu
re
s
th
a
n
th
e
Bo
W
a
n
d
T
F
–
IDF
d
a
tas
e
t.
W
e
e
x
p
e
rim
e
n
ted
w
it
h
th
re
e
c
a
n
d
id
a
te
c
las
sif
ier
s:
S
u
p
p
o
rt
v
e
c
to
r
m
a
c
h
in
e
(S
V
M
),
d
e
c
isio
n
tree
(J4
8
)
a
n
d
n
a
iv
e
b
a
y
e
s
(NB).
W
e
f
o
u
n
d
t
h
a
t
th
e
J4
8
c
las
sif
ie
r
w
it
h
th
e
BW
F
d
a
tas
e
t
p
ro
v
id
e
d
t
h
e
b
e
st
p
e
rf
o
rm
a
n
c
e
f
o
r
th
e
T
M
TA
in
term
s
o
f
a
c
c
u
ra
c
y
(0
.
8
1
5
)
,
F
-
m
e
a
su
re
(0
.
8
1
8
),
Ka
p
p
a
(0
.
5
2
8
)
,
M
a
tt
h
e
w
s
c
o
rre
latio
n
c
o
e
ff
icie
n
t
(0
.
5
2
9
)
a
n
d
h
ig
h
a
re
a
u
n
d
e
r
th
e
ROC
Cu
rv
e
(0
.
7
6
3
)
.
M
o
re
o
v
e
r,
T
M
TA
p
ro
v
id
e
d
th
e
lo
w
e
st
ru
n
ti
m
e
(3
.
4
8
0
se
c
o
n
d
s
)
u
sin
g
t
h
e
J4
8
w
it
h
t
h
e
BW
F
d
a
tas
e
t.
K
ey
w
o
r
d
s
:
Data
p
r
ep
r
o
ce
s
s
in
g
Featu
r
e
s
elec
t
io
n
Me
th
a
m
p
h
eta
m
i
n
e
T
ex
t c
lass
if
icatio
n
T
w
ee
t
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
:
Nar
o
n
g
s
a
k
C
h
a
y
an
g
k
o
o
n
Dep
ar
t
m
en
t o
f
C
o
m
p
u
ter
Scie
n
ce
,
Facu
l
t
y
o
f
Scie
n
ce
Kasets
ar
t U
n
i
v
er
s
it
y
B
an
g
k
o
k
1
0
9
0
0
,
T
h
ailan
d
E
m
ail:
n
ar
o
n
g
s
a
k
.
ch
a
y
@
k
u
.
th
1.
I
NT
RO
D
UCT
I
O
N
So
u
t
h
ea
s
t
A
s
ia
is
co
n
s
id
er
ed
a
ce
n
tr
e
o
f
m
et
h
a
m
p
h
eta
m
i
n
e
p
r
o
d
u
ct
io
n
b
ec
au
s
e
o
f
m
a
n
y
r
elate
d
ar
r
ests
,
w
h
ic
h
co
n
tin
u
e
to
r
is
e
an
n
u
all
y
a
f
ter
in
cr
ea
s
i
n
g
f
o
u
r
-
f
o
ld
f
r
o
m
1
9
9
8
to
2
0
1
4
[
1
]
.
Dr
u
g
ad
d
icts
o
f
te
n
talk
ab
o
u
t
ac
ti
v
ities
r
elate
d
t
o
m
et
h
a
m
p
h
eta
m
i
n
e
o
n
p
o
p
u
lar
s
o
cial
n
et
w
o
r
k
i
n
g
s
er
v
ice
s
.
So
m
e
t
w
ee
ts
ar
e
p
u
b
lis
h
ed
o
n
s
o
cial
m
ed
ia
f
o
r
th
e
p
u
r
p
o
s
es
o
f
b
u
y
in
g
a
n
d
s
ellin
g
d
r
u
g
s
o
n
li
n
e.
Ho
w
e
v
er
,
litt
le
r
esear
c
h
h
as
ex
a
m
in
ed
t
h
e
d
e
v
elo
p
m
e
n
t
o
f
tex
t
clas
s
if
icatio
n
m
o
d
els
f
o
r
t
w
ee
ts
r
elati
n
g
to
m
et
h
a
m
p
h
eta
m
in
e
[
2
]
.
T
h
is
s
tu
d
y
’
s
o
b
j
ec
tiv
e
is
to
p
r
o
p
o
s
e
a
n
e
w
d
ata
p
r
ep
r
o
ce
s
s
in
g
t
e
ch
n
iq
u
e
f
o
r
m
et
h
a
m
p
h
eta
m
i
n
e
-
r
elate
d
t
w
ee
ts
in
So
u
t
h
ea
s
t
Asi
a.
Fo
r
th
is
p
u
r
p
o
s
e,
w
e
h
av
e
in
tr
o
d
u
ce
d
a
m
o
d
el
ca
ll
ed
th
e
“
te
x
t
cla
s
s
i
f
icat
io
n
m
o
d
el
o
f
m
et
h
a
m
p
h
eta
m
i
n
e
t
w
ee
ts
i
n
So
u
t
h
ea
s
t
A
s
ia
u
s
in
g
d
u
al
d
ata
p
r
ep
r
o
ce
s
s
in
g
tech
n
iq
u
e
s
(
T
M
T
A
)
”.
A
cr
itica
l
p
r
o
ce
s
s
in
th
e
d
ev
e
lo
p
m
en
t
o
f
th
e
T
MT
A
w
a
s
d
ata
p
r
ep
r
o
ce
s
s
i
n
g
u
s
i
n
g
th
e
b
a
g
-
of
-
w
o
r
d
s
(
B
o
W
)
m
o
d
el,
a
b
asic,
class
ical,
s
tr
aig
h
t
f
o
r
w
ar
d
tech
n
iq
u
e,
p
o
p
u
lar
f
o
r
d
ata
p
r
ep
r
o
ce
s
s
in
g
in
te
x
t
class
i
f
ic
atio
n
.
T
h
is
m
eth
o
d
co
n
s
id
er
s
t
h
e
f
r
eq
u
e
n
c
y
o
f
ea
ch
w
o
r
d
as
a
class
i
f
ica
tio
n
f
ea
tu
r
e
k
n
o
w
n
a
s
o
n
e
-
h
o
t
r
ep
r
esen
tatio
n
.
E
ac
h
w
o
r
d
is
r
ep
r
esen
ted
b
y
a
s
p
ar
s
e
v
ec
t
o
r
co
n
s
is
ti
n
g
o
f
its
i
n
d
ex
a
n
d
f
r
eq
u
en
c
y
[
3
,
4
]
.
A
s
f
ea
t
u
r
es
m
a
y
p
o
te
n
tiall
y
r
u
n
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
.
11
,
No
.
4
,
A
u
g
u
s
t
2021
:
3
6
1
7
-
3628
3618
in
to
m
u
ltip
le
v
ec
to
r
s
,
B
o
w
’
s
w
ea
k
n
ess
is
th
e
li
k
eli
h
o
o
d
o
f
r
esu
lt
in
g
i
n
a
lar
g
er
s
ize
i
n
th
e
f
o
r
m
o
f
h
i
g
h
-
d
i
m
en
s
io
n
al
v
ec
to
r
s
[
5
]
.
T
h
e
cu
r
r
en
t
s
t
u
d
y
r
ed
u
ce
d
t
h
is
w
e
ak
n
e
s
s
o
f
B
o
W
b
y
p
r
o
p
o
s
in
g
th
e
B
W
F
tech
n
iq
u
e,
a
n
o
v
el
ap
p
r
o
ac
h
co
n
s
is
t
in
g
o
f
B
o
W
an
d
W
o
r
d
2
Vec
f
ea
tu
r
e
s
elec
tio
n
co
m
p
r
i
s
i
n
g
t
w
o
s
tep
s
.
T
h
e
f
ir
s
t
s
tep
cr
ea
ted
a
tex
t
r
ep
r
esen
tatio
n
d
ataset
u
s
i
n
g
B
o
W
.
T
h
e
s
ec
o
n
d
in
v
o
lv
ed
a
d
o
m
ai
n
-
b
ase
d
f
ea
t
u
r
e
s
elec
t
io
n
,
p
er
f
o
r
m
ed
u
s
i
n
g
th
e
B
o
W
d
ataset
an
d
W
o
r
d
2
Vec
.
W
e
b
eg
an
b
y
co
llecti
n
g
t
w
ee
t
s
f
r
o
m
T
w
itter
th
at
o
r
ig
i
n
ated
f
r
o
m
So
u
t
h
ea
s
t
A
s
ia
an
d
d
i
v
i
d
in
g
th
e
m
in
to
t
w
o
cla
s
s
es,
n
a
m
el
y
ab
u
s
e
an
d
n
o
n
-
ab
u
s
e.
W
e
th
e
n
e
x
p
er
i
m
e
n
ted
w
it
h
t
h
r
ee
clas
s
i
f
icatio
n
alg
o
r
it
h
m
s
,
in
cl
u
d
in
g
s
u
p
p
o
r
t
v
ec
to
r
m
a
ch
in
e
(
SV
M)
,
d
ec
is
io
n
tr
ee
(
J
4
8
)
an
d
n
aiv
e
b
a
y
es
(
NB
)
.
W
e
m
ea
s
u
r
ed
th
e
p
er
f
o
r
m
a
n
ce
o
f
ea
c
h
m
o
d
el
b
ased
o
n
ac
cu
r
ac
y
,
F
-
m
ea
s
u
r
e,
th
e
A
r
ea
u
n
d
er
th
e
R
OC
c
u
r
v
e
(
A
U
C
)
,
Kap
p
a,
Ma
tth
e
w
s
co
r
r
elatio
n
co
ef
f
ici
en
t
(
MCC
)
an
d
r
u
n
ti
m
e.
Fin
a
ll
y
,
w
e
co
m
p
ar
ed
o
u
r
m
o
d
el
w
it
h
t
h
r
ee
d
if
f
er
e
n
t
d
ata
p
r
ep
r
o
ce
s
s
in
g
tec
h
n
iq
u
e
s
(
B
o
W
,
T
F
–
I
DF,
B
W
F).
T
h
e
ex
p
er
i
m
e
n
tal
r
esu
lts
s
h
o
w
e
d
th
at
t
h
e
T
MT
A
,
u
s
i
n
g
th
e
J
4
8
an
d
B
W
F d
ataset,
p
r
o
v
id
ed
th
e
h
ig
h
est p
er
f
o
r
m
an
ce
m
ea
s
u
r
e
m
en
ts
.
T
h
is
r
esear
ch
co
n
tr
ib
u
te
s
to
th
e
liter
atu
r
e
a
n
e
w
d
ata
p
r
ep
r
o
ce
s
s
in
g
tec
h
n
iq
u
e
f
o
r
class
i
f
y
in
g
m
et
h
a
m
p
h
eta
m
i
n
e
-
r
elate
d
t
w
e
ets.
B
W
F
p
r
o
v
id
es
a
s
m
a
ller
d
ataset
th
a
n
tr
ad
itio
n
al
o
r
w
id
el
y
u
s
ed
tech
n
iq
u
e
s
s
u
c
h
a
s
B
o
W
an
d
T
F
–
I
DF.
F
u
r
th
er
m
o
r
e,
th
e
T
MT
A
m
o
d
el
ca
n
ac
c
u
r
atel
y
id
en
ti
f
y
n
ar
co
tic
m
et
h
a
m
p
h
eta
m
i
n
e
t
w
ee
t
s
.
He
n
ce
,
th
is
m
o
d
el
ca
n
b
e
d
e
v
el
o
p
ed
as
an
ap
p
licatio
n
s
y
s
te
m
to
m
o
n
ito
r
t
w
ee
t
s
r
elate
d
to
m
et
h
a
m
p
h
eta
m
i
n
e
o
n
th
e
T
w
i
tter
p
latf
o
r
m
i
n
So
u
t
h
ea
s
t
Asi
a.
A
lt
h
o
u
g
h
a
h
an
d
f
u
l
o
f
r
esea
r
ch
er
s
h
a
v
e
u
s
ed
d
if
f
er
e
n
t
c
lass
i
f
ier
s
to
d
ev
e
lo
p
tex
t
cla
s
s
i
f
icatio
n
m
o
d
el
s
f
o
r
t
w
ee
t
s
r
elate
d
to
illeg
al
d
r
u
g
s
,
f
e
w
r
esear
ch
s
t
u
d
ies
ar
e
av
ailab
le.
P
h
a
n
et
a
l.
[
2
]
d
ev
elo
p
ed
a
m
o
d
el
to
d
etec
t
t
h
e
s
h
ar
in
g
o
f
t
w
ee
t
s
r
elate
d
to
illeg
al
d
r
u
g
s
,
i
n
cl
u
d
in
g
m
ar
ij
u
an
a,
co
ca
in
e
an
d
h
er
o
in
.
T
h
e
au
th
o
r
s
co
n
d
u
cted
th
eir
r
esear
ch
in
a
r
u
r
al
r
eg
io
n
o
f
th
e
Un
ited
States
o
f
Am
er
ica
(
US
A
)
.
T
h
eir
d
ataset
w
a
s
d
iv
id
ed
b
y
ex
p
er
ts
i
n
to
2
class
es:
A
b
u
s
e
o
r
n
o
n
-
ab
u
s
e.
B
o
W
an
d
T
F
-
I
DF
w
er
e
u
s
ed
as
d
ata
p
r
ep
r
o
ce
s
s
in
g
tech
n
iq
u
es,
a
n
d
3
class
i
f
ier
s
w
er
e
u
s
ed
:
SV
M,
J
4
8
an
d
NB
.
T
h
e
s
tu
d
y
f
i
n
d
in
g
s
r
e
v
ea
led
th
at
t
h
e
b
est
m
o
d
el
w
a
s
th
e
J
4
8
alg
o
r
ith
m
u
s
in
g
t
h
e
T
F
–
I
DF
m
et
h
o
d
,
w
h
ic
h
p
r
o
v
id
ed
th
e
h
ig
h
est
F
-
m
ea
s
u
r
e
o
f
0
.
7
4
8
0
.
R
ag
in
i
an
d
An
an
d
[
6
]
,
i
n
a
s
t
u
d
y
a
d
d
r
ess
in
g
t
h
e
m
u
lti
-
clas
s
c
las
s
i
f
ica
tio
n
p
r
o
b
lem
f
o
r
a
d
is
a
s
ter
ev
e
n
t
i
n
I
n
d
ia,
co
llected
7
0
,
8
1
7
r
elev
an
t
t
w
e
ets
f
r
o
m
2
0
1
4
to
2
0
1
5
.
T
h
ey
d
iv
id
ed
th
e
t
w
ee
t
s
i
n
to
7
class
es:
f
o
o
d
,
w
ater
,
s
h
elter
,
a
n
d
m
ed
ical
e
m
er
g
en
c
y
,
p
eo
p
le
tr
ap
p
ed
,
co
llap
s
ed
s
tr
u
ctu
r
e
a
n
d
elec
tr
ici
t
y
.
Nex
t,
t
h
e
a
u
t
h
o
r
s
cr
ea
ted
m
o
d
el
s
u
s
i
n
g
S
VM
an
d
NB
cl
ass
i
f
ier
s
.
T
h
e
b
est
-
p
er
f
o
r
m
i
n
g
m
o
d
el
in
t
h
is
ca
s
e
u
s
ed
t
h
e
S
VM
class
i
f
ier
w
it
h
th
e
T
F
-
I
DF
d
ataset.
W
an
g
et
a
l
.
[
7
]
co
m
p
ar
ed
th
e
e
f
f
icie
n
c
y
o
f
d
ata
p
r
ep
r
o
ce
s
s
in
g
tech
n
iq
u
es
co
n
s
is
tin
g
o
f
B
o
W
,
T
F
–
I
DF,
P
V
-
DM
an
d
P
V
-
DB
OW
.
T
h
e
d
ataset
u
s
ed
in
th
e
e
x
p
er
i
m
e
n
t,
b
ased
o
n
th
e
Sh
a
n
g
h
ai
a
n
d
Sh
e
n
zh
e
n
Sto
ck
E
x
c
h
a
n
g
e
s
,
w
a
s
d
i
v
id
ed
in
to
2
d
atasets
:
s
m
all
clas
s
a
n
d
b
ig
clas
s
.
T
h
e
class
i
f
icatio
n
m
o
d
els
w
er
e
NB
,
lo
g
is
t
ic
r
eg
r
ess
io
n
,
SVM,
K
-
n
ea
r
est
n
eig
h
b
o
u
r
(
K
NN)
an
d
Dec
is
io
n
T
r
ee
.
T
h
e
r
esear
ch
er
s
r
ep
o
r
ted
th
at
t
h
e
s
m
all
c
lass
d
ataset,
u
s
i
n
g
t
h
e
S
VM
al
g
o
r
ith
m
w
i
th
t
h
e
T
F
-
I
D
F
d
ataset,
d
e
m
o
n
s
tr
ated
t
h
e
h
i
g
h
e
s
t
ac
cu
r
ac
y
o
f
0
.
8
3
5
5
.
Gh
o
s
h
et
a
l
.
[
8
]
ad
d
r
ess
ed
t
h
e
m
u
lti
-
cla
s
s
c
lass
if
ica
tio
n
p
r
o
b
lem
f
o
r
d
is
a
s
ter
e
v
en
ts
co
n
s
is
ti
n
g
o
f
ea
r
th
q
u
a
k
es,
h
u
r
r
ican
e
s
,
elec
tr
ical
o
u
tag
es
a
n
d
d
r
o
u
g
h
t.
T
h
e
ex
p
er
i
m
en
ta
l
t
w
ee
ts
i
n
th
e
2
0
1
5
d
ataset
r
elate
d
to
th
e
Nep
al
ea
r
th
q
u
a
k
e
i
n
Ap
r
il
o
f
th
at
y
ea
r
.
T
h
e
T
F
–
I
DF
m
et
h
o
d
p
r
o
v
id
ed
th
e
d
ataset
f
o
r
th
e
m
o
d
el
s
th
at
w
er
e
cr
ea
ted
u
s
i
n
g
th
e
f
o
llo
w
i
n
g
clas
s
if
ier
s
: N
B
,
SV
M,
De
ci
s
io
n
T
r
ee
,
A
d
aB
o
o
s
t,
r
an
d
o
m
f
o
r
est a
n
d
g
r
ad
ien
t
b
o
o
s
tin
g
.
Acc
o
r
d
in
g
to
t
h
e
r
esu
lt
s
,
t
h
e
m
o
d
el
cr
ea
ted
u
s
i
n
g
SVM
w
it
h
th
e
T
F
-
I
DF
d
ataset
p
r
o
v
id
ed
t
h
e
h
ig
h
e
s
t
F
-
m
ea
s
u
r
e
o
f
0
.
9
1
7
8
.
B
u
r
el
an
d
A
la
n
i
[
9
]
also
ad
d
r
ess
ed
d
is
aster
ev
e
n
t
s
w
ith
a
d
ataset
th
at
co
n
s
i
s
ted
o
f
2
8
,
0
0
0
t
w
ee
ts
o
n
v
ar
io
u
s
cr
is
es
b
et
w
ee
n
2
0
1
2
an
d
2
0
1
3
.
T
h
eir
t
w
o
m
o
d
els
w
er
e
b
ased
o
n
th
e
co
n
v
o
lu
tio
n
al
n
e
u
r
al
n
e
t
w
o
r
k
(
C
NN)
clas
s
if
ier
u
s
in
g
a
w
o
r
d
-
e
m
b
ed
d
in
g
d
atase
t
an
d
t
h
e
S
VM
class
if
ier
u
s
in
g
th
e
T
F
-
I
DF
d
ataset.
T
h
e
r
es
u
lts
s
h
o
w
ed
t
h
at
C
N
N
w
it
h
a
w
o
r
d
-
e
m
b
ed
d
i
n
g
d
ataset
d
i
d
n
o
t
s
ig
n
i
f
ica
n
tl
y
o
u
tp
er
f
o
r
m
SVM
w
it
h
t
h
e
T
F
-
I
DF
d
ataset.
T
h
e
liter
atu
r
e
r
ev
ie
w
also
co
v
er
s
clas
s
i
f
ier
s
f
o
r
th
e
d
ev
elo
p
m
en
t
o
f
tex
t
clas
s
i
f
icatio
n
m
o
d
els
u
s
i
n
g
t
w
ee
ted
d
ata
w
i
th
clas
s
i
f
ier
s
co
n
s
i
s
ti
n
g
o
f
S
VM
,
J
4
8
an
d
NB
.
T
h
e
S
VM
class
i
f
ier
w
it
h
T
F
–
I
DF
w
a
s
w
id
el
y
u
s
ed
to
d
ev
elo
p
th
e
te
x
t
clas
s
i
f
icatio
n
m
o
d
el.
A
d
d
itio
n
all
y
,
r
esear
c
h
er
s
ch
o
s
e
th
e
J
4
8
an
d
NB
class
i
f
ie
r
s
to
d
ev
elo
p
th
e
tex
t c
las
s
i
f
ic
atio
n
m
o
d
el.
T
ex
t
r
ep
r
esen
tatio
n
is
a
p
ar
t
o
f
n
at
u
r
al
la
n
g
u
a
g
e
p
r
o
ce
s
s
in
g
(
N
L
P
)
,
w
h
i
ch
co
n
v
er
t
s
tex
t
d
ata
in
to
n
u
m
er
ic
v
ec
to
r
s
t
h
at
th
e
m
ac
h
in
e
ca
n
m
a
n
ip
u
late.
Nu
m
er
o
u
s
m
et
h
o
d
s
ca
n
p
er
f
o
r
m
tex
t
d
ata
co
n
v
er
s
io
n
.
O
n
e
s
i
m
p
le
ap
p
r
o
ac
h
g
i
v
es
ea
c
h
w
o
r
d
a
o
n
e
-
h
o
t
r
ep
r
esen
ta
tio
n
,
s
u
c
h
as
B
o
W
.
I
n
ad
d
itio
n
,
T
F
-
I
DF
tex
t
r
ep
r
esen
tatio
n
i
s
a
p
o
p
u
lar
t
ec
h
n
iq
u
e
f
o
r
d
ev
elo
p
in
g
a
te
x
t c
l
ass
i
f
icatio
n
m
o
d
el.
As
m
en
tio
n
ed
,
B
o
W
in
v
o
lv
e
s
a
co
llectio
n
o
f
w
o
r
d
s
th
at
r
ep
r
esen
t
s
th
e
f
ea
tu
r
e
s
o
f
th
e
te
x
t
b
y
t
h
e
w
o
r
d
f
r
eq
u
e
n
c
y
.
Fo
r
ex
a
m
p
le,
a
w
o
r
d
h
as
a
v
al
u
e
o
f
o
n
e
if
it
ap
p
ea
r
s
o
n
ce
i
n
th
e
tex
t.
T
h
e
v
ec
to
r
r
ep
r
esen
tatio
n
o
f
te
x
t
u
s
i
n
g
B
o
W
is
an
u
n
s
tr
u
ctu
r
ed
tex
t
d
o
cu
m
e
n
t
[
3
,
4
]
.
Fu
r
th
er
m
o
r
e,
ter
m
f
r
eq
u
en
c
y
(
T
F)
is
a
ca
lcu
latio
n
o
f
th
e
f
r
eq
u
e
n
c
y
o
f
a
w
o
r
d
th
at
ap
p
ea
r
s
in
th
e
d
o
cu
m
e
n
t
r
elati
v
e
to
th
e
to
tal
n
u
m
b
er
o
f
w
o
r
d
s
in
th
e
d
o
cu
m
e
n
t.
A
h
i
g
h
T
F
v
alu
e
in
d
icate
s
t
h
e
i
m
p
o
r
tan
ce
o
f
th
e
w
o
r
d
.
I
n
ad
d
itio
n
,
in
v
er
s
e
d
o
cu
m
e
n
t
f
r
eq
u
en
c
y
(
I
DF)
is
th
e
i
n
v
er
s
e
o
f
th
e
w
o
r
d
f
r
eq
u
e
n
c
y
in
t
h
e
d
o
cu
m
e
n
t.
A
h
i
g
h
I
D
F
v
alu
e
in
d
icate
s
an
i
m
p
o
r
tan
t
w
o
r
d
,
w
h
ic
h
s
h
o
u
ld
ap
p
ea
r
o
n
l
y
in
t
h
at
ca
te
g
o
r
y
an
d
n
o
t
in
o
t
h
er
ca
teg
o
r
ies.
T
h
er
ef
o
r
e,
ter
m
f
r
eq
u
en
c
y
-
i
n
v
e
r
s
e
d
o
cu
m
e
n
t
f
r
eq
u
e
n
c
y
(
T
F
-
I
DF)
is
th
e
w
e
ig
h
t
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
Text
cla
s
s
ifica
tio
n
mo
d
el
fo
r
meth
a
mp
h
eta
min
e
-
r
ela
ted
tw
e
ets in
S
o
u
th
ea
s
t..
.
(
N
a
r
o
n
g
s
a
k
C
h
a
ya
n
g
ko
o
n
)
3619
in
d
icati
n
g
t
h
e
i
m
p
o
r
ta
n
ce
o
f
t
h
e
w
o
r
d
.
T
F
-
I
DF
d
eter
m
i
n
es
t
h
e
w
eig
h
t
o
f
th
e
w
o
r
d
(
w
)
in
a
d
o
cu
m
en
t
(
d
)
th
at
ap
p
ea
r
s
in
th
e
d
o
cu
m
e
n
t,
b
ase
d
o
n
(
1
)
[
1
0
,
1
1
]
.
−
=
(
,
×
)
(
1
)
T
o
m
as
M
ik
o
lo
v
d
ev
elo
p
ed
W
o
r
d
2
Vec
as
a
to
o
l
f
o
r
N
L
P
.
T
h
is
to
o
l,
w
h
ic
h
e
m
p
lo
y
s
d
ee
p
n
e
u
r
al
n
et
w
o
r
k
s
t
h
at
tr
ai
n
w
o
r
d
ass
o
ciatio
n
s
to
s
y
n
o
n
y
m
o
u
s
w
o
r
d
s
,
is
u
s
ed
to
cr
ea
te
a
p
r
e
-
tr
ain
ed
w
o
r
d
e
m
b
ed
d
in
g
m
o
d
el
t
h
at
i
s
tr
ai
n
e
d
f
r
o
m
th
e
co
r
p
u
s
.
W
o
r
d
2
Vec
h
as
t
w
o
d
if
f
er
en
t
al
g
o
r
ith
m
s
:
T
h
e
S
k
ip
-
g
r
a
m
m
o
d
el
a
n
d
co
n
tin
u
o
u
s
b
ag
-
of
-
w
o
r
d
s
(
C
B
OW
)
.
T
h
o
s
e
m
o
d
els
r
ep
r
esen
t
f
ea
tu
r
es
t
h
at
u
s
e
th
e
v
ec
to
r
n
u
m
b
er
.
S
y
n
o
n
y
m
w
o
r
d
s
ca
n
b
e
f
o
u
n
d
b
y
u
s
i
n
g
t
h
e
co
s
in
e
s
i
m
ilar
it
y
f
u
n
ctio
n
b
et
w
ee
n
t
h
e
t
w
o
v
ec
to
r
s
[
1
2
]
.
C
o
s
i
n
e
s
i
m
ilar
it
y
i
s
a
s
ta
tis
tic
al
tech
n
iq
u
e
u
s
ed
to
m
ea
s
u
r
e
th
e
s
i
m
ilar
it
y
b
et
w
ee
n
t
w
o
d
o
cu
m
e
n
t
s
(
1
,
2
)
r
ep
r
esen
ted
b
y
n
u
m
er
ic
v
ec
to
r
s
in
th
e
p
r
o
j
ec
tio
n
s
p
ac
e.
A
co
s
i
n
e
s
i
m
ilar
it
y
v
a
lu
e
clo
s
er
to
o
n
e
s
u
g
g
e
s
ts
s
i
m
ilar
d
o
cu
m
e
n
ts
;
alter
n
ati
v
el
y
,
a
v
al
u
e
t
h
at
i
s
clo
s
er
to
ze
r
o
s
u
g
g
est
s
d
is
s
i
m
ilar
o
n
e
s
.
C
o
s
i
n
e
s
i
m
ilar
it
y
i
s
ca
lcu
lated
as s
h
o
w
n
i
n
(
2
)
[
1
3
]
.
(
1
,
2
)
=
1
,
2
(
1
.
2
)
1
/
2
+
(
1
.
2
)
1
/
2
(
2
)
Data
clas
s
if
icatio
n
is
th
e
p
r
o
c
ess
o
f
cr
ea
tin
g
m
ac
h
i
n
e
lear
n
i
n
g
m
o
d
els
in
w
h
ich
a
r
elatio
n
s
h
ip
e
x
is
ts
b
et
w
ee
n
t
h
e
f
ea
tu
r
es
a
n
d
clas
s
es
o
f
a
d
ataset.
P
o
p
u
lar
d
a
ta
class
i
f
icatio
n
al
g
o
r
ith
m
s
ar
e
S
VM
[
1
4
]
,
J
4
8
[
1
5
]
an
d
NB
[
1
6
]
.
SVM
is
a
cla
s
s
i
f
icatio
n
al
g
o
r
ith
m
d
e
s
i
g
n
ed
f
o
r
b
in
ar
y
-
cla
s
s
p
r
o
b
lem
s
.
S
VM
class
i
f
ier
s
cr
ea
te
a
d
ec
is
io
n
b
o
u
n
d
ar
y
i
n
a
h
y
p
er
p
lan
e
th
at
d
iv
id
es
t
h
e
d
ata
in
to
t
w
o
class
e
s
in
th
e
f
ea
t
u
r
e
s
p
ac
e
u
s
i
n
g
a
n
o
n
-
p
r
o
b
a
b
ilis
tic
b
in
ar
y
b
ased
o
n
a
lin
ea
r
f
u
n
ctio
n
.
T
h
e
f
u
n
ctio
n
d
eter
m
i
n
es
a
d
ec
is
io
n
b
o
u
n
d
ar
y
t
h
at
m
a
x
i
m
ize
s
th
e
m
ar
g
in
b
et
w
ee
n
t
h
e
s
u
p
p
o
r
t
v
ec
to
r
s
.
Ho
w
e
v
er
,
f
u
n
ctio
n
s
d
ef
i
n
i
n
g
t
h
e
d
ec
is
io
n
b
o
u
n
d
ar
y
ca
n
b
e
p
o
ly
n
o
m
ia
l
an
d
r
ad
ial
b
ased
.
T
h
e
ad
v
an
tag
e
o
f
t
h
e
SVM
class
i
f
ier
is
t
h
at
it
d
o
es
n
o
t
c
au
s
e
a
n
o
v
er
f
it
tin
g
p
r
o
b
lem
f
r
o
m
t
h
e
m
o
d
el
m
e
m
o
r
iz
in
g
to
o
m
an
y
o
f
t
h
e
tr
ai
n
in
g
s
et.
T
h
er
ef
o
r
e,
th
e
m
o
d
e
l
ca
n
n
o
t
clas
s
i
f
y
th
e
test
d
ataset
to
its
b
est
ab
ilit
y
[
1
4
]
.
I
n
co
m
p
ar
is
o
n
,
J
4
8
i
s
a
Dec
i
s
io
n
T
r
ee
class
i
f
ic
at
io
n
al
g
o
r
ith
m
.
J
4
8
class
i
f
ier
s
s
elec
t
t
h
e
f
ea
t
u
r
e
w
it
h
t
h
e
h
ig
h
es
t
in
f
o
r
m
at
io
n
g
ain
v
al
u
e,
w
h
ich
i
s
t
h
en
u
s
e
d
as
th
e
r
o
o
t
n
o
d
e
o
f
th
e
tr
ee
.
T
h
e
m
o
d
el
is
cr
ea
ted
u
s
i
n
g
a
to
p
-
d
o
w
n
g
r
ee
d
y
s
ea
r
ch
th
at
s
e
lects
f
ea
tu
r
e
s
f
r
o
m
th
e
r
o
o
t
n
o
d
e.
T
h
e
J
4
8
class
if
ier
i
s
s
u
itab
le
f
o
r
l
ar
g
e
d
atasets
b
ec
a
u
s
e
o
f
its
l
o
w
er
r
u
n
ti
m
e
[
1
5
]
.
Fin
all
y
,
NB
class
i
f
ier
s
u
s
e
a
co
n
d
itio
n
al
p
r
o
b
ab
ilit
y
ca
lc
u
l
atio
n
.
P
(
A
|
B
)
is
t
h
e
co
n
d
itio
n
al
p
r
o
b
ab
ilit
y
o
r
p
r
o
b
ab
ilit
y
t
h
at
e
v
en
t
B
o
cc
u
r
s
f
ir
s
t
a
n
d
is
f
o
llo
w
ed
b
y
e
v
en
t
A
.
P
(
A
∩
B
)
is
t
h
e
j
o
in
t
p
r
o
b
ab
ilit
y
o
r
th
e
p
r
o
b
ab
ilit
y
th
at
e
v
en
t
A
a
n
d
ev
en
t
B
w
il
l
b
o
th
o
cc
u
r
.
P
(
B
)
is
th
e
p
r
o
b
a
b
ilit
y
th
at
ev
e
n
t
B
w
ill
o
cc
u
r
.
T
h
e
NB
class
if
ier
m
a
k
es
it
ea
s
y
to
tr
ain
m
o
d
el
s
u
s
i
n
g
a
d
ataset
w
i
th
a
lar
g
e
n
u
m
b
er
o
f
f
ea
t
u
r
es,
s
u
ch
a
s
tex
t
d
ata
s
ets.
T
h
e
co
n
d
itio
n
al
p
r
o
b
ab
ilit
y
ca
lcu
latio
n
is
s
h
o
w
n
i
n
(
3
)
[
1
6
]
.
P
(
A
|
B
)
=
P
(
A
∩
B
)
P
(
B
)
(
3
)
P
er
f
o
r
m
a
n
ce
m
ea
s
u
r
e
m
e
n
ts
a
r
e
th
e
m
ea
s
u
r
e
m
en
ts
o
f
te
x
t
class
i
f
icatio
n
m
o
d
els
t
h
at
a
s
s
es
s
th
e
ir
ac
cu
r
ac
y
.
Ho
w
e
v
er
,
th
i
s
p
r
o
ce
s
s
m
a
y
s
o
m
e
ti
m
e
s
e
n
d
u
p
r
e
v
is
i
n
g
t
h
e
m
o
d
el
a
n
d
ev
al
u
ati
n
g
th
e
te
x
t
m
i
n
i
n
g
p
r
o
ce
s
s
u
n
t
il
t
h
e
m
o
d
el
i
s
th
e
m
o
s
t
ac
c
u
r
ate.
A
cc
u
r
ac
y
is
ca
lcu
lated
f
r
o
m
t
h
e
co
r
r
ec
t
class
i
f
icatio
n
o
f
t
h
e
m
o
d
el
t
h
at
co
n
s
id
er
s
all
clas
s
e
s
d
iv
id
ed
b
y
al
l d
ata,
as sh
o
w
n
in
(
4
)
[
1
7
]
.
=
+
+
+
+
(
4
)
F
-
m
ea
s
u
r
e
is
an
o
v
er
all
v
a
lu
e
th
at
m
ea
s
u
r
es
t
h
e
co
r
r
elatio
n
b
et
w
ee
n
p
r
ec
is
io
n
a
n
d
r
ec
all
v
al
u
es,
a
s
s
h
o
w
n
in
(
5
)
[
1
8
]
.
=
/
(
+
)
,
=
/
(
+
)
−
=
2
×
×
+
(5
)
A
U
C
i
s
t
h
e
ar
ea
u
n
d
er
th
e
r
e
ce
iv
er
o
p
er
atin
g
c
h
ar
ac
ter
is
t
i
c
(
R
OC
)
c
u
r
v
e
g
r
ap
h
.
A
U
C
i
s
th
e
ar
ea
u
n
d
er
t
h
e
2
D
g
r
ap
h
to
th
e
x
-
a
x
is
(
r
ep
r
esen
ti
n
g
t
h
e
FP
)
an
d
th
e
y
-
ax
is
(
r
ep
r
esen
tin
g
th
e
T
P
)
,
as
s
h
o
w
n
in
(
6
)
[
1
9
].
=
1
+
−
2
(
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
.
11
,
No
.
4
,
A
u
g
u
s
t
2021
:
3
6
1
7
-
3628
3620
T
h
e
Kap
p
a
co
ef
f
icien
t
is
a
s
t
atis
tic
u
s
ed
to
ex
a
m
in
e
t
h
e
co
n
s
i
s
te
n
c
y
o
f
th
e
r
es
u
lts
o
f
cla
s
s
i
f
icatio
n
b
et
w
ee
n
t
w
o
clas
s
es.
T
h
e
d
ata
s
et
u
s
ed
in
th
e
e
x
p
er
i
m
e
n
t
d
o
es
n
o
t
h
av
e
to
h
a
v
e
a
n
o
r
m
al
d
is
tr
ib
u
tio
n
o
r
n
o
n
-
p
ar
am
etr
ic
s
ta
tis
t
ics.
P
o
is
t
h
e
o
b
s
er
v
ed
p
r
o
b
a
b
ilit
y
o
f
ag
r
ee
m
e
n
t,
an
d
P
e
is
th
e
h
y
p
o
th
etica
l
ex
p
ec
ted
p
r
o
b
a
b
ilit
y
o
f
ag
r
ee
m
e
n
t,
as s
h
o
w
n
i
n
(
7
)
[
2
0
]
.
=
−
1
−
(
7
)
MCC
is
a
m
ea
s
u
r
e
o
f
t
h
e
e
f
f
i
cien
c
y
clas
s
i
f
icatio
n
r
es
u
lts
t
h
at
i
s
u
s
ed
w
it
h
t
w
o
-
cla
s
s
d
at
asets
.
T
h
e
MCC
v
al
u
e
d
eter
m
in
e
s
t
h
e
b
a
lan
ce
o
f
clas
s
if
icatio
n
r
e
s
u
l
ts
w
it
h
a
v
al
u
e
b
et
w
ee
n
-
1
an
d
+1
b
ein
g
ca
lc
u
lated
u
s
i
n
g
T
P
,
T
N,
FP
an
d
FN,
as sh
o
w
n
i
n
(
8
)
[
2
1
,
2
2
]
.
=
×
−
×
√
(
+
)
(
+
)
(
+
)
(
+
)
(
8
)
R
u
n
ti
m
e
p
er
f
o
r
m
a
n
ce
is
ca
lc
u
lated
f
r
o
m
t
h
e
3
co
m
p
o
n
e
n
t
s
o
f
th
e
ac
t
u
al
w
o
r
k
i
n
g
ti
m
e:
tr
ain
ti
m
e
,
test
ti
m
e
an
d
m
o
d
el
ti
m
e
[
2
3
]
.
2.
P
RO
P
O
SE
D
AL
G
O
R
I
T
H
M
T
h
e
B
W
F
alg
o
r
ith
m
w
as
a
d
o
m
ai
n
-
b
ased
f
ea
t
u
r
e
s
elec
ti
o
n
tech
n
iq
u
e
p
er
f
o
r
m
ed
u
s
i
n
g
th
e
B
o
W
d
atase
t
an
d
W
o
r
d
2
Vec
.
T
h
is
alg
o
r
ith
m
f
il
ter
ed
th
e
f
ea
t
u
r
es
o
f
t
h
e
B
o
W
d
ataset
to
p
r
o
d
u
ce
a
n
e
w
d
ata
s
et
f
o
r
th
e
cr
ea
tio
n
o
f
a
te
x
t
cla
s
s
i
f
ic
atio
n
m
o
d
el.
T
h
e
ad
v
an
ta
g
e
o
f
t
h
is
al
g
o
r
ith
m
w
a
s
t
h
at
it
cr
e
ated
a
B
W
F
d
ataset
s
m
al
ler
th
a
n
t
h
e
B
o
W
d
ataset.
T
h
e
B
W
F
alg
o
r
i
th
m
in
cl
u
d
e
d
t
w
o
s
tep
s
.
T
h
e
f
ir
s
t
s
tep
i
n
v
o
lv
ed
cr
ea
ti
n
g
th
e
B
o
W
d
ataset,
co
n
s
is
tin
g
o
f
th
e
s
et
o
f
a
n
in
s
ta
n
ce
w
h
er
e
a
b
o
w
s
u
c
h
t
h
at
ea
c
h
b
o
w
w
as
i
n
s
tan
ce
1
to
i
n
s
ta
n
ce
n
,
as sh
o
w
n
i
n
(
9
)
.
=
{
1
,
2
,
…
,
}
(
9
)
W
w
as
a
s
et
o
f
f
ea
t
u
r
es
i
n
t
h
e
B
o
W
d
ataset
w
h
er
e
W
co
n
tain
ed
th
e
s
et
o
f
f
ea
t
u
r
es
s
tar
tin
g
f
r
o
m
f
ea
t
u
r
e
1
to
f
ea
tu
r
e
w
,
as s
h
o
wn
in
(
10
)
.
=
{
1
,
2
,
…
,
}
(
1
0
)
T
h
e
s
ec
o
n
d
s
tep
in
v
o
lv
ed
a
d
o
m
a
in
-
b
ased
f
ea
t
u
r
e
s
elec
tio
n
tech
n
iq
u
e,
p
er
f
o
r
m
ed
u
s
i
n
g
B
o
W
an
d
W
o
r
d
2
Vec
.
T
h
e
d
o
m
ai
n
-
b
ase
d
f
ea
tu
r
e
s
elec
t
io
n
tech
n
iq
u
e
u
s
ed
th
r
ee
s
tep
s
:
W
o
r
d
2
Vec
w
as
u
s
ed
to
p
r
o
d
u
ce
a
p
r
e
-
tr
ain
ed
w
o
r
d
e
m
b
ed
d
in
g
m
o
d
el
f
r
o
m
th
e
m
et
h
a
m
p
h
eta
m
in
e
t
w
ee
t
d
ataset.
W
e
u
s
ed
th
e
S
k
ip
-
g
r
a
m
m
o
d
el,
an
al
g
o
r
it
h
m
t
h
at
g
en
er
ated
th
e
p
r
e
-
tr
ain
ed
wo
r
d
em
b
ed
d
in
g
m
o
d
el
u
s
i
n
g
W
o
r
d
2
Vec
.
T
o
m
as
Mi
k
o
lo
v
s
u
g
g
ested
th
is
alg
o
r
ith
m
,
w
h
ic
h
w
as
s
u
p
er
io
r
f
o
r
in
f
r
eq
u
en
t
w
o
r
d
s
.
T
h
o
s
e
w
o
r
d
s
co
n
s
i
s
ted
o
f
tec
h
n
ical
ter
m
s
,
s
la
n
g
n
a
m
e
an
d
s
y
n
o
n
y
m
n
a
m
e.
T
h
e
S
k
ip
-
g
r
a
m
m
o
d
el
s
elec
ted
in
f
r
eq
u
e
n
t
w
o
r
d
s
to
ca
lcu
late
th
e
v
ec
to
r
n
u
m
b
er
.
T
h
u
s
,
i
n
f
r
eq
u
e
n
t
w
o
r
d
s
h
ad
a
h
i
g
h
er
-
q
u
al
it
y
v
ec
to
r
n
u
m
b
er
th
a
n
w
h
en
u
s
i
n
g
C
B
OW
[
1
2
]
.
T
h
e
p
r
e
-
tr
ain
ed
w
o
r
d
em
b
ed
d
in
g
m
o
d
el
co
n
s
i
s
ted
o
f
1
0
0
-
d
i
m
en
s
io
n
a
l
f
ea
t
u
r
es
r
ep
r
esen
t
ed
b
y
v
ec
to
r
n
u
m
b
er
.
W
e
d
ef
i
n
ed
th
e
1
0
0
-
d
i
m
e
n
s
io
n
al
f
ea
tu
r
es
in
f
o
cu
s
i
n
g
o
n
r
u
n
t
i
m
e
co
m
p
ete
n
cie
s
th
at
w
er
e
u
s
ed
to
cr
ea
te
th
e
p
r
e
-
tr
ain
ed
w
o
r
d
e
m
b
ed
d
in
g
m
o
d
el
f
r
o
m
a
lar
g
e
co
r
p
u
s
.
T
h
e
s
et
o
f
d
o
m
ai
n
-
b
ased
f
ea
t
u
r
es (
SDB
F)
w
a
s
cr
ea
ted
b
y
m
e
asu
r
i
n
g
th
e
co
s
i
n
e
s
i
m
ilar
it
y
b
et
w
ee
n
d
o
m
ai
n
k
e
y
w
o
r
d
s
i
n
t
h
e
p
r
e
-
tr
ain
ed
wo
r
d
em
b
ed
d
in
g
m
o
d
el.
O
u
r
r
e
s
ea
r
ch
u
s
ed
t
h
e
k
e
y
w
o
r
d
“
met
h
a
mp
h
eta
min
e
”
as th
e
co
m
m
o
n
n
a
m
e
o
f
m
et
h
a
m
p
h
eta
m
i
n
e.
=
(
−
,
“
”
)
T
h
e
SDB
F
w
as
s
o
r
ted
b
y
d
e
s
ce
n
d
i
n
g
co
s
i
n
e
s
i
m
ilar
it
y
.
I
f
th
e
co
s
in
e
s
i
m
ilar
it
y
w
as
e
q
u
al
to
o
r
g
r
ea
ter
th
a
n
0
.
8
,
th
o
s
e
f
ea
t
u
r
e
s
w
er
e
s
elec
ted
f
o
r
in
cl
u
s
io
n
as
f
il
ter
f
ea
t
u
r
es
o
f
t
h
e
B
o
W
d
ataset.
T
h
e
SDB
F
co
n
tain
ed
t
h
e
s
et
o
f
f
ea
t
u
r
es st
ar
tin
g
f
r
o
m
f
ea
t
u
r
e
1
to
f
ea
tu
r
e
w'
,
as sh
o
w
n
in
(
11
)
.
S
DB
F
=
{
1
,
2
,
…
,
′
}
(
1
1
)
T
h
e
B
o
W
d
ataset
w
a
s
f
i
lter
e
d
to
k
ee
p
o
n
ly
t
h
e
f
ea
tu
r
es
i
n
th
e
SDB
F.
Nex
t,
t
h
e
B
o
W
d
a
taset
w
as
co
n
s
id
er
ed
b
ased
o
n
t
h
e
s
u
m
m
ed
f
r
eq
u
en
c
y
in
ea
ch
i
n
s
ta
n
ce
o
f
t
h
e
d
atase
t.
I
f
t
h
e
s
u
m
f
r
eq
u
e
n
c
y
o
f
a
n
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
Text
cla
s
s
ifica
tio
n
mo
d
el
fo
r
meth
a
mp
h
eta
min
e
-
r
ela
ted
tw
e
ets in
S
o
u
th
ea
s
t..
.
(
N
a
r
o
n
g
s
a
k
C
h
a
ya
n
g
ko
o
n
)
3621
in
s
ta
n
ce
w
as
eq
u
al
to
ze
r
o
,
th
at
in
s
tan
ce
w
a
s
d
elete
d
f
r
o
m
t
h
e
B
o
W
d
ataset.
T
h
is
r
e
s
ea
r
ch
u
s
ed
th
e
R
p
r
o
g
r
am
m
i
n
g
p
ac
k
a
g
e
to
i
m
p
le
m
e
n
t
t
h
e
B
W
F
al
g
o
r
ith
m
[
2
4
]
.
T
h
e
p
r
o
p
o
s
ed
d
ata
p
r
ep
r
o
ce
s
s
in
g
tec
h
n
iq
u
e
co
n
s
is
ted
o
f
th
e
B
W
F a
lg
o
r
it
h
m
,
as s
h
o
w
n
i
n
Fi
g
u
r
e
1
.
In
p
u
t
:
=
{
1
,
2
,
…
,
}
M
e
t
h
a
m
p
h
e
t
a
m
i
n
e
T
w
e
e
t
D
a
t
a
set
Ou
t
p
u
t
:
=
{
′
1
,
′
2
,
…
,
′
}
.
1
Pre
-
t
r
a
i
n
e
d
w
o
rd
e
m
b
e
d
d
i
n
g
m
o
d
e
l
=
W
o
r
d
2
V
e
c
(
Me
t
h
a
m
p
h
e
t
a
m
i
n
e
T
w
e
e
t
D
a
t
a
se
t
)
2
S
D
B
F
=
C
o
si
n
e
S
i
m
i
l
a
r
i
t
y
(
P
re
-
t
ra
i
n
e
d
w
o
r
d
e
m
b
e
d
d
i
n
g
m
o
d
e
l
,
“
k
e
y
w
o
r
d
s
”
)
3
S
D
B
F
i
s so
r
t
e
d
b
y
d
e
sce
n
d
i
n
g
c
o
si
n
e
si
mi
l
a
r
i
t
y
4
Fo
r
e
a
c
h
in
S
D
B
F
5
I
f
c
o
si
n
e
si
mi
l
a
r
i
t
y
v
a
l
u
e
l
e
ss t
h
a
n
0
.
8
6
…….
.
.
is
r
e
mo
v
e
d
f
r
o
m
S
D
B
F
7
E
n
d
I
f
8
E
n
d
F
o
r
9
R
e
t
u
r
n
S
D
B
F
d
a
t
a
se
t
S
D
BF
=
{
1
,
2
,
…
,
′
}
10
d
a
t
a
se
t
=
C
o
p
y
o
f
11
d
a
t
a
se
t
i
s
IN
N
E
R
J
o
i
n
a
n
d
S
D
B
F
=
{
1
,
2
,
…
,
}
12
Fo
r
e
a
c
h
′
i
n
d
a
t
a
s
e
t
13
|
′
|
i
s
t
h
e
s
u
m fr
e
q
u
e
n
c
y
o
f
a
n
i
n
s
t
a
n
c
e
i
n
′
14
If
|
′
|
e
q
u
a
l
t
o
0
1
5
…….
.
.
′
i
s re
mo
v
e
d
f
r
o
m
d
a
t
a
se
t
1
6
E
n
d
I
f
1
7
E
n
d
F
o
r
18
R
e
t
u
r
n
d
a
t
a
se
t
Fig
u
r
e
1
.
B
W
F a
lg
o
r
ith
m
Fro
m
Fi
g
u
r
e
1
,
t
h
e
r
esu
lt
o
f
t
h
e
B
W
F
alg
o
r
ith
m
w
as
a
n
e
w
d
ataset,
ca
lled
th
e
“
B
W
F
d
atas
et”,
w
h
ic
h
u
s
ed
th
e
s
a
m
e
tex
t
r
ep
r
esen
tatio
n
o
u
tco
m
es
f
r
o
m
th
e
B
o
W
d
ataset.
T
h
is
d
ataset
w
a
s
u
s
ed
f
o
r
tex
t
class
i
f
icatio
n
in
t
h
at
t
h
e
w
o
r
d
f
r
eq
u
en
c
y
w
a
s
u
s
ed
f
o
r
th
e
f
e
atu
r
e
o
f
th
e
tr
ai
n
i
n
g
w
i
th
t
h
e
c
lass
i
f
ier
alg
o
r
it
h
m
.
Ho
w
e
v
er
,
th
e
B
W
F
d
ataset
h
a
d
f
e
w
er
f
ea
t
u
r
es
an
d
in
s
ta
n
ce
s
th
a
n
th
e
B
o
W
d
ataset.
T
h
e
B
W
F
co
n
tain
ed
th
e
s
et
o
f
v
ec
to
r
izatio
n
(
b
o
w
'
)
,
w
h
er
e
ea
ch
v
ec
to
r
izatio
n
w
as
f
r
o
m
in
s
ta
n
ce
1
to
in
s
ta
n
ce
m
,
as
s
h
o
w
n
in
(
12
)
:
=
{
′
1
,
′
2
,
…
,
′
}
(
1
2
)
P
r
o
o
f:
L
et
w
b
e
th
e
n
u
m
b
er
o
f
f
ea
tu
r
es
in
B
o
W
.
L
et
S
DB
F
b
e
t
h
e
s
et
o
f
f
ea
t
u
r
es.
S
DB
F
d
er
iv
e
s
f
r
o
m
th
e
co
s
in
e
s
i
m
i
lar
it
y
u
s
in
g
t
h
e
t
h
r
esh
o
ld
o
f
0
.
8
.
L
et
w'
b
e
th
e
n
u
m
b
er
o
f
f
ea
t
u
r
es
in
S
DB
F
.
T
h
e
B
W
F
d
ataset
i
s
d
er
iv
ed
f
r
o
m
B
o
W
w
it
h
o
n
l
y
t
h
e
f
ea
t
u
r
es
i
n
S
DB
F
.
T
h
u
s
,
th
e
n
u
m
b
er
o
f
f
ea
t
u
r
es
in
t
h
e
B
W
F
d
ataset
m
u
s
t
b
e
at
m
o
s
t
w'
.
Mo
r
eo
v
er
,
th
e
B
WF
d
ataset
is
p
r
o
d
u
ce
d
b
y
r
e
m
o
v
in
g
(
in
s
tan
ce
o
f
)
B
o
W
in
wh
ich
t
h
e
s
u
m
s
o
f
al
l
f
ea
t
u
r
e
f
r
eq
u
en
cie
s
ar
e
eq
u
al
to
0
.
T
h
er
ef
o
r
e,
th
e
n
u
m
b
er
o
f
in
s
tan
ce
s
in
t
h
e
B
W
F
d
atase
t
m
u
s
t
b
e
less
th
a
n
th
at
o
f
B
o
W
.
3.
RE
S
E
ARCH
M
E
T
H
O
D
T
h
is
r
esear
ch
co
n
s
is
ted
o
f
t
w
o
o
b
j
ec
tiv
es.
T
h
e
f
ir
s
t
w
a
s
th
e
d
ev
elo
p
m
en
t
o
f
th
e
“
B
W
F”
d
ataset.
T
h
e
s
ec
o
n
d
w
a
s
th
e
d
ev
elo
p
m
e
n
t
o
f
th
e
T
M
T
A
,
w
h
ich
co
n
s
is
t
ed
o
f
th
e
f
o
llo
w
i
n
g
s
tep
s
:
t
wee
t
co
llectio
n
,
d
ata
p
r
ep
r
o
ce
s
s
in
g
,
cla
s
s
i
f
icatio
n
,
p
er
f
o
r
m
a
n
ce
te
s
ti
n
g
an
d
h
y
p
o
th
esi
s
tes
tin
g
,
a
s
s
h
o
w
n
i
n
t
h
e
o
v
er
v
ie
w
o
f
t
h
e
r
esear
ch
f
r
a
m
e
w
o
r
k
i
n
Fi
g
u
r
e
2
.
3
.
1
.
T
w
ee
t
co
llect
io
n
3
.
1
.
1
.
Sy
no
ny
m
ide
ntif
ica
t
io
n
T
h
is
p
r
o
ce
d
u
r
e
in
v
o
lv
ed
th
e
id
en
tif
icatio
n
o
f
k
ey
w
o
r
d
s
r
elate
d
to
m
eth
am
p
h
etam
in
e
co
n
s
is
tin
g
o
f
th
e
co
m
m
o
n
n
am
e,
s
lan
g
n
am
e
an
d
s
tr
ee
t
n
am
e.
T
h
ese
w
er
e
co
llected
an
d
id
en
tif
ied
b
y
th
e
UK
p
o
lice
[
2
5
]
.
I
n
ad
d
itio
n
,
w
e
u
s
ed
th
e
co
m
m
o
n
n
am
e
o
f
m
eth
am
p
h
etam
in
e
to
m
ea
s
u
r
e
co
s
in
e
s
im
ilar
ity
w
ith
Go
o
g
le
New
s
v
ec
to
r
s
[
2
6
]
to
lo
o
k
f
o
r
ad
d
itio
n
al
s
lan
g
n
am
es
th
at
h
ad
n
o
t
b
ee
n
co
llected
an
d
id
en
tif
ied
b
y
th
e
UK
p
o
lice.
3
.
1
.
2
.
T
w
ee
t
re
t
riev
a
l
T
w
ee
t
r
etr
iev
al
i
s
th
e
s
elec
tio
n
o
f
s
h
o
r
t
tex
t
o
n
T
w
itter
r
elat
ed
to
m
eth
a
m
p
h
e
ta
m
in
e
t
h
at
w
a
s
p
o
s
ted
b
y
u
s
er
s
i
n
So
u
th
ea
s
t
A
s
ia,
s
p
ec
if
icall
y
T
h
ailan
d
,
I
n
d
o
n
e
s
ia,
an
d
M
y
an
m
ar
.
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
.
11
,
No
.
4
,
A
u
g
u
s
t
2021
:
3
6
1
7
-
3628
3622
Fig
u
r
e
2
.
R
esear
ch
f
r
a
m
e
w
o
r
k
3
.
1
.
3
.
T
w
ee
t
la
belin
g
T
w
ee
t
s
w
er
e
lab
eled
b
y
an
ex
p
er
t
f
r
o
m
t
h
e
R
o
y
a
l
T
h
ai
P
o
li
ce
Fo
r
en
s
ics
O
f
f
ice
in
to
2
class
es:
No
n
-
ab
u
s
e
o
r
ab
u
s
e.
No
n
-
ab
u
s
e
t
wee
ts
m
e
n
tio
n
ed
th
e
p
en
alt
y
f
o
r
u
s
i
n
g
m
et
h
a
m
p
h
eta
m
i
n
e
o
r
its
u
s
e
a
s
a
m
ed
ici
n
e.
T
h
e
ab
u
s
e
cla
s
s
co
n
tai
n
ed
t
wee
ts
ab
o
u
t
th
e
ille
g
al
u
s
e
o
f
m
eth
a
m
p
h
eta
m
i
n
e,
i
n
cl
u
d
in
g
t
wee
ts
p
r
o
m
o
tin
g
th
e
u
s
e
o
f
m
et
h
a
m
p
h
eta
m
i
n
e,
s
u
c
h
as e
n
co
u
r
ag
i
n
g
s
u
b
s
ta
n
ce
ab
u
s
e
to
r
ed
u
ce
o
b
esit
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
Text
cla
s
s
ifica
tio
n
mo
d
el
fo
r
meth
a
mp
h
eta
min
e
-
r
ela
ted
tw
e
ets in
S
o
u
th
ea
s
t..
.
(
N
a
r
o
n
g
s
a
k
C
h
a
ya
n
g
ko
o
n
)
3623
3
.
1
.
4
.
M
et
ha
m
ph
et
a
m
ine t
w
ee
t
da
t
a
s
et
(
M
T
D)
W
e
co
llected
2
,
8
9
9
t
w
ee
t
s
f
r
o
m
o
n
lin
e
s
o
cial
m
ed
ia
r
elate
d
to
m
et
h
a
m
p
h
eta
m
i
n
e
i
n
So
u
t
h
ea
s
t
Asi
a
th
at
an
e
x
p
er
t
f
r
o
m
t
h
e
R
o
y
a
l
T
h
ai
P
o
lice
Fo
r
en
s
ics
O
f
f
ic
e
s
u
b
s
eq
u
e
n
tl
y
lab
eled
.
T
h
ese
d
ata
w
er
e
d
iv
id
ed
in
to
t
w
o
cla
s
s
e
s
:
2
,
1
7
0
in
s
tan
ce
s
o
f
n
o
n
-
ab
u
s
e
an
d
7
2
9
in
s
t
an
ce
s
o
f
ab
u
s
e,
f
o
r
a
to
tal
o
f
2
3
,
1
7
5
w
o
r
d
s
.
T
h
e
o
u
tp
u
t o
f
t
h
is
s
tep
w
as M
T
D,
w
h
o
s
e
p
r
o
p
er
ties
ar
e
s
h
o
w
n
i
n
T
a
b
le
1
.
T
ab
le
1
.
C
h
ar
ac
ter
is
tics
o
f
M
T
D
I
n
st
a
n
c
e
s
2
,
8
9
9
N
u
mb
e
r
o
f
C
l
a
sse
s
2
N
u
mb
e
r
o
f
C
l
a
ss
M
e
mb
e
r
s
N
o
n
-
A
b
u
se
A
b
u
se
2
,
1
7
0
7
2
9
T
o
t
a
l
F
e
a
t
u
r
e
s (w
o
r
d
s)
2
3
,
1
7
5
3
.
2
.
Da
t
a
prepro
ce
s
s
ing
T
h
is
p
r
o
ce
s
s
co
n
s
is
ted
o
f
co
r
p
u
s
p
r
ep
ar
atio
n
,
tex
t r
ep
r
esen
ta
tio
n
an
d
B
W
F.
3
.
2
.
1
.
Co
rpus
prepa
ra
t
io
n
C
o
r
p
u
s
p
r
ep
ar
atio
n
i
n
clu
d
ed
s
to
p
w
o
r
d
eli
m
in
at
io
n
a
n
d
s
te
m
m
in
g
.
Sto
p
w
o
r
d
eli
m
i
n
atio
n
i
n
v
o
l
v
ed
r
e
m
o
v
i
n
g
s
o
m
e
w
o
r
d
s
t
h
at
w
e
r
e
n
o
t i
m
p
o
r
tan
t
an
d
d
id
n
o
t
n
ee
d
to
b
e
f
u
r
th
er
an
al
y
ze
d
.
Sto
p
w
o
r
d
eli
m
i
n
atio
n
co
n
s
is
ted
o
f
m
a
k
i
n
g
a
ll
w
o
r
d
s
lo
w
er
ca
s
e,
cu
tti
n
g
m
ar
k
er
s
,
cu
tti
n
g
tab
s
,
cu
tt
in
g
s
t
o
p
p
o
in
ts
an
d
cu
tti
n
g
s
to
p
w
o
r
d
s
,
s
u
c
h
as
“
o
n
”,
“
i
n
”,
“
to
”
an
d
“
t
h
e”
.
Ste
m
m
i
n
g
w
as
th
e
m
o
d
if
ica
tio
n
o
f
w
o
r
d
s
t
h
at
h
ad
th
e
s
a
m
e
s
te
m
m
ea
n
in
g
b
u
t
w
er
e
w
r
i
tten
d
i
f
f
er
en
tl
y
,
s
u
c
h
as
“
ea
t”
an
d
“
ea
tin
g
”.
Ste
m
m
in
g
r
ed
u
ce
d
t
h
e
n
u
m
b
er
o
f
f
ea
t
u
r
es
o
f
th
e
m
et
h
a
m
p
h
eta
m
in
e
d
ata
s
et
[
2
7
]
.
3
.
2
.
2
.
T
ex
t
re
presenta
t
io
n
T
h
e
p
r
o
ce
s
s
o
f
tex
t
r
ep
r
ese
n
t
atio
n
w
as
a
p
ar
t
o
f
N
L
P
th
at
co
n
v
er
ted
te
x
t
to
v
ec
to
r
.
Vec
to
r
izatio
n
cr
ea
ted
a
s
et
o
f
v
ec
to
r
s
n
u
m
b
er
r
ep
r
esen
tin
g
te
x
t
t
w
ee
t
s
t
h
a
t
w
er
e
u
s
ed
to
cr
ea
te
a
te
x
t
cl
ass
i
f
icatio
n
m
o
d
el.
T
h
e
class
if
ier
co
u
ld
o
p
er
ate
o
n
th
e
tex
t
v
ec
to
r
s
.
W
e
u
s
ed
d
ata
p
r
ep
r
o
ce
s
s
in
g
tech
n
iq
u
e
s
co
n
s
is
tin
g
o
f
B
o
W
,
TF
–
I
DF
an
d
B
W
F,
u
s
i
n
g
B
o
W
,
a
p
o
p
u
lar
tex
t
v
ec
to
r
izatio
n
m
o
d
el,
as
a
b
aseli
n
e.
I
f
w
o
r
d
s
ap
p
ea
r
ed
in
th
e
t
w
ee
t
s
,
t
h
en
t
h
e
f
r
eq
u
e
n
c
y
w
a
s
co
u
n
ted
as
1
;
o
th
er
w
i
s
e,
it
w
a
s
co
u
n
ted
as
0
[
3
,
4
]
.
T
h
e
T
F
-
I
DF
alg
o
r
it
h
m
,
a
d
ata
p
r
ep
r
o
ce
s
s
in
g
tech
n
iq
u
e
th
at
r
ep
lace
d
t
h
e
te
x
t
w
it
h
w
e
ig
h
t
v
alu
e
s
,
ca
lc
u
lated
t
h
e
w
e
ig
h
t
o
f
i
m
p
o
r
ta
n
ce
th
at
w
o
r
d
s
u
s
ed
a
s
a
f
ea
t
u
r
e
f
o
r
ea
ch
t
w
ee
t.
W
e
d
eter
m
in
e
d
th
at
a
n
i
m
p
o
r
tan
t
f
ea
t
u
r
e
s
h
o
u
ld
n
o
t
ap
p
ea
r
i
n
ev
er
y
t
w
ee
t.
T
h
e
T
F
–
I
DF
m
et
h
o
d
is
w
id
el
y
u
s
ed
i
n
te
x
t
m
i
n
in
g
r
esear
ch
[
1
0
,
1
1
]
,
w
h
ile
B
W
F
r
ep
r
esen
ts
t
h
e
n
e
w
d
ata
p
r
ep
r
o
ce
s
s
in
g
tec
h
n
i
q
u
e
th
at
o
u
r
r
esear
ch
p
r
o
p
o
s
ed
.
T
h
is
alg
o
r
it
h
m
p
er
f
o
r
m
ed
th
e
d
o
m
ain
f
ea
t
u
r
es
s
elec
tio
n
o
f
t
h
e
B
o
W
d
ata
s
et.
3
.
3
.
Cla
s
s
if
ica
t
io
n
C
las
s
i
f
icatio
n
w
as
t
h
e
p
r
o
ce
s
s
o
f
cr
ea
tin
g
te
x
t
clas
s
i
f
icatio
n
m
o
d
els.
I
n
t
h
i
s
s
t
u
d
y
,
t
h
e
cla
s
s
i
f
icatio
n
alg
o
r
ith
m
s
SVM
[
1
4
]
,
J
4
8
[
1
5
]
an
d
NB
[
1
6
]
,
class
i
f
ier
s
f
o
u
n
d
i
n
t
h
e
W
ek
a
s
o
f
t
w
ar
e,
w
er
e
u
s
ed
to
cr
ea
te
t
h
e
tex
t
clas
s
i
f
icati
o
n
m
o
d
el
s
.
T
h
e
W
ek
a
v
er
s
io
n
3
.
9
p
r
o
g
r
a
m
,
w
h
ic
h
is
o
p
en
s
o
u
r
ce
an
d
w
id
el
y
u
s
ed
in
r
esear
ch
f
o
r
th
is
p
u
r
p
o
s
e,
w
as
u
s
ed
to
d
ev
elo
p
th
e
tex
t c
lass
if
ica
tio
n
m
o
d
el
s
[
2
8
,
2
9
]
.
3
.
4
.
P
er
f
o
r
m
a
nces t
esting
W
e
u
s
ed
1
0
-
f
o
ld
cr
o
s
s
-
v
a
lid
atio
n
f
o
r
th
e
m
ea
s
u
r
e
m
e
n
t o
f
T
MT
A
p
er
f
o
r
m
an
ce
u
s
i
n
g
v
ar
io
u
s
m
etr
ics
:
ac
cu
r
ac
y
[
1
7
]
,
F
-
m
ea
s
u
r
e
[
1
8
]
,
A
UC
[
1
9
]
,
Kap
p
a
[
2
0
]
,
M
C
C
[
2
1
,
2
2
]
an
d
r
u
n
ti
m
e
[
2
3
]
.
T
h
e
1
0
-
f
o
ld
cr
o
s
s
-
v
alid
atio
n
tec
h
n
iq
u
e
i
s
a
p
o
p
u
lar
m
e
th
o
d
to
o
b
tain
r
eliab
le
test
r
es
u
lts
b
ec
a
u
s
e
all
d
ata
p
o
in
ts
ar
e
u
s
ed
f
o
r
tr
ain
i
n
g
a
n
d
v
alid
atio
n
; e
ac
h
d
ata
p
o
in
t is u
s
ed
to
b
e
test
ed
ex
ac
tl
y
o
n
ce
[
2
8
]
.
3
.
5
.
H
y
po
t
hes
is
t
esting
T
h
e
W
ilco
x
o
n
R
an
k
Su
m
T
est
w
a
s
u
s
ed
to
in
v
est
ig
ate
5
d
if
f
er
e
n
t
p
er
f
o
r
m
an
ce
m
ea
s
u
r
e
m
en
t
s
(
ac
cu
r
ac
y
,
F
-
m
ea
s
u
r
e,
A
U
C
,
Kap
p
a,
MCC
)
b
et
w
ee
n
th
e
p
r
o
p
o
s
ed
an
d
ca
n
d
id
ate
m
o
d
els
to
d
eter
m
i
n
e
t
h
e
d
if
f
er
e
n
ce
s
i
n
5
p
er
f
o
r
m
an
ce
m
ea
s
u
r
e
m
e
n
t
s
at
a
s
i
g
n
i
f
ica
n
c
e
lev
el
o
f
0
.
0
5
[
3
0
,
3
1
]
.
4.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
T
h
is
s
ec
tio
n
d
escr
ib
es
a
n
d
d
is
cu
s
s
es
th
e
e
x
p
er
i
m
e
n
tal
r
esu
lt
s
.
I
t
i
n
clu
d
es
f
o
u
r
s
u
b
-
ch
ap
ter
s
,
p
r
esen
ted
ac
co
r
d
in
g
to
th
e
t
wo
o
b
j
ec
tiv
es
an
d
b
ased
o
n
th
e
ch
ar
ac
ter
is
tics
o
f
t
h
e
B
W
F
d
ataset,
in
f
o
r
m
atio
n
g
ain
,
cla
s
s
i
f
icat
io
n
p
er
f
o
r
m
a
n
ce
an
d
h
y
p
o
th
e
s
is
te
s
ti
n
g
.
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
.
11
,
No
.
4
,
A
u
g
u
s
t
2021
:
3
6
1
7
-
3628
3624
4
.
1
.
Cha
ra
ct
er
is
t
ics o
f
B
WF
da
t
a
s
et
T
h
e
f
ea
tu
r
e
r
ed
u
ctio
n
p
er
f
o
r
m
an
ce
u
s
in
g
t
h
e
B
W
F
alg
o
r
ith
m
w
as
co
m
p
ar
ed
w
i
th
t
wo
p
o
p
u
lar
tech
n
iq
u
es
:
B
o
W
an
d
T
F
–
I
D
F.
As
T
ab
le
2
s
h
o
w
s
,
th
e
B
W
F
d
ataset
h
ad
f
e
w
er
f
ea
t
u
r
es
(
9
6
9
)
an
d
in
s
tan
ce
s
(
2
,
4
4
6
)
th
an
th
e
B
o
W
an
d
th
e
T
F
-
I
DF
d
atasets
.
T
h
e
B
W
F
alg
o
r
ith
m
w
as
h
i
g
h
l
y
e
f
f
icie
n
t
a
t
f
ea
t
u
r
e
r
ed
u
ctio
n
.
T
h
e
ex
p
er
im
e
n
tal
r
es
u
lt
s
d
e
m
o
n
s
tr
ated
th
at
t
h
e
B
W
F
d
ataset
in
cl
u
d
ed
9
6
9
f
ea
tu
r
es
o
u
t
o
f
th
e
to
tal
2
3
,
1
7
5
f
ea
t
u
r
es
in
t
h
e
m
e
th
a
m
p
h
e
ta
m
in
e
t
w
ee
t
d
ataset.
T
ab
le
2
s
h
o
w
s
th
e
B
W
F
d
ataset,
w
h
ic
h
h
a
d
a
s
m
a
ller
n
u
m
b
e
r
o
f
f
ea
t
u
r
es
a
n
d
in
s
ta
n
ce
s
t
h
a
n
t
h
e
B
o
W
an
d
T
F
-
I
DF
d
atasets
.
T
h
o
s
e
f
ea
t
u
r
es
w
er
e
f
il
t
er
ed
f
ea
tu
r
es
o
f
th
e
B
o
W
d
ataset
u
s
i
n
g
S
DB
F.
T
h
er
ef
o
r
e,
th
e
B
W
F
alg
o
r
ith
m
w
as
e
f
f
ec
ti
v
e
at
h
a
n
d
lin
g
th
e
s
e
m
a
n
tic
w
o
r
d
s
ass
o
ciate
d
w
ith
m
et
h
a
m
p
h
et
a
m
i
n
e,
s
u
c
h
as
s
la
n
g
n
a
m
es
o
r
s
y
n
o
n
y
m
s
f
o
r
m
eth
a
m
p
h
eta
m
i
n
e.
T
h
is
i
m
p
le
m
en
ta
tio
n
w
a
s
d
if
f
er
en
t
f
r
o
m
t
h
e
B
o
W
,
as
th
e
latter
r
ed
u
ce
s
f
ea
t
u
r
es b
y
r
e
m
o
v
i
n
g
in
f
r
eq
u
en
t
w
o
r
d
s
.
T
ab
le
2
.
C
o
m
p
ar
is
o
n
o
f
d
ata
p
r
ep
r
o
ce
s
s
in
g
t
ec
h
n
iq
u
es
C
h
a
r
a
c
t
e
r
i
st
i
c
D
a
t
a
P
r
e
p
r
o
c
e
ssi
n
g
T
e
c
h
n
i
q
u
e
B
o
W
TF
-
I
D
F
B
W
F
N
u
mb
e
r
o
f
F
e
a
t
u
r
e
s
1
0
,
9
2
6
1
0
,
4
6
4
9
6
9
N
u
mb
e
r
o
f
I
n
st
a
n
c
e
s
2
,
8
9
9
2
,
8
9
9
2
,
4
4
6
4
.
2
.
I
nfo
r
m
a
t
io
n g
a
in
I
n
f
o
r
m
a
tio
n
g
a
in
w
a
s
ap
p
lied
to
m
ea
s
u
r
e
t
h
e
q
u
alit
y
o
f
t
h
e
f
ea
t
u
r
es
u
s
ed
to
cr
ea
te
a
D
ec
i
s
io
n
T
r
ee
.
T
h
e
in
f
o
r
m
atio
n
g
a
in
test
s
f
o
r
th
e
B
W
F d
atase
t id
en
ti
f
ied
s
e
v
er
al
i
m
p
o
r
tan
t
f
ea
t
u
r
es,
i
n
cl
u
d
in
g
“m
e
th
”,
“lab
”,
“
cr
y
s
tal”,
“
ice”
,
“
s
m
o
k
e”
,
“
p
o
lice”
,
“
n
e
w
s
”,
“
r
ep
o
r
t”,
“
s
e
x
y
”
an
d
“f
at”.
T
h
e
w
o
r
d
s
“m
e
th
”
a
n
d
“
lab
”
w
er
e
i
m
p
o
r
tan
t
f
ea
t
u
r
es
i
n
t
h
e
B
W
F
d
ataset
as
t
h
e
y
w
er
e
u
s
ed
in
t
w
ee
t
s
th
a
t
m
en
tio
n
ed
lab
o
r
ato
r
y
-
p
r
o
d
u
ce
d
m
et
h
a
m
p
h
eta
m
i
n
e.
T
h
e
w
o
r
d
s
“
cr
y
s
tal”
an
d
“ice
”
ar
e
s
lan
g
n
a
m
e
s
f
o
r
m
et
h
a
m
p
h
eta
m
in
e;
b
o
th
h
ad
h
ig
h
in
f
o
r
m
atio
n
g
ai
n
,
i
n
d
icati
n
g
t
h
e
f
ea
tu
r
e
s
’
p
o
te
n
tial
f
o
r
th
e
p
r
ed
ictio
n
class
es
u
s
i
n
g
t
h
e
D
ec
is
io
n
T
r
ee
.
T
en
i
m
p
o
r
tan
t
f
ea
tu
r
e
s
ar
e
s
h
o
w
n
i
n
T
ab
le
3
.
T
ab
le
3
s
h
o
w
s
t
h
e
ex
p
er
i
m
e
n
tal
r
esu
lts
o
f
t
h
e
in
f
o
r
m
at
io
n
g
ai
n
th
at
w
a
s
u
s
ed
to
test
th
e
f
ea
t
u
r
e
q
u
alit
y
o
f
t
h
e
B
W
F
d
ataset.
H
ig
h
in
f
o
r
m
at
io
n
g
ain
i
n
d
icate
d
th
e
i
m
p
o
r
ta
n
t
f
ea
tu
r
es
f
o
r
th
e
p
r
ed
ictio
n
class
es
b
ased
o
n
th
e
D
ec
is
io
n
T
r
ee
.
T
h
o
s
e
f
ea
t
u
r
es
h
ad
s
tr
o
n
g
p
o
w
e
r
in
class
i
f
y
i
n
g
th
e
clas
s
es
b
as
ed
o
n
th
e
D
ec
is
io
n
T
r
ee
.
I
n
f
o
r
m
atio
n
g
a
in
s
h
o
w
e
d
i
m
p
o
r
tan
t
f
ea
t
u
r
es
s
u
c
h
as
“n
e
w
s
”,
“
p
o
lice”
a
n
d
“
r
ep
o
r
t”
in
t
h
e
n
o
n
-
ab
u
s
e
class
t
w
ee
t
s
; i
n
co
n
tr
ast,
“f
at”
an
d
“sex
y
”
w
er
e
f
ea
t
u
r
es o
f
t
h
e
ab
u
s
e
class
t
w
ee
ts
.
T
ab
le
3
.
I
m
p
o
r
tan
t f
ea
t
u
r
es o
f
B
W
F d
ataset
u
s
i
n
g
i
n
f
o
r
m
atio
n
g
ai
n
R
a
n
k
e
d
F
e
a
t
u
r
e
I
n
f
o
r
mat
i
o
n
G
a
i
n
(
d
e
sce
n
d
i
n
g
o
r
d
e
r
)
me
t
h
0
.
0
9
5
1
4
l
a
b
0
.
0
3
7
0
1
c
r
y
st
a
l
0
.
0
3
5
1
1
i
c
e
0
.
0
2
5
4
2
smo
k
e
0
.
0
2
3
2
9
p
o
l
i
c
e
0
.
0
2
1
4
3
n
e
w
s
0
.
0
1
6
4
0
r
e
p
o
r
t
0
.
0
1
0
6
5
se
x
y
0
.
0
1
0
4
8
f
a
t
0
.
0
0
5
5
5
4
.
3
.
Cla
s
s
if
ica
t
io
n per
f
o
r
m
a
nce
T
h
e
class
if
icatio
n
p
er
f
o
r
m
a
n
c
e
co
m
p
ar
is
o
n
o
f
t
h
e
th
r
ee
p
r
ep
r
o
ce
s
s
in
g
tec
h
n
iq
u
e
s
u
s
ed
to
p
r
o
d
u
ce
B
o
W
,
T
F
–
I
DF
an
d
B
W
F
d
a
tasets
ar
e
s
h
o
w
n
i
n
T
ab
le
s
4
,
5
an
d
6
.
First,
th
e
p
er
f
o
r
m
an
ce
o
f
t
h
e
SV
M
class
i
f
ier
w
it
h
t
h
e
B
o
W
d
ata
s
et
h
ad
t
h
e
h
i
g
h
est
ac
c
u
r
ac
y
(
0
.
8
1
3
)
,
F
-
m
ea
s
u
r
e
(
0
.
8
0
3
)
an
d
MC
C
(
0
.
4
6
5
)
.
Ho
w
e
v
er
,
th
i
s
clas
s
i
f
ier
u
s
ed
w
i
th
t
h
e
B
W
F
d
ataset
h
ad
th
e
h
ig
h
e
s
t
A
U
C
(
0
.
7
2
0
)
a
n
d
Kap
p
a
(
0
.
4
6
1
)
.
Mo
r
eo
v
er
,
th
e
B
W
F d
ataset
h
ad
th
e
lo
w
e
s
t r
u
n
ti
m
e
(
0
.
8
2
0
s
ec
o
n
d
s
)
w
ith
th
e
SVM
clas
s
i
f
i
er
.
T
ab
le
4
d
is
p
lay
s
th
e
clas
s
i
f
icatio
n
p
er
f
o
r
m
an
ce
co
m
p
ar
is
o
n
s
o
f
t
h
e
t
h
r
ee
p
r
ep
r
o
ce
s
s
in
g
tech
n
iq
u
es c
o
m
b
i
n
e
d
w
it
h
SVM.
T
h
e
d
ec
is
io
n
t
r
ee
u
s
in
g
th
e
J
4
8
class
if
ier
w
it
h
th
e
B
W
F
d
ataset
h
ad
th
e
h
ig
h
es
t
s
co
r
es
in
all
m
ea
s
u
r
es,
i
n
clu
d
i
n
g
ac
cu
r
ac
y
(
0
.
8
1
5
)
,
F
-
m
ea
s
u
r
e
(
0
.
8
1
8
)
,
A
U
C
(
0
.
7
6
3
)
,
Ka
p
p
a
(
0
.
5
2
8
)
an
d
MCC
(
0
.
5
2
9
)
,
an
d
th
e
lo
w
e
s
t
r
u
n
t
i
m
e
(
3
.
4
8
0
s
ec
o
n
d
s
)
.
T
ab
le
5
p
r
esen
ts
th
e
class
if
icatio
n
p
er
f
o
r
m
a
n
ce
co
m
p
ar
is
o
n
s
f
o
r
th
i
s
class
i
f
ier
.
T
h
e
NB
class
if
ier
w
it
h
t
h
e
B
o
W
d
ataset
h
ad
th
e
h
ig
h
est
ac
c
u
r
ac
y
(
0
.
7
9
5
)
,
F
-
m
ea
s
u
r
e
(
0
.
7
8
9
)
,
Kap
p
a
(
0
.
4
2
8
)
an
d
MCC
(
0
.
4
3
0
)
.
Ho
w
e
v
er
,
w
h
en
co
m
b
i
n
e
d
w
i
th
th
e
B
W
F
d
ataset,
t
h
i
s
class
if
ier
h
ad
t
h
e
h
ig
h
e
s
t
A
U
C
(
0
.
8
1
9
)
an
d
th
e
lo
w
e
s
t
r
u
n
ti
m
e
(
0
.
4
0
0
s
ec
o
n
d
s
)
.
T
h
e
class
i
f
icatio
n
p
er
f
o
r
m
an
ce
co
m
p
ar
is
o
n
s
ar
e
s
h
o
w
n
i
n
T
ab
le
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
Text
cla
s
s
ifica
tio
n
mo
d
el
fo
r
meth
a
mp
h
eta
min
e
-
r
ela
ted
tw
e
ets in
S
o
u
th
ea
s
t..
.
(
N
a
r
o
n
g
s
a
k
C
h
a
ya
n
g
ko
o
n
)
3625
T
ab
le
4
.
C
lass
if
icatio
n
p
er
f
o
r
m
an
ce
c
o
m
p
ar
i
s
o
n
u
s
i
n
g
SV
M
M
e
a
su
r
e
me
n
t
S
V
M
C
l
a
ss
i
f
i
e
r
B
o
W
D
a
t
a
se
t
(
b
a
se
l
i
n
e
)
TF
-
I
D
F
D
a
t
a
se
t
B
W
F
D
a
t
a
se
t
A
c
c
u
r
a
c
y
0
.
8
1
3
0
.
8
1
2
0
.
8
0
5
F
-
m
e
a
su
r
e
0
.
8
0
3
0
.
7
9
4
0
.
8
0
0
A
U
C
0
.
7
0
8
0
.
6
8
4
0
.
7
2
0
K
a
p
p
a
0
.
4
5
6
0
.
4
2
4
0
.
4
6
1
M
C
C
0
.
4
6
5
0
.
4
4
6
0
.
4
6
3
R
u
n
t
i
me
(
se
c
o
n
d
s)
3
3
.
3
1
0
1
1
.
5
8
0
0
.
8
2
0
B
ased
o
n
th
e
cla
s
s
i
f
icatio
n
p
e
r
f
o
r
m
an
ce
co
m
p
ar
is
o
n
s
i
n
T
ab
les
4
,
5
an
d
6
,
th
e
p
r
o
p
o
s
ed
m
o
d
el
th
at
co
m
b
i
n
ed
t
h
e
J
4
8
class
i
f
ier
with
t
h
e
B
W
F
d
ataset
s
h
o
w
ed
t
h
e
b
est
p
er
f
o
r
m
a
n
ce
f
o
r
th
e
T
MT
A
b
ased
o
n
th
e
f
o
u
r
m
ea
s
u
r
es
o
f
ac
cu
r
ac
y
,
F
-
m
ea
s
u
r
e,
Kap
p
a
an
d
MCC
.
I
n
co
m
p
ar
is
o
n
,
th
e
SV
M
class
if
ier
w
it
h
t
h
e
B
W
F
d
ataset
w
a
s
th
e
b
es
t b
ased
o
n
r
u
n
t
i
m
e,
a
n
d
th
e
NB
clas
s
if
ier
w
it
h
t
h
e
B
W
F d
ataset
p
r
o
v
id
ed
th
e
h
i
g
h
e
s
t
A
U
C
.
T
ab
le
5
.
C
lass
if
icatio
n
p
er
f
o
r
m
an
ce
c
o
m
p
ar
i
s
o
n
u
s
i
n
g
J
4
8
M
e
a
su
r
e
me
n
t
J4
8
C
l
a
ss
i
f
i
e
r
B
o
W
D
a
t
a
se
t
(
b
a
se
l
i
n
e
)
TF
-
I
D
F
D
a
t
a
se
t
B
W
F
D
a
t
a
se
t
A
c
c
u
r
a
c
y
0
.
8
0
7
0
.
8
0
7
0
.
8
1
5
F
-
m
e
a
su
r
e
0
.
8
0
4
0
.
8
0
5
0
.
8
1
8
A
U
C
0
.
7
2
3
0
.
7
3
5
0
.
7
6
3
K
a
p
p
a
0
.
4
7
4
0
.
4
7
4
0
.
5
2
8
M
C
C
0
.
4
7
5
0
.
4
7
5
0
.
5
2
9
R
u
n
t
i
me
(
se
c
o
n
d
s)
6
1
.
0
6
0
6
2
.
5
5
0
3
.
4
8
0
T
h
e
r
esu
lt
s
f
r
o
m
T
ab
les
4
,
5
an
d
6
co
m
p
ar
e
t
h
e
p
er
f
o
r
m
an
ce
m
ea
s
u
r
e
m
e
n
ts
f
o
r
S
VM
,
J
4
8
an
d
NB
,
r
ev
ea
lin
g
t
h
at
th
e
m
o
d
el
b
u
ilt
o
n
th
e
J
4
8
class
if
ier
an
d
u
s
i
n
g
t
h
e
B
W
F
d
ataset
w
a
s
th
e
b
est.
I
n
s
h
o
r
t,
th
i
s
m
o
d
el
p
r
o
v
id
ed
t
h
e
b
est
p
er
f
o
r
m
a
n
ce
m
ea
s
u
r
e
m
e
n
t
s
(
ac
cu
r
ac
y
,
F
-
m
ea
s
u
r
e,
Kap
p
a,
MCC
)
.
T
h
e
h
i
g
h
es
t
ac
cu
r
ac
y
w
as
s
h
o
w
n
i
n
ter
m
s
o
f
th
e
co
r
r
ec
tn
ess
o
f
t
h
e
d
ata
class
i
f
icatio
n
u
s
in
g
th
i
s
m
o
d
el
.
T
h
e
B
W
F
d
ataset
in
cl
u
d
ed
1
,
8
2
7
in
s
tan
ce
s
o
f
n
o
n
-
ab
u
s
e
t
w
ee
ts
an
d
6
1
9
i
n
s
ta
n
ce
s
o
f
ab
u
s
e
t
w
ee
ts
.
T
h
is
m
o
d
el
co
u
ld
b
e
p
r
ed
icted
to
c
o
r
r
ec
t
1
,
5
6
5
n
o
n
-
ab
u
s
e
t
w
ee
ts
a
n
d
4
2
8
ab
u
s
e
t
w
ee
t
s
.
A
d
d
itio
n
all
y
,
th
is
m
o
d
el
p
r
o
v
id
ed
t
h
e
h
ig
h
e
s
t
F
-
m
ea
s
u
r
e
v
al
u
e
s
.
T
h
is
r
es
u
lt
s
h
o
w
ed
t
h
at
th
e
m
o
d
el
d
em
o
n
s
tr
ated
ac
cu
r
ate
c
lass
i
f
icatio
n
o
f
th
e
in
ter
est cla
s
s
,
w
h
ich
w
a
s
t
h
e
ab
u
s
e
t
w
ee
t
s
.
T
ab
le
6
.
C
lass
if
icatio
n
p
er
f
o
r
m
an
ce
c
o
m
p
ar
i
s
o
n
u
s
i
n
g
NB
M
e
a
su
r
e
me
n
t
N
B
C
l
a
ss
i
f
i
e
r
B
o
W
D
a
t
a
se
t
(
b
a
se
l
i
n
e
)
TF
-
I
D
F
D
a
t
a
se
t
B
W
F
D
a
t
a
se
t
A
c
c
u
r
a
c
y
0
.
7
9
5
0
.
4
9
0
0
.
7
9
4
F
-
m
e
a
su
r
e
0
.
7
8
9
0
.
4
9
5
0
.
7
8
5
A
U
C
0
.
7
9
7
0
.
7
6
2
0
.
8
1
9
K
a
p
p
a
0
.
4
2
8
0
.
1
6
5
0
.
4
1
4
M
C
C
0
.
4
3
0
0
.
2
6
0
0
.
4
1
9
R
u
n
t
i
me
(
se
c
o
n
d
s)
6
.
1
9
0
7
.
8
7
0
0
.
4
0
0
T
h
e
A
U
C
v
al
u
es
o
f
J
4
8
w
it
h
th
e
B
W
F
d
ataset
w
er
e
clo
s
e
t
o
1
as
s
h
o
w
n
in
T
ab
le
5
,
in
d
icatin
g
th
a
t
th
e
clas
s
if
icatio
n
r
es
u
lts
o
f
J
4
8
w
it
h
th
e
B
W
F
d
ataset
h
ad
h
ig
h
tr
u
e
p
o
s
iti
v
e
v
a
lu
e
s
.
T
h
e
f
in
d
i
n
g
s
r
ev
ea
led
th
at
J
4
8
w
it
h
t
h
e
B
W
F
d
ata
s
et
h
ig
h
l
y
c
lass
if
ied
th
e
ab
u
s
e
clas
s
(
h
er
e,
an
in
v
ita
tio
n
t
w
ee
t
to
co
n
s
u
m
e
m
et
h
a
m
p
h
eta
m
i
n
e)
.
T
a
b
le
5
s
h
o
w
s
t
h
e
m
o
d
el
g
e
n
er
ated
u
s
in
g
J
4
8
w
it
h
th
e
B
W
F
d
ataset,
w
h
ic
h
h
ad
th
e
h
ig
h
e
s
t
Kap
p
a
a
n
d
MC
C
v
al
u
es,
s
u
g
g
es
tin
g
h
i
g
h
co
n
s
is
te
n
c
y
i
n
cla
s
s
i
f
icat
io
n
b
et
w
ee
n
t
h
e
t
w
o
clas
s
es
(
ab
u
s
e
o
r
n
o
n
-
ab
u
s
e)
.
T
h
e
B
W
F
d
ataset
w
a
s
f
itted
to
th
e
J
4
8
class
i
f
ier
b
ec
au
s
e
th
e
f
ea
tu
r
e
s
i
n
t
h
e
B
W
F
d
ataset
w
er
e
s
i
m
ilar
to
th
e
k
e
y
w
o
r
d
“m
et
h
a
m
p
h
eta
m
i
n
e”
.
T
ab
le
3
s
h
o
w
s
t
h
e
f
ea
tu
r
es
th
at
h
ad
h
ig
h
i
n
f
o
r
m
atio
n
g
ai
n
.
T
h
er
ef
o
r
e,
th
o
s
e
f
ea
t
u
r
es
w
er
e
u
s
ed
as
a
co
n
d
itio
n
f
o
r
clas
s
i
f
icatio
n
b
ased
o
n
t
h
e
D
ec
is
io
n
T
r
ee
,
an
d
th
en
t
h
e
J
4
8
class
if
ier
w
a
s
u
s
ed
as a
s
u
b
s
et
o
f
th
e
D
ec
i
s
io
n
T
r
ee
.
4
.
4
.
H
y
po
t
hes
is
t
esting
As
d
ep
icted
in
T
ab
le
7
th
e
W
ilco
x
o
n
r
an
k
s
u
m
te
s
t
r
es
u
lts
s
u
g
g
es
ted
th
at
t
h
e
p
r
o
p
o
s
ed
m
o
d
el
b
ase
d
o
n
th
e
J
4
8
class
i
f
ier
u
s
in
g
t
h
e
B
W
F
d
ataset
w
as
t
h
e
b
est.
T
h
is
m
o
d
e
l
w
as
p
r
esen
ted
as
T
M
T
A
b
ec
au
s
e
t
h
e
f
i
v
e
p
er
f
o
r
m
a
n
ce
m
ea
s
u
r
e
m
e
n
ts
(
ac
cu
r
ac
y
,
F
-
m
ea
s
u
r
e,
AUC,
Kap
p
a,
MC
C
)
w
er
e
s
i
g
n
if
ica
n
tl
y
h
i
g
h
er
th
a
n
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
.
11
,
No
.
4
,
A
u
g
u
s
t
2021
:
3
6
1
7
-
3628
3626
f
o
r
th
e
s
i
x
-
ca
n
d
id
ate
m
o
d
el
s
w
ith
a
P
-
Val
u
e
o
f
0
.
0
4
3
.
Ho
w
e
v
er
,
J
4
8
w
it
h
t
h
e
B
W
F
d
ataset
y
ield
ed
p
er
f
o
r
m
a
n
ce
m
ea
s
u
r
e
m
e
n
ts
t
h
at
w
er
e
n
o
t
s
i
g
n
if
ica
n
tl
y
h
ig
h
e
r
th
an
NB
u
s
i
n
g
th
e
B
OW
an
d
B
W
F
d
ataset
w
it
h
a
P
-
Valu
e
o
f
0
.
2
2
5
.
T
h
e
W
ilco
x
o
n
r
an
k
s
u
m
te
s
t
r
es
u
lt
s
f
o
r
th
e
p
er
f
o
r
m
a
n
ce
m
ea
s
u
r
em
en
ts
ar
e
s
h
o
w
n
i
n
T
ab
le
7
.
T
ab
le
7
.
W
ilco
x
o
n
r
an
k
s
u
m
t
est f
o
r
p
er
f
o
r
m
an
ce
m
ea
s
u
r
e
m
en
ts
P
r
o
p
o
se
d
M
o
d
e
l
C
a
n
d
i
d
a
t
e
M
o
d
e
l
P
-
V
a
l
u
e
J4
8
w
i
t
h
B
W
F
S
V
M
w
i
t
h
T
F
-
I
D
F
0
.
0
4
3
S
V
M
w
i
t
h
B
O
W
S
V
M
w
i
t
h
B
W
F
J4
8
w
i
t
h
T
F
-
I
D
F
J4
8
w
i
t
h
B
O
W
N
B
w
i
t
h
T
F
-
I
D
F
N
B
w
i
t
h
B
O
W
0
.
2
2
5
N
B
w
i
t
h
B
W
F
T
ab
le
7
s
h
o
w
s
t
h
e
r
es
u
lt
s
o
f
t
h
e
W
ilco
x
o
n
r
an
k
s
u
m
test
,
wh
ich
w
as
tes
ted
at
a
s
i
g
n
if
ica
n
ce
lev
el
o
f
0
.
0
5
.
T
h
e
m
ea
s
u
r
ed
v
al
u
es
f
o
r
th
e
ac
cu
r
ac
y
,
F
-
m
ea
s
u
r
e,
A
U
C
,
Kap
p
a
an
d
MCC
o
f
th
e
p
r
o
p
o
s
ed
m
o
d
el
w
er
e
co
m
p
ar
ed
w
it
h
th
e
ei
g
h
t
ca
n
d
id
ate
m
o
d
els.
T
h
e
ex
p
er
im
en
tal
r
es
u
lts
s
u
g
g
ested
t
h
e
s
e
f
i
v
e
p
er
f
o
r
m
a
n
ce
m
ea
s
u
r
e
m
e
n
t
s
o
f
th
e
p
r
o
p
o
s
ed
m
o
d
el
w
er
e
b
etter
th
an
f
o
r
th
e
s
ix
ca
n
d
id
ate
m
o
d
els
at
a
s
ig
n
if
ican
ce
le
v
el
o
f
0
.
0
5
w
it
h
a
s
tati
s
tical
co
n
f
id
e
n
ce
lev
el
o
f
9
5
p
er
ce
n
t.
T
h
er
ef
o
r
e,
th
e
J
4
8
class
if
ier
u
s
in
g
t
h
e
B
W
F
d
ataset
w
as
u
s
e
d
in
d
ev
elo
p
in
g
t
h
e
T
MT
A
b
e
ca
u
s
e
t
h
i
s
m
o
d
el
p
r
o
v
i
d
ed
th
e
h
ig
h
es
t
f
o
u
r
p
er
f
o
r
m
a
n
ce
m
ea
s
u
r
e
m
en
ts
(
ac
cu
r
ac
y
,
F
-
m
ea
s
u
r
e,
Kap
p
a
an
d
M
C
C
)
a
n
d
p
r
o
v
id
ed
a
lo
w
r
u
n
t
i
m
e
as
s
h
o
w
n
i
n
T
ab
le
5
.
Fu
r
t
h
er
m
o
r
e,
th
i
s
m
o
d
el
p
r
o
v
id
ed
s
ig
n
i
f
ica
n
tl
y
h
ig
h
e
r
p
er
f
o
r
m
a
n
ce
m
ea
s
u
r
e
m
e
n
t
s
th
an
th
e
s
ix
-
ca
n
d
id
ate
m
o
d
els
as
s
h
o
w
n
i
n
T
ab
l
e
7
.
P
r
ev
io
u
s
r
esear
ch
cr
ea
ted
te
x
t
class
i
f
icat
io
n
m
o
d
el
s
u
s
i
n
g
t
w
ee
t
d
ata
b
ased
o
n
SVM,
J
4
8
an
d
NB
class
i
f
ier
s
.
A
lt
h
o
u
g
h
SVM
w
it
h
T
F
–
I
DF
is
s
till
w
id
el
y
u
s
ed
f
o
r
th
e
d
ev
elo
p
m
e
n
t
o
f
te
x
t
cl
ass
i
f
icatio
n
m
o
d
els
[6
-
9
]
,
w
e
f
o
u
n
d
t
h
at
t
h
e
T
MT
A
,
u
s
i
n
g
J
4
8
w
ith
t
h
e
B
W
F
d
ataset,
p
r
o
v
id
ed
h
ig
h
er
v
al
u
es
f
o
r
p
er
f
o
r
m
an
ce
m
ea
s
u
r
e
m
e
n
t
s
th
a
n
SVM
w
it
h
T
F
–
I
DF.
I
n
p
a
r
ticu
lar
,
th
e
T
MT
A
u
s
i
n
g
J
4
8
w
i
th
t
h
e
B
W
F
d
ataset
h
ad
a
lo
w
er
r
u
n
ti
m
e
t
h
a
n
s
u
c
h
w
id
el
y
u
s
ed
tech
n
iq
u
es a
s
B
o
W
an
d
T
F
–
I
DF.
5.
CO
NCLU
SI
O
N
W
e
p
r
o
p
o
s
ed
a
n
e
w
m
o
d
el,
ca
lled
th
e
T
M
T
A
,
to
id
en
tify
w
h
et
h
er
a
T
w
itter
t
w
ee
t
w
as
r
elate
d
to
m
et
h
a
m
p
h
eta
m
i
n
e
u
s
e
o
r
ab
u
s
e
b
ased
o
n
d
ata
ex
tr
ac
ted
f
r
o
m
T
w
itter
i
n
So
u
t
h
ea
s
t
Asi
a.
A
v
ital
p
r
o
ce
s
s
i
n
t
h
e
T
M
T
A
is
d
ata
p
r
ep
r
o
ce
s
s
in
g
.
T
h
is
r
esear
ch
ad
d
r
ess
ed
th
e
w
ea
k
n
ess
o
f
B
o
W
in
ter
m
s
o
f
f
ea
tu
r
e
s
e
lectio
n
u
s
i
n
g
t
h
e
B
o
W
d
ataset
an
d
W
o
r
d
2
Vec
.
A
n
o
v
el
d
ata
p
r
ep
r
o
ce
s
s
i
n
g
tech
n
iq
u
e,
th
e
B
W
F
a
lg
o
r
ith
m
,
u
s
ed
t
h
e
tex
t
v
ec
to
r
izatio
n
m
et
h
o
d
in
th
e
s
a
m
e
w
a
y
a
s
th
e
B
o
W
d
at
aset;
h
o
w
ev
er
,
th
e
p
r
o
p
o
s
ed
B
W
F
alg
o
r
ith
m
w
as
ap
p
lied
u
s
in
g
t
h
e
f
ea
t
u
r
e
s
ele
c
tio
n
o
f
th
e
B
o
W
d
ataset
to
p
r
o
d
u
ce
a
B
W
F
d
ataset.
T
h
is
ap
p
r
o
ac
h
r
esu
lted
in
a
s
m
al
ler
n
u
m
b
er
o
f
f
ea
t
u
r
es
t
h
an
s
u
ch
w
id
el
y
u
s
ed
tec
h
n
i
q
u
es
as
B
o
W
an
d
th
e
T
F
-
I
D
F
d
atasets
.
T
h
e
n
e
w
d
ataset
w
as
u
s
ed
f
o
r
th
e
T
MT
A
d
ataset.
T
h
e
d
ev
e
lo
p
m
en
t
o
f
t
h
e
T
MT
A
co
n
s
is
ted
o
f
f
o
u
r
s
tep
s
.
First,
w
e
co
llected
d
ata
w
it
h
k
e
y
w
o
r
d
s
r
elate
d
to
m
et
h
a
m
p
h
eta
m
i
n
e
f
r
o
m
t
h
e
T
w
itter
d
ata
s
tr
ea
m
.
Seco
n
d
,
d
ata
p
r
ep
r
o
ce
s
s
in
g
tec
h
n
iq
u
es
w
er
e
ap
p
lied
,
in
clu
d
in
g
co
r
p
u
s
p
r
ep
ar
atio
n
an
d
tex
t
r
ep
r
esen
t
atio
n
co
n
s
is
ti
n
g
o
f
B
o
W
,
T
F
-
I
DF
an
d
B
W
F.
T
h
ir
d
,
w
e
e
x
p
er
i
m
e
n
ted
an
d
p
r
o
p
o
s
ed
a
tex
t
class
if
icatio
n
m
o
d
el
u
s
i
n
g
t
h
r
ee
ca
n
d
id
ate
clas
s
i
f
ier
s
:
SVM,
J
4
8
an
d
NB
.
L
astl
y
,
w
e
co
m
p
ar
ed
t
h
e
p
er
f
o
r
m
an
ce
o
f
t
h
e
v
ar
io
u
s
te
x
t
class
i
f
icatio
n
m
o
d
els
t
h
at
wer
e
cr
ea
ted
f
r
o
m
th
e
ab
o
v
e
th
r
ee
class
i
f
ier
s
u
s
in
g
th
r
ee
d
ata
p
r
ep
r
o
ce
s
s
in
g
tech
n
iq
u
es.
T
h
e
p
er
f
o
r
m
an
ce
m
ea
s
u
r
e
m
e
n
t
s
in
cl
u
d
ed
ac
cu
r
ac
y
,
F
-
m
ea
s
u
r
e,
A
UC
,
Kap
p
a,
MCC
an
d
r
u
n
ti
m
e.
A
d
d
itio
n
al
l
y
,
t
h
e
T
MT
A
m
o
d
el
d
ev
elo
p
m
en
t
u
s
ed
t
h
e
J
4
8
class
i
f
ier
w
it
h
t
h
e
B
W
F
d
ataset.
T
h
is
m
o
d
el
p
r
o
d
u
ce
d
th
e
h
ig
h
e
s
t
v
a
lu
e
s
f
o
r
ac
cu
r
ac
y
(
0
.
8
1
5
)
,
F
-
m
ea
s
u
r
e
(
0
.
8
1
8
)
,
Kap
p
a
(
0
.
5
2
8
)
an
d
MCC
(
0
.
5
2
9
)
,
h
ig
h
A
U
C
(
0
.
7
6
3
)
an
d
lo
w
r
u
n
ti
m
e
(
3
.
4
8
0
s
ec
o
n
d
s
)
u
s
in
g
t
h
e
J
4
8
class
if
ier
.
T
h
ese
r
esu
lt
s
s
h
o
w
ed
t
h
at
t
h
e
p
r
o
p
o
s
ed
T
M
T
A
w
a
s
f
i
tted
to
th
e
T
w
itter
d
ataset
co
llected
in
th
i
s
s
t
u
d
y
.
T
h
e
T
M
T
A
u
s
i
n
g
J
4
8
w
i
th
t
h
e
B
W
F
d
ataset
p
r
o
v
id
ed
h
ig
h
er
p
er
f
o
r
m
an
ce
m
ea
s
u
r
e
m
en
ts
t
h
an
s
u
ch
tr
ad
itio
n
al
tec
h
n
iq
u
es
a
s
SVM
w
it
h
T
F
–
I
DF.
C
o
n
s
eq
u
en
tl
y
,
t
h
e
T
MT
A
u
s
i
n
g
t
h
e
J
4
8
class
if
ier
co
u
ld
b
e
co
n
v
er
ted
to
an
if
-
t
h
en
r
u
le
-
b
ased
d
ec
is
io
n
tr
ee
.
T
h
is
r
u
le
m
ig
h
t
b
e
i
m
p
le
m
e
n
ted
f
o
r
p
r
o
to
t
y
p
e
s
o
f
t
w
ar
e
t
o
h
elp
th
e
p
o
lice
o
f
t
h
e
n
ar
c
o
tics
co
n
tr
o
l
b
o
ar
d
id
en
ti
f
y
s
h
o
r
t
m
e
s
s
a
g
es r
elate
d
to
d
r
u
g
ab
u
s
e.
T
h
e
B
W
F
alg
o
r
ith
m
ca
n
b
e
u
s
ed
f
o
r
d
ata
p
r
ep
ar
atio
n
s
te
m
m
in
g
f
r
o
m
t
h
e
d
ev
e
lo
p
m
en
t
o
f
a
tex
t
class
i
f
icatio
n
m
o
d
el
b
ased
o
n
a
d
if
f
er
e
n
t
d
o
m
ai
n
,
s
u
c
h
as
a
m
p
h
eta
m
i
n
e
u
s
e
in
T
h
ailan
d
o
r
illeg
al
ad
v
er
tis
e
m
en
t
s
f
o
r
n
u
tr
itio
n
al
s
u
p
p
le
m
e
n
t
s
.
P
o
lice
h
a
v
e
f
o
u
n
d
ten
s
o
f
t
h
o
u
s
an
d
s
o
f
a
m
p
h
e
t
a
m
i
n
e
n
et
w
o
r
k
s
o
n
Evaluation Warning : The document was created with Spire.PDF for Python.