T
E
L
K
O
M
NIKA
T
elec
o
mm
un
ica
t
io
n Co
m
pu
t
i
ng
E
lect
ro
nics
a
nd
Co
ntr
o
l
Vo
l.
2
3
,
No
.
2
,
A
p
r
il
2
0
2
5
,
p
p
.
416
~
425
I
SS
N:
1
6
9
3
-
6
9
3
0
,
DOI
: 1
0
.
1
2
9
2
8
/
T
E
L
KOM
NI
K
A
.
v
2
3
i
2
.
26223
416
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//telko
mn
ika
.
u
a
d
.
a
c.
i
d
Co
m
pa
riso
n of
w
o
rd e
m
bedding
fe
a
ture
s
using
d
e
ep
l
ea
rning
in
s
enti
m
en
t
a
na
ly
sis
J
a
s
m
ir
1
,
E
rr
is
s
y
a
Ra
s
y
w
ir
2
,
H
er
t
i Y
a
ni
3
,
Ag
us
Nug
ro
ho
2
1
D
e
p
a
r
t
e
me
n
t
o
f
C
o
mp
u
t
e
r
E
n
g
i
n
e
e
r
i
n
g
,
F
a
c
u
l
t
y
o
f
C
o
mp
u
t
e
r
S
c
i
e
n
c
e
,
U
n
i
v
e
r
si
t
a
s Di
n
a
m
i
k
a
B
a
n
g
sa,
Jam
b
i
,
I
n
d
o
n
e
si
a
2
D
e
p
a
r
t
me
n
t
o
f
I
n
f
o
r
mat
i
c
En
g
i
n
e
e
r
i
n
g
,
F
a
c
u
l
t
y
o
f
C
o
mp
u
t
e
r
S
c
i
e
n
c
e
,
U
n
i
v
e
r
si
t
a
s Di
n
a
mi
k
a
B
a
n
g
sa
,
Jam
b
i
,
I
n
d
o
n
e
si
a
3
D
e
p
a
r
t
me
n
t
o
f
I
n
f
o
r
mat
i
o
n
S
y
st
e
m
,
F
a
c
u
l
t
y
o
f
C
o
mp
u
t
e
r
S
c
i
e
n
c
e
,
U
n
i
v
e
r
si
t
a
s
D
i
n
a
m
i
k
a
B
a
n
g
sa
,
Jam
b
i
,
I
n
d
o
n
e
si
a
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Ma
r
2
4
,
2
0
2
4
R
ev
i
s
ed
Dec
4
,
2
0
2
4
A
cc
ep
ted
J
an
2
2
,
2
0
2
5
In
t
h
is
re
se
a
rc
h
,
w
e
u
se
se
v
e
r
a
l
d
e
e
p
lea
rn
in
g
m
e
th
o
d
s
w
it
h
th
e
w
o
r
d
e
m
b
e
d
d
in
g
f
e
a
tu
re
t
o
se
e
t
h
e
ir
e
ff
e
c
t
o
n
i
n
c
re
a
sin
g
th
e
e
v
a
lu
a
ti
o
n
v
a
lu
e
o
f
c
las
si
f
ica
ti
o
n
p
e
rf
o
rm
a
n
c
e
f
ro
m
p
ro
c
e
ss
in
g
se
n
ti
m
e
n
t
a
n
a
ly
sis
d
a
ta.
T
h
e
d
e
e
p
lea
rn
in
g
m
e
th
o
d
s
u
se
d
a
re
c
o
n
d
it
i
o
n
a
l
ra
n
d
o
m
f
ield
(CRF
),
b
i
d
irec
ti
o
n
a
l
l
o
n
g
sh
o
rt
term
m
e
m
o
r
y
(B
L
S
T
M
)
a
n
d
c
o
n
v
o
l
u
ti
o
n
a
l
n
e
u
ra
l
n
e
tw
o
rk
(
CNN
).
Ou
r
tes
t
u
se
s
so
c
ial
m
e
d
ia
d
a
ta
f
ro
m
Ne
tf
li
x
a
p
p
li
c
a
ti
o
n
u
se
r
c
o
m
m
e
n
ts.
T
h
ro
u
g
h
e
x
p
e
ri
m
e
n
tatio
n
o
n
d
if
f
e
r
e
n
t
it
e
ra
ti
o
n
s
o
f
v
a
rio
u
s
d
e
e
p
lea
rn
in
g
t
e
c
h
n
iq
u
e
s
a
lo
n
g
sid
e
m
u
lt
ip
le
w
o
rd
e
m
b
e
d
d
in
g
c
h
a
ra
c
teristics
,
th
e
BL
S
T
M
a
lg
o
rit
h
m
a
c
h
iev
e
d
th
e
m
o
st
n
o
tab
le
a
c
c
u
ra
c
y
ra
te
o
f
7
9
.
5
%
p
rio
r
to
i
n
teg
ra
ti
n
g
w
o
rd
e
m
b
e
d
d
in
g
f
e
a
tu
re
s.
On
th
e
o
th
e
r
h
a
n
d
,
t
h
e
h
ig
h
e
st
a
c
c
u
ra
c
y
v
a
l
u
e
re
su
lt
s
w
h
e
n
u
sin
g
th
e
w
o
rd
e
m
b
e
d
d
i
n
g
f
e
a
tu
re
c
a
n
b
e
se
e
n
in
th
e
BL
S
T
M
a
lg
o
rit
h
m
w
h
ich
u
se
s
th
e
w
o
rd
to
v
e
c
to
r
(W
o
rd
2
V
e
c
)
f
e
a
tu
re
w
it
h
a
v
a
lu
e
o
f
8
7
.
1
%
.
M
e
a
n
w
h
il
e
,
a
v
e
r
y
sig
n
if
ic
a
n
t
c
h
a
n
g
e
in
v
a
lu
e
in
c
re
a
se
w
a
s
o
b
tain
e
d
f
ro
m
th
e
F
a
stT
e
x
t
f
e
a
tu
re
in
th
e
CNN
a
l
g
o
rit
h
m
.
Af
ter
a
ll
th
e
e
v
a
lu
a
ti
o
n
p
ro
c
e
ss
e
s
we
re
c
a
rried
o
u
t,
t
h
e
b
e
st
c
las
sif
ica
ti
o
n
e
v
a
lu
a
ti
o
n
re
su
l
ts
w
e
re
o
b
tain
e
d
,
n
a
m
e
l
y
th
e
BL
S
T
M
a
lg
o
rit
h
m
w
it
h
sta
b
le v
a
lu
e
s o
n
a
ll
w
o
rd
e
m
b
e
d
d
in
g
f
e
a
tu
re
s.
K
ey
w
o
r
d
s
:
Dee
p
lear
n
in
g
Sen
ti
m
e
n
t a
n
a
l
y
s
is
So
cial
m
ed
ia
T
ex
t c
lass
if
icatio
n
W
o
r
d
em
b
ed
d
in
g
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
:
J
as
m
ir
Dep
ar
t
m
en
t o
f
C
o
m
p
u
ter
E
n
g
i
n
ee
r
in
g
,
Fac
u
lt
y
o
f
C
o
m
p
u
ter
Scien
ce
,
U
n
iv
er
s
itas
D
in
a
m
i
k
a
B
an
g
s
a
St.
J
en
d
r
al
Su
d
ir
m
a
n
,
T
eh
o
k
,
J
am
b
i Sela
tan
,
J
a
m
b
i,
I
n
d
o
n
es
ia
E
m
ail: ij
a
y
_
j
as
m
ir
@
y
a
h
o
o
.
co
m
1.
I
NT
RO
D
UCT
I
O
N
I
n
r
ec
en
t
d
ec
ad
es,
tech
n
o
lo
g
i
ca
l
d
ev
elo
p
m
en
t
s
h
a
v
e
ex
p
er
i
en
ce
d
a
r
ap
id
s
u
r
g
e,
esp
ec
iall
y
s
i
n
ce
th
e
e
m
er
g
e
n
ce
o
f
th
e
i
n
ter
n
et
an
d
p
er
s
o
n
al
co
m
p
u
ter
s
in
t
h
e
1
9
8
0
s
.
T
h
ese
tech
n
o
lo
g
ical
ad
v
an
ce
s
h
a
v
e
ca
u
s
ed
m
aj
o
r
ch
an
g
e
s
i
n
v
ar
io
u
s
s
ec
t
o
r
s
,
in
clu
d
i
n
g
in
f
o
r
m
a
tio
n
a
n
d
co
m
m
u
n
icatio
n
[
1
]
,
[
2
]
.
T
h
e
s
ig
n
i
f
ica
n
t
i
n
cr
ea
s
e
in
i
n
ter
n
e
t
tec
h
n
o
lo
g
y
h
a
s
ex
p
a
n
d
ed
th
e
r
ea
c
h
o
f
in
f
o
r
m
a
tio
n
d
is
tr
ib
u
tio
n
.
On
e
a
s
p
ec
t
t
h
at
s
u
p
p
o
r
ts
th
is
i
n
cr
ea
s
e
is
s
o
cial
m
ed
ia,
w
h
er
e
u
s
er
s
n
o
t
o
n
l
y
f
u
n
ctio
n
as
r
ec
ip
ien
ts
o
f
in
f
o
r
m
atio
n
b
u
t
al
s
o
as
cr
ea
to
r
s
o
f
in
f
o
r
m
a
tio
n
.
T
h
e
in
cr
ea
s
e
i
n
t
h
e
n
u
m
b
er
o
f
in
ter
n
et
u
s
er
s
i
n
I
n
d
o
n
esia
is
d
u
e
to
th
e
v
ar
io
u
s
co
n
v
e
n
ien
c
es
o
f
f
er
ed
b
y
s
o
cia
l
m
ed
ia
an
d
th
e
i
n
ter
n
et.
T
h
r
o
u
g
h
s
o
cial
m
ed
ia,
p
eo
p
le
ca
n
ac
ce
s
s
i
n
f
o
r
m
a
tio
n
a
n
d
co
m
m
u
n
icate
v
er
y
q
u
ick
l
y
.
T
h
e
u
s
e
o
f
d
ata
f
r
o
m
s
o
cial
m
ed
ia
is
th
e
late
s
t
i
n
n
o
v
ati
v
e
s
te
p
th
at
p
r
o
v
id
es
an
al
ter
n
ati
v
e
d
ata
s
o
u
r
ce
o
u
ts
id
e
o
f
tr
ad
itio
n
al
d
ata
co
ll
ec
tio
n
m
et
h
o
ds
[
3
]
,
[
4
]
.
Data
co
llectio
n
v
ia
s
o
cial
m
ed
ia
is
co
n
s
id
er
ed
to
p
r
o
v
id
e
ef
f
icien
c
y
in
m
an
y
w
a
y
s
.
T
h
is
ef
f
icien
c
y
in
c
lu
d
es
t
h
e
co
s
ts
th
at
m
u
s
t
b
e
in
cu
r
r
ed
f
o
r
d
at
a
ac
q
u
is
itio
n
,
b
ein
g
ab
le
to
o
b
tain
d
ata
in
r
ea
l
tim
e,
an
d
p
r
o
d
u
cin
g
d
ata
th
a
t
h
as
m
o
r
e
d
etailed
in
f
o
r
m
atio
n
to
d
escr
ib
e
th
e
tr
u
e
o
p
in
io
n
o
f
th
e
co
m
m
u
n
it
y
[
5
]
.
A
cti
v
itie
s
s
u
ch
as
t
h
o
s
e
ab
o
v
e
th
at
ar
e
r
elate
d
to
an
aly
zi
n
g
an
d
r
esp
o
n
d
in
g
to
p
u
b
lic
o
p
in
io
n
u
s
i
n
g
d
ata
s
o
u
r
ce
d
f
r
o
m
s
o
cial
m
ed
ia
ar
e
ca
ll
ed
s
en
ti
m
en
t a
n
al
y
s
i
s
[
6
]
,
[
7
]
.
Sen
ti
m
e
n
t
an
al
y
s
i
s
,
w
h
ic
h
is
a
s
u
b
s
et
o
f
n
at
u
r
al
lan
g
u
a
g
e
p
r
o
ce
s
s
in
g
(
NL
P
)
,
u
s
es
m
ac
h
i
n
e
lear
n
in
g
m
et
h
o
d
s
to
r
ec
o
g
n
ize
an
d
ex
tr
ac
t
f
ac
tu
a
l
in
f
o
r
m
atio
n
f
r
o
m
w
r
itte
n
te
x
t
[
8
]
.
T
h
is
an
al
y
s
i
s
in
v
o
l
v
es
id
e
n
ti
f
y
i
n
g
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KOM
NI
K
A
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l
C
o
n
tr
o
l
C
o
mp
a
r
is
o
n
o
f wo
r
d
emb
ed
d
i
n
g
fea
tu
r
es u
s
in
g
d
ee
p
lea
r
n
in
g
in
s
en
timen
t a
n
a
lysi
s
(
Ja
s
m
i
r
)
417
e
m
o
tio
n
al
n
u
a
n
ce
s
a
n
d
d
eter
m
in
i
n
g
t
h
e
o
v
er
all
s
e
n
ti
m
e
n
t
—
w
h
et
h
er
p
o
s
itiv
e,
n
e
u
tr
al,
o
r
n
eg
ati
v
e
—
e
x
p
r
ess
ed
b
y
th
e
au
t
h
o
r
.
A
p
p
l
y
i
n
g
s
e
n
ti
m
en
t
an
a
l
y
s
is
to
lar
g
er
d
ata
s
ets
allo
w
s
f
o
r
a
m
o
r
e
co
m
p
r
eh
en
s
i
v
e
a
n
d
in
-
d
ep
t
h
lev
el
o
f
a
n
al
y
s
i
s
[
9
]
.
I
n
NL
P
,
co
m
p
u
ter
s
d
o
n
o
t
h
av
e
an
in
n
ate
u
n
d
er
s
tan
d
i
n
g
o
f
tex
tu
a
l
lan
g
u
a
g
e,
s
o
th
e
y
n
ee
d
tech
n
iq
u
e
s
to
co
n
v
er
t
w
o
r
d
s
i
n
to
v
ec
to
r
s
f
o
r
ea
s
ier
u
n
d
er
s
tan
d
i
n
g
.
T
h
e
p
r
o
ce
s
s
o
f
r
ep
r
esen
tin
g
w
o
r
d
v
ec
to
r
s
r
e
m
ai
n
s
a
n
in
ter
esti
n
g
ar
ea
o
f
r
esear
ch
.
T
h
is
r
ep
r
esen
tatio
n
h
o
ld
s
g
r
ea
t
s
ig
n
i
f
ican
ce
a
s
it
p
r
o
f
o
u
n
d
ly
i
n
f
l
u
en
ce
s
th
e
ac
cu
r
ac
y
a
n
d
ef
f
icac
y
o
f
t
h
e
c
o
n
s
tr
u
cted
lear
n
i
n
g
m
o
d
els.
T
h
is
w
o
r
d
r
ep
r
esen
tatio
n
tech
n
i
q
u
e
is
i
n
cl
u
d
ed
in
th
e
f
ea
t
u
r
e
en
g
i
n
ee
r
in
g
s
ec
ti
o
n
.
Feat
u
r
e
en
g
i
n
ee
r
in
g
i
n
t
ex
tu
a
l
d
ata
h
a
s
it
s
o
w
n
ch
a
llen
g
e
s
d
u
e
to
t
h
e
ch
ar
ac
ter
is
tic
s
o
f
u
n
s
tr
u
ct
u
r
ed
tex
t.
T
h
e
f
ea
t
u
r
e
e
n
g
i
n
ee
r
i
n
g
s
tr
ateg
y
f
o
r
tex
tu
al
d
ata
t
h
at
i
s
p
o
p
u
lar
l
y
u
s
ed
is
k
n
o
w
n
as t
h
e
w
o
r
d
e
m
b
ed
d
in
g
f
ea
tu
r
e
[
1
0
]
–
[
1
2
]
.
T
h
is
w
o
r
d
e
m
b
ed
d
in
g
f
ea
t
u
r
e
is
co
llab
o
r
ate
d
w
it
h
s
e
v
er
al
class
i
f
icatio
n
m
et
h
o
d
s
.
T
h
er
e
ar
e
m
a
n
y
t
y
p
es o
f
clas
s
if
ier
s
th
at
ar
e
co
m
m
o
n
l
y
u
s
ed
to
class
if
y
s
en
ti
m
en
t a
n
al
y
s
i
s
.
T
h
e
m
et
h
o
d
s
th
at
ar
e
o
f
ten
u
s
ed
ar
e
m
ac
h
in
e
lear
n
i
n
g
m
et
h
o
d
s
[
1
3
]
,
[
1
4
]
an
d
d
ee
p
lear
n
in
g
[
1
5
]
.
I
n
th
i
s
r
esear
ch
,
t
h
e
t
y
p
e
s
o
f
m
eth
o
d
s
u
s
ed
ar
e
d
ee
p
lear
n
in
g
m
et
h
o
d
s
,
n
a
m
el
y
co
n
d
itio
n
al
r
a
n
d
o
m
f
ield
(
C
R
F)
[
1
6
]
,
b
i
d
ir
ec
tio
n
al
lo
n
g
s
h
o
r
t
ter
m
m
e
m
o
r
y
(
B
L
ST
M)
[
1
7
]
,
an
d
co
n
v
o
lu
ti
o
n
al
n
e
u
r
al
n
et
w
o
r
k
(
C
NN)
[
1
8
]
.
CRF
s
ar
e
u
s
ed
to
b
u
ild
p
r
o
b
ab
ilis
tic
m
o
d
els
f
o
r
s
eq
u
en
tial d
ata
s
e
g
m
e
n
tatio
n
a
n
d
lab
elin
g
.
B
ec
au
s
e
it i
s
co
n
d
itio
n
al,
C
R
F
is
a
ls
o
u
s
ed
to
en
s
u
r
e
t
h
at
i
n
f
er
en
ce
is
ea
s
y
to
d
o
an
d
also
av
o
id
s
th
e
p
r
o
b
lem
o
f
lab
el
b
ias.
B
L
S
T
M
is
u
s
ed
to
f
in
d
o
u
t
t
h
e
p
r
ev
io
u
s
in
f
o
r
m
at
io
n
p
r
o
ce
s
s
an
d
f
in
d
o
u
t
th
e
in
f
o
r
m
atio
n
p
r
o
ce
s
s
af
ter
w
ar
d
.
Me
an
w
h
ile,
C
NN
is
u
s
ed
to
s
ee
p
r
o
ce
s
s
in
g
ca
p
ab
ilit
ies
an
d
ev
al
u
ate
clas
s
i
f
icatio
n
p
er
f
o
r
m
a
n
ce
o
n
tex
t d
ata.
W
e
ev
alu
ate
t
h
e
ef
f
ec
ti
v
e
n
es
s
o
f
d
if
f
er
e
n
t
clas
s
i
f
icatio
n
m
et
h
o
d
s
b
y
test
i
n
g
th
e
ir
p
er
f
o
r
m
a
n
ce
u
s
in
g
s
ev
er
al
t
y
p
es
o
f
w
o
r
d
r
ep
r
es
en
tatio
n
s
,
n
a
m
el
y
w
o
r
d
to
v
ec
to
r
(
W
o
r
d
2
Vec
)
[
1
9
]
,
g
lo
b
al
v
ec
to
r
s
f
o
r
w
o
r
d
r
ep
r
esen
tatio
n
(
Glo
Ve
)
[
2
0
]
,
an
d
Fas
tT
ex
t
[
2
1
]
.
T
h
e
test
s
w
er
e
co
n
d
u
cted
o
n
a
s
en
ti
m
e
n
t
an
al
y
s
i
s
d
ataset
co
n
s
is
tin
g
o
f
Net
f
li
x
u
s
er
co
m
m
en
ts
.
Net
f
li
x
w
as
c
h
o
s
en
a
s
th
e
o
b
j
ec
t
o
f
s
tu
d
y
d
u
e
to
its
h
ig
h
p
o
p
u
lar
it
y
as
a
s
tr
ea
m
i
n
g
p
latf
o
r
m
,
its
lar
g
e
u
s
er
b
ase,
an
d
th
e
v
ar
iet
y
o
f
co
n
ten
t
it
o
f
f
er
s
.
T
h
is
m
a
k
es
it
a
r
ele
v
an
t
to
p
ic
f
o
r
u
n
d
er
s
ta
n
d
in
g
u
s
er
p
r
ef
er
en
ce
s
f
o
r
d
ig
ital
e
n
ter
tai
n
m
e
n
t
s
er
v
ices.
An
al
y
s
is
o
f
u
s
er
s
e
n
ti
m
e
n
t,
b
o
th
p
o
s
itiv
e
an
d
n
eg
at
iv
e,
ca
n
p
r
o
v
id
e
v
al
u
ab
l
e
in
s
i
g
h
ts
i
n
to
t
h
eir
v
ie
w
s
o
n
th
e
q
u
ali
t
y
o
f
th
e
s
er
v
ice,
i
n
t
er
f
ac
e,
an
d
co
n
te
n
t
p
r
o
v
id
ed
.
Si
m
i
la
r
s
t
u
d
ies
t
h
at
h
av
e
b
ee
n
d
is
cu
s
s
ed
in
c
lu
d
e
b
y
Al
-
S
m
a
d
i
et
a
l
.
[
2
2
]
u
s
in
g
s
e
v
er
al
d
ee
p
lear
n
in
g
m
et
h
o
d
s
s
u
ch
as
B
L
ST
M
-
C
R
F
co
m
b
i
n
ed
w
it
h
W
o
r
d
2
Vec
f
ea
tu
r
es
a
n
d
p
r
o
d
u
cin
g
a
n
F1
-
s
co
r
e
o
f
6
6
.
3
2
%.
th
en
B
L
ST
M
C
R
F
co
m
b
i
n
ed
w
it
h
Fas
tT
ex
t
f
ea
t
u
r
es
p
r
o
d
u
ci
n
g
an
F1
-
s
co
r
e
o
f
6
9
.
9
8
%.
T
h
e
n
,
J
a
n
g
e
t
a
l
.
[
2
3
]
p
r
o
p
o
s
e
d
a
h
y
b
r
i
d
m
o
d
el
o
f
B
i
-
L
S
T
M
+
C
N
N
w
ith
W
o
r
d
2
V
e
c
,
t
h
e
t
es
t
r
e
s
u
lt
s
s
h
o
w
e
d
th
a
t
th
e
p
r
o
p
o
s
e
d
m
o
d
el
p
r
o
d
u
c
e
d
m
o
r
e
a
cc
u
r
at
e
c
l
ass
i
f
i
c
a
ti
o
n
r
esu
l
ts
,
a
s
w
e
ll
a
s
h
i
g
h
e
r
r
e
c
al
l
an
d
F
1
s
c
o
r
e
s
,
th
an
th
e
m
u
l
ti
-
l
ay
e
r
p
e
r
c
e
p
t
r
o
n
(
ML
P)
m
o
d
e
l
,
C
N
N
o
r
i
n
d
iv
i
d
u
al
L
S
T
M
a
n
d
h
y
b
r
i
d
m
o
d
e
ls
.
Fu
r
th
e
r
m
o
r
e
,
I
f
t
ik
h
a
r
e
t
a
l
.
[
2
4
]
c
o
n
d
u
c
t
e
d
ex
p
e
r
im
en
t
s
w
it
h
s
ev
e
r
a
l
d
e
e
p
l
ea
r
n
i
n
g
m
o
d
el
s
c
o
m
b
in
e
d
w
i
th
s
ev
e
r
al
w
o
r
d
em
b
e
d
d
i
n
g
f
e
at
u
r
e
s
s
u
ch
a
s
C
N
N
+G
l
o
v
e
,
C
N
N
+
W
o
r
d
2
V
e
c
,
L
S
T
M
+
G
l
o
v
e
,
an
d
L
S
T
M
+
W
o
r
d
2
V
e
c
.
T
h
e
r
esu
lts
o
f
th
eir
r
esear
ch
s
tated
th
at
th
e
r
es
u
lts
o
f
t
h
e
co
m
b
in
atio
n
o
f
d
ee
p
lear
n
in
g
w
it
h
th
e
w
o
r
d
e
m
b
ed
d
in
g
f
ea
t
u
r
e
p
r
o
d
u
ce
d
b
etter
p
er
f
o
r
m
a
n
ce
.
B
ased
o
n
t
h
e
p
r
o
b
le
m
s
,
w
e
co
n
d
u
cted
r
esear
ch
as
w
ell
a
s
t
h
e
co
n
tr
ib
u
tio
n
o
f
t
h
is
r
esear
ch
,
n
a
m
el
y
to
i
m
p
r
o
v
e
th
e
ev
a
lu
at
io
n
v
a
l
u
e
o
f
t
h
e
class
i
f
ica
tio
n
p
er
f
o
r
m
an
ce
o
f
d
ee
p
lear
n
in
g
m
et
h
o
d
s
,
n
a
m
el
y
C
R
F
,
B
L
ST
M
,
an
d
C
NN
b
y
u
s
i
n
g
wo
r
d
em
b
ed
d
in
g
f
ea
t
u
r
es,
n
a
m
e
l
y
W
o
r
d
2
Vec
,
Glo
Ve,
an
d
Fas
tT
ex
t
as
tech
n
iq
u
es
to
im
p
r
o
v
e
th
e
e
v
alu
a
tio
n
v
al
u
e
o
f
d
ee
p
lear
n
in
g
clas
s
i
f
icat
io
n
p
er
f
o
r
m
a
n
ce
o
n
m
ac
h
i
n
e
l
ea
r
n
in
g
d
atasets
o
n
s
o
cial
m
ed
ia
d
ata
f
r
o
m
Ne
t
f
l
ix
ap
p
licatio
n
u
s
er
co
m
m
e
n
ts
.
2.
M
AT
E
RIAL
A
ND
M
E
T
H
O
D
I
n
o
r
d
er
f
o
r
th
is
r
esear
ch
to
ac
h
iev
e
m
a
x
i
m
u
m
r
es
u
lts
,
w
e
h
av
e
co
m
p
iled
a
s
er
ies
o
f
i
m
p
o
r
tan
t
s
tep
s
th
at
ca
n
p
r
o
d
u
ce
th
e
r
i
g
h
t
m
o
d
el
an
d
n
o
t
w
id
e
n
th
e
d
ir
ec
tio
n
in
ac
h
ie
v
in
g
t
h
e
g
o
al.
T
h
e
s
tep
s
tak
e
n
to
o
b
tain
r
esu
lt
s
th
at
ar
e
in
ac
co
r
d
an
ce
w
it
h
ex
p
ec
tatio
n
s
ar
e
co
m
p
ile
d
in
th
e
f
o
r
m
o
f
a
r
esear
ch
f
r
am
e
w
o
r
k
.
T
h
e
r
esear
ch
f
r
a
m
e
w
o
r
k
r
ef
er
r
ed
is
p
r
esen
t
ed
in
Fig
u
r
e
1.
2
.
1
.
Da
t
a
s
et
T
h
e
d
ataset
w
as
o
b
tain
ed
th
r
o
u
g
h
a
d
ata
co
llectio
n
p
r
o
ce
s
s
ca
r
r
ied
o
u
t
b
y
cr
a
w
l
in
g
.
W
e
u
tili
ze
t
h
e
Go
o
g
le
P
la
y
Scr
ap
er
P
y
t
h
o
n
li
b
r
ar
y
.
T
o
cr
a
w
l
d
ata,
t
h
e
I
D
o
f
th
e
ap
p
licatio
n
f
r
o
m
w
h
ic
h
d
a
ta
is
to
b
e
r
etr
ie
v
ed
is
f
ir
s
t
r
eq
u
ir
ed
.
I
n
th
is
ca
s
e,
Netf
li
x
h
as
th
e
I
D
‘
co
m
.
n
etf
li
x
.
m
ed
iaclie
n
t
’
.
F
u
r
t
h
er
m
o
r
e,
t
h
e
s
elec
tio
n
o
f
t
h
e
lan
g
u
a
g
e
i
n
t
h
e
r
e
v
ie
w
i
s
a
n
i
m
p
o
r
tan
t
s
tep
,
w
h
er
e
th
is
s
t
u
d
y
o
n
l
y
co
n
s
id
er
s
r
ev
ie
w
s
i
n
I
n
d
o
n
e
s
ia
n
.
Af
ter
s
elec
ti
n
g
th
e
lan
g
u
ag
e,
th
e
s
el
ec
tio
n
o
f
r
ev
ie
w
s
i
s
b
ased
o
n
th
e
s
co
r
e.
I
n
t
h
is
s
t
u
d
y
,
t
h
e
r
e
v
ie
w
s
ta
k
en
h
a
v
e
a
s
co
r
e
r
an
g
e
o
f
1
to
5
.
Fu
r
t
h
er
m
o
r
e,
th
e
o
r
d
er
o
f
r
ev
ie
w
s
u
s
e
d
is
m
o
s
t r
ele
v
an
t.
T
h
e
a
m
o
u
n
t o
f
d
ata
to
b
e
tak
e
n
also
n
ee
d
s
to
b
e
d
eter
m
i
n
ed
.
T
h
e
d
ata
o
b
tain
ed
h
as
s
e
v
er
al
attr
ib
u
tes,
in
c
lu
d
i
n
g
:
r
ev
ie
w
I
d
,
u
s
er
n
a
m
e,
u
s
er
I
m
a
g
e,
co
n
te
n
t,
s
co
r
e,
th
u
m
b
s
Up
C
o
u
n
t,
r
ev
ie
w
C
r
ea
t
ed
Ver
s
io
n
,
at,
r
ep
ly
C
o
n
ten
t,
an
s
w
er
ed
At,
an
d
ap
p
Ver
s
io
n
.
Ho
w
e
v
er
,
n
o
t
all
o
f
th
ese
attr
ib
u
tes
ar
e
r
eq
u
ir
ed
f
o
r
th
is
s
t
u
d
y
.
T
h
er
ef
o
r
e,
ir
r
elev
an
t
o
r
u
n
u
s
ed
attr
ib
u
tes
ar
e
r
e
m
o
v
ed
to
s
i
m
p
lify
t
h
e
d
ata.
T
h
er
e
ar
e
4
attr
ib
u
tes
th
a
t
w
ill
b
e
u
s
ed
,
n
a
m
e
l
y
u
s
er
n
a
m
e,
s
co
r
e,
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
1
6
9
3
-
6930
T
E
L
KOM
NI
K
A
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l
C
o
n
tr
o
l
,
Vo
l.
2
3
,
No
.
2
,
A
p
r
il
20
2
5
:
4
1
6
-
425
418
d
ate
,
an
d
co
n
te
n
t
.
Fi
g
u
r
e
2
is
a
f
lo
w
d
ia
g
r
a
m
o
f
d
ata
co
lle
ctio
n
.
T
h
e
d
ata
u
s
ed
is
a
r
a
w
d
ataset
th
a
t
w
il
l
g
o
th
r
o
u
g
h
s
ev
er
al
p
r
e
-
p
r
o
ce
s
s
i
n
g
p
r
o
ce
s
s
es b
ef
o
r
e
b
ec
o
m
in
g
a
d
ataset
th
at
is
r
ea
d
y
to
u
s
e.
Fig
u
r
e
1
.
R
esear
ch
f
r
a
m
e
w
o
r
k
2
.
2
.
P
re
pro
ce
s
s
ing
Af
ter
g
e
tti
n
g
t
h
e
Net
f
li
x
ap
p
li
ca
tio
n
u
s
er
r
ev
ie
w
d
ata,
t
h
e
n
ex
t
s
tep
i
s
to
ca
r
r
y
o
u
t
t
h
e
p
r
ep
r
o
ce
s
s
in
g
s
tag
e
b
e
f
o
r
e
en
ter
i
n
g
t
h
e
s
en
t
i
m
en
t
clas
s
i
f
icatio
n
s
ta
g
e.
T
h
is
p
r
o
ce
s
s
is
i
m
p
o
r
ta
n
t
to
en
s
u
r
e
th
at
t
h
e
d
ata
u
s
ed
b
y
s
e
n
ti
m
en
t
clas
s
i
f
icatio
n
m
o
d
els
is
c
lea
n
,
s
tr
u
ct
u
r
ed
,
an
d
r
ea
d
y
to
u
s
e.
T
h
e
p
r
ep
r
o
ce
s
s
in
g
s
ta
g
es
ca
r
r
ied
o
u
t
ar
e
d
ata
clea
n
in
g
,
ca
s
e
f
o
ld
in
g
,
to
k
en
izatio
n
,
s
to
p
w
o
r
d
r
e
m
o
v
al,
s
te
m
m
i
n
g
,
a
n
d
lab
elin
g
.
2
.
3
.
Wo
rd
e
m
bedd
i
ng
E
v
er
y
w
o
r
d
is
d
ep
icted
as a
n
u
m
er
ica
l lo
w
-
d
i
m
e
n
s
io
n
a
l v
ec
to
r
.
W
o
r
d
em
b
ed
d
in
g
e
n
ab
les
th
e
ca
p
tu
r
e
o
f
s
e
m
a
n
tic
d
etails
f
r
o
m
ex
te
n
s
i
v
e
tex
t
co
r
p
o
r
a.
T
h
ese
em
b
ed
d
in
g
s
f
in
d
ap
p
licatio
n
in
d
i
v
er
s
e
N
L
P
task
s
f
o
r
o
p
tim
a
l
w
o
r
d
r
ep
r
esen
tatio
n
.
No
tab
l
y
,
s
e
v
er
al
al
g
o
r
ith
m
s
ex
is
t
f
o
r
w
o
r
d
e
m
b
ed
d
in
g
,
in
cl
u
d
in
g
G
lo
Ve,
W
o
r
d
2
Vec
,
an
d
Fas
tT
ex
t.
Fo
r
th
is
r
esear
c
h
,
w
e
u
ti
lize
p
r
e
-
tr
ain
ed
m
o
d
els e
n
co
m
p
as
s
in
g
al
l th
r
ee
f
ea
t
u
r
es.
2
.
3
.
1
.
G
lo
Ve
Glo
Ve
is
a
co
-
o
cc
u
r
r
en
ce
m
atr
ix
-
b
ased
w
o
r
d
r
ep
r
esen
tatio
n
l
ea
r
n
in
g
tech
n
iq
u
e
th
at
ca
p
tu
r
e
s
s
e
m
a
n
ti
c
r
elatio
n
s
h
ip
s
b
et
w
ee
n
w
o
r
d
s
in
a
co
r
p
u
s
.
Glo
Ve
co
m
b
i
n
es
g
lo
b
al
s
ta
tis
tic
s
-
b
ased
a
p
p
r
o
ac
h
es
(
s
u
c
h
as
co
-
o
cc
u
r
r
en
ce
m
atr
ices)
a
n
d
lo
ca
l
co
n
tex
t
-
b
ased
m
e
th
o
d
s
(
s
u
ch
a
s
W
o
r
d
2
Vec
)
to
g
en
er
ate
w
o
r
d
e
m
b
ed
d
in
g
s
in
a
v
ec
to
r
s
p
ac
e,
allo
w
in
g
f
o
r
m
o
r
e
ef
f
ec
tiv
e
m
o
d
elin
g
o
f
li
n
ea
r
r
elatio
n
s
h
ip
s
b
et
w
ee
n
w
o
r
d
s
[
2
0
]
,
[
2
5
]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KOM
NI
K
A
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l
C
o
n
tr
o
l
C
o
mp
a
r
is
o
n
o
f wo
r
d
emb
ed
d
i
n
g
fea
tu
r
es u
s
in
g
d
ee
p
lea
r
n
in
g
in
s
en
timen
t a
n
a
lysi
s
(
Ja
s
m
i
r
)
419
Fig
u
r
e
2
.
Data
co
llectio
n
f
lo
w
ch
ar
t
2
.
3
.
2
.
Wo
rd2
Vec
W
o
r
d
2
V
e
c
u
t
i
li
z
es
t
h
e
o
c
cu
r
r
en
c
e
o
f
w
o
r
d
s
i
n
t
ex
t
t
o
e
s
t
a
b
li
s
h
c
o
n
n
e
ct
i
o
n
s
b
e
t
w
e
en
t
h
em
.
Fo
r
i
n
s
ta
n
c
e
,
i
t
m
ig
h
t
a
s
s
o
c
i
at
e
w
o
r
d
s
li
k
e
“
f
em
a
l
e
”
an
d
“
m
a
l
e
”
b
e
c
au
s
e
th
ey
f
r
e
q
u
en
t
ly
o
c
cu
r
in
c
o
m
p
a
r
a
b
l
e
c
o
n
t
ex
ts
.
W
o
r
d
2
V
e
c
o
p
e
r
a
t
e
s
th
r
o
u
g
h
tw
o
a
r
ch
i
t
ec
tu
r
a
l
f
o
r
m
s
:
c
o
n
t
ex
t
p
r
e
d
i
c
ti
o
n
,
w
h
i
ch
f
o
r
e
ca
s
t
s
th
e
s
u
r
r
o
u
n
d
in
g
w
o
r
d
s
b
a
s
e
d
o
n
a
g
iv
en
w
o
r
d
,
a
n
d
co
n
t
ex
t
-
b
as
e
d
p
r
e
d
i
c
t
i
o
n
(
B
ag
-
of
-
w
o
r
d
s
)
,
w
h
ic
h
p
r
e
d
i
ct
s
w
o
r
d
s
g
iv
en
a
c
o
n
t
ex
t.
E
s
s
en
t
i
a
lly
,
W
o
r
d
2
V
e
c
t
ak
es
a
t
e
x
t
u
a
l
c
o
r
p
u
s
as
in
p
u
t
an
d
g
en
e
r
a
t
e
s
a
w
o
r
d
v
ec
t
o
r
a
s
o
u
t
p
u
t
[
1
9
]
,
[
2
6
]
.
2
.
3
.
3
.
F
a
s
t
T
ex
t
E
v
er
y
w
o
r
d
is
d
ep
icted
as
a
c
o
llectio
n
o
f
n
-
g
r
a
m
c
h
ar
ac
ter
s
,
aid
in
g
in
ca
p
tu
r
i
n
g
th
e
ess
e
n
ce
o
f
s
h
o
r
ter
w
o
r
d
s
an
d
f
ac
i
litati
n
g
t
h
e
e
m
b
ed
d
in
g
’
s
u
n
d
er
s
ta
n
d
in
g
o
f
w
o
r
d
p
r
ef
ix
es a
n
d
s
u
f
f
i
x
es.
E
ac
h
n
-
g
r
a
m
c
h
ar
ac
ter
i
s
lin
k
ed
w
ith
a
v
ec
to
r
r
ep
r
esen
tatio
n
,
w
h
ile
w
o
r
d
s
ar
e
d
ep
icted
as
th
e
s
u
m
o
f
th
e
s
e
v
ec
t
o
r
r
ep
r
esen
tatio
n
s
.
Fas
tT
ex
t
d
e
m
o
n
s
tr
ate
s
s
tr
o
n
g
p
er
f
o
r
m
a
n
ce
,
e
n
ab
li
n
g
r
ap
id
m
o
d
el
tr
ain
in
g
o
n
e
x
ten
s
i
v
e
d
atasets
a
n
d
o
f
f
er
i
n
g
r
ep
r
esen
tatio
n
s
f
o
r
w
o
r
d
s
ab
s
e
n
t i
n
t
h
e
tr
ai
n
in
g
d
ata.
I
n
ca
s
es
w
h
er
e
a
w
o
r
d
is
ab
s
e
n
t d
u
r
i
n
g
m
o
d
el
tr
ai
n
i
n
g
,
it
ca
n
b
e
d
ec
o
m
p
o
s
ed
in
to
n
-
g
r
a
m
s
to
ac
q
u
ir
e
it
s
e
m
b
ed
d
in
g
v
ec
to
r
[
2
1
]
,
[
2
7
]
.
2
.
4
.
Dee
p
lea
rning
2
.
4
.
1
.
Co
nd
it
io
na
l
ra
nd
o
m
f
i
elds
C
R
F
s
b
elo
n
g
to
a
class
o
f
d
is
cr
i
m
i
n
ati
v
e
m
o
d
els
id
ea
ll
y
s
u
ited
f
o
r
class
if
icat
io
n
tas
k
s
wh
er
ein
t
h
e
cu
r
r
en
t
clas
s
i
f
icatio
n
i
s
i
m
p
ac
t
ed
b
y
co
n
tex
tu
al
f
ac
to
r
s
o
r
ad
jace
n
t
s
tate
s
[
2
8
]
.
C
R
F
f
in
d
s
ap
p
licatio
n
in
n
a
m
ed
en
tit
y
r
ec
o
g
n
itio
n
[
2
9
]
,
p
ar
t
-
of
-
s
p
ee
ch
ta
g
g
i
n
g
,
g
en
e
p
r
ed
ic
tio
n
,
n
o
is
e
r
ed
u
ctio
n
,
a
n
d
o
b
ject
d
etec
tio
n
tas
k
s
.
Dis
cr
i
m
in
at
iv
e
m
o
d
els,
also
k
n
o
w
n
as
co
n
d
itio
n
al
m
o
d
els
,
ar
e
a
s
u
b
s
et
o
f
m
o
d
els
co
m
m
o
n
l
y
e
m
p
lo
y
ed
i
n
s
tatis
t
ical
cla
s
s
i
f
icati
o
n
,
p
ar
tic
u
lar
l
y
i
n
s
u
p
er
v
i
s
ed
m
ac
h
i
n
e
l
ea
r
n
in
g
.
Di
s
cr
i
m
in
a
tiv
e
clas
s
i
f
ier
s
a
i
m
to
m
o
d
el
th
e
o
b
s
er
v
ed
d
ata
ex
clu
s
iv
el
y
,
lear
n
in
g
class
i
f
icat
io
n
f
r
o
m
p
r
o
v
id
ed
s
tatis
tic
s
.
A
p
p
r
o
ac
h
es
in
s
u
p
er
v
is
ed
lear
n
in
g
ar
e
t
y
p
icall
y
clas
s
i
f
ie
d
in
to
d
is
cr
i
m
i
n
ati
v
e
m
o
d
els
o
r
g
en
er
ativ
e
m
o
d
els.
Dis
cr
i
m
i
n
ati
v
e
m
o
d
els,
i
n
co
n
tr
ast
to
g
en
er
ativ
e
m
o
d
el
s
,
m
a
k
e
f
e
w
er
ass
u
m
p
t
io
n
s
ab
o
u
t
d
is
tr
ib
u
tio
n
s
an
d
p
lace
g
r
ea
ter
r
elian
ce
o
n
d
ata
q
u
alit
y
[
3
0
]
,
[
3
1
]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
1
6
9
3
-
6930
T
E
L
KOM
NI
K
A
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l
C
o
n
tr
o
l
,
Vo
l.
2
3
,
No
.
2
,
A
p
r
il
20
2
5
:
4
1
6
-
425
420
2
.
4
.
2
.
B
idi
re
ct
io
na
l
lo
ng
s
ho
r
t
-
t
er
m
m
e
m
o
ry
Der
iv
ed
f
r
o
m
t
h
e
r
ec
u
r
r
en
t
n
e
u
r
al
n
et
w
o
r
k
(
R
NN)
,
B
L
ST
M
[
3
2
]
en
h
an
ce
s
t
h
e
R
N
N
ar
ch
i
tectu
r
e
b
y
in
tr
o
d
u
ci
n
g
a
“
g
ate
w
a
y
”
m
ec
h
an
i
s
m
to
r
eg
u
late
th
e
f
lo
w
o
f
d
ata.
P
r
im
ar
il
y
,
th
e
lo
n
g
s
h
o
r
t
-
ter
m
m
e
m
o
r
y
(
L
ST
M)
ar
ch
itect
u
r
e
co
m
p
r
is
es
m
e
m
o
r
y
ce
ll
s
alo
n
g
w
i
th
i
n
p
u
t
g
ate
s
,
o
u
tp
u
t
g
ates,
an
d
f
o
r
g
et
g
ate
s
.
T
h
ese
ele
m
e
n
ts
ar
e
s
tr
u
ctu
r
ed
in
to
a
ch
ain
-
l
ik
e
ar
r
an
g
e
m
e
n
t
co
m
p
o
s
ed
o
f
R
NN
m
o
d
u
les,
w
h
ic
h
en
ab
les
th
e
s
m
o
o
t
h
tr
an
s
f
er
o
f
m
e
m
o
r
y
ce
ll
s
alo
n
g
th
e
ch
ai
n
.
Mo
r
eo
v
er
,
th
r
ee
s
ep
ar
ate
g
ates
ar
e
in
teg
r
ated
to
o
v
er
s
ee
an
d
r
eg
u
late
th
e
in
c
lu
s
io
n
o
r
in
h
ib
itio
n
o
f
i
n
f
o
r
m
atio
n
i
n
to
th
e
m
e
m
o
r
y
c
ell
[
3
3
]
.
2
.
4
.
3
.
Co
nv
o
lutio
na
l
neura
l
net
rw
o
rk
T
h
e
C
NN
is
a
f
o
r
m
o
f
r
eg
u
lated
f
ee
d
-
f
o
r
w
ar
d
n
e
u
r
al
n
et
w
o
r
k
th
at
au
to
n
o
m
o
u
s
l
y
lear
n
s
f
ea
t
u
r
e
en
g
i
n
ee
r
i
n
g
v
ia
t
h
e
o
p
ti
m
izati
o
n
o
f
f
ilter
s
,
al
s
o
k
n
o
w
n
as
k
er
n
els.
U
n
li
k
e
lo
w
er
la
y
er
f
ea
tu
r
es,
h
ig
h
er
la
y
er
f
ea
t
u
r
es
ar
e
e
x
tr
ac
ted
f
r
o
m
a
b
r
o
ad
e
r
co
n
tex
t
w
i
n
d
o
w
.
C
N
Ns
ar
e
s
o
m
eti
m
es
ca
lled
s
h
i
f
t
in
v
ar
ian
t
o
r
s
p
ac
e
in
v
ar
ia
n
t
ar
ti
f
icial
n
e
u
r
al
n
et
wo
r
k
s
(
SIA
NN)
b
ec
au
s
e
o
f
th
eir
ar
ch
itectu
r
e,
w
h
ic
h
in
v
o
lv
e
s
co
n
v
o
l
u
tio
n
k
er
n
el
s
o
r
f
ilter
s
w
it
h
s
h
ar
ed
w
ei
g
h
ts
m
o
v
i
n
g
ac
r
o
s
s
in
p
u
t
f
ea
t
u
r
es
.
T
h
is
m
o
v
e
m
en
t
p
r
o
d
u
ce
s
a
f
ea
t
u
r
e
m
ap
th
at
i
s
eq
u
iv
ale
n
t
to
tr
a
n
s
lat
io
n
.
Ho
w
e
v
er
,
d
esp
ite
t
h
e
t
er
m
in
o
lo
g
y
,
m
a
n
y
C
N
Ns
ar
e
n
o
t
in
h
er
en
tl
y
tr
an
s
latio
n
in
v
ar
ia
n
t,
m
ai
n
l
y
b
ec
au
s
e
o
f
t
h
e
d
o
w
n
s
a
m
p
li
n
g
o
p
er
atio
n
a
p
p
lied
to
th
e
in
p
u
t
[
3
4
]
–
[
3
6
]
.
3.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
ON
T
h
is
s
ec
tio
n
s
u
m
m
ar
izes
t
h
e
r
e
s
u
lt
s
o
f
t
h
e
e
x
p
er
i
m
e
n
ts
co
n
d
u
cted
ac
co
r
d
in
g
to
th
e
p
r
ev
io
u
s
l
y
p
lan
n
ed
r
esear
ch
f
lo
w
.
T
h
is
e
x
p
er
i
m
e
n
t
f
o
cu
s
es
o
n
a
n
al
y
zi
n
g
te
x
t
d
a
ta
f
r
o
m
s
o
cial
m
ed
ia
u
s
i
n
g
s
e
v
er
al
d
ee
p
lear
n
in
g
m
et
h
o
d
s
co
m
b
in
ed
w
i
th
w
o
r
d
e
m
b
ed
d
in
g
f
ea
t
u
r
es.
T
r
ain
in
g
an
d
test
i
n
g
d
ata
ar
e
d
iv
id
ed
w
i
th
a
n
8
0
:2
0
d
iv
is
io
n
s
ch
e
m
e.
T
h
is
s
t
u
d
y
tes
ts
d
ee
p
lear
n
in
g
m
eth
o
d
s
w
i
th
v
ar
io
u
s
v
ar
iatio
n
s
o
f
w
o
r
d
e
m
b
ed
d
in
g
f
ea
tu
r
e
s
.
T
h
e
d
ee
p
lear
n
in
g
m
e
th
o
d
is
ap
p
lied
as
an
ap
p
r
o
ac
h
to
s
en
ti
m
en
t
cla
s
s
if
ica
tio
n
o
n
tex
t
d
ata.
T
h
e
ty
p
es
o
f
d
ee
p
lear
n
in
g
m
et
h
o
d
s
u
s
ed
in
cl
u
d
e
C
R
F,
B
L
ST
M
,
an
d
C
NN.
I
n
ad
d
iti
o
n
,
th
e
w
o
r
d
e
m
b
ed
d
in
g
f
ea
t
u
r
es
u
s
ed
i
n
clu
d
e
W
o
r
d
2
Vec
,
Glo
Ve,
an
d
Fas
tTe
x
t.
T
ab
le
1
ex
p
lain
s
th
e
co
n
f
u
s
io
n
m
atr
i
x
o
f
C
R
F
w
i
th
t
h
r
ee
w
o
r
d
em
b
ed
d
in
g
f
ea
t
u
r
es
an
d
o
n
e
w
i
th
o
u
t
f
ea
t
u
r
es.
I
n
C
R
F
w
it
h
o
u
t f
ea
tu
r
es,
th
e
r
esu
lts
ar
e
T
P
=4
0
6
,
F
P
=1
0
4
,
FN=9
1
,
an
d
T
N=
2
9
9
.
T
h
is
m
ea
n
s
t
h
at
th
i
s
m
o
d
el
h
a
s
a
f
air
l
y
lo
w
n
u
m
b
er
o
f
T
P
co
m
p
ar
ed
to
th
e
u
s
e
o
f
W
o
r
d
2
Vec
,
Glo
Ve,
an
d
F
astT
ex
t
f
ea
t
u
r
es.
I
n
ad
d
itio
n
,
th
e
FP
v
alu
e
is
q
u
ite
h
ig
h
,
in
d
icatin
g
th
at
th
e
m
o
d
el
ten
d
s
to
in
co
r
r
ec
tly
id
en
ti
f
y
n
e
g
ati
v
e
d
ata
as
p
o
s
itiv
e.
T
h
en
C
R
F
w
it
h
W
o
r
d
2
Vec
p
r
o
d
u
ce
s
T
P
=
5
1
2
,
FP
=9
8
,
FN=6
9
,
an
d
T
N=
2
2
1
.
T
h
e
ad
d
itio
n
o
f
th
e
W
o
r
d
2
Vec
f
ea
tu
r
e
s
i
g
n
i
f
ica
n
t
l
y
in
cr
ea
s
es
T
P
(
f
r
o
m
4
0
6
to
5
1
2
)
,
in
d
icatin
g
t
h
at
th
e
m
o
d
el
is
b
etter
a
b
le
to
r
ec
o
g
n
ize
p
o
s
itiv
e
d
ata
co
r
r
e
ctl
y
.
Ho
w
e
v
er
,
FP
is
s
till
q
u
ite
h
ig
h
(
9
8
)
,
an
d
th
e
n
u
m
b
e
r
o
f
T
N
d
ec
r
ea
s
es
co
m
p
ar
ed
to
th
e
m
o
d
el
w
it
h
o
u
t f
ea
tu
r
es.
T
h
is
s
h
o
w
s
th
a
t W
o
r
d
2
Vec
im
p
r
o
v
es th
e
r
ec
o
g
n
iti
o
n
o
f
p
o
s
itiv
e
d
ata
b
u
t
s
lig
h
tl
y
d
ec
r
ea
s
e
s
th
e
ab
ili
t
y
to
r
ec
o
g
n
ize
n
eg
at
iv
e
d
ata.
I
n
th
e
C
R
F
w
i
th
Glo
Ve
s
ec
tio
n
,
th
er
e
ar
e
r
esu
lts
o
f
T
P
0
=4
4
8
,
F
P
=9
1
,
FN=8
1
,
an
d
T
N=
2
8
0
.
T
h
is
in
d
icate
s
th
at
Glo
Ve
p
r
o
v
id
es
m
o
r
e
b
al
an
ce
d
r
esu
lts
t
h
a
n
W
o
r
d
2
Vec
.
T
P
is
lo
w
er
th
a
n
W
o
r
d
2
Vec
,
b
u
t
FP
is
also
lo
w
er
(
9
1
)
,
in
d
icatin
g
t
h
at
t
h
e
m
o
d
el
is
b
etter
at
m
i
n
i
m
izi
n
g
er
r
o
r
s
in
cla
s
s
i
f
y
i
n
g
n
e
g
ativ
e
d
ata.
T
h
e
n
u
m
b
er
o
f
T
Ns
in
cr
ea
s
es
co
m
p
ar
ed
to
W
o
r
d
2
Vec
.
T
h
en
C
R
F
w
i
th
Fa
s
tT
ex
t
w
h
ic
h
p
r
o
d
u
ce
s
T
P
=4
7
3
,
F
P
=7
9
,
FN=5
9
,
an
d
T
N=
2
8
9
.
I
t
ca
n
b
e
s
ee
n
th
at
Fa
s
tT
ex
t
p
r
o
v
id
es
th
e
b
est
o
v
er
all
p
er
f
o
r
m
an
ce
.
T
P
an
d
T
N
in
cr
ea
s
e
c
o
m
p
ar
ed
to
Glo
Ve,
w
h
ile
FP
an
d
FN
ar
e
th
e
lo
w
e
s
t
a
m
o
n
g
all
m
o
d
els.
T
h
is
s
h
o
ws
th
at
Fa
s
tT
ex
t
is
v
er
y
e
f
f
ec
ti
v
e
in
i
m
p
r
o
v
in
g
t
h
e
r
ec
o
g
n
iti
o
n
o
f
p
o
s
itiv
e
an
d
n
eg
at
iv
e
d
ata,
w
it
h
th
e
lea
s
t c
l
ass
i
f
icatio
n
er
r
o
r
s
.
T
ab
l
e
1
.
C
o
n
f
u
s
io
n
m
atr
i
x
o
f
C
R
F
Ex
p
e
r
i
me
n
t
TP
FP
FN
TN
C
R
F
w
i
t
h
o
u
t
f
e
a
t
u
r
e
4
0
6
1
0
4
91
2
9
9
C
R
F
w
i
t
h
W
o
r
d
2
V
e
c
5
1
2
98
69
2
2
1
C
R
F
w
i
t
h
G
l
o
V
e
4
4
8
91
81
2
8
0
C
R
F
w
i
t
h
F
a
st
T
e
x
t
4
7
3
79
59
2
8
9
T
ab
le
2
is
a
C
R
F
te
s
t
w
i
th
3
w
o
r
d
e
m
b
ed
d
in
g
v
ar
ian
t
s
a
n
d
o
n
e
w
it
h
o
u
t
w
o
r
d
e
m
b
ed
d
i
n
g
.
I
t
ca
n
b
e
s
ee
n
t
h
at
th
e
ac
cu
r
ac
y
w
it
h
o
u
t
u
s
i
n
g
f
ea
tu
r
e
s
is
lo
w
er
th
an
t
h
e
m
o
d
el
th
at
u
s
es
f
ea
t
u
r
es.
T
h
is
s
h
o
w
s
th
e
i
m
p
o
r
tan
ce
o
f
th
e
e
m
b
ed
d
in
g
f
ea
t
u
r
e.
I
n
t
h
e
C
R
F w
i
th
W
o
r
d
2
Vec
s
ec
tio
n
,
t
h
er
e
ar
e
th
e
b
es
t
r
esu
lt
s
f
o
r
R
ec
all,
w
h
ic
h
m
ea
n
s
t
h
at
th
i
s
m
o
d
el
is
ab
le
to
ca
p
t
u
r
e
m
o
r
e
ac
tu
al
p
o
s
itiv
e
ca
s
es.
Me
an
w
h
i
le,
C
R
F
w
i
th
Glo
V
e
p
r
o
d
u
ce
s
m
o
r
e
s
tab
le
p
er
f
o
r
m
an
ce
i
n
all
m
etr
ics,
al
th
o
u
g
h
n
o
t
th
e
b
est.
T
h
en
C
R
F
w
it
h
Fa
s
tT
ex
t
g
i
v
es
th
e
b
es
t
v
alu
e
in
ac
cu
r
ac
y
,
p
r
ec
is
io
n
,
an
d
F1
-
Sco
r
e.
T
h
is
m
o
d
el
is
t
h
e
m
o
s
t
o
p
ti
m
a
l
i
n
p
r
o
d
u
cin
g
co
r
r
ec
t
p
r
ed
ictio
n
s
an
d
m
ai
n
tai
n
i
n
g
a
b
ala
n
ce
b
etw
ee
n
p
r
ec
is
io
n
a
n
d
r
ec
all.
T
h
ese
r
esu
lt
s
s
h
o
w
th
a
t th
e
Fas
tT
ex
t f
ea
tu
r
e
p
r
o
v
id
es
a
s
ig
n
i
f
ican
t in
cr
ea
s
e
i
n
acc
u
r
ac
y
(
8
.
0
9
%)
an
d
r
ec
all
(
8
.
8
4
%),
m
a
k
i
n
g
it a
n
ex
ce
l
len
t c
h
o
ic
e
f
o
r
i
m
p
r
o
v
in
g
t
h
e
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KOM
NI
K
A
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l
C
o
n
tr
o
l
C
o
mp
a
r
is
o
n
o
f wo
r
d
emb
ed
d
i
n
g
fea
tu
r
es u
s
in
g
d
ee
p
lea
r
n
in
g
in
s
en
timen
t a
n
a
lysi
s
(
Ja
s
m
i
r
)
421
m
o
d
el
’
s
ab
ilit
y
to
ca
p
tu
r
e
tr
u
e
p
o
s
itiv
es.
P
r
ec
is
io
n
an
d
F1
-
Sc
o
r
e
also
in
cr
ea
s
e
q
u
ite
s
ig
n
i
f
ic
an
tl
y
,
s
u
p
p
o
r
tin
g
a
b
alan
ce
b
et
w
ee
n
co
r
r
ec
t
p
o
s
iti
v
e
p
r
ed
ictio
n
s
an
d
th
e
ab
ilit
y
to
ca
p
tu
r
e
p
o
s
itiv
e
ca
s
es.
Fo
cu
s
o
n
th
e
F1
-
Sco
r
e
m
etr
ic,
s
i
n
ce
F1
-
Sco
r
e
is
a
m
e
tr
ic
th
at
co
m
b
i
n
es
p
r
ec
is
io
n
a
n
d
r
ec
all
,
it
is
v
er
y
r
elev
a
n
t f
o
r
ca
s
es
t
h
at
r
eq
u
ir
e
a
b
alan
ce
b
et
w
ee
n
th
e
t
w
o
m
e
t
r
ics,
esp
ec
iall
y
i
n
class
i
f
icat
io
n
tas
k
s
in
v
o
l
v
in
g
d
ata
w
ith
a
n
i
m
b
ala
n
ce
d
class
d
is
tr
ib
u
tio
n
o
r
ca
s
es
w
h
er
e
th
e
b
alan
ce
b
et
w
ee
n
p
r
ec
is
io
n
an
d
r
ec
all
is
a
p
r
io
r
ity
.
I
n
t
h
ese
i
m
b
a
lan
ce
d
d
ataset
s
w
h
er
e
o
n
e
class
i
s
v
er
y
d
o
m
in
an
t,
ac
cu
r
ac
y
m
a
y
ap
p
ea
r
h
i
g
h
b
ec
au
s
e
th
e
m
o
d
el
ca
n
ig
n
o
r
e
th
e
m
i
n
o
r
it
y
cla
s
s
.
F1
-
Sco
r
e
ad
d
r
ess
e
s
t
h
is
p
r
o
b
lem
b
y
ta
k
i
n
g
th
e
m
i
n
o
r
it
y
cla
s
s
in
to
ac
co
u
n
t.
C
R
F
w
i
th
Fas
tT
ex
t h
a
s
t
h
e
h
ig
h
es
t
F1
-
Sco
r
e,
in
d
icati
n
g
t
h
at
t
h
i
s
f
ea
tu
r
e
is
o
p
ti
m
al
f
o
r
p
r
o
d
u
cin
g
a
g
o
o
d
b
alan
ce
.
T
ab
l
e
2
.
C
o
m
p
ar
is
o
n
o
f
C
R
F
ev
alu
a
tio
n
v
alu
e
s
w
it
h
w
o
r
d
em
b
ed
d
i
n
g
Ex
p
e
r
i
me
n
t
A
c
c
u
r
acy
P
r
e
c
i
ssi
o
n
R
e
c
a
l
l
F1
-
S
c
o
r
e
C
R
F
w
i
t
h
o
u
t
f
e
a
t
u
r
e
7
8
.
3
3
3
3
3
3
3
3
7
9
.
6
0
7
8
4
8
1
.
6
9
0
1
4
8
0
.
6
3
5
5
5
C
R
F
w
i
t
h
W
o
r
d
2
V
e
c
8
1
.
4
4
4
4
4
4
4
4
8
3
.
9
3
4
4
3
8
8
.
1
2
3
9
2
8
5
.
9
7
8
1
7
C
R
F
w
i
t
h
G
l
o
V
e
8
0
.
8
8
8
8
8
8
8
9
8
3
.
1
1
6
8
8
8
4
.
6
8
8
0
9
8
3
.
8
9
5
1
3
C
R
F
w
i
t
h
F
a
st
T
e
x
t
84
.
6
6
6
6
6
6
6
7
8
5
.
6
8
8
4
1
8
8
.
9
0
9
7
7
8
7
.
2
6
9
3
7
T
ab
le
3
ex
p
lain
s
th
e
co
n
f
u
s
io
n
m
atr
i
x
o
f
B
L
ST
M
w
ith
th
r
ee
wo
r
d
em
b
ed
d
in
g
f
ea
tu
r
es
an
d
o
n
e
w
i
th
o
u
t
f
ea
t
u
r
es.
I
n
B
L
ST
M
w
i
th
o
u
t
f
ea
tu
r
es,
th
er
e
ar
e
v
alu
es
o
f
T
P=4
8
1
,
FP
=1
0
1
,
FN=8
3
,
an
d
T
N=
2
3
5
.
T
h
is
m
ea
n
s
th
at
t
h
e
m
o
d
el
w
it
h
o
u
t
f
ea
tu
r
e
s
p
r
o
d
u
ce
s
q
u
ite
g
o
o
d
p
er
f
o
r
m
an
ce
,
w
ith
a
T
P
o
f
4
8
1
.
Ho
w
e
v
er
,
th
e
FP
is
q
u
ite
h
ig
h
(
1
0
1
)
,
in
d
icatin
g
t
h
at
t
h
e
m
o
d
el
o
f
te
n
m
is
cla
s
s
i
f
ie
s
n
eg
ativ
e
d
ata
as
p
o
s
iti
v
e.
I
n
ad
d
iti
o
n
,
th
e
T
N
v
al
u
e
i
s
lo
w
er
th
a
n
t
h
e
m
o
d
el
u
s
in
g
f
ea
tu
r
es,
in
d
icat
in
g
a
w
ea
k
er
ab
ili
t
y
to
r
ec
o
g
n
ize
n
e
g
ati
v
e
d
a
ta.
I
n
th
e
B
L
ST
M
w
it
h
W
o
r
d
2
Vec
s
ec
tio
n
,
th
er
e
ar
e
v
alu
es
o
f
T
P
=
5
0
1
,
F
P
=9
7
,
FN=8
1
,
an
d
T
N=
2
2
1
.
W
ith
t
h
e
ad
d
itio
n
o
f
th
e
W
o
r
d
2
Vec
f
ea
tu
r
e,
t
h
e
n
u
m
b
er
o
f
T
P
in
cr
ea
s
es
to
5
0
1
,
in
d
icatin
g
t
h
at
t
h
e
m
o
d
el
i
s
b
ette
r
ab
le
to
r
ec
o
g
n
ize
p
o
s
itiv
e
d
ata
co
r
r
ec
tl
y
th
a
n
t
h
e
m
o
d
el
w
it
h
o
u
t
f
ea
tu
r
es.
Ho
w
e
v
er
,
th
e
FP
v
al
u
e
i
s
s
till
q
u
ite
h
ig
h
(
9
7
)
,
m
ea
n
i
n
g
th
at
t
h
e
m
i
s
clas
s
i
f
icatio
n
o
f
n
e
g
ati
v
e
d
ata
as
p
o
s
iti
v
e
r
e
m
ain
s
q
u
ite
s
i
g
n
i
f
ican
t.
T
h
e
d
ec
r
ea
s
e
i
n
t
h
e
n
u
m
b
er
o
f
T
N
also
in
d
icate
s
t
h
at
n
e
g
ati
v
e
d
ata
r
ec
o
g
n
itio
n
i
s
s
li
g
h
tl
y
i
m
p
a
ir
ed
.
Nex
t,
i
n
B
L
ST
M
w
it
h
Glo
Ve,
t
h
er
e
ar
e
v
alu
e
s
o
f
T
P
=4
8
0
,
FP
=
9
9
,
FN
=7
9
,
an
d
T
N=
2
4
2
.
T
h
e
Glo
Ve
f
ea
tu
r
e
p
r
o
d
u
ce
s
a
s
li
g
h
tl
y
lo
w
er
n
u
m
b
er
o
f
T
P
th
an
W
o
r
d
2
Vec
(
4
8
0
v
s
.
5
0
1
)
,
b
u
t
th
e
FN
is
also
lo
w
er
(
7
9
v
s
.
8
1
)
.
I
n
ad
d
itio
n
,
th
e
n
u
m
b
er
o
f
FP
is
s
m
aller
th
an
W
o
r
d
2
Vec
(
9
9
v
s
.
9
7
)
,
an
d
T
N
in
cr
ea
s
es
to
2
4
2
,
in
d
icatin
g
t
h
at
th
e
m
o
d
el
is
b
etter
at
r
e
co
g
n
izi
n
g
n
e
g
ati
v
e
d
ata
th
an
W
o
r
d
2
Vec
.
T
h
en
B
L
ST
M
w
i
th
Fa
s
tT
ex
t
p
r
o
d
u
ce
s
v
al
u
es
o
f
T
P
=5
0
5
,
FP
=
6
6
,
F
N=
5
0
,
an
d
T
N=
2
7
9
.
T
h
e
Fas
tT
ex
t
m
o
d
el
g
iv
es
t
h
e
b
est
r
esu
lts
a
m
o
n
g
all
m
eth
o
d
s
.
W
ith
th
e
h
ig
h
est
T
P
(
5
0
5
)
an
d
th
e
lo
w
e
s
t
FN
(
5
0
)
,
th
is
m
o
d
el
is
v
er
y
e
f
f
ec
ti
v
e
in
r
ec
o
g
n
izi
n
g
p
o
s
itiv
e
d
ata
.
I
n
ad
d
itio
n
,
FP
is
th
e
lo
w
e
s
t
(
6
6
)
,
an
d
T
N
is
th
e
h
ig
h
e
s
t
(
2
7
9
)
,
in
d
icatin
g
t
h
at
th
is
m
o
d
el
is
al
s
o
v
er
y
g
o
o
d
at
r
ec
o
g
n
i
zi
n
g
n
e
g
ati
v
e
d
ata.
T
h
is
co
n
f
ir
m
s
t
h
a
t
Fas
tT
ex
t i
m
p
r
o
v
e
s
o
v
er
all
p
er
f
o
r
m
a
n
ce
.
T
ab
l
e
3
.
C
o
n
f
u
s
io
n
m
atr
i
x
o
f
B
L
ST
M
Ex
p
e
r
i
me
n
t
TP
FP
FN
TN
B
L
S
T
M
w
i
t
h
o
u
t
f
e
a
t
u
r
e
4
8
1
1
0
1
83
2
3
5
B
L
S
T
M
w
i
t
h
W
o
r
d
2
V
e
c
5
0
1
97
81
2
2
1
B
L
S
T
M
w
i
t
h
G
l
o
V
e
4
8
0
99
79
2
4
2
B
L
S
T
M
w
i
t
h
F
a
st
T
e
x
t
5
0
5
66
50
2
7
9
T
ab
le
4
is
a
s
u
m
m
ar
y
tab
le
o
f
th
e
ex
p
er
i
m
e
n
tal
r
esu
lt
s
o
f
th
e
B
L
ST
M
m
eth
o
d
w
ith
t
h
r
ee
v
ar
ian
t
s
o
f
w
o
r
d
e
m
b
ed
d
in
g
f
ea
tu
r
es
a
n
d
o
n
e
w
it
h
o
u
t
f
ea
t
u
r
es.
I
n
t
h
e
B
L
ST
M
w
ith
o
u
t
f
ea
t
u
r
e
s
ec
tio
n
;
t
h
i
s
m
o
d
el
s
er
v
es
as
a
b
aselin
e,
an
d
its
p
er
f
o
r
m
an
ce
is
r
elati
v
el
y
g
o
o
d
w
it
h
o
u
t
ad
d
itio
n
al
f
ea
t
u
r
es,
b
u
t
it
ca
n
s
till
b
e
f
u
r
th
er
i
m
p
r
o
v
ed
b
y
ad
d
in
g
w
o
r
d
e
m
b
ed
d
in
g
f
ea
t
u
r
es.
I
n
th
e
B
L
S
T
M
w
it
h
W
o
r
d
2
Vec
s
ec
tio
n
,
th
e
m
o
d
el
is
s
li
g
h
tl
y
b
etter
th
an
th
e
b
aselin
e,
w
it
h
s
m
all
i
m
p
r
o
v
e
m
e
n
ts
i
n
ac
cu
r
ac
y
,
p
r
ec
is
io
n
,
r
ec
all
,
an
d
F1
-
Sco
r
e.
T
h
e
u
s
e
o
f
W
o
r
d
2
Vec
as
an
em
b
ed
d
in
g
f
e
atu
r
e
i
m
p
r
o
v
es
th
e
m
o
d
el
’
s
u
n
d
er
s
tan
d
in
g
o
f
w
o
r
d
r
elatio
n
s
h
ip
s
.
I
n
th
e
B
L
ST
M
w
it
h
Glo
Ve
s
ec
tio
n
,
t
h
e
r
esu
lt
s
ar
e
v
er
y
s
i
m
ilar
to
W
o
r
d
2
Vec
,
b
u
t
s
lig
h
tl
y
lo
w
er
in
p
r
ec
is
io
n
.
T
h
is
m
o
d
el
s
h
o
w
s
b
etter
p
er
f
o
r
m
an
ce
i
n
ter
m
s
o
f
R
ec
all,
b
u
t
n
o
t
as
g
o
o
d
as
th
e
m
o
d
el
w
it
h
W
o
r
d
2
Vec
.
T
h
en
in
th
e
B
L
ST
M
w
it
h
Fas
tT
ex
t sectio
n
: T
h
is
is
th
e
b
est
m
o
d
el,
w
i
th
s
ig
n
i
f
ica
n
t i
m
p
r
o
v
e
m
en
t
s
in
all
m
etr
ics.
Fa
s
tT
ex
t p
r
o
v
id
es c
lear
i
m
p
r
o
v
e
m
en
t
s
in
p
r
ec
is
io
n
,
r
ec
a
ll,
an
d
F1
-
Sco
r
e,
m
a
k
i
n
g
it
a
v
er
y
ef
f
ec
t
iv
e
m
o
d
el
in
t
h
is
class
i
f
icatio
n
.
T
h
e
u
s
e
o
f
e
m
b
ed
d
in
g
f
ea
tu
r
e
s
s
u
c
h
as
W
o
r
d
2
Vec
,
Glo
Ve,
an
d
Fas
tT
ex
t
af
f
ec
ts
t
h
e
i
m
p
r
o
v
e
m
e
n
t
o
f
m
o
d
el
p
er
f
o
r
m
a
n
ce
co
m
p
ar
ed
to
th
e
b
aselin
e
m
o
d
el
w
ith
o
u
t
f
ea
t
u
r
es.
T
h
e
m
o
d
el
w
it
h
Fas
tT
ex
t
s
h
o
w
s
t
h
e
g
r
ea
test
i
m
p
r
o
v
e
m
e
n
t,
esp
ec
i
all
y
i
n
r
ec
all
a
n
d
F1
-
Sco
r
e.
B
L
ST
M
w
i
th
Fas
tT
ex
t
h
as
t
h
e
h
ig
h
es
t
F1
-
Sco
r
e
(
8
9
.
6
9
%),
in
d
icatin
g
t
h
at
th
is
m
o
d
el
i
s
th
e
b
est
ch
o
ice
esp
e
ciall
y
i
n
th
e
b
alan
ce
b
et
w
ee
n
ac
cu
r
ate
p
r
ed
ictio
n
an
d
th
e
m
o
d
el
’
s
ab
ilit
y
to
ca
p
tu
r
e
p
o
s
itiv
e
ca
s
es.
T
h
e
u
s
e
o
f
e
m
b
ed
d
in
g
f
ea
t
u
r
es
s
u
ch
as
Fas
tT
ex
t
ca
n
s
ig
n
i
f
ica
n
tl
y
i
m
p
r
o
v
e
p
er
f
o
r
m
an
ce
co
m
p
ar
ed
to
n
o
t
u
s
in
g
f
e
atu
r
es
o
r
u
s
in
g
o
th
er
f
ea
t
u
r
es
s
u
c
h
as
W
o
r
d
2
Ve
c
an
d
Glo
Ve
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
1
6
9
3
-
6930
T
E
L
KOM
NI
K
A
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l
C
o
n
tr
o
l
,
Vo
l.
2
3
,
No
.
2
,
A
p
r
il
20
2
5
:
4
1
6
-
425
422
T
ab
l
e
4
.
C
o
m
p
ar
is
o
n
o
f
B
L
ST
M
ev
alu
a
tio
n
v
alu
e
s
w
it
h
w
o
r
d
em
b
ed
d
in
g
Ex
p
e
r
i
me
n
t
A
c
c
u
r
a
c
y
P
r
e
c
i
ssi
o
n
R
e
c
a
l
l
F1
-
S
c
o
r
e
B
L
S
T
M
w
i
t
h
o
u
t
f
e
a
t
u
r
e
79
.
5
5
5
5
5
5
5
6
82
.
6
4
6
0
5
85
.
2
8
3
6
9
83
.
9
4
4
1
5
B
L
S
T
M
w
i
t
h
W
o
r
d
2
V
e
c
80
.
2
2
2
2
2
2
2
2
83
.
7
7
9
2
6
86
.
0
8
2
4
7
84
.
9
1
5
2
5
B
L
S
T
M
w
i
t
h
G
l
o
V
e
80
.
2
2
2
2
2
2
2
2
82
.
9
0
1
5
5
85
.
8
6
7
6
2
84
.
3
5
8
5
2
B
L
S
T
M
w
i
t
h
F
a
st
T
e
x
t
87
.
1
1
1
1
1
1
1
1
88
.
4
4
1
3
3
90
.
9
9
0
9
9
89
.
6
9
8
0
5
T
ab
le
5
ex
p
lain
s
t
h
e
co
n
f
u
s
io
n
m
a
tr
ix
o
f
C
NN
w
it
h
th
r
ee
wo
r
d
em
b
ed
d
in
g
f
ea
t
u
r
es
a
n
d
o
n
e
w
it
h
o
u
t
f
ea
t
u
r
es.
I
n
C
NN
w
it
h
o
u
t f
ea
t
u
r
es,
th
er
e
ar
e
v
al
u
e
s
o
f
T
P
=4
0
4
,
FP
=
1
2
8
,
FN=1
1
9
,
an
d
T
N
=2
4
9
.
W
ith
o
u
t w
o
r
d
e
m
b
ed
d
in
g
f
ea
t
u
r
es,
C
NN
p
r
o
d
u
ce
s
th
e
lo
w
es
t
p
er
f
o
r
m
an
ce
.
T
h
e
n
u
m
b
er
o
f
T
P
is
th
e
lo
w
est
(
4
0
4
)
,
w
h
i
le
FP
an
d
FN a
r
e
th
e
h
ig
h
es
t (
1
2
8
a
n
d
1
1
9
)
.
T
h
is
s
h
o
w
s
t
h
at
th
e
m
o
d
el
h
as
m
a
n
y
er
r
o
r
s
in
r
ec
o
g
n
iz
in
g
b
o
th
p
o
s
itiv
e
an
d
n
e
g
ati
v
e
d
ata.
I
n
ad
d
itio
n
,
th
e
T
N
v
al
u
e
i
s
also
q
u
ite
lo
w
co
m
p
ar
ed
to
th
e
m
o
d
el
w
it
h
f
ea
tu
r
e
s
.
Nex
t,
th
e
C
NN
w
it
h
W
o
r
d
2
Vec
s
ec
tio
n
h
a
s
T
P
=5
1
2
,
F
P
=9
8
,
FN=
7
0
,
an
d
T
N=
2
2
0
.
T
h
e
ad
d
it
io
n
o
f
W
o
r
d
2
Vec
s
ig
n
i
f
ica
n
tl
y
i
n
cr
ea
s
e
s
th
e
n
u
m
b
er
o
f
T
P
to
5
1
2
,
in
d
icatin
g
th
at
th
e
m
o
d
el
is
b
etter
ab
le
to
r
ec
o
g
n
ize
p
o
s
itiv
e
d
ata
th
an
w
it
h
o
u
t
f
ea
t
u
r
es.
Ho
w
e
v
er
,
th
e
FP
(
9
8
)
an
d
FN (
7
0
)
v
alu
e
s
ar
e
s
ti
ll q
u
i
te
h
i
g
h
,
wh
ich
m
ea
n
s
th
er
e
i
s
r
o
o
m
f
o
r
i
m
p
r
o
v
e
m
e
n
t
in
r
ec
o
g
n
iz
in
g
n
e
g
ati
v
e
d
ata.
T
h
e
d
e
cr
ea
s
e
in
th
e
n
u
m
b
er
o
f
T
Ns
co
m
p
ar
ed
to
w
it
h
o
u
t
f
ea
t
u
r
es
also
s
h
o
w
s
t
h
at
th
e
m
o
d
el
is
s
lig
h
tl
y
les
s
ef
f
ec
ti
v
e
in
r
ec
o
g
n
izi
n
g
n
eg
at
iv
e
d
at
a.
I
n
th
e
C
NN
w
it
h
Glo
Ve
s
ec
tio
n
,
t
h
er
e
ar
e
v
al
u
es
o
f
T
P
=4
4
6
,
FP
=8
8
,
FN=8
7
,
an
d
T
N=
2
7
9
.
T
h
is
m
ea
n
s
th
at
Glo
Ve
p
r
o
v
id
es
m
o
r
e
b
alan
ce
d
r
es
u
lts
t
h
an
W
o
r
d
2
Vec
.
FP
d
ec
r
ea
s
es
to
8
8
,
w
h
ile
T
N
i
n
cr
ea
s
es si
g
n
if
ica
n
t
l
y
to
2
7
9
,
in
d
icati
n
g
a
b
etter
ab
ilit
y
to
r
ec
o
g
n
ize
n
eg
ati
v
e
d
ata.
Ho
w
e
v
er
,
T
P
is
lo
w
er
t
h
a
n
W
o
r
d
2
Vec
(
4
4
6
v
s
.
5
1
2
)
,
an
d
FN
i
s
s
lig
h
tl
y
h
i
g
h
er
th
a
n
W
o
r
d
2
Vec
.
Fu
r
th
er
m
o
r
e,
C
NN
w
it
h
Fas
tT
ex
t
p
r
o
d
u
ce
s
v
alu
e
s
o
f
T
P
=4
6
3
,
F
P
=7
6
,
FN=7
4
,
an
d
T
N=
2
8
7
.
Fas
t
T
ex
t p
r
o
d
u
c
es th
e
b
est r
esu
lts
a
m
o
n
g
all
m
eth
o
d
s
.
W
ith
h
i
g
h
T
P (
4
6
3
)
an
d
lo
w
F
N
(
7
4
)
,
th
e
m
o
d
el
i
s
v
er
y
e
f
f
ec
ti
v
e
in
r
ec
o
g
n
izi
n
g
p
o
s
iti
v
e
d
ata.
I
n
ad
d
itio
n
,
FP
is
t
h
e
lo
w
est
(
7
6
)
,
an
d
T
N
is
th
e
h
ig
h
e
s
t
(
2
8
7
)
,
in
d
icatin
g
t
h
at
t
h
i
s
m
o
d
el
is
also
v
er
y
g
o
o
d
at
r
ec
o
g
n
i
zin
g
n
e
g
ati
v
e
d
ata.
T
ab
l
e
5
.
C
o
n
f
u
s
io
n
m
atr
i
x
o
f
C
NN
Ex
p
e
r
i
me
n
t
TP
FP
FN
TN
C
N
N
w
i
t
h
o
u
t
f
e
a
t
u
r
e
4
0
4
1
2
8
1
1
9
2
4
9
C
N
N
w
i
t
h
W
o
r
d
2
V
e
c
5
1
2
98
70
2
2
0
C
N
N
w
i
t
h
G
l
o
V
e
4
4
6
88
87
2
7
9
C
N
N
w
i
t
h
F
a
st
T
e
x
t
4
6
3
76
74
2
8
7
T
ab
le
6
is
a
s
u
m
m
ar
y
tab
le
o
f
th
e
r
es
u
lts
o
f
C
N
N
ex
p
er
i
m
en
t
s
w
it
h
th
r
ee
v
ar
ian
t
s
o
f
w
o
r
d
em
b
ed
d
i
n
g
f
ea
t
u
r
es
an
d
o
n
e
w
it
h
o
u
t
f
ea
t
u
r
es.
T
h
is
m
o
d
el
s
h
o
w
s
t
h
e
b
est
i
m
p
r
o
v
e
m
e
n
t
co
m
p
ar
ed
to
th
e
b
aseli
n
e
m
o
d
el
.
Fas
tT
ex
t
p
r
o
v
id
es
a
v
er
y
g
o
o
d
b
alan
ce
b
et
w
ee
n
p
r
ec
is
io
n
,
r
ec
all
,
an
d
F1
-
Sco
r
e,
w
it
h
e
x
ce
llen
t
r
es
u
lt
s
i
n
all
m
etr
ics.
T
h
e
u
s
e
o
f
e
m
b
ed
d
in
g
f
ea
t
u
r
es
s
u
c
h
as
W
o
r
d
2
Vec
,
Glo
Ve,
an
d
Fas
tT
ex
t
s
ig
n
i
f
ica
n
t
l
y
i
m
p
r
o
v
es
m
o
d
el
p
er
f
o
r
m
a
n
ce
co
m
p
ar
ed
to
th
e
b
aselin
e
m
o
d
el
th
at
d
o
es
n
o
t
u
s
e
ad
d
itio
n
al
f
ea
t
u
r
es.
T
h
e
C
NN
w
i
t
h
Fas
tT
ex
t
m
o
d
el
h
as
th
e
h
ig
h
es
t
F1
-
Sco
r
e
(
8
6
.
0
6
%),
w
h
ic
h
s
h
o
w
s
a
n
o
p
tim
a
l
b
alan
ce
b
et
w
ee
n
p
r
ed
ictio
n
ac
cu
r
ac
y
a
n
d
th
e
ab
ilit
y
to
ca
p
tu
r
e
p
o
s
itiv
e
ca
s
es,
in
d
icati
n
g
th
at
th
is
m
o
d
el
is
v
er
y
ef
f
ec
ti
v
e
in
b
alan
ci
n
g
b
o
th
asp
ec
ts
.
T
h
e
u
s
e
o
f
F1
-
Sco
r
e
in
t
h
is
ca
s
e
is
b
ec
au
s
e
w
e
w
a
n
t to
m
ain
tain
a
b
alan
ce
b
et
w
ee
n
ac
c
u
r
ac
y
a
n
d
p
r
ec
is
io
n
an
d
th
e
m
o
d
el
’
s
ab
ilit
y
to
f
i
n
d
all
p
o
s
itiv
e
c
lass
e
s
.
T
h
e
C
N
N
w
it
h
Fas
tT
ex
t
m
o
d
el
is
t
h
e
b
es
t
c
h
o
ice
f
o
r
th
i
s
m
o
d
el,
w
it
h
s
i
g
n
if
ican
t i
m
p
r
o
v
e
m
e
n
t
s
in
all
m
etr
ic
s
,
esp
ec
iall
y
in
r
ec
a
ll a
n
d
p
r
ec
is
io
n
.
T
h
u
s
,
Fas
t
T
ex
t p
r
o
v
id
es b
etter
r
esu
lt
s
t
h
an
o
t
h
er
e
m
b
ed
d
i
n
g
f
ea
tu
r
e
s
s
u
c
h
as
W
o
r
d
2
Vec
an
d
Glo
Ve
i
n
o
p
ti
m
izi
n
g
tex
t
c
lass
if
ica
tio
n
p
er
f
o
r
m
a
n
ce
.
T
ab
l
e
6
.
C
o
m
p
ar
is
o
n
o
f
C
NN
ev
alu
a
tio
n
v
alu
e
s
w
it
h
w
o
r
d
em
b
ed
d
i
n
g
Ex
p
e
r
i
me
n
t
A
c
c
u
r
a
c
y
P
r
e
c
i
ssi
o
n
R
e
c
a
l
l
F1
-
S
c
o
r
e
C
N
N
w
i
t
h
o
u
t
f
e
a
t
u
r
e
72
.
5
5
5
5
5
5
5
6
75
.
9
3
9
8
5
77
.
2
4
6
6
5
76
.
5
8
7
6
8
C
N
N
w
i
t
h
W
o
r
d
2
V
e
c
81
.
3
3
3
3
3
3
3
3
83
.
9
3
4
4
3
87
.
9
7
2
5
1
85
.
9
0
6
0
4
C
N
N
w
i
t
h
G
l
o
V
e
80
.
5
5
5
5
5
5
5
6
83
.
5
2
0
6
83
.
6
7
7
3
83
.
5
9
8
8
8
C
N
N
w
i
t
h
F
a
st
T
e
x
t
83
.
3
3
3
3
3
3
3
3
85
.
8
9
9
8
1
86
.
2
1
9
7
4
86
.
0
5
9
4
8
T
h
is
s
tu
d
y
e
x
a
m
i
n
es
t
h
e
i
m
p
ac
t
o
f
p
er
f
o
r
m
an
ce
i
m
p
r
o
v
e
m
en
ts
,
co
m
p
u
tatio
n
a
ll
y
B
L
S
T
M
is
v
er
y
ef
f
icien
t,
th
is
is
b
ec
au
s
e
t
h
e
B
L
ST
M
p
r
o
ce
s
s
o
cc
u
r
s
s
eq
u
en
t
iall
y
a
n
d
r
eg
u
lar
l
y
,
m
ak
in
g
it
s
u
itab
le
f
o
r
p
r
o
ce
s
s
in
g
lo
n
g
te
x
ts
a
n
d
lar
g
e
d
atasets
.
W
ith
t
h
e
w
o
r
d
e
m
b
ed
d
in
g
f
ea
tu
r
e,
B
L
ST
M
ca
n
ca
p
tu
r
e
m
o
r
e
in
ter
ac
tio
n
s
b
et
w
ee
n
f
ea
t
u
r
es
t
h
at
m
a
y
b
e
i
g
n
o
r
ed
b
y
C
R
F
a
n
d
C
NN.
W
h
ile
p
r
ev
io
u
s
s
t
u
d
ies
h
a
v
e
in
v
est
ig
ate
d
th
e
i
m
p
ac
t
o
f
o
th
er
f
ea
t
u
r
es
o
f
th
e
s
a
m
e
m
et
h
o
d
.
th
e
s
tu
d
y
d
id
n
o
t
ex
p
licitl
y
d
is
c
u
s
s
th
eir
ef
f
ec
t
on
co
m
p
u
tatio
n
al
p
er
f
o
r
m
a
n
ce
.
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KOM
NI
K
A
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l
C
o
n
tr
o
l
C
o
mp
a
r
is
o
n
o
f wo
r
d
emb
ed
d
i
n
g
fea
tu
r
es u
s
in
g
d
ee
p
lea
r
n
in
g
in
s
en
timen
t a
n
a
lysi
s
(
Ja
s
m
i
r
)
423
B
ased
o
n
th
e
r
es
u
lt
s
o
f
t
h
e
t
h
r
e
e
ex
p
er
i
m
e
n
ts
,
t
h
e
B
L
ST
M
alg
o
r
ith
m
ac
h
ie
v
ed
th
e
h
i
g
h
est
ac
cu
r
ac
y
o
f
7
9
.
5
%,
w
h
ile
t
h
e
C
NN
al
g
o
r
it
h
m
r
ec
o
r
d
ed
th
e
lo
w
est
ac
c
u
r
ac
y
o
f
7
2
.
5
%
b
ef
o
r
e
th
e
w
o
r
d
e
m
b
ed
d
in
g
f
ea
t
u
r
e
w
a
s
ap
p
lied
.
A
f
ter
co
m
b
i
n
i
n
g
w
o
r
d
em
b
ed
d
in
g
,
B
L
ST
M
w
it
h
th
e
W
o
r
d
2
Vec
f
ea
tu
r
e
a
ch
iev
ed
th
e
h
ig
h
es
t
ac
cu
r
ac
y
o
f
8
7
.
1
%,
w
h
ile
th
e
l
o
w
est
ac
c
u
r
ac
y
p
o
s
t
-
e
m
b
ed
d
in
g
w
as
also
s
ee
n
in
B
L
ST
M
u
s
in
g
th
e
Glo
Ve
an
d
Fas
tT
ex
t
f
ea
tu
r
e
s
.
B
y
r
e
v
ie
win
g
all
cla
s
s
if
ica
tio
n
e
v
al
u
ati
o
n
m
e
tr
ics
—
ac
c
u
r
ac
y
,
p
r
ec
is
io
n
,
r
ec
all,
an
d
F1
s
co
r
e
—
B
L
ST
M
e
m
er
g
ed
a
s
t
h
e
b
est
p
er
f
o
r
m
in
g
al
g
o
r
ith
m
,
co
n
s
is
te
n
tl
y
p
r
o
d
u
cin
g
s
tab
l
e
r
esu
lt
s
ac
r
o
s
s
all
e
m
b
ed
d
in
g
s
.
Ho
w
e
v
er
,
all
test
s
s
till
a
llo
w
s
o
m
e
f
alse
p
o
s
iti
v
es
a
n
d
f
al
s
e
n
eg
at
iv
e
s
,
in
d
icati
n
g
p
o
ten
tia
l
ar
ea
s
f
o
r
f
u
r
t
h
er
r
esear
c
h
,
s
u
ch
as
m
i
n
i
m
izi
n
g
t
h
ese
er
r
o
r
s
.
A
d
d
itio
n
al
ac
cu
r
ac
y
i
m
p
r
o
v
e
m
e
n
ts
ca
n
b
e
ac
h
ie
v
ed
b
y
t
u
n
i
n
g
h
y
p
er
p
ar
a
m
eter
s
.
An
i
m
p
o
r
ta
n
t
o
b
s
er
v
atio
n
is
t
h
at,
b
ef
o
r
e
e
m
b
ed
d
in
g
,
C
NN
h
as
t
h
e
lo
west
p
er
f
o
r
m
a
n
ce
,
b
u
t
af
ter
ap
p
l
y
in
g
e
m
b
ed
d
i
n
g
,
esp
ec
iall
y
W
o
r
d
2
Vec
w
it
h
B
L
ST
M,
th
e
p
er
f
o
r
m
a
n
ce
i
m
p
r
o
v
es
s
ig
n
i
f
ica
n
tl
y
.
T
h
i
s
m
a
y
b
e
d
u
e
to
th
e
ch
ar
ac
ter
is
t
ics
o
f
C
NN
w
h
ic
h
ar
e
n
o
t
w
el
l
s
u
ited
f
o
r
tex
t
d
ata,
w
h
ile
B
L
ST
M,
w
h
ich
r
ea
d
s
s
eq
u
en
ce
s
b
id
ir
ec
tio
n
all
y
,
s
h
o
w
s
a
h
i
g
h
ab
ili
t
y
to
p
r
o
ce
s
s
te
x
t in
d
etail,
r
es
u
lti
n
g
in
s
u
p
er
i
o
r
p
er
f
o
r
m
an
ce
.
4.
CO
NCLU
SI
O
N
Ou
r
s
t
u
d
y
h
as
h
i
g
h
li
g
h
ted
t
h
e
ef
f
icac
y
o
f
p
r
e
-
tr
ai
n
ed
w
o
r
d
e
m
b
ed
d
i
n
g
m
o
d
els
i
n
s
en
t
i
m
e
n
t
an
al
y
s
i
s
.
T
h
r
o
u
g
h
a
s
er
ies
o
f
e
x
p
er
i
m
e
n
ts
,
w
e
h
a
v
e
d
e
m
o
n
s
tr
ated
th
e
ab
ilit
y
o
f
th
e
s
e
m
o
d
els
to
ac
h
iev
e
h
i
g
h
lev
el
s
o
f
ac
cu
r
ac
y
ac
r
o
s
s
d
i
v
er
s
e
tex
tu
a
l d
atasets
.
I
n
o
u
r
e
v
alu
a
tio
n
,
v
ar
io
u
s
d
ee
p
lear
n
in
g
m
eth
o
d
s
w
it
h
d
i
f
f
er
e
n
t
w
o
r
d
e
m
b
ed
d
in
g
f
ea
t
u
r
es
w
er
e
test
e
d
w
it
h
C
R
F,
B
L
ST
M,
an
d
C
N
N
alg
o
r
it
h
m
s
.
T
h
e
u
s
e
o
f
w
o
r
d
e
m
b
ed
d
in
g
f
ea
t
u
r
es
s
u
c
h
as
Fas
tT
ex
t,
W
o
r
d
2
Vec
,
an
d
Glo
Ve
co
n
s
is
te
n
tl
y
i
m
p
r
o
v
ed
th
e
p
er
f
o
r
m
a
n
ce
o
f
v
ar
io
u
s
tex
t
class
i
f
icat
io
n
m
o
d
el
s
o
n
C
R
F,
B
L
ST
M,
an
d
C
NN
co
m
p
ar
ed
to
m
o
d
els
w
i
th
o
u
t
f
ea
tu
r
e
s
.
Fas
tT
ex
t
w
a
s
i
d
en
tifie
d
as
th
e
b
est
f
ea
t
u
r
e
b
ased
o
n
th
e
tab
le
r
esu
lts
as
it
p
r
o
d
u
ce
d
th
e
m
o
s
t
b
al
an
ce
d
class
i
f
icat
io
n
w
it
h
m
i
n
i
m
al
er
r
o
r
.
Fas
tT
e
x
t
also
p
r
o
d
u
ce
d
h
ig
h
l
y
ac
cu
r
at
e
class
i
f
icatio
n
o
n
b
o
t
h
p
o
s
i
tiv
e
a
n
d
n
e
g
ati
v
e
d
ata.
W
o
r
d
2
Vec
ex
ce
lled
in
r
ec
o
g
n
izi
n
g
p
o
s
itiv
e
d
ata
b
u
t
t
en
d
ed
to
b
e
less
ac
c
u
r
ate
o
n
n
eg
ati
v
e
d
ata.
Fo
r
li
m
ited
co
m
p
u
tatio
n
a
l
r
eso
u
r
ce
s
,
Glo
Ve
ca
n
b
e
ch
o
s
en
as
it
p
r
o
v
id
es
b
ala
n
ce
d
r
esu
lt
s
w
it
h
lo
w
er
er
r
o
r
c
o
m
p
ar
ed
to
W
o
r
d
2
Vec
.
Glo
Ve
o
f
f
er
s
a
g
o
o
d
b
alan
ce
w
it
h
li
g
h
ter
co
m
p
u
tatio
n
al
r
eq
u
ir
e
m
e
n
t
s
,
s
u
itab
le
f
o
r
r
ed
u
ci
n
g
er
r
o
r
s
o
n
n
eg
at
i
v
e
d
ata.
T
h
e
ch
o
ice
o
f
w
o
r
d
e
m
b
ed
d
i
n
g
f
ea
t
u
r
es
u
s
ed
ca
n
b
e
tailo
r
ed
to
th
e
s
p
ec
if
ic
n
ee
d
s
o
f
t
h
e
m
o
d
el
an
d
th
e
cla
s
s
if
ica
tio
n
o
b
j
ec
tiv
es.
ACK
NO
WL
E
D
G
E
M
E
NT
S
W
e
w
o
u
ld
lik
e
to
th
an
k
Ya
y
asan
Di
n
a
m
ik
a
B
an
g
s
a
J
a
m
b
i
f
o
r
th
e
m
o
r
al
an
d
f
i
n
an
c
ial
s
u
p
p
o
r
t
in
co
m
p
let
in
g
t
h
i
s
r
esear
ch
,
a
n
d
w
o
u
ld
li
k
e
to
t
h
an
k
t
h
e
r
esear
ch
a
n
d
co
m
m
u
n
it
y
s
er
v
ice
in
s
titu
tio
n
,
U
n
iv
er
s
itas
Din
a
m
i
k
a
B
an
g
s
a
J
a
m
b
i f
o
r
it
s
f
ac
ili
ties
a
n
d
an
n
u
al
w
o
r
k
p
r
o
g
r
a
m
s
.
RE
F
E
R
E
NC
E
S
[
1
]
A
.
L
.
G
u
z
man
a
n
d
S
.
C
.
L
e
w
i
s,
“
A
r
t
i
f
i
c
i
a
l
i
n
t
e
l
l
i
g
e
n
c
e
a
n
d
c
o
mm
u
n
i
c
a
t
i
o
n
:
A
H
u
man
–
M
a
c
h
i
n
e
C
o
mm
u
n
i
c
a
t
i
o
n
r
e
se
a
r
c
h
a
g
e
n
d
a
,
”
N
e
w
M
e
d
i
a
&
S
o
c
i
e
t
y
,
v
o
l
.
2
2
,
n
o
.
1
,
p
p
.
7
0
–
8
6
,
2
0
2
0
,
d
o
i
:
1
0
.
1
1
7
7
/
1
4
6
1
4
4
4
8
1
9
8
5
8
6
9
1
.
[
2
]
B
.
J
i
ma
d
a
-
O
j
u
o
l
a
p
e
a
n
d
J.
T
e
h
,
“
I
mp
a
c
t
o
f
t
h
e
I
n
t
e
g
r
a
t
i
o
n
o
f
I
n
f
o
r
mat
i
o
n
a
n
d
C
o
mm
u
n
i
c
a
t
i
o
n
T
e
c
h
n
o
l
o
g
y
o
n
P
o
w
e
r
S
y
st
e
m
R
e
l
i
a
b
i
l
i
t
y
:
A
R
e
v
i
e
w
,
”
I
EEE
A
c
c
e
ss
,
v
o
l
.
8
,
p
p
.
2
4
6
0
0
–
2
4
6
1
5
,
2
0
2
0
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
0
.
2
9
7
0
5
9
8
.
[
3
]
R
.
L
o
z
a
n
o
-
B
l
a
sco
,
M
.
M
i
r
a
-
A
l
a
d
r
é
n
,
a
n
d
M
.
G
i
l
-
L
a
mat
a
,
“
S
o
c
i
a
l
me
d
i
a
i
n
f
l
u
e
n
c
e
o
n
y
o
u
n
g
p
e
o
p
l
e
a
n
d
c
h
i
l
d
r
e
n
:
A
n
a
l
y
si
s
o
n
I
n
st
a
g
r
a
m,
T
w
i
t
t
e
r
a
n
d
Y
o
u
T
u
b
e
,
”
C
o
m
u
n
i
c
a
r
,
v
o
l
.
3
0
,
n
o
.
7
4
,
p
p
.
1
1
7
–
1
2
8
,
2
0
2
3
,
d
o
i
:
1
0
.
3
9
1
6
/
C
7
4
-
2
0
2
3
-
1
0
.
[
4
]
B
.
T
.
K
.
,
C
.
S
.
R
.
A
n
n
a
v
a
r
a
p
u
,
a
n
d
A
.
B
a
b
l
a
n
i
,
“
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
so
c
i
a
l
me
d
i
a
a
n
a
l
y
si
s:
A
su
r
v
e
y
,
”
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
Re
v
i
e
w
,
v
o
l
.
4
0
,
p
.
1
0
0
3
9
5
,
2
0
2
1
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
c
o
sr
e
v
.
2
0
2
1
.
1
0
0
3
9
5
.
[
5
]
S
.
M
.
F
e
r
n
á
n
d
e
z
-
M
i
g
u
é
l
e
z
,
M
.
D
í
a
z
-
P
u
c
h
e
,
J.
A
.
C
a
m
p
o
s
-
S
o
r
i
a
,
a
n
d
F
.
G
a
l
á
n
-
V
a
l
d
i
v
i
e
so
,
“
T
h
e
i
mp
a
c
t
o
f
so
c
i
a
l
m
e
d
i
a
o
n
r
e
st
a
u
r
a
n
t
c
o
r
p
o
r
a
t
i
o
n
s
’
f
i
n
a
n
c
i
a
l
p
e
r
f
o
r
man
c
e
,
”
S
u
s
t
a
i
n
a
b
i
l
i
t
y
,
v
o
l
.
1
2
,
n
o
.
4
,
p
p
.
1
–
1
4
,
2
0
2
0
,
d
o
i
:
1
0
.
3
3
9
0
/
s
u
1
2
0
4
1
6
4
6
.
[
6
]
H
.
R
.
A
l
h
a
k
i
e
m
a
n
d
E
.
B
.
S
e
t
i
a
w
a
n
,
“
A
s
p
e
c
t
-
B
a
s
1
e
d
S
e
n
t
i
m
e
n
t
A
n
a
l
y
s
i
s
o
n
T
w
i
t
t
e
r
U
s
i
n
g
L
o
g
i
s
t
i
c
R
e
g
r
e
s
s
i
o
n
w
i
t
h
F
a
s
t
T
e
x
t
F
e
a
t
u
r
e
E
x
p
a
n
s
i
o
n
,
”
J
u
r
n
a
l
R
E
S
T
I
(
R
e
k
a
y
a
sa
S
i
st
e
m
d
a
n
T
e
k
n
o
l
o
g
i
I
n
f
o
rm
a
si
)
,
v
o
l
.
6
,
n
o
.
5
,
p
p
.
8
4
0
–
8
4
6
,
2
0
2
2
,
d
o
i
:
1
0
.
2
9
2
0
7
/
r
e
s
t
i
.
v
6
i
5
.
4
4
2
9
.
[
7
]
M
.
B
i
r
j
a
l
i
,
M
.
K
a
sr
i
,
a
n
d
A
.
B
e
n
i
-
H
s
san
e
,
“
A
c
o
mp
r
e
h
e
n
si
v
e
su
r
v
e
y
o
n
se
n
t
i
me
n
t
a
n
a
l
y
si
s:
A
p
p
r
o
a
c
h
e
s,
c
h
a
l
l
e
n
g
e
s
a
n
d
t
r
e
n
d
s,
”
K
n
o
w
l
e
d
g
e
-
B
a
se
d
S
y
s
t
e
m
s
,
v
o
l
.
2
2
6
,
p
.
1
0
7
1
3
4
,
2
0
2
1
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
k
n
o
sy
s.2
0
2
1
.
1
0
7
1
3
4
.
[
8
]
A
.
P
a
l
a
n
i
v
i
n
a
y
a
g
a
m,
C
.
Z
.
El
-
B
a
y
e
h
,
a
n
d
R
.
D
a
maše
v
i
č
i
u
s
,
“
T
w
e
n
t
y
Y
e
a
r
s
o
f
M
a
c
h
i
n
e
-
L
e
a
r
n
i
n
g
-
B
a
se
d
T
e
x
t
C
l
a
ssi
f
i
c
a
t
i
o
n
:
A
S
y
st
e
mat
i
c
R
e
v
i
e
w
,
”
Al
g
o
r
i
t
h
m
s
,
v
o
l
.
1
6
,
n
o
.
5
,
p
p
.
1
–
2
8
,
2
0
2
3
,
d
o
i
:
1
0
.
3
3
9
0
/
a
1
6
0
5
0
2
3
6
.
[
9
]
G
.
X
u
,
Y
.
M
e
n
g
,
X
.
Q
i
u
,
Z
.
Y
u
,
a
n
d
X
.
W
u
,
“
S
e
n
t
i
me
n
t
a
n
a
l
y
si
s
o
f
c
o
mm
e
n
t
t
e
x
t
s
b
a
se
d
o
n
B
i
L
S
T
M
,
”
I
E
EE
A
c
c
e
ss
,
v
o
l
.
7
,
p
p
.
5
1
5
2
2
–
5
1
5
3
2
,
2
0
1
9
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
1
9
.
2
9
0
9
9
1
9
.
[
1
0
]
J.
J
a
smir,
W
.
R
i
y
a
d
i
,
S
.
R
.
A
g
u
st
i
n
i
,
Y
.
A
r
v
i
t
a
,
D
.
M
e
i
sa
k
,
a
n
d
L
.
A
r
y
a
n
i
,
“
B
i
d
i
r
e
c
t
i
o
n
a
l
L
o
n
g
S
h
o
r
t
-
T
e
r
m
M
e
mo
r
y
a
n
d
W
o
r
d
Emb
e
d
d
i
n
g
F
e
a
t
u
r
e
f
o
r
,
”
J
u
rn
a
l
R
E
S
T
I
(
R
e
k
a
y
a
sa
S
i
s
t
e
m
d
a
n
T
e
k
n
o
l
o
g
i
I
n
f
o
rm
a
s
i
)
,
v
o
l
.
6
,
n
o
.
4
,
p
p
.
5
0
5
–
5
1
0
,
2
0
2
2
,
d
o
i
:
1
0
.
2
9
2
0
7
/
r
e
st
i
.
v
6
i
4
.
4
0
0
5
.
[1
1]
S
.
R
u
d
e
r
,
I
.
V
u
l
i
ć
,
a
n
d
A
.
S
ø
g
a
a
r
d
,
“
A
S
u
r
v
e
y
o
f
C
r
o
ss
-
l
i
n
g
u
a
l
W
o
r
d
E
mb
e
d
d
i
n
g
M
o
d
e
l
s,
”
J
o
u
r
n
a
l
o
f
Ar
t
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
Re
se
a
rc
h
,
v
o
l
.
6
5
,
p
p
.
5
6
9
–
6
3
0
,
A
u
g
.
2
0
1
9
,
d
o
i
:
1
0
.
1
6
1
3
/
j
a
i
r
.
1
.
1
1
6
4
0
.
[
1
2
]
Z
.
Z
h
u
a
n
g
,
Z
.
L
i
a
n
g
,
Y
.
R
a
o
,
H
.
X
i
e
,
a
n
d
F
.
L
.
W
a
n
g
,
“
O
u
t
-
of
-
v
o
c
a
b
u
l
a
r
y
w
o
r
d
e
mb
e
d
d
i
n
g
l
e
a
r
n
i
n
g
b
a
se
d
o
n
r
e
a
d
i
n
g
c
o
mp
r
e
h
e
n
si
o
n
me
c
h
a
n
i
sm,
”
N
a
t
u
ra
l
L
a
n
g
u
a
g
e
Pro
c
e
ss
i
n
g
J
o
u
r
n
a
l
,
v
o
l
.
5
,
n
o
.
A
u
g
u
s
t
,
p
.
1
0
0
0
3
8
,
2
0
2
3
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
n
l
p
.
2
0
2
3
.
1
0
0
0
3
8
.
[
1
3
]
S
.
R
a
p
a
c
z
,
P
.
C
h
o
ł
d
a
,
a
n
d
M
.
N
a
t
k
a
n
i
e
c
,
“
A
me
t
h
o
d
f
o
r
f
a
st
se
l
e
c
t
i
o
n
o
f
mac
h
i
n
e
-
l
e
a
r
n
i
n
g
c
l
a
ss
i
f
i
e
r
s
f
o
r
sp
a
m
f
i
l
t
e
r
i
n
g
,
”
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
1
6
9
3
-
6930
T
E
L
KOM
NI
K
A
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l
C
o
n
tr
o
l
,
Vo
l.
2
3
,
No
.
2
,
A
p
r
il
20
2
5
:
4
1
6
-
425
424
El
e
c
t
r
o
n
i
c
s
,
v
o
l
.
1
0
,
n
o
.
1
7
,
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
1
7
2
0
8
3
.
[
1
4
]
F
.
N
.
N
.
H
.
R
.
P
a
ssare
l
l
a
,
S
.
N
u
r
ma
i
n
i
,
M
.
N
.
R
a
c
h
ma
t
u
l
l
a
h
,
a
n
d
H
.
V
e
n
y
,
“
D
e
v
e
l
o
p
me
n
t
o
f
a
mac
h
i
n
e
l
e
a
r
n
i
n
g
mo
d
e
l
f
o
r
p
r
e
d
i
c
t
i
n
g
a
b
n
o
r
mal
i
t
i
e
s o
f
c
o
mm
e
r
c
i
a
l
a
i
r
p
l
a
n
e
s,
”
D
a
t
a
S
c
i
e
n
c
e
a
n
d
Ma
n
a
g
e
m
e
n
t
,
p
.
1
0
0
1
3
7
,
2
0
2
3
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
j
sa
md
.
2
0
2
3
.
1
0
0
6
1
3
.
[1
5
]
A
.
T
a
v
a
n
a
e
i
,
M
.
G
h
o
d
r
a
t
i
,
S
.
R
.
K
h
e
r
a
d
p
i
s
h
e
h
,
T
.
M
a
s
q
u
e
l
i
e
r
,
a
n
d
A
.
M
a
i
d
a
,
“
D
e
e
p
l
e
a
r
n
i
n
g
i
n
sp
i
k
i
n
g
n
e
u
r
a
l
n
e
t
w
o
r
k
s,
”
N
e
u
ra
l
N
e
t
w
o
rks
,
v
o
l
.
1
1
1
,
p
p
.
4
7
–
6
3
,
2
0
1
9
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
n
e
u
n
e
t
.
2
0
1
8
.
1
2
.
0
0
2
.
[1
6
]
J.
Jasm
i
r
,
S
.
N
u
r
ma
i
n
i
,
R
.
F
.
M
a
l
i
k
,
a
n
d
B
.
T
u
t
u
k
o
,
“
B
i
g
r
a
m
f
e
a
t
u
r
e
e
x
t
r
a
c
t
i
o
n
a
n
d
c
o
n
d
i
t
i
o
n
a
l
r
a
n
d
o
m
f
i
e
l
d
s
mo
d
e
l
t
o
i
mp
r
o
v
e
t
e
x
t
c
l
a
ssi
f
i
c
a
t
i
o
n
c
l
i
n
i
c
a
l
t
r
i
a
l
d
o
c
u
m
e
n
t
,
”
T
ELKO
M
N
I
K
A
(
T
e
l
e
c
o
m
m
u
n
i
c
a
t
i
o
n
,
C
o
m
p
u
t
i
n
g
,
E
l
e
c
t
r
o
n
i
c
s
a
n
d
C
o
n
t
r
o
l
),
v
o
l
.
1
9
,
n
o
.
3
,
p
p
.
8
8
6
–
8
9
2
,
2
0
2
1
,
d
o
i
:
1
0
.
1
2
9
2
8
/
T
EL
K
O
M
N
I
K
A
.
v
1
9
i
3
.
1
8
3
5
7
.
[1
7
]
G
.
L
i
u
a
n
d
J.
G
u
o
,
“
B
i
d
i
r
e
c
t
i
o
n
a
l
L
S
TM
w
i
t
h
a
t
t
e
n
t
i
o
n
me
c
h
a
n
i
sm
a
n
d
c
o
n
v
o
l
u
t
i
o
n
a
l
l
a
y
e
r
f
o
r
t
e
x
t
c
l
a
ssi
f
i
c
a
t
i
o
,
”
N
e
u
ro
c
o
m
p
u
t
i
n
g
,
2
0
1
9
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
n
e
u
c
o
m.2
0
1
9
.
0
1
.
0
7
8
.
[1
8
]
M
.
A
k
b
a
r
,
S
.
N
u
r
ma
i
n
i
,
a
n
d
R
.
U
.
P
a
r
t
a
n
,
“
T
h
e
d
e
e
p
c
o
n
v
o
l
u
t
i
o
n
a
l
n
e
t
w
o
r
k
s
f
o
r
t
h
e
c
l
a
ssi
f
i
c
a
t
i
o
n
o
f
mu
l
t
i
-
c
l
a
ss
a
r
r
h
y
t
h
m
i
a
,
”
Bu
l
l
e
t
i
n
o
f
El
e
c
t
ri
c
a
l
E
n
g
i
n
e
e
ri
n
g
a
n
d
I
n
f
o
rm
a
t
i
c
s
,
v
o
l
.
1
3
,
n
o
.
2
,
p
p
.
1
3
2
5
–
1
3
3
3
,
2
0
2
4
,
d
o
i
:
1
0
.
1
1
5
9
1
/
e
e
i
.
v
1
3
i
2
.
6
1
0
2
.
[
19
]
R
.
R
a
h
m
a
n
d
a
a
n
d
E
.
B
.
S
e
t
i
a
w
a
n
,
“
W
o
r
d
2
V
e
c
o
n
S
e
n
t
i
me
n
t
A
n
a
l
y
si
s
w
i
t
h
S
y
n
t
h
e
t
i
c
M
i
n
o
r
i
t
y
O
v
e
r
samp
l
i
n
g
T
e
c
h
n
i
q
u
e
a
n
d
B
o
o
st
i
n
g
A
l
g
o
r
i
t
h
m,
”
J
u
r
n
a
l
RE
S
T
I
(
Re
k
a
y
a
s
a
S
i
st
e
m
d
a
n
T
e
k
n
o
l
o
g
i
I
n
f
o
rm
a
s
i
)
,
v
o
l
.
6
,
n
o
.
4
,
p
p
.
5
9
9
–
6
0
5
,
2
0
2
2
,
d
o
i
:
1
0
.
2
9
2
0
7
/
r
e
st
i
.
v
6
i
4
.
4
1
8
6
.
[2
0
]
A
.
G
e
o
r
g
e
,
H
.
B
.
B
.
G
a
n
e
sh
,
M
.
A
.
K
u
mar,
a
n
d
K
.
P
.
S
o
ma
n
,
S
i
g
n
i
f
i
c
a
n
c
e
o
f
g
l
o
b
a
l
v
e
c
t
o
rs
re
p
res
e
n
t
a
t
i
o
n
i
n
p
ro
t
e
i
n
se
q
u
e
n
c
e
s
a
n
a
l
y
si
s
,
S
p
r
i
n
g
e
r
I
n
t
e
r
n
a
t
i
o
n
a
l
P
u
b
l
i
s
h
i
n
g
,
v
o
l
.
3
1
,
2
0
1
9
,
d
o
i
:
1
0
.
1
0
0
7
/
9
7
8
-
3
-
0
3
0
-
0
4
0
6
1
-
1
_
2
7
.
[2
1
]
I
.
N
.
K
h
a
sa
n
a
h
,
“
S
e
n
t
i
me
n
t
C
l
a
ss
i
f
i
c
a
t
i
o
n
U
s
i
n
g
f
a
s
t
T
e
x
t
Emb
e
d
d
i
n
g
a
n
d
D
e
e
p
L
e
a
r
n
i
n
g
M
o
d
e
l
,
”
Pr
o
c
e
d
i
a
C
I
RP
,
v
o
l
.
1
8
9
,
p
p
.
3
4
3
–
3
5
0
,
2
0
2
1
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
p
r
o
c
s.2
0
2
1
.
0
5
.
1
0
3
.
[2
2
]
M
.
A
l
-
S
mad
i
,
B
.
T
a
l
a
f
h
a
,
M
.
A
l
-
A
y
y
o
u
b
,
a
n
d
Y
.
Jar
a
r
w
e
h
,
“
U
si
n
g
l
o
n
g
sh
o
r
t
-
t
e
r
m
me
mo
r
y
d
e
e
p
n
e
u
r
a
l
n
e
t
w
o
r
k
s
f
o
r
a
sp
e
c
t
-
b
a
se
d
se
n
t
i
m
e
n
t
a
n
a
l
y
si
s o
f
A
r
a
b
i
c
r
e
v
i
e
w
s,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
rn
a
l
o
f
Ma
c
h
i
n
e
L
e
a
rn
i
n
g
a
n
d
C
y
b
e
rn
e
t
i
c
s
,
v
o
l
.
1
0
,
n
o
.
8
,
p
p
.
2
1
6
3
–
2
1
7
5
,
2
0
1
9
,
d
o
i
:
1
0
.
1
0
0
7
/
s1
3
0
4
2
-
0
1
8
-
0
7
9
9
-
4.
[2
3
]
B
.
J
a
n
g
,
M
.
K
i
m,
G
.
H
a
r
e
r
i
man
a
,
S
.
U
.
K
a
n
g
,
a
n
d
J.
W
.
K
i
m,
“
Bi
-
L
S
T
M
mo
d
e
l
t
o
i
n
c
r
e
a
se
a
c
c
u
r
a
c
y
i
n
t
e
x
t
c
l
a
ssi
f
i
c
a
t
i
o
n
:
C
o
mb
i
n
i
n
g
w
o
r
d
2
v
e
c
C
N
N
a
n
d
a
t
t
e
n
t
i
o
n
me
c
h
a
n
i
sm
,
”
A
p
p
l
i
e
d
S
c
i
e
n
c
e
s
,
v
o
l
.
1
0
,
n
o
.
1
7
,
p
.
5
8
4
1
,
2
0
2
0
,
d
o
i
:
1
0
.
3
3
9
0
/
a
p
p
1
0
1
7
5
8
4
1
.
[2
4
]
S
.
I
f
t
i
k
h
a
r
,
B
.
A
l
l
u
h
a
y
b
i
,
M
.
S
u
l
i
ma
n
,
A
.
S
a
e
e
d
,
a
n
d
K
.
F
a
t
i
ma
,
“
A
maz
o
n
p
r
o
d
u
c
t
s
r
e
v
i
e
w
s
c
l
a
ssi
f
i
c
a
t
i
o
n
b
a
se
d
o
n
mac
h
i
n
e
l
e
a
r
n
i
n
g
,
d
e
e
p
l
e
a
r
n
i
n
g
me
t
h
o
d
s
a
n
d
B
ER
T
,
”
T
ELKO
M
N
I
K
A
(
T
e
l
e
c
o
m
m
u
n
i
c
a
t
i
o
n
C
o
m
p
u
t
i
n
g
E
l
e
c
t
r
o
n
i
c
s
a
n
d
C
o
n
t
r
o
l
)
,
v
o
l
.
2
1
,
n
o
.
5
,
p
p
.
1
0
8
4
–
1
1
0
1
,
2
0
2
3
,
d
o
i
:
1
0
.
1
2
9
2
8
/
T
EL
K
O
M
N
I
K
A
.
v
2
1
i
5
.
2
4
0
4
6
.
[2
5
]
N
.
B
a
d
r
i
,
F
.
K
b
o
u
b
i
,
a
n
d
A
.
H
.
C
h
a
i
b
i
,
“
C
o
mb
i
n
i
n
g
F
a
st
T
e
x
t
a
n
d
G
l
o
v
e
W
o
r
d
Emb
e
d
d
i
n
g
f
o
r
O
f
f
e
n
si
v
e
a
n
d
H
a
t
e
sp
e
e
c
h
T
e
x
t
D
e
t
e
c
t
i
o
n
,
”
Pr
o
c
e
d
i
a
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
,
v
o
l
.
2
0
7
,
p
p
.
7
6
9
–
7
7
8
,
2
0
2
2
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
p
r
o
c
s
.
2
0
2
2
.
0
9
.
1
3
2
.
[2
6
]
D
.
Ja
t
n
i
k
a
,
M
.
A
.
B
i
j
a
k
sa
n
a
,
a
n
d
A
.
A
.
S
u
r
y
a
n
i
,
“
W
o
r
d
2
v
e
c
mo
d
e
l
a
n
a
l
y
s
i
s
f
o
r
se
man
t
i
c
s
i
mi
l
a
r
i
t
i
e
s
i
n
E
n
g
l
i
sh
w
o
r
d
s,
”
Pr
o
c
e
d
i
a
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
,
v
o
l
.
1
5
7
,
p
p
.
1
6
0
–
1
6
7
,
2
0
1
9
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
p
r
o
c
s.
2
0
1
9
.
0
8
.
1
5
3
.
[2
7
]
M
.
A
.
R
a
i
h
a
n
a
n
d
E
.
B
.
S
e
t
i
a
w
a
n
,
“
A
sp
e
c
t
B
a
se
d
S
e
n
t
i
me
n
t
A
n
a
l
y
si
s
w
i
t
h
F
a
st
T
e
x
t
F
e
a
t
u
r
e
Ex
p
a
n
si
o
n
a
n
d
S
u
p
p
o
r
t
V
e
c
t
o
r
M
a
c
h
i
n
e
M
e
t
h
o
d
o
n
T
w
i
t
t
e
r
,
”
J
u
r
n
a
l
RE
S
T
I
(
Re
k
a
y
a
sa
S
i
s
t
e
m
d
a
n
T
e
k
n
o
l
o
g
i
I
n
f
o
rm
a
s
i
)
,
v
o
l
.
6
,
n
o
.
4
,
p
p
.
5
9
1
–
5
9
8
,
2
0
2
2
,
d
o
i
:
1
0
.
2
9
2
0
7
/
r
e
st
i
.
v
6
i
4
.
4
1
8
7
.
[2
8
]
Q
.
Z
h
a
n
g
,
Y
.
C
a
o
,
a
n
d
H
.
Y
u
,
“
P
a
r
si
n
g
c
i
t
a
t
i
o
n
s
i
n
b
i
o
me
d
i
c
a
l
a
r
t
i
c
l
e
s
u
si
n
g
c
o
n
d
i
t
i
o
n
a
l
r
a
n
d
o
m
f
i
e
l
d
s,
”
C
o
m
p
u
t
e
rs
i
n
b
i
o
l
o
g
y
a
n
d
m
e
d
i
c
i
n
e
,
v
o
l
.
4
1
,
n
o
.
4
,
p
p
.
1
9
0
–
1
9
4
,
2
0
1
1
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
c
o
m
p
b
i
o
me
d
.
2
0
1
1
.
0
2
.
0
0
5
.
[
29
]
W
.
L
e
e
,
K
.
K
i
m,
E.
Y
.
L
e
e
,
a
n
d
J.
C
h
o
i
,
“
C
o
n
d
i
t
i
o
n
a
l
r
a
n
d
o
m
f
i
e
l
d
s
f
o
r
c
l
i
n
i
c
a
l
n
a
me
d
e
n
t
i
t
y
r
e
c
o
g
n
i
t
i
o
n
:
A
c
o
mp
a
r
a
t
i
v
e
st
u
d
y
u
si
n
g
K
o
r
e
a
n
c
l
i
n
i
c
a
l
t
e
x
t
s,
”
C
o
m
p
u
t
e
rs
i
n
Bi
o
l
o
g
y
a
n
d
Me
d
i
c
i
n
e
,
v
o
l
.
1
0
,
p
p
.
7
–
1
4
,
2
0
1
8
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
c
o
mp
b
i
o
me
d
.
2
0
1
8
.
0
7
.
0
1
9
.
[3
0
]
P
.
C
o
r
c
o
r
a
n
,
P
.
M
o
o
n
e
y
,
a
n
d
M
.
B
e
r
t
o
l
o
t
t
o
,
“
L
i
n
e
a
r
st
r
e
e
t
e
x
t
r
a
c
t
i
o
n
u
si
n
g
a
C
o
n
d
i
t
i
o
n
a
l
R
a
n
d
o
m
F
i
e
l
d
mo
d
e
l
,
”
S
p
a
t
i
a
l
S
t
a
t
i
st
i
c
s
v
o
l
,
v
o
l
.
1
4
,
p
p
.
5
3
2
–
5
4
5
,
2
0
1
5
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
s
p
a
s
t
a
.
2
0
1
5
.
1
0
.
0
0
3
.
[3
1
]
C
.
J
i
a
n
g
,
M
.
M
a
d
d
e
l
a
,
W
.
L
a
n
,
Y
.
Z
h
o
n
g
,
a
n
d
W
.
X
u
,
“
N
e
u
r
a
l
C
R
F
mo
d
e
l
f
o
r
se
n
t
e
n
c
e
a
l
i
g
n
me
n
t
i
n
t
e
x
t
si
m
p
l
i
f
i
c
a
t
i
o
n
,
”
Pro
c
e
e
d
i
n
g
s
o
f
t
h
e
5
8
t
h
A
n
n
u
a
l
Me
e
t
i
n
g
o
f
t
h
e
Ass
o
c
i
a
t
i
o
n
f
o
r
C
o
m
p
u
t
a
t
i
o
n
a
l
L
i
n
g
u
i
s
t
i
c
s
,
2
0
2
0
,
p
p
.
7
9
4
3
–
7
9
6
0
,
d
o
i
:
1
0
.
1
8
6
5
3
/
v
1
/
2
0
2
0
.
a
c
l
-
mai
n
.
7
0
9
.
[3
2
]
K
.
S
.
T
a
i
,
R
.
S
o
c
h
e
r
,
a
n
d
C
.
D
.
M
a
n
n
i
n
g
,
“
I
mp
r
o
v
e
d
S
e
man
t
i
c
R
e
p
r
e
se
n
t
a
t
i
o
n
s
F
r
o
m
T
r
e
e
-
S
t
r
u
c
t
u
r
e
d
L
o
n
g
S
h
o
r
t
-
T
e
r
m
M
e
mo
r
y
N
e
t
w
o
r
k
s
,”
a
rXi
v
p
re
p
ri
n
t
,
2
0
1
5
,
d
o
i
:
1
0
.
4
8
5
5
0
/
a
r
X
i
v
.
1
5
0
3
.
0
0
0
7
5
.
[
33
]
Z
.
D
a
i
,
X
.
W
a
n
g
,
P
.
N
i
,
Y
.
L
i
,
G
.
L
i
,
a
n
d
X
.
B
a
i
,
“
N
a
me
d
E
n
t
i
t
y
R
e
c
o
g
n
i
t
i
o
n
U
si
n
g
B
ER
T
B
i
L
S
T
M
C
R
F
f
o
r
C
h
i
n
e
se
El
e
c
t
r
o
n
i
c
H
e
a
l
t
h
R
e
c
o
r
d
s,
”
2
0
1
9
1
2
t
h
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
g
re
ss
o
n
I
m
a
g
e
a
n
d
S
i
g
n
a
l
Pro
c
e
ssi
n
g
,
B
i
o
Me
d
i
c
a
l
E
n
g
i
n
e
e
ri
n
g
a
n
d
I
n
f
o
rm
a
t
i
c
s
(
C
I
S
P
-
B
ME
I
)
,
S
u
z
h
o
u
,
C
h
i
n
a
,
2
0
1
9
,
p
p
.
1
–
5
,
d
o
i
:
1
0
.
1
1
0
9
/
C
I
S
P
-
B
M
E
I
4
8
8
4
5
.
2
0
1
9
.
8
9
6
5
8
2
3
.
[3
4
]
D
.
T
.
P
u
t
r
a
a
n
d
E
.
B
.
S
e
t
i
a
w
a
n
,
“
S
e
n
t
i
me
n
t
A
n
a
l
y
si
s
o
n
S
o
c
i
a
l
M
e
d
i
a
w
i
t
h
G
l
o
v
e
U
si
n
g
C
o
mb
i
n
a
t
i
o
n
C
N
N
a
n
d
R
o
B
E
R
T
a
,
”
J
u
r
n
a
l
RES
T
I
(
R
e
k
a
y
a
s
a
S
i
s
t
e
m
d
a
n
T
e
k
n
o
l
o
g
i
I
n
f
o
rm
a
si
)
,
v
o
l
.
7
,
n
o
.
3
,
p
p
.
4
5
7
–
5
6
3
,
2
0
2
3
,
d
o
i
:
1
0
.
2
9
2
0
7
/
r
e
st
i
.
v
7
i
3
.
4
8
9
2
.
[3
5
]
J.
Z
h
a
n
g
,
F
.
L
i
u
,
W
.
X
u
,
a
n
d
H
.
Y
u
,
“
F
e
a
t
u
r
e
f
u
si
o
n
t
e
x
t
c
l
a
ss
i
f
i
c
a
t
i
o
n
mo
d
e
l
c
o
mb
i
n
i
n
g
C
N
N
a
n
d
B
i
G
R
U
w
i
t
h
mu
l
t
i
-
a
t
t
e
n
t
i
o
n
me
c
h
a
n
i
sm
,
”
Fu
t
u
re
I
n
t
e
rn
e
t
,
v
o
l
.
1
1
,
n
o
.
1
1
,
2
0
1
9
,
d
o
i
:
1
0
.
3
3
9
0
/
f
i
1
1
1
1
0
2
3
7
.
[3
6
]
J.
Y
a
o
,
C
.
W
a
n
g
,
C
.
H
u
,
a
n
d
X
.
H
u
a
n
g
,
“
C
h
i
n
e
se
S
p
a
m
D
e
t
e
c
t
i
o
n
U
si
n
g
a
H
y
b
r
i
d
B
i
G
R
U
-
C
N
N
N
e
t
w
o
r
k
w
i
t
h
Jo
i
n
t
T
e
x
t
u
a
l
a
n
d
P
h
o
n
e
t
i
c
Em
b
e
d
d
i
n
g
,
”
El
e
c
t
ro
n
i
c
s
,
v
o
l
.
1
1
,
n
o
.
1
5
,
2
0
2
2
,
d
o
i
:
1
0
.
3
3
9
0
/
e
l
e
c
t
r
o
n
i
c
s1
1
1
5
2
4
1
8
.
B
I
O
G
RAP
H
I
E
S O
F
AUTH
O
RS
J
a
s
m
ir
is
se
n
i
o
r
lec
tu
re
a
t
Un
i
v
e
rsitas
Din
a
m
ik
a
Ba
n
g
s
a
Ja
m
b
i,
In
d
o
n
e
sia
.
He
re
c
e
iv
e
d
h
is
Ba
c
h
e
lo
r
in
Co
m
p
u
t
e
r
En
g
in
e
e
rin
g
in
1
9
9
5
a
n
d
M
a
ste
r
d
e
g
re
e
in
In
f
o
rm
a
ti
o
n
T
e
c
h
n
o
lo
g
y
in
2
0
0
6
f
ro
m
Un
iv
e
rs
it
a
s
P
u
tra
In
d
o
n
e
sia
YP
T
K
P
a
d
a
n
g
,
In
d
o
n
e
sia
.
He
re
c
e
iv
e
s
a
Do
c
to
r
i
n
In
f
o
rm
a
ti
c
s E
n
g
in
e
e
rin
g
a
t
Un
iv
e
rsitas
S
riw
ij
a
y
a
P
a
lem
b
a
n
g
,
In
d
o
n
e
sia
i
n
2
0
2
2
.
His
re
se
a
rc
h
in
tere
st
is
d
a
ta
m
in
in
g
,
m
a
c
h
in
e
lea
rn
i
n
g
a
n
d
d
e
e
p
lea
r
n
in
g
f
o
r
n
a
tu
ra
l
lan
g
u
a
g
e
p
ro
c
e
ss
in
g
,
a
n
d
it
s a
p
p
li
c
a
ti
o
n
.
H
e
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
ij
a
y
_
jas
m
ir@
y
a
h
o
o
.
c
o
m
.
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KOM
NI
K
A
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l
C
o
n
tr
o
l
C
o
mp
a
r
is
o
n
o
f wo
r
d
emb
ed
d
i
n
g
fea
tu
r
es u
s
in
g
d
ee
p
lea
r
n
in
g
in
s
en
timen
t a
n
a
lysi
s
(
Ja
s
m
i
r
)
425
Er
r
issy
a
R
a
sy
w
ir
re
c
e
i
v
e
d
th
e
Ba
c
h
e
lo
r
d
e
g
re
e
(S
.
Ko
m
)
in
C
o
m
p
u
ter
S
c
ien
c
e
f
ro
m
th
e
S
riw
ij
a
y
a
Un
iv
e
rsit
y
.
S
h
e
re
c
e
iv
e
d
th
e
M
a
ste
r
d
e
g
re
e
(M
.
T
)
in
In
f
o
rm
a
ti
c
s
M
a
ste
r
S
T
EI
f
ro
m
th
e
I
n
stit
u
t
T
e
k
n
o
lo
g
i
Ba
n
d
u
n
g
(IT
B).
S
h
e
is a
l
e
c
tu
re
o
f
c
o
m
p
u
ter sc
ien
c
e
in
t
h
e
In
f
o
rm
a
ti
c
s
En
g
in
e
e
rin
g
,
Din
a
m
i
k
a
Ba
n
g
s
a
Un
iv
e
rsit
y
(UN
A
M
A
).
S
h
e
is
c
u
rre
n
tl
y
stu
d
y
in
g
f
o
r
a
Do
c
to
ra
te
in
Co
m
p
u
ter
S
c
ie
n
c
e
a
t
S
riw
ij
a
y
a
Un
iv
e
r
sit
y
.
In
a
d
d
it
i
o
n
,
sh
e
is
se
rv
in
g
a
s
h
e
a
d
o
f
th
e
re
se
a
rc
h
g
ro
u
p
(L
P
P
M
)
o
n
UN
A
M
A
.
He
r
re
se
a
rc
h
in
tere
sts
a
re
in
d
a
ta
m
in
in
g
,
a
rti
f
icia
l
in
telli
g
e
n
t
(A
I),
n
a
t
u
ra
l
l
a
n
g
u
a
n
g
e
p
ro
c
c
e
ss
in
g
(NL
P
),
m
a
c
h
in
e
lea
rn
i
n
g
,
a
n
d
d
e
e
p
lea
rn
in
g
.
S
h
e
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
e
rriss
y
a
.
ra
s
y
w
ir@
g
m
a
il
.
c
o
m
.
H
e
r
ti
Ya
n
i
is
a
lec
tu
re
a
t
Un
i
v
e
rsitas
Din
a
m
ik
a
Ba
n
g
s
a
Ja
m
b
i,
In
d
o
n
e
sia
.
S
h
e
re
c
e
iv
e
d
h
is
Ba
c
h
e
lo
r
in
In
f
o
rm
a
t
io
n
S
y
ste
m
in
Un
iv
e
rsitas
Din
a
m
i
k
a
Ba
n
g
s
a
Ja
m
b
i
in
2
0
0
9
a
n
d
M
a
ste
r
d
e
g
re
e
in
M
a
g
ister
S
y
ste
m
In
f
o
r
m
a
ti
o
n
i
n
U
n
iv
e
rsitas
Din
a
m
ik
a
Ba
n
g
sa
Ja
m
b
i,
In
d
o
n
e
sia
in
2
0
1
1
.
S
h
e
is
c
u
rre
n
t
ly
stu
d
y
in
g
f
o
r
a
Do
c
to
r
a
te
i
n
Co
m
p
u
ter
S
c
ien
c
e
a
t
S
a
ty
a
W
a
c
a
n
a
Ch
risti
a
n
Un
iv
e
rsity
.
H
e
r
re
se
a
rc
h
in
tere
st
a
re
in
d
a
tab
a
se
,
a
rti
f
icia
l
in
telli
g
e
n
c
e
,
a
n
d
m
a
c
h
in
e
lea
rn
in
g
.
S
h
e
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
a
d
e
h
e
rti
@u
n
a
m
a
.
a
c
.
id
.
Ag
u
s Nu
g
r
o
h
o
is l
e
c
tu
re
a
t
Un
iv
e
rsitas
Din
a
m
ik
a
Ba
n
g
s
a
Ja
m
b
i,
In
d
o
n
e
sia
.
He
re
c
e
iv
e
d
h
is
Ba
c
h
e
lo
r
in
In
f
o
rm
a
t
ics
En
g
in
e
e
rin
g
in
Un
iv
e
rsitas
Din
a
m
ik
a
Ba
n
g
s
a
Ja
m
b
i
in
2
0
1
1
a
n
d
M
a
ste
r
d
e
g
re
e
in
M
a
g
ister
o
f
In
f
o
rm
a
ti
c
s
En
g
in
e
e
rin
g
in
S
T
M
IK
A
M
IKO
M
Yo
g
y
a
k
a
rta,
In
d
o
n
e
sia
in
2
0
1
3
.
H
is
re
se
a
rc
h
in
tere
st
a
re
in
m
u
lt
ime
d
ia,
a
rti
f
icia
l
in
telli
g
e
n
c
e
,
a
n
d
m
a
c
h
in
e
lea
rn
in
g
.
S
h
e
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
a
g
u
sn
u
g
ro
h
o
0
8
8
8
@g
m
a
il
.
c
o
m
.
Evaluation Warning : The document was created with Spire.PDF for Python.