I
AE
S In
t
er
na
t
io
na
l J
o
urna
l o
f
Art
if
icia
l In
t
ellig
ence
(
I
J
-
AI
)
Vo
l.
13
,
No
.
4
,
Dece
m
b
er
2
0
2
4
,
p
p
.
4
9
7
6
~
4
9
8
7
I
SS
N:
2
2
5
2
-
8
9
3
8
,
DOI
: 1
0
.
1
1
5
9
1
/ijai.v
13
.i
4
.
p
p
4
9
7
6
-
4
9
8
7
4976
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
a
i
.
ia
esco
r
e.
co
m
A neural ma
ch
ine
t
r
a
nsla
tion sy
ste
m for
K
reol
R
epi
blik Mo
ris
a
nd Eng
lish
Sa
m
ee
rc
ha
nd
P
ud
a
rut
h
1
,
Sh
ee
ba
Arm
o
o
g
um
1
,
Nirma
l K
um
a
r
B
et
cho
o
2
,
Aneer
a
v
Su
k
ho
o
3
,
Va
nd
a
na
h G
o
o
ria
4
,
Abda
lla
h P
ee
ra
lly
1
,
M
o
ha
m
ma
d Z
a
f
a
r
K
ho
da
bo
cus
1
1
D
e
p
a
r
t
me
n
t
o
f
I
n
f
o
r
mat
i
o
n
a
n
d
C
o
m
mu
n
i
c
a
t
i
o
n
Te
c
h
n
o
l
o
g
y
,
F
a
c
u
l
t
y
o
f
I
n
f
o
r
mat
i
o
n
,
C
o
mm
u
n
i
c
a
t
i
o
n
a
n
d
D
i
g
i
t
a
l
Te
c
h
n
o
l
o
g
i
e
s,
U
n
i
v
e
r
si
t
y
o
f
M
a
u
r
i
t
i
u
s
,
M
o
k
a
,
M
a
u
r
i
t
i
u
s
2
D
e
p
a
r
t
me
n
t
o
f
M
a
n
a
g
e
me
n
t
,
F
a
c
u
l
t
y
o
f
B
u
s
i
n
e
ss a
n
d
M
a
n
a
g
e
me
n
t
,
U
n
i
v
e
r
si
t
d
e
s
M
a
s
c
a
r
e
i
g
n
e
s,
B
e
a
u
B
a
ssi
n
-
R
o
se
H
i
l
l
,
M
a
u
r
i
t
i
u
s
3
C
e
n
t
r
a
l
I
n
f
o
r
m
a
t
i
o
n
S
y
st
e
ms D
i
v
i
si
o
n
,
M
i
n
i
st
r
y
o
f
I
n
f
o
r
ma
t
i
o
n
Te
c
h
n
o
l
o
g
y
,
C
o
mm
u
n
i
c
a
t
i
o
n
a
n
d
I
n
n
o
v
a
t
i
o
n
,
P
o
r
t
L
o
u
i
s
,
M
a
u
r
i
t
i
u
s
4
A
c
a
d
e
m
i
c
A
f
f
a
i
r
s
D
i
v
i
s
i
o
n
,
O
p
e
n
U
n
i
v
e
r
si
t
y
o
f
M
a
u
r
i
t
i
u
s,
R
e
d
u
i
t
,
M
a
u
r
i
t
i
u
s
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
May
8
,
2
0
2
4
R
ev
is
ed
J
u
n
22
,
2
0
2
4
Acc
ep
ted
J
u
n
28
,
2
0
2
4
Alth
o
u
g
h
G
o
o
g
le
Tra
n
sla
te
is
a
wid
e
ly
u
se
d
m
a
c
h
in
e
tran
sla
ti
o
n
se
rv
ice
th
a
t
su
p
p
o
rts
1
3
3
la
n
g
u
a
g
e
s,
it
d
o
e
s
n
o
t
in
c
o
rp
o
ra
te
su
p
p
o
rt
f
o
r
t
h
e
Kre
o
l
Re
p
ib
li
k
M
o
ris
(KRM)
lan
g
u
a
g
e
.
A
d
d
re
s
sin
g
t
h
is
li
m
it
a
ti
o
n
,
t
h
e
c
u
rre
n
t
re
se
a
rc
h
fo
c
u
se
s
o
n
e
n
h
a
n
c
in
g
t
h
e
a
c
c
u
ra
c
y
a
n
d
f
lu
e
n
c
y
o
f
m
a
c
h
in
e
t
ra
n
sla
ti
o
n
b
e
twe
e
n
KRM
a
n
d
E
n
g
li
sh
th
r
o
u
g
h
n
a
t
u
ra
l
lan
g
u
a
g
e
p
r
o
c
e
ss
in
g
a
n
d
d
e
e
p
n
e
u
ra
l
m
a
c
h
in
e
tran
sla
ti
o
n
tec
h
n
iq
u
e
s.
In
t
h
is
stu
d
y
,
a
m
a
c
h
in
e
tran
sla
ti
o
n
sy
ste
m
u
sin
g
a
tran
sfo
rm
e
r
m
o
d
e
l
train
e
d
with
a
d
a
tas
e
t
o
f
5
0
,
0
0
0
p
a
ra
ll
e
l
c
o
rp
o
ra
h
a
s
b
e
e
n
d
e
v
e
lo
p
e
d
.
Th
e
m
o
d
e
l
wa
s
e
v
a
lu
a
ted
u
si
n
g
m
a
n
u
a
l
tran
sla
ti
o
n
s
a
n
d
th
e
b
il
i
n
g
u
a
l
e
v
a
lu
a
ti
o
n
u
n
d
e
rstu
d
y
(
BLE
U
)
sc
o
re
.
A
sc
o
re
o
f
3
1
.
4
6
fo
r
tran
sla
ti
n
g
fro
m
KRM
to
E
n
g
li
s
h
a
n
d
2
8
.
1
5
f
o
r
tran
sla
ti
n
g
fro
m
En
g
li
s
h
t
o
KRM
wa
s
a
c
h
iev
e
d
.
To
o
u
r
k
n
o
wle
d
g
e
,
th
e
se
a
re
th
e
h
ig
h
e
st
BLE
U
sc
o
re
s
fo
r
tran
sla
ti
o
n
b
e
twe
e
n
t
h
e
se
two
lan
g
u
a
g
e
s.
Th
is
is
d
u
e
to
u
t
il
isin
g
th
e
larg
e
st
d
a
tas
e
t
a
n
d
e
x
te
n
siv
e
a
t
o
m
ic
wo
rd
s
fr
o
m
t
h
e
KRM
d
icti
o
n
a
ry
.
Th
is
su
c
c
e
ss
fu
l
in
terd
isc
ip
li
n
a
ry
fu
n
d
e
d
p
ro
jec
t
led
to
t
h
e
se
tt
i
n
g
u
p
o
f
a
f
re
e
o
n
li
n
e
tran
sla
ti
o
n
se
rv
ice
a
n
d
a
sm
a
rtp
h
o
n
e
a
p
p
fo
r
M
a
u
ri
ti
a
n
c
it
ize
n
s a
n
d
to
u
rists.
K
ey
w
o
r
d
s
:
Dee
p
lear
n
in
g
Kr
eo
l Rep
ib
lik
Mo
r
is
Ma
ch
in
e
tr
an
s
latio
n
Ma
u
r
itiu
s
T
r
an
s
f
o
r
m
e
r
m
o
d
el
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
:
Sam
ee
r
ch
an
d
Pu
d
ar
u
th
Dep
ar
tm
en
t o
f
I
n
f
o
r
m
atio
n
an
d
C
o
m
m
u
n
icatio
n
T
ec
h
n
o
lo
g
y
Facu
lty
o
f
I
n
f
o
r
m
atio
n
,
C
o
m
m
u
n
icatio
n
a
n
d
Dig
ital T
ec
h
n
o
lo
g
ies,
Un
iv
er
s
ity
o
f
Ma
u
r
iti
u
s
Mo
k
a,
Ma
u
r
itiu
s
E
m
ail: s.p
u
d
ar
u
th
@
u
o
m
.
ac
.
m
u
1.
I
NT
RO
D
UCT
I
O
N
As
p
er
Statis
tics
Ma
u
r
itiu
s
[
1
]
,
C
r
eo
le
is
s
p
o
k
en
b
y
9
0
%
o
f
th
e
Ma
u
r
itian
p
o
p
u
latio
n
.
T
h
e
C
r
eo
le
lan
g
u
ag
e
is
f
o
r
m
ally
k
n
o
wn
as
Kr
eo
l
Mo
r
is
ien
(
KM
)
o
r
as
Kr
eo
l
R
ep
ib
lik
Mo
r
is
(
K
R
M)
.
I
n
2
0
1
2
,
th
e
g
o
v
er
n
m
en
t
o
f
Ma
u
r
itiu
s
in
cl
u
d
ed
KM
as
o
n
e
o
f
th
e
s
u
b
jects
in
p
r
im
a
r
y
ed
u
ca
tio
n
,
a
n
d
in
2
0
1
7
,
4
,
0
0
0
s
tu
d
en
ts
to
o
k
th
is
lan
g
u
ag
e
f
o
r
th
eir
p
r
im
ar
y
s
ch
o
o
l
ac
h
iev
e
m
en
t
ce
r
tific
ate
(
PS
AC
)
.
A
s
ig
n
if
ican
t
p
r
o
p
o
r
tio
n
o
f
th
e
Ma
u
r
itian
p
o
p
u
latio
n
is
n
o
t
p
r
o
f
icien
t
in
E
n
g
lis
h
wh
ich
lead
s
to
d
if
f
icu
lties
in
u
n
d
er
s
tan
d
i
n
g
E
n
g
lis
h
u
s
ed
in
p
u
b
lic
co
m
m
u
n
icatio
n
an
d
th
e
m
ed
ia.
T
h
is
lim
itatio
n
also
h
in
d
er
s
t
h
eir
in
ter
ac
tio
n
wit
h
to
u
r
is
ts
an
d
o
th
e
r
f
o
r
eig
n
e
r
s
.
Desp
ite
th
e
av
ailab
ilit
y
o
f
n
u
m
er
o
u
s
o
n
lin
e
tr
an
s
latio
n
s
er
v
ices,
th
ey
h
av
e
y
et
to
ad
d
r
ess
tr
an
s
latin
g
b
etwe
en
KR
M
an
d
E
n
g
lis
h
ef
f
ec
tiv
ely
.
Ad
v
o
ca
cy
f
o
r
u
s
in
g
KR
M
as a
n
atio
n
al
lan
g
u
a
g
e
h
as b
ee
n
ad
v
an
ce
d
b
y
Dev
Vir
ah
s
awm
y
,
an
au
th
o
r
,
p
o
et,
an
d
p
o
liti
ca
l
f
ig
u
r
e
[
2
]
.
Vir
ah
s
awm
y
h
as
au
th
o
r
ed
n
u
m
er
o
u
s
tex
ts
an
d
p
o
em
s
in
Kr
eo
l
Mo
r
is
ien
.
T
h
e
d
ev
elo
p
m
e
n
t
o
f
a
h
ar
m
o
n
is
ed
wr
itin
g
s
y
s
tem
ca
lled
Gr
af
i
-
lar
m
o
n
i
aim
ed
at
estab
lis
h
in
g
a
s
tan
d
ar
d
is
ed
f
o
r
m
f
o
r
wr
itin
g
th
e
lan
g
u
ag
e.
Fu
r
t
h
er
m
o
r
e
,
C
ar
p
o
o
r
an
in
tr
o
d
u
ce
d
an
u
p
d
ated
Kr
e
o
l
Mo
r
is
ien
d
ictio
n
ar
y
in
2
0
1
1
,
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
A
n
eu
r
a
l m
a
ch
in
e
t
r
a
n
s
la
tio
n
s
ystem
fo
r
K
r
eo
l
R
ep
ib
lik
Mo
r
is
a
n
d
E
n
g
lis
h
…
(
S
a
mee
r
ch
a
n
d
P
u
d
a
r
u
th
)
4977
an
d
s
u
b
s
eq
u
e
n
t
v
er
s
io
n
s
h
av
e
in
clu
d
ed
n
ew
wo
r
d
s
[
3
]
.
T
h
e
s
e
ef
f
o
r
ts
to
war
d
s
s
tan
d
ar
d
is
atio
n
an
d
lin
g
u
is
tic
d
ev
elo
p
m
e
n
t in
KR
M
h
av
e
c
o
n
tr
ib
u
ted
to
its
r
ec
o
g
n
itio
n
an
d
in
clu
s
io
n
in
e
d
u
ca
tio
n
.
KR
M
o
r
KM
h
as
a
s
en
te
n
ce
s
tr
u
ctu
r
e
m
ain
ly
in
f
lu
en
ce
d
b
y
its
h
is
to
r
ical
f
o
r
m
atio
n
an
d
e
v
o
lu
tio
n
f
r
o
m
a
p
id
g
in
la
n
g
u
a
g
e
d
u
r
in
g
t
h
e
Fre
n
ch
co
lo
n
is
atio
n
p
er
io
d
to
an
estab
lis
h
ed
o
n
e.
K
r
eo
l
Mo
r
is
ien
’
s
s
tr
u
ctu
r
e
is
s
h
ap
ed
b
y
its
p
ar
en
t
la
n
g
u
a
g
es,
m
ain
ly
Fre
n
ch
a
n
d
th
e
lan
g
u
a
g
es
o
f
th
e
Af
r
ican
s
lav
es
an
d
i
n
d
en
tu
r
e
d
lab
o
u
r
er
s
f
r
o
m
I
n
d
ia
wh
o
co
n
tr
i
b
u
ted
t
o
it
i
s
f
o
r
m
atio
n
[
4
]
.
Ho
wev
e
r
,
m
an
y
E
n
g
lis
h
wo
r
d
s
h
av
e
a
ls
o
f
o
u
n
d
th
eir
way
in
to
th
e
lan
g
u
ag
e.
T
h
e
s
en
ten
c
e
s
tr
u
ctu
r
e
o
f
Kr
e
o
l
Mo
r
is
ien
c
lo
s
ely
r
esem
b
les
E
n
g
lis
h
b
u
t
w
ith
s
o
m
e
d
is
tin
ctiv
e
v
ar
iatio
n
s
.
Ho
wev
er
,
u
n
lik
e
E
n
g
lis
h
,
Kr
eo
l
Mo
r
is
ien
d
o
es
n
o
t
h
av
e
p
lu
r
al
f
o
r
m
s
f
o
r
w
o
r
d
s
an
d
f
r
e
q
u
en
tly
p
o
s
itio
n
s
ad
jectiv
es
b
e
h
in
d
th
e
o
b
ject.
T
h
e
tr
an
s
latio
n
p
r
o
ce
s
s
o
f
ten
en
tails
o
m
itti
n
g
s
tate
-
of
-
b
ein
g
v
er
b
s
.
Fo
r
in
s
tan
ce
,
“
He
is
g
o
o
d
at
d
an
ci
n
g
”
wo
u
ld
b
e
tr
a
n
s
lated
as
“
L
i
b
o
n
d
a
n
d
a
n
s
e
”
,
wh
er
e
“h
e”
b
ec
o
m
es
“L
i”,
an
d
“g
o
o
d
at
d
an
cin
g
”
b
ec
o
m
es “
b
o
n
d
an
d
an
s
e”
.
T
h
e
v
er
b
“is” i
s
o
m
itted
in
th
e
t
r
an
s
latio
n
.
B
y
u
n
d
er
s
tan
d
in
g
th
e
d
is
tin
ct
s
en
ten
ce
s
tr
u
ctu
r
e
o
f
Kr
eo
l
Mo
r
is
ien
,
ed
u
ca
to
r
s
c
an
ef
f
ec
tiv
ely
teac
h
a
n
d
i
n
clu
d
e
th
e
lan
g
u
a
g
e
in
ed
u
ca
tio
n
.
T
h
e
f
o
llo
win
g
a
r
e
th
e
ch
ar
ac
ter
is
tics
o
f
th
e
s
en
ten
ce
s
tr
u
ctu
r
e:
s
u
b
ject
-
wo
r
d
-
o
b
ject
(
SVO)
o
r
d
er
,
p
r
e
-
v
er
b
al
m
ar
k
er
s
f
o
r
ten
s
e
an
d
asp
ec
t,
n
e
g
atio
n
an
d
q
u
esti
o
n
f
o
r
m
atio
n
.
L
ik
e
E
n
g
lis
h
a
n
d
Fre
n
ch
,
Kr
eo
l
Mo
r
is
ien
u
s
u
ally
u
s
es
th
e
s
u
b
ject
-
v
er
b
-
o
b
ject
wo
r
d
o
r
d
er
.
T
h
e
s
u
b
ject
is
p
lace
d
f
ir
s
t,
f
o
llo
wed
b
y
th
e
v
er
b
an
d
th
en
th
e
o
b
ject
th
at
r
ec
eiv
es
th
e
ac
tio
n
.
T
h
is
s
en
ten
c
e
f
o
r
m
at
f
ac
ilit
ates
clea
r
co
m
m
u
n
icatio
n
an
d
ef
f
o
r
tles
s
u
n
d
er
s
tan
d
in
g
in
Kr
e
o
l
Mo
r
is
ien
,
th
u
s
f
ac
ilit
atin
g
ef
f
e
ctiv
e
in
s
tr
u
ctio
n
a
n
d
in
teg
r
atio
n
in
to
th
e
ed
u
ca
tio
n
s
y
s
tem
.
I
n
ad
d
itio
n
to
th
e
SVO
wo
r
d
o
r
d
er
,
Kr
eo
l
Mo
r
is
ien
ex
h
ib
its
d
is
tin
ctiv
e
s
en
ten
ce
s
tr
u
ctu
r
e
ch
ar
ac
te
r
is
tics
.
Pre
-
v
er
b
al
m
a
r
k
er
s
in
Kr
eo
l
Mo
r
is
ien
ex
p
r
ess
ten
s
e
an
d
asp
ec
t
with
in
s
en
ten
ce
s
.
T
h
ese
t
ag
s
p
r
ec
ed
e
th
e
v
er
b
an
d
o
f
f
er
v
ital
d
etails
ab
o
u
t
th
e
tim
in
g
o
f
th
e
ac
tio
n
o
r
its
d
u
r
atio
n
.
T
en
s
e,
m
o
o
d
,
an
d
asp
ec
t
m
a
r
k
er
s
in
Kr
eo
l
Mo
r
is
ien
in
d
icate
an
ac
tio
n
’
s
tim
e,
p
er
s
p
ec
tiv
e,
an
d
d
u
r
atio
n
.
Fo
r
in
s
tan
ce
:
“
Mo
p
e
m
an
ze
.
”
(
I
am
ea
tin
g
)
-
“
pe
”
s
ig
n
if
ies
a
co
n
tin
u
o
u
s
asp
ec
t.
“
L
i
ti
p
e
ale.
”
(
He
was
leav
in
g
.
)
-
“
ti
”
is
a
n
ex
p
r
ess
io
n
o
f
th
e
p
ast
ten
s
e.
“
Z
o
t
p
o
u
v
in
i
”
-
“
pou
”
d
en
o
tes
th
e
f
u
tu
r
e
ten
s
e.
Kr
eo
l
Mo
r
is
ien
co
m
m
o
n
ly
u
s
e
s
th
e
m
ar
k
er
“
pa
”
b
ef
o
r
e
th
e
v
er
b
t
o
ex
p
r
ess
n
eg
atio
n
.
F
o
r
in
s
tan
ce
,
“
L
i p
a
p
e
d
a
n
s
e.
”
tr
an
s
lates to
“
He/Sh
e
is
n
o
t d
an
cin
g
.
”
T
h
e
f
o
r
m
atio
n
o
f
q
u
esti
o
n
s
in
Kr
eo
l
Mo
r
is
ien
d
if
f
er
s
b
etwe
en
E
n
g
lis
h
an
d
Fre
n
c
h
.
I
t
o
f
te
n
in
v
o
lv
es
s
u
b
ject
-
v
er
b
in
v
er
s
io
n
o
r
th
e
u
s
e
o
f
q
u
esti
o
n
wo
r
d
s
.
Yes/n
o
q
u
esti
o
n
s
ca
n
f
r
eq
u
e
n
tly
b
e
f
o
r
m
ed
b
y
in
t
o
n
atio
n
alo
n
e,
r
aisi
n
g
th
e
p
itch
o
f
th
e
v
o
ice
to
war
d
s
th
e
en
d
o
f
th
e
s
en
ten
ce
.
W
h
-
q
u
esti
o
n
s
u
s
e
q
u
esti
o
n
wo
r
d
s
lik
e
“
k
if
er
”
(
wh
y
)
,
“
ki
”
(
wh
at)
,
a
n
d
“
kot
”
(
wh
er
e)
,
wh
ich
ar
e
p
lace
d
at
th
e
b
e
g
in
n
in
g
o
f
th
e
s
en
ten
ce
.
Kr
eo
l
Mo
r
is
ien
’
s
s
en
ten
ce
s
tr
u
ctu
r
e
is
s
tr
aig
h
tf
o
r
war
d
an
d
p
r
ac
ti
ca
l,
allo
win
g
s
p
ea
k
er
s
t
o
co
m
m
u
n
i
ca
te
in
tr
icate
co
n
ce
p
ts
with
in
a
v
er
s
atile
an
d
m
in
im
alis
t
g
r
am
m
atica
l
s
y
s
tem
.
T
h
e
r
esear
ch
d
e
m
o
n
s
tr
ates
th
at
Kr
eo
l
Mo
r
is
ien
d
is
p
lay
s
a
u
n
iq
u
e
s
en
ten
ce
s
tr
u
ctu
r
e
d
is
tin
g
u
is
h
ed
b
y
it
i
s
s
u
b
ject
-
v
er
b
-
o
b
ject
w
o
r
d
o
r
d
e
r
,
p
r
e
-
v
er
b
al
m
ar
k
er
s
f
o
r
ten
s
e,
asp
ec
t a
n
d
n
e
g
atio
n
,
an
d
d
is
tin
ct
m
eth
o
d
s
f
o
r
f
o
r
m
in
g
q
u
esti
o
n
s
.
R
ec
en
t
ad
v
an
ce
m
e
n
ts
in
m
ac
h
in
e
tr
an
s
latio
n
(
MT
)
tech
n
o
lo
g
y
h
av
e
b
ee
n
s
u
b
s
tan
tial,
b
u
t
its
ab
ilit
y
to
tr
an
s
late
Kr
eo
l
Mo
r
is
ien
is
c
o
n
s
tr
ain
ed
b
y
th
e
lan
g
u
ag
e
’
s
d
is
tin
ct
s
en
ten
ce
s
tr
u
ctu
r
e,
g
r
am
m
ar
an
d
lack
o
f
r
eso
u
r
ce
s
.
T
h
u
s
,
it
is
v
ital
to
en
h
an
ce
th
e
d
ev
elo
p
m
en
t
o
f
m
ac
h
in
e
tr
an
s
latio
n
s
y
s
tem
s
t
h
at
ca
n
ef
f
ec
tiv
ely
m
an
ag
e
t
h
e
u
n
iq
u
e
ch
ar
ac
te
r
is
tics
an
d
s
u
b
tleties
p
r
esen
t
in
Kr
eo
l
M
o
r
is
ien
to
attain
im
p
r
o
v
ed
p
r
e
cisi
o
n
i
n
tr
an
s
latio
n
s
.
MT
is
a
b
r
an
ch
o
f
co
m
p
u
tatio
n
al
lin
g
u
is
tics
th
at
ex
am
in
es
th
e
u
tili
s
atio
n
o
f
s
o
f
t
war
e
f
o
r
co
n
v
er
tin
g
tex
t
o
r
s
p
ee
ch
f
r
o
m
o
n
e
lan
g
u
ag
e
to
an
o
th
er
.
L
iter
al
m
ac
h
in
e
tr
an
s
latio
n
in
v
o
lv
es
r
ep
l
ac
in
g
wo
r
d
s
in
o
n
e
lan
g
u
ag
e
with
eq
u
iv
ale
n
t
wo
r
d
s
in
an
o
th
er
.
Ho
wev
er
,
m
o
r
e
is
n
ee
d
ed
to
g
en
er
ate
an
ac
cu
r
ate
tr
an
s
latio
n
.
I
t
is
ess
en
tial to
r
e
co
g
n
is
e
en
tire
p
h
r
ases
an
d
f
in
d
th
eir
m
o
s
t a
p
p
r
o
p
r
iate
eq
u
iv
alen
ts
in
th
e
tar
g
et
lan
g
u
ag
e.
No
t a
ll
ter
m
s
in
a
p
ar
ticu
lar
lan
g
u
ag
e
h
av
e
d
ir
ec
t
co
u
n
ter
p
ar
ts
in
an
o
th
er
lan
g
u
ag
e
,
an
d
n
u
m
er
o
u
s
wo
r
d
s
ca
r
r
y
m
u
ltip
le
in
ter
p
r
etatio
n
s
.
W
h
ile
th
e
p
r
in
cip
les
o
f
m
ac
h
i
n
e
tr
an
s
latio
n
m
ay
s
ee
m
s
tr
aig
h
tf
o
r
war
d
,
th
e
u
n
d
er
ly
i
n
g
s
cien
ce
an
d
tech
n
o
lo
g
ies
in
v
o
l
v
ed
a
r
e
h
ig
h
l
y
in
tr
icate
.
Sin
ce
th
e
b
eg
in
n
in
g
o
f
t
h
e
2
0
1
0
s
,
a
n
e
w
f
o
r
m
o
f
ar
tific
ial
in
tellig
en
ce
tech
n
o
l
o
g
y
k
n
o
wn
as
d
ee
p
n
eu
r
al
n
etwo
r
k
s
h
a
s
p
r
o
v
id
e
d
s
p
ee
ch
r
ec
o
g
n
itio
n
tech
n
o
l
o
g
ies
an
d
m
ac
h
in
e
tr
a
n
s
latio
n
s
y
s
tem
s
with
th
e
ca
p
a
b
ilit
y
to
ac
h
iev
e
a
s
atis
f
ac
to
r
y
lev
el
o
f
q
u
ality
[
5
]
.
T
h
is
r
esear
ch
f
u
lly
s
u
p
p
o
r
ts
m
o
r
e
f
lu
id
c
o
m
m
u
n
i
ca
tio
n
ac
r
o
s
s
b
o
r
d
er
s
an
d
d
if
f
e
r
en
t c
u
ltu
r
es.
T
h
e
r
u
le
-
b
ased
m
ac
h
in
e
tr
a
n
s
latio
n
(
R
B
MT
)
m
o
d
el
en
co
m
p
ass
ed
tr
an
s
f
er
-
b
ased
,
in
ter
-
lin
g
u
al,
an
d
d
ictio
n
ar
y
-
b
ased
m
ac
h
in
e
tr
a
n
s
latio
n
ap
p
r
o
ac
h
es.
T
h
is
f
o
r
m
o
f
in
ter
p
r
etatio
n
was
p
r
i
m
ar
ily
u
tili
s
ed
in
d
ev
elo
p
in
g
lex
ico
n
s
an
d
lin
g
u
is
tic
s
o
f
twar
e.
R
B
MT
in
v
o
lv
es
m
o
r
e
in
f
o
r
m
atio
n
ab
o
u
t
th
e
lin
g
u
is
tics
o
f
th
e
s
o
u
r
ce
an
d
tar
g
et
la
n
g
u
a
g
es.
T
h
e
f
u
n
d
am
e
n
tal
s
tr
ateg
y
co
n
n
ec
ts
th
e
in
p
u
t
s
en
ten
ce
’
s
f
o
r
m
at
with
th
e
o
u
tp
u
t
s
en
ten
ce
’
s
s
tr
u
ctu
r
e.
T
h
e
R
B
MT
ap
p
r
o
ac
h
f
ac
e
d
lim
itatio
n
s
s
u
ch
as
in
ad
eq
u
ate
h
ig
h
-
q
u
ality
d
ictio
n
ar
ies
,
m
an
u
al
s
etu
p
o
f
lin
g
u
is
tic
in
f
o
r
m
atio
n
,
an
d
ch
allen
g
es
with
r
u
le
i
n
ter
ac
tio
n
s
in
co
m
p
l
ex
s
y
s
tem
s
s
u
ch
as
am
b
ig
u
ity
a
n
d
id
io
m
atic
ex
p
r
e
s
s
io
n
s
[
6
]
.
Statis
t
ical
m
ac
h
in
e
tr
an
s
latio
n
(
SMT
)
is
th
e
au
to
m
ated
tr
an
s
f
o
r
m
atio
n
o
f
s
en
ten
ce
s
f
r
o
m
o
n
e
h
u
m
an
lan
g
u
ag
e
to
an
o
t
h
er
u
s
in
g
s
tatis
tics
an
d
p
r
o
b
ab
ilit
y
.
T
h
e
o
r
i
g
in
al
lan
g
u
ag
e
is
ter
m
ed
th
e
“
s
o
u
r
ce
,
”
wh
ile
th
e
s
ec
o
n
d
ar
y
lan
g
u
ag
e
is
lab
elled
th
e
“
tar
g
et
”.
T
h
is
p
r
o
ce
d
u
r
e
ca
n
b
e
co
n
ce
p
tu
alis
ed
as
a
p
r
o
b
ab
ilis
tic
p
r
o
ce
s
s
.
W
illi
am
s
et
a
l.
[
7
]
s
u
g
g
est
t
h
at
v
ar
io
u
s
SMT
v
ar
ian
ts
ex
is
t,
d
ep
en
d
in
g
o
n
th
eir
ap
p
r
o
ac
h
to
tr
an
s
latio
n
m
o
d
ellin
g
.
T
h
ese
ap
p
r
o
ac
h
e
s
in
clu
d
e
s
tr
in
g
-
to
-
s
tr
in
g
m
a
p
p
in
g
,
t
r
ee
-
to
-
s
tr
in
g
s
,
a
n
d
tr
ee
-
to
-
tr
ee
m
o
d
els.
Ho
wev
er
,
th
e
y
all
s
h
ar
e
t
h
e
f
a
m
iliar
p
r
in
cip
le
th
at
tr
an
s
latio
n
is
au
to
m
ated
a
n
d
in
v
o
l
v
es
m
o
d
els
d
er
i
v
ed
f
r
o
m
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
13
,
No
.
4
,
Dec
em
b
er
2
0
2
4
:
4
9
7
6
-
4
9
8
7
4978
p
ar
allel
co
r
p
o
r
a
(
s
o
u
r
ce
-
tar
g
et
p
air
s
)
an
d
m
o
n
o
lin
g
u
al
co
r
p
o
r
a
(
tar
g
et
s
en
ten
ce
e
x
am
p
les).
SMT
f
ac
es
s
ev
er
al
s
ig
n
if
ican
t
o
b
s
tacle
s
:
i
)
g
en
er
atin
g
an
d
tr
ain
in
g
p
ar
allel
d
ata
i
s
ex
p
en
s
iv
e
an
d
tim
e
-
co
n
s
u
m
in
g
,
ii
)
it
n
ec
ess
itates
lar
g
e
v
o
lu
m
es
o
f
p
ar
allel
d
at
a
co
m
p
r
is
in
g
at
least
2
m
illi
o
n
wo
r
d
s
,
iii
)
an
ticip
atin
g
an
d
r
ec
tify
in
g
s
p
ec
if
ic
tr
an
s
latio
n
er
r
o
r
s
ca
n
b
e
c
h
all
en
g
in
g
,
an
d
iv
)
SMT
is
n
o
t
w
ell
-
s
u
ited
f
o
r
lan
g
u
ag
e
p
air
s
with
d
if
f
er
in
g
wo
r
d
o
r
d
er
s
.
Neu
r
al
m
ac
h
in
e
tr
an
s
latio
n
(
NM
T
)
r
ep
r
esen
ts
a
s
ig
n
if
ican
t
s
h
if
t
f
r
o
m
tr
ad
itio
n
al
ap
p
r
o
ac
h
es
to
m
ac
h
in
e
tr
a
n
s
latio
n
,
as
it
u
til
is
es
co
n
tin
u
o
u
s
r
ep
r
esen
tatio
n
s
r
ath
er
th
an
d
is
cr
ete
s
y
m
b
o
lic
r
ep
r
esen
tatio
n
s
em
p
lo
y
ed
in
SMT
.
C
o
n
v
er
s
ely
,
NM
T
em
p
lo
y
s
a
s
in
g
le
e
x
ten
s
iv
e
n
eu
r
al
n
etwo
r
k
to
r
ep
r
esen
t
th
e
en
tire
tr
an
s
latio
n
p
r
o
ce
s
s
,
elim
in
atin
g
th
e
n
ec
ess
ity
f
o
r
ab
u
n
d
an
t f
e
atu
r
e
en
g
in
ee
r
i
n
g
[
8
]
.
NM
T
e
m
p
lo
y
s
d
ee
p
n
e
u
r
al
n
etwo
r
k
s
to
m
ap
t
h
e
s
o
u
r
ce
a
n
d
tar
g
et
lan
g
u
ag
es
d
ir
ec
tly
,
s
ig
n
if
ican
tly
a
d
v
an
cin
g
tr
a
n
s
latio
n
p
e
r
f
o
r
m
an
ce
.
I
t
h
as
n
o
w
b
ec
o
m
e
th
e
d
o
m
i
n
an
t
p
ar
ad
ig
m
o
f
m
ac
h
in
e
tr
an
s
latio
n
[
9
]
.
R
ag
n
i
a
n
d
Vieir
a
[
1
0
]
ex
p
r
ess
p
ar
ticu
lar
co
n
ce
r
n
s
r
e
g
ar
d
in
g
NM
T
.
Fir
s
t,
it
is
im
p
o
r
tan
t
to
d
elv
e
m
o
r
e
d
ee
p
ly
in
to
th
e
NM
T
e
d
itin
g
p
r
o
ce
s
s
an
d
its
s
p
ec
if
ic
asp
ec
ts
f
r
o
m
th
e
p
er
s
p
ec
ti
v
e
o
f
tr
an
s
lato
r
s
.
Seco
n
d
,
t
h
er
e
s
h
o
u
l
d
b
e
a
s
tr
o
n
g
er
em
p
h
asis
o
n
h
ig
h
lig
h
tin
g
th
e
u
s
ef
u
ln
ess
o
f
NM
T
as
a
to
o
l
f
o
r
p
r
o
f
ess
io
n
als.
T
h
ir
d
,
th
e
r
e
h
as
b
ee
n
a
ten
d
en
c
y
to
n
ar
r
o
wly
co
n
ce
p
tu
alis
e
tr
an
s
latio
n
p
r
o
d
u
ctiv
ity
b
ased
s
o
lely
o
n
p
r
o
ce
s
s
in
g
tim
e
o
r
th
r
o
u
g
h
p
u
t
m
ea
s
u
r
es.
L
astl
y
,
it
h
as
b
ec
o
m
e
ap
p
ar
en
t
th
at
in
v
esti
g
atio
n
s
r
elate
d
to
N
MT
in
v
o
lv
in
g
en
d
-
u
s
er
s
ar
e
s
till
r
elativ
ely
r
ar
e.
T
h
e
in
itial
s
u
cc
ess
o
f
Mo
r
is
ia
1
.
0
[
1
1
]
,
a
m
u
lti
-
d
is
cip
lin
ar
y
p
r
o
ject
f
u
n
d
ed
b
y
th
e
h
ig
h
e
r
ed
u
ca
tio
n
co
m
m
is
s
io
n
(
HE
C
)
o
f
Ma
u
r
itiu
s
,
led
to
th
e
d
ev
elo
p
m
en
t
o
f
Mo
r
is
ia
2
.
0
.
T
h
e
p
r
im
ar
y
f
o
c
u
s
in
itially
was
o
n
estab
lis
h
in
g
a
p
latf
o
r
m
th
at
w
o
u
ld
in
co
r
p
o
r
ate
s
p
ec
if
ic
p
ar
a
llel
co
r
p
o
r
a.
T
h
e
cu
r
r
e
n
t
s
tu
d
y
b
u
ild
s
u
p
o
n
p
r
i
o
r
r
esear
ch
b
y
s
ig
n
if
ican
tly
e
x
p
a
n
d
in
g
t
h
e
co
r
p
u
s
to
e
n
h
an
ce
u
s
er
tr
an
s
latio
n
ef
f
icien
cy
.
I
n
th
is
co
n
tex
t,
th
er
e
is
a
clea
r
r
eq
u
ir
em
e
n
t
f
o
r
ex
ten
s
iv
e
co
llectio
n
s
o
f
p
a
r
allel
tex
ts
,
o
r
p
ar
allel
co
r
p
o
r
a,
to
s
u
p
p
o
r
t
v
ar
io
u
s
lin
g
u
is
tic
s
tu
d
ies
th
at
r
ely
o
n
s
en
ten
ce
-
l
ev
el
alig
n
m
en
ts
with
in
th
ese
co
r
p
o
r
a.
W
h
en
tr
an
s
latin
g
,
th
e
tr
an
s
lato
r
h
as
th
e
f
lex
ib
ilit
y
to
s
p
lit,
co
m
b
in
e,
r
e
m
o
v
e,
ad
d
o
r
r
ea
r
r
an
g
e
s
en
ten
ce
s
.
Alig
n
m
en
t
is
a
s
ig
n
if
ica
n
t
u
n
d
er
tak
in
g
.
T
h
is
p
ap
er
ad
d
r
ess
es
th
e
ef
f
icien
t
tr
an
s
latio
n
b
etwe
en
KR
M
an
d
E
n
g
lis
h
an
d
v
ice
-
v
er
s
a,
ai
m
in
g
to
en
s
u
r
e
u
s
er
s
atis
f
ac
tio
n
an
d
s
u
cc
ess
f
u
l m
ess
ag
e
in
ter
p
r
etatio
n
th
r
o
u
g
h
a
h
ig
h
-
q
u
ality
tr
an
s
latio
n
s
y
s
tem
.
T
h
e
r
esear
ch
p
r
o
ject
aim
s
to
d
ev
elo
p
an
au
to
m
ate
d
tr
an
s
l
atio
n
s
y
s
tem
th
at
s
ea
m
less
ly
tr
an
s
lates
b
etwe
en
E
n
g
lis
h
an
d
KR
M
u
s
in
g
d
ee
p
n
e
u
r
al
n
etwo
r
k
s
.
T
h
e
p
r
o
ject
in
clu
d
es th
e
d
e
v
elo
p
m
en
t o
f
a
web
p
o
r
tal
tr
an
s
lato
r
with
s
u
p
p
o
r
t
f
r
o
m
a
m
o
b
ile
ap
p
.
I
t
s
ee
k
s
to
ac
h
iev
e
s
ev
er
al
o
b
jectiv
es,
in
clu
d
in
g
b
u
ild
in
g
a
p
a
r
allel
co
r
p
u
s
o
f
5
0
,
0
0
0
s
en
ten
ce
s
t
o
im
p
r
o
v
e
tr
a
n
s
latio
n
ac
cu
r
ac
y
b
etwe
en
KM
/KR
M
an
d
E
n
g
lis
h
,
ass
ess
in
g
th
e
s
y
s
tem
’
s
p
er
f
o
r
m
an
ce
u
s
in
g
m
etr
ics
s
u
ch
as
th
e
b
il
in
g
u
al
ev
alu
atio
n
u
n
d
er
s
tu
d
y
(
B
L
E
U
)
s
co
r
e
,
d
esig
n
in
g
a
web
p
o
r
tal
ap
p
with
ac
ce
s
s
ib
ilit
y
f
ea
tu
r
es
tailo
r
ed
f
o
r
in
d
iv
id
u
als
with
d
is
ab
ilit
ies
(
e.
g
.
,
lar
g
er
f
o
n
t
s
ize
an
d
co
lo
u
r
d
if
f
er
en
tiatio
n
)
,
as we
ll
as in
co
r
p
o
r
atin
g
v
o
ice
-
to
-
tex
t
an
d
tex
t
-
to
-
v
o
ice
f
u
n
ctio
n
alit
y
in
th
e
m
o
b
ile
ap
p
s
p
ec
if
ically
f
o
r
E
n
g
lis
h
lan
g
u
a
g
e
tr
an
s
latio
n
s
.
T
h
e
u
ltima
te
g
o
al
o
f
th
is
p
r
o
ject
is
to
p
r
o
v
id
e
a
u
s
er
-
f
r
ien
d
ly
a
n
d
ef
f
icien
t
tr
an
s
latio
n
s
y
s
tem
th
at
b
r
id
g
es
th
e
lan
g
u
ag
e
g
ap
b
etwe
en
E
n
g
lis
h
an
d
Kr
eo
l
M
o
r
is
ien
to
f
ac
ilit
ate
ef
f
ec
tiv
e
co
m
m
u
n
icatio
n
a
n
d
p
r
o
m
o
te
c
u
ltu
r
al
ex
c
h
an
g
e
b
et
wee
n
th
e
u
s
er
s
o
f
th
ese
two
la
n
g
u
ag
es.
Af
ter
p
r
esen
tin
g
th
e
to
p
ic
an
d
h
ig
h
lig
h
tin
g
th
e
ch
allen
g
es
ass
o
ciate
d
with
th
e
d
ev
elo
p
m
en
t
o
f
Mo
r
is
ia
2
.
0
,
s
ec
tio
n
2
ex
am
i
n
es
th
e
r
ele
v
an
t
liter
atu
r
e
o
n
m
ac
h
in
e
tr
an
s
latio
n
an
d
em
p
h
asis
es
th
e
im
p
o
r
tan
ce
o
f
u
tili
s
in
g
m
etr
ics,
s
u
ch
as
B
L
E
U,
to
ev
al
u
ate
tr
an
s
latio
n
q
u
ality
.
Sectio
n
3
d
elv
es
in
to
th
e
tech
n
ical
as
p
ec
ts
b
y
d
is
cu
s
s
in
g
th
e
m
eth
o
d
o
lo
g
y
f
o
r
m
ac
h
in
e
tr
an
s
latio
n
i
n
t
h
is
p
r
o
ject,
f
o
l
lo
wed
b
y
a
d
etailed
ex
p
lo
r
atio
n
o
f
ex
p
er
im
en
tal
d
esig
n
.
Sectio
n
4
o
u
tlin
es
th
e
r
esu
lts
an
d
ev
alu
atio
n
o
f
th
e
s
y
s
tem
,
wh
ile
s
ec
tio
n
5
p
r
esen
ts
th
e
co
n
clu
d
in
g
p
ar
t
o
f
th
is
r
esear
ch
.
2.
L
I
T
E
R
AT
U
RE
R
E
VI
E
W
T
h
is
s
ec
tio
n
aim
s
to
co
m
p
r
eh
e
n
s
iv
ely
u
n
d
e
r
s
tan
d
th
e
e
x
is
tin
g
r
esear
ch
a
n
d
estab
lis
h
a
f
o
u
n
d
atio
n
f
o
r
d
ev
elo
p
in
g
a
n
au
t
o
m
ated
tr
a
n
s
latio
n
s
y
s
tem
f
o
r
E
n
g
lis
h
an
d
Kr
eo
l
M
o
r
is
ien
.
I
t
p
r
o
v
id
es
an
in
-
d
ep
th
ex
am
in
atio
n
o
f
th
e
B
L
E
U
s
co
r
e,
m
etr
ic
f
o
r
ev
alu
atio
n
o
f
tr
a
n
s
latio
n
with
ex
p
licit o
r
d
er
in
g
(
ME
T
E
OR
)
Sco
r
e,
an
d
th
e
NST
m
eth
o
d
,
alo
n
g
with
a
co
m
p
ar
is
o
n
.
T
h
e
r
esear
c
h
er
s
h
av
e
also
ex
p
lo
r
ed
p
r
io
r
r
esear
ch
o
n
m
ac
h
in
e
tr
an
s
latio
n
b
etwe
en
E
n
g
lis
h
an
d
Kr
eo
l M
o
r
is
ien
to
id
en
tif
y
d
ef
icien
cies a
n
d
p
o
ten
tial a
r
ea
s
f
o
r
en
h
a
n
ce
m
en
t.
Nath
et
a
l.
[
1
2
]
d
ef
in
e
m
ac
h
in
e
tr
an
s
latio
n
as
a
c
o
m
p
u
tatio
n
al
p
r
o
ce
s
s
f
o
r
tr
an
s
latin
g
a
g
iv
en
s
et
o
f
wo
r
d
s
f
r
o
m
o
n
e
h
u
m
a
n
-
r
ea
d
ab
le
lan
g
u
a
g
e
in
to
a
n
o
th
er
.
Ma
ch
in
e
tr
an
s
latio
n
m
o
d
els
ca
n
f
all
in
to
th
r
ee
ca
teg
o
r
ies:
r
u
le
-
b
ased
,
s
tatis
tical
-
b
ased
,
o
r
n
eu
r
al
n
etwo
r
k
-
b
ased
.
T
h
e
em
e
r
g
en
ce
o
f
th
e
NM
T
m
o
d
el
was
d
r
iv
en
b
y
th
e
co
n
s
tr
ain
ts
en
co
u
n
ter
ed
with
r
u
le
-
b
ased
an
d
s
t
atis
tical
-
b
ased
m
ac
h
in
e
tr
an
s
latio
n
m
o
d
els.
Neu
r
al
n
etwo
r
k
-
b
ased
m
ac
h
in
e
tr
an
s
latio
n
m
o
d
els
h
a
v
e
d
em
o
n
s
tr
ated
p
o
ten
tial
f
o
r
v
a
r
io
u
s
h
u
m
an
l
an
g
u
ag
es
,
f
ea
t
u
r
in
g
an
ex
ten
s
iv
e
v
o
ca
b
u
lar
y
ac
q
u
ir
ed
f
r
o
m
a
s
u
b
s
tan
tial d
ataset
[
1
2
]
.
I
n
th
e
b
e
g
in
n
in
g
,
m
ac
h
in
e
tr
a
n
s
latio
n
r
elied
h
ea
v
ily
o
n
R
B
MT
to
f
o
r
m
u
late
g
r
a
m
m
atica
l
r
u
l
es
f
o
r
b
o
t
h
th
e
s
o
u
r
ce
an
d
tar
g
et
lan
g
u
a
g
es
[
1
3
]
.
SMT
was
th
u
s
d
ev
elo
p
ed
to
ad
d
r
ess
th
is
is
s
u
e
[
1
4
]
.
A
s
tat
is
tical
m
o
d
el
was
d
ev
elo
p
ed
b
y
a
n
aly
s
in
g
a
m
atch
ed
c
o
llectio
n
o
f
s
en
ten
ce
s
in
th
e
s
o
u
r
ce
an
d
tar
g
et
lan
g
u
ag
es
(
tr
ain
in
g
s
et)
,
wh
ich
was
th
en
u
tili
s
ed
to
g
e
n
er
ate
a
tr
an
s
latio
n
.
NM
T
is
t
h
e
m
o
s
t
r
ec
en
t
s
y
s
tem
,
wh
ich
co
n
s
id
er
s
th
e
en
tire
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
A
n
eu
r
a
l m
a
ch
in
e
t
r
a
n
s
la
tio
n
s
ystem
fo
r
K
r
eo
l
R
ep
ib
lik
Mo
r
is
a
n
d
E
n
g
lis
h
…
(
S
a
mee
r
ch
a
n
d
P
u
d
a
r
u
th
)
4979
s
en
ten
ce
an
d
ca
n
id
en
tify
co
n
n
ec
tio
n
s
b
etwe
en
p
h
r
ases
ev
en
wh
en
lo
ca
ted
f
u
r
th
er
ap
ar
t.
T
h
i
s
lead
s
to
en
h
an
ce
d
g
r
am
m
atica
l p
r
ec
is
io
n
i
n
co
m
p
ar
is
o
n
to
SMT
.
2
.
1
.
B
L
E
U
s
co
re
s
Ph
an
-
Vu
et
a
l.
[
1
5
]
n
o
ted
a
tr
a
n
s
itio
n
in
m
ac
h
in
e
tr
an
s
latio
n
to
war
d
s
an
en
d
-
to
-
e
n
d
s
tr
ateg
y
u
tili
s
in
g
d
ee
p
n
eu
r
al
n
etwo
r
k
s
.
Sig
n
if
i
ca
n
t
ad
v
an
ce
m
e
n
ts
h
av
e
b
ee
n
m
ad
e
in
th
e
s
tate
o
f
th
e
ar
t
f
o
r
wid
ely
s
p
o
k
e
n
lan
g
u
ag
e
p
air
s
lik
e
E
n
g
lis
h
-
Fr
en
ch
o
r
E
n
g
lis
h
-
C
h
in
ese.
T
h
ei
r
s
t
u
d
y
f
o
c
u
s
ed
o
n
e
n
h
an
ci
n
g
E
n
g
lis
h
-
Vietn
am
ese
tr
an
s
latio
n
s
:
i
)
co
n
s
tr
u
ctin
g
th
e
m
o
s
t
ex
ten
s
iv
e
o
p
en
Vietn
am
ese
-
E
n
g
lis
h
co
r
p
u
s
an
d
ii
)
co
n
d
u
ctin
g
co
m
p
r
eh
e
n
s
iv
e
tr
ials
u
s
in
g
s
tate
-
of
-
th
e
-
a
r
t
n
eu
r
al
m
o
d
els
to
attain
th
e
h
i
g
h
est
B
L
E
U
s
co
r
es.
Kr
ü
g
e
r
[
1
6
]
ex
p
lo
r
ed
th
e
co
g
n
itiv
e
lin
g
u
is
tic
v
iewp
o
in
t
t
o
ex
a
m
in
e
h
o
w
h
u
m
a
n
tr
a
n
s
latio
n
co
u
ld
b
e
m
o
d
elled
i
n
ter
m
s
o
f
co
n
tex
t
an
d
m
ea
n
i
n
g
.
T
h
is
r
esear
ch
d
em
o
n
s
tr
ated
h
o
w
NM
T
in
ter
p
r
ets
lin
g
u
is
tic
m
ea
n
in
g
an
d
to
w
h
at
d
eg
r
ee
it in
co
r
p
o
r
ates c
o
n
tex
tu
al
in
f
o
r
m
atio
n
in
to
t
h
is
p
r
o
ce
s
s
.
B
atsu
k
h
[
1
7
]
e
x
am
in
ed
th
e
ad
v
a
n
tag
es
an
d
d
is
ad
v
an
ta
g
es
o
f
co
n
te
m
p
o
r
a
r
y
n
eu
r
al
m
ac
h
in
e
tr
an
s
latio
n
.
W
h
ile
NM
T
p
r
o
v
i
d
ed
a
m
o
r
e
s
tr
aig
h
tf
o
r
war
d
a
p
p
r
o
ac
h
to
m
o
d
ellin
g
,
lead
in
g
t
o
ef
f
ec
tiv
e
wo
r
d
an
d
s
en
ten
ce
s
tr
u
ctu
r
e
im
p
lem
en
t
atio
n
,
it
s
o
m
etim
es
le
d
to
d
is
to
r
t
ed
s
en
ten
ce
s
tr
u
ctu
r
es
a
n
d
b
o
u
n
d
ar
ies
wh
en
tr
an
s
latin
g
u
n
tr
ai
n
ed
d
ata.
Z
h
an
g
et
a
l.
[
1
8
]
s
u
g
g
ested
th
at
d
esp
ite
th
e
im
p
r
ess
iv
e
p
e
r
f
o
r
m
an
ce
o
f
NM
T
,
it
n
ee
d
s
h
elp
to
ac
cu
r
ately
ca
p
t
u
r
e
th
e
alig
n
m
en
t
b
etwe
en
t
h
e
in
p
u
ts
an
d
th
e
o
u
tp
u
ts
d
u
r
in
g
th
e
tr
an
s
latio
n
p
r
o
ce
s
s
.
T
h
is
lack
o
f
alig
n
m
e
n
t
g
iv
es
r
is
e
to
th
r
ee
c
h
allen
g
es:
in
ter
p
r
etin
g
th
e
tr
a
n
s
latio
n
p
r
o
ce
s
s
,
im
p
o
s
in
g
lex
ical
co
n
s
tr
ain
ts
,
an
d
ap
p
l
y
in
g
s
tr
u
ctu
r
al
co
n
s
tr
ain
ts
.
T
h
ese
is
s
u
es
co
m
p
licate
th
e
d
ev
elo
p
m
en
t
o
f
n
ew
NM
T
ar
ch
itectu
r
es a
n
d
r
estrict
th
eir
p
r
ac
tical
ap
p
licatio
n
s
.
E
m
n
a
et
a
l.
[
1
9
]
s
tu
d
ied
th
e
im
p
o
r
ta
n
ce
o
f
co
m
p
r
eh
e
n
d
i
n
g
an
d
p
r
o
ce
s
s
in
g
u
n
clea
r
ly
ar
ticu
lated
s
p
ee
ch
.
T
h
e
y
d
e
v
elo
p
ed
a
N
MT
s
y
s
tem
f
o
r
tr
a
n
s
latin
g
th
e
tu
n
is
ian
d
ialec
t
(
T
D)
to
m
o
d
er
n
s
tan
d
ar
d
Ar
ab
ic
(
MSA)
.
T
h
is
ty
p
e
o
f
NM
T
t
ask
f
ac
ed
ch
allen
g
es
d
u
e
t
o
lim
ited
tr
ain
in
g
d
ata
av
ailab
l
e
f
o
r
lo
w
-
r
eso
u
r
ce
lan
g
u
ag
es
s
u
ch
as
T
D.
B
y
b
u
il
d
in
g
a
p
a
r
allel
co
r
p
u
s
o
f
T
D
-
MSA
an
d
ef
f
ec
tiv
ely
u
tili
s
in
g
it,
th
ey
f
o
r
m
u
lated
a
s
etu
p
f
o
r
a
n
eu
r
al
tr
a
n
s
latio
n
m
o
d
el
th
at
ac
h
ie
v
ed
an
im
p
r
es
s
iv
e
B
L
E
U
s
co
r
e
o
f
6
7
.
5
6
%.
2
.
2
.
T
he
pu
rpo
s
e
o
f
t
he
B
L
E
U
a
lg
o
rit
hm
Ad
lao
n
an
d
Ma
r
co
s
[
2
0
]
s
o
u
g
h
t
to
cr
ea
te
a
p
ar
allel
co
r
p
u
s
as
an
ess
en
tial
to
o
l
in
m
ac
h
in
e
lear
n
in
g
-
b
ased
tr
an
s
latio
n
b
y
em
p
lo
y
i
n
g
a
r
ec
u
r
r
en
t
n
e
u
r
al
n
etwo
r
k
(
R
NN)
with
in
th
e
Op
en
N
MT
f
r
am
ewo
r
k
.
T
h
e
q
u
ality
o
f
th
e
tr
a
n
s
latio
n
was
ass
es
s
ed
u
s
in
g
th
e
B
L
E
U
s
co
r
e.
A
s
u
b
wo
r
d
u
n
it
tr
an
s
latio
n
was
co
n
d
u
cted
to
r
ec
tify
in
c
o
n
s
is
ten
cies
in
th
e
o
r
ig
in
al
d
ataset,
lead
in
g
to
a
n
im
p
r
o
v
ed
B
L
E
U
s
co
r
e
o
f
2
2
.
8
7
co
m
p
ar
e
d
to
th
e
in
itial
2
0
.
0
1
.
Villan
u
ev
a
et
a
l.
[
2
1
]
i
n
v
esti
g
ated
th
e
i
n
tr
icac
ies
o
f
co
n
v
er
tin
g
th
e
tr
ad
itio
n
al
Ph
il
ip
p
in
e
lan
g
u
a
g
e
to
E
n
g
lis
h
.
T
h
ey
p
r
o
p
o
s
ed
a
m
o
b
ile
-
o
r
ien
te
d
tr
an
s
latio
n
s
y
s
tem
eq
u
ip
p
ed
with
o
b
ject
d
etec
tio
n
to
aid
tr
a
v
eller
s
.
T
h
e
s
y
s
tem
u
tili
s
ed
NM
T
to
co
n
v
er
t
Fil
ip
in
o
to
C
eb
u
an
o
lan
g
u
ag
e
a
n
d
v
ice
v
er
s
a,
d
r
aw
in
g
in
p
u
t
f
r
o
m
th
e
u
s
er
’
s
k
ey
b
o
ar
d
an
d
ex
tr
ac
ti
n
g
tex
t
s
tr
in
g
s
f
r
o
m
id
en
tifie
d
o
b
jects
in
im
ag
es.
T
h
ese
s
tu
d
ies
h
ig
h
lig
h
t
th
e
ch
allen
g
es
f
ac
ed
in
n
eu
r
al
m
a
ch
in
e
tr
an
s
latio
n
,
s
u
ch
as
ca
p
tu
r
in
g
alig
n
m
e
n
t
b
etwe
en
in
p
u
t
an
d
o
u
tp
u
t,
d
ea
li
n
g
with
lim
ited
tr
ain
in
g
d
ata
f
o
r
lo
w
-
r
eso
u
r
ce
la
n
g
u
a
g
es,
an
d
im
p
r
o
v
in
g
th
e
q
u
ality
o
f
t
r
an
s
latio
n
th
r
o
u
g
h
tech
n
iq
u
es
lik
e
s
u
b
wo
r
d
u
n
it
tr
an
s
latio
n
an
d
o
b
ject
d
etec
ti
o
n
.
T
h
e
m
o
d
el
f
o
r
Fil
ip
in
o
-
C
eb
u
an
o
ac
h
ie
v
ed
a
B
L
E
U
s
co
r
e
o
f
3
1
.
1
,
wh
ile
th
e
C
eb
u
an
o
-
Fil
ip
in
o
m
o
d
el
s
co
r
ed
3
1
.
6
.
T
h
e
B
L
E
U
s
co
r
e
m
e
asu
r
es
th
e
p
r
ec
is
io
n
o
f
m
at
ch
in
g
wo
r
d
s
eq
u
en
ce
s
b
etwe
en
a
“
ca
n
d
id
ate
”
m
ac
h
i
n
e
tr
an
s
latio
n
an
d
o
n
e
o
r
m
o
r
e
“
r
ef
er
en
ce
”
h
u
m
an
tr
an
s
latio
n
s
[
2
2
]
.
T
h
e
alg
o
r
ith
m
was
cr
ea
ted
to
co
m
p
a
r
e
s
en
ten
ce
s
with
in
a
co
r
p
u
s
,
ca
lcu
lat
in
g
n
-
g
r
am
m
atch
es
at
th
e
s
en
ten
ce
lev
el
an
d
th
en
ag
g
r
eg
atin
g
th
em
i
n
to
an
o
v
er
al
l
s
co
r
e
f
o
r
th
e
co
r
p
u
s
.
Usi
n
g
th
e
B
L
E
U
alg
o
r
ith
m
in
th
ese
s
tu
d
ies d
em
o
n
s
tr
ates its
ef
f
ec
tiv
en
ess
in
ev
alu
atin
g
t
h
e
q
u
ality
a
n
d
ac
cu
r
ac
y
o
f
m
a
ch
in
e
tr
an
s
latio
n
s
.
2
.
3
.
E
m
pirica
l r
ev
iew
o
f
t
r
a
ns
la
t
io
n qua
lity
wit
h B
L
E
U
s
co
re
s
An
em
p
ir
ical
ev
alu
atio
n
co
m
p
ar
ed
th
e
tr
an
s
latio
n
q
u
ality
o
f
d
if
f
er
en
t a
v
ailab
le
s
y
s
tem
s
f
o
r
en
d
u
s
er
s
in
T
h
ailan
d
to
g
ai
n
in
s
ig
h
t
in
t
o
th
e
o
v
er
all
q
u
ality
o
f
tr
an
s
la
tio
n
s
u
s
ed
[
2
3
]
.
T
h
e
d
if
f
ic
u
lty
o
f
tr
an
s
latin
g
T
h
ai
to
E
n
g
lis
h
is
ev
id
en
t
f
r
o
m
th
e
h
ig
h
er
r
o
r
r
ate
o
f
4
7
.
2
%
an
d
a
lo
w
B
L
E
U
s
co
r
e
o
f
2
1
.
Ho
we
v
er
,
d
esp
ite
th
e
h
ig
h
tr
an
s
latio
n
er
r
o
r
r
ate,
u
s
er
s
co
r
r
ec
tly
an
s
wer
ed
a
p
p
r
o
x
im
ately
6
0
%
o
f
q
u
esti
o
n
s
in
r
ea
d
in
g
c
o
m
p
r
eh
e
n
s
io
n
test
s
u
s
in
g
o
u
tp
u
t
f
r
o
m
m
ac
h
in
e
tr
a
n
s
latio
n
s
y
s
tem
s
.
Pap
in
en
i
et
a
l.
[
2
2
]
p
r
ev
io
u
s
ly
claim
ed
t
h
at
B
L
E
U
co
u
ld
s
p
ee
d
u
p
m
ac
h
i
n
e
tr
an
s
latio
n
b
y
en
a
b
lin
g
r
esear
ch
e
r
s
to
f
o
c
u
s
o
n
p
r
ac
tical
m
o
d
ellin
g
id
ea
s
q
u
ic
k
ly
.
T
h
e
r
esu
lts
o
f
B
L
E
U
wer
e
clo
s
ely
ass
o
ciate
d
with
h
u
m
an
e
v
alu
atio
n
s
as it a
v
er
ag
ed
o
u
t in
d
iv
id
u
al
s
en
ten
ce
ju
d
g
m
en
t e
r
r
o
r
s
ac
r
o
s
s
a
test
co
r
p
u
s
r
ath
er
th
an
s
tr
iv
in
g
f
o
r
p
r
ec
is
e
h
u
m
an
ju
d
g
m
en
t
f
o
r
ea
ch
s
en
ten
ce
,
d
em
o
n
s
tr
atin
g
th
at
q
u
an
tity
ca
n
lead
to
q
u
ality
.
Usi
n
g
B
L
E
U
s
co
r
es
to
ev
alu
a
te
m
ac
h
in
e
tr
a
n
s
latio
n
s
h
as
p
r
o
v
en
e
f
f
ec
tiv
e
a
n
d
r
eliab
le.
T
h
er
e
f
o
r
e,
it
is
ap
p
r
o
p
r
iate
to
c
o
n
s
id
er
t
h
e
B
L
E
U
s
co
r
e
as
a
v
iab
le
m
et
r
ic
f
o
r
ass
ess
in
g
th
e
q
u
ality
an
d
ac
cu
r
ac
y
o
f
m
ac
h
i
n
e
tr
an
s
latio
n
s
.
2
.
4
.
Co
m
pa
riso
n bet
wee
n B
L
E
U
a
n
d M
E
T
E
O
R
B
an
er
jee
an
d
L
a
v
ie
[
2
4
]
s
u
g
g
e
s
ted
th
at
th
e
s
tatic
b
r
e
v
ity
p
e
n
a
lty
in
B
L
E
U
d
id
n
o
t
s
u
f
f
icien
tl
y
ad
d
r
ess
th
e
r
ec
all
is
s
u
e.
T
h
ey
p
r
o
p
o
s
ed
th
at
a
d
ir
ec
t
ass
es
s
m
en
t
o
f
g
r
am
m
atica
lity
(
o
r
wo
r
d
o
r
d
er
)
c
o
u
ld
m
o
r
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
13
,
No
.
4
,
Dec
em
b
er
2
0
2
4
:
4
9
7
6
-
4
9
8
7
4980
ef
f
ec
tiv
ely
ca
p
tu
r
e
th
e
s
ig
n
i
f
ican
ce
o
f
g
r
am
m
atica
lity
as
a
co
m
p
o
n
e
n
t
in
th
e
m
ac
h
in
e
tr
an
s
l
atio
n
m
etr
ic,
lead
in
g
to
im
p
r
o
v
ed
alig
n
m
en
t
with
h
u
m
an
ev
alu
atio
n
s
o
f
tr
a
n
s
latio
n
q
u
ality
.
C
o
n
s
eq
u
e
n
tly
,
s
en
ten
c
e
o
r
s
eg
m
en
t
-
lev
el
B
L
E
U
s
co
r
es
m
ay
n
o
t
b
e
co
n
s
id
er
ed
r
elev
a
n
t.
ME
T
E
OR
w
as
d
esig
n
ed
to
r
ec
tify
th
e
s
h
o
r
tco
m
in
g
s
o
f
B
L
E
U
b
y
ass
ess
in
g
a
tr
an
s
latio
n
b
y
ca
lcu
latin
g
a
s
co
r
e
d
er
iv
e
d
f
r
o
m
ex
p
licit
wo
r
d
-
to
-
wo
r
d
c
o
r
r
e
s
p
o
n
d
en
ce
s
b
etwe
e
n
th
e
tr
an
s
latio
n
a
n
d
a
r
ef
e
r
en
ce
v
er
s
io
n
[
2
4
]
.
A
g
ar
wal
an
d
L
a
v
ie
[
2
5
]
e
n
d
o
r
s
ed
th
e
ME
T
E
OR
m
ea
s
u
r
e,
wh
ich
in
v
o
lv
es
an
in
itial
p
h
ase
o
f
p
r
e
cisely
m
ap
p
in
g
wo
r
d
s
f
r
o
m
tw
o
tex
ts
,
f
o
llo
wed
b
y
a
s
ec
o
n
d
p
h
ase
wh
er
e
m
a
p
p
ed
n
-
g
r
am
s
ar
e
d
iv
id
e
d
in
to
s
u
b
s
ets.
T
h
e
s
eq
u
en
ce
th
en
s
ele
cts
th
e
m
o
s
t
s
ig
n
if
ican
t
s
u
b
s
et
as
th
e
r
esu
ltin
g
alig
n
m
en
t
s
et.
E
ac
h
n
-
g
r
am
f
r
o
m
th
e
ca
n
d
id
ate
tex
t
ca
n
b
e
p
air
ed
with
its
clo
s
est
co
r
r
esp
o
n
d
in
g
n
-
g
r
am
f
r
o
m
th
e
r
ef
er
en
ce
tex
t [
2
4
]
.
2
.
5
.
NIS
T
m
et
ho
d o
f
e
v
a
lua
t
io
n
Sin
ce
2
0
0
2
,
Natio
n
al
I
n
s
titu
te
o
f
Stan
d
ar
d
s
an
d
T
ec
h
n
o
lo
g
y
(
NI
ST
)
h
as
b
ee
n
lead
in
g
o
p
e
n
e
v
alu
atio
n
s
lik
e
Op
en
MT
,
wh
ich
s
er
v
e
as
a
p
latf
o
r
m
f
o
r
ex
p
e
r
im
en
tin
g
with
ev
alu
atio
n
m
eth
o
d
s
ap
p
li
ca
b
le
to
s
p
o
n
s
o
r
ed
MT
tech
n
o
lo
g
y
ass
ess
m
en
ts
.
NI
ST
’
s
Me
tr
ics
f
o
r
m
ac
h
in
e
tr
an
s
latio
n
ch
allen
g
e
o
f
f
e
r
s
an
o
p
p
o
r
tu
n
ity
to
ex
p
l
o
r
e
an
d
ad
v
o
ca
te
f
o
r
n
ew
tec
h
n
iq
u
es
th
at
en
h
an
ce
t
h
e
m
ea
s
u
r
em
en
t
s
cien
ce
s
in
MT
ev
alu
atio
n
s
.
NI
ST
co
o
r
d
in
ate
d
an
d
ex
ec
u
ted
th
e
d
ef
en
s
e
ad
v
a
n
ce
d
r
esear
ch
p
r
o
jects
ag
en
cy
(
DARP
A
)
b
r
o
ad
o
p
er
atio
n
al
la
n
g
u
ag
e
tr
a
n
s
latio
n
(
B
OL
T
)
ass
es
s
m
en
ts
o
f
s
p
ee
ch
-
to
-
tex
t
a
n
d
tex
t
-
to
-
tex
t
MT
tech
n
o
lo
g
y
an
d
th
e
en
d
-
to
-
e
n
d
MT
s
y
s
tem
s
th
at
f
ac
ilit
ate
r
ea
l
-
tim
e
s
p
o
k
e
n
co
m
m
u
n
icatio
n
b
etwe
en
s
p
ea
k
er
s
o
f
d
if
f
er
en
t
lan
g
u
a
g
es [
2
6
]
.
T
h
e
NI
ST
m
eth
o
d
o
f
ev
alu
atio
n
,
e
x
em
p
lifie
d
th
r
o
u
g
h
in
itiativ
es
s
u
ch
as
Op
e
n
MT
an
d
th
e
m
etr
ics
f
o
r
m
ac
h
in
e
tr
an
s
latio
n
ch
allen
g
e
,
h
as p
lay
ed
a
p
iv
o
tal
r
o
le
in
ad
v
an
cin
g
t
h
e
m
ea
s
u
r
em
e
n
t scien
ce
s
in
m
ac
h
in
e
tr
a
n
s
latio
n
ev
alu
atio
n
s
.
T
h
e
B
OL
T
p
r
o
g
r
am
’
s
g
o
al
o
f
b
r
id
g
in
g
th
e
lan
g
u
a
g
e
b
ar
r
ie
r
b
etwe
en
E
n
g
lis
h
-
s
p
ea
k
in
g
in
d
iv
id
u
als
an
d
non
-
E
n
g
lis
h
-
s
p
ea
k
in
g
p
o
p
u
latio
n
s
u
n
d
er
s
co
r
es
th
e
im
p
o
r
t
an
ce
o
f
ef
f
ec
tiv
e
co
m
m
u
n
ic
atio
n
an
d
ef
f
icien
t
in
f
o
r
m
atio
n
r
etr
iev
al
th
r
o
u
g
h
m
ac
h
in
e
tr
an
s
latio
n
tech
n
o
lo
g
y
.
W
ith
a
f
o
cu
s
o
n
en
ab
lin
g
m
u
lti
-
tu
r
n
co
m
m
u
n
icatio
n
in
b
o
th
tex
t
an
d
s
p
ee
ch
,
t
h
e
p
r
o
g
r
am
aim
ed
to
f
ac
ilit
ate
s
ea
m
less
i
n
ter
ac
tio
n
s
b
etwe
e
n
in
d
iv
id
u
als
ac
r
o
s
s
d
if
f
er
en
t
la
n
g
u
ag
es.
T
h
is
in
v
o
lv
ed
en
a
b
lin
g
E
n
g
lis
h
s
p
ea
k
er
s
to
co
m
p
r
eh
en
d
a
wid
e
r
an
g
e
o
f
f
o
r
e
ig
n
-
lan
g
u
ag
e
s
o
u
r
ce
s
,
s
u
ch
as
ch
at
co
n
v
e
r
s
atio
n
s
an
d
in
f
o
r
m
al
m
ess
ag
in
g
;
eq
u
i
p
p
in
g
th
em
with
th
e
ca
p
ac
ity
to
lo
ca
te
s
p
ec
if
ic
in
f
o
r
m
atio
n
i
n
th
ese
s
o
u
r
ce
s
u
s
in
g
n
atu
r
al
-
la
n
g
u
a
g
e
q
u
er
ies s
wif
tly
; a
n
d
f
ac
ilit
atin
g
m
u
lti
-
tu
r
n
c
o
m
m
u
n
icatio
n
in
b
o
th
tex
t a
n
d
s
p
e
ec
h
with
n
o
n
-
E
n
g
lis
h
s
p
ea
k
er
s
[
2
6
]
.
T
h
e
ex
is
tin
g
b
o
d
y
o
f
r
esear
c
h
p
r
o
v
id
es
v
alu
a
b
le
in
s
ig
h
ts
in
to
th
e
ch
allen
g
es
an
d
o
p
p
o
r
tu
n
ities
o
f
d
ev
elo
p
in
g
an
a
u
to
m
ated
tr
a
n
s
latio
n
s
y
s
tem
f
o
r
E
n
g
lis
h
an
d
Kr
eo
l
Mo
r
is
ien
.
As
in
d
icate
d
b
y
p
r
e
v
io
u
s
s
tu
d
ies,
b
u
ild
in
g
a
u
to
m
ated
tr
an
s
latio
n
s
y
s
tem
s
f
o
r
d
iv
er
s
e
lan
g
u
a
g
es
lik
e
E
n
g
lis
h
an
d
Kr
eo
l
Mo
r
is
ien
p
r
esen
ts
b
o
th
h
u
r
d
les
a
n
d
p
o
ten
tial
b
en
ef
its
.
C
o
n
s
is
ten
t
en
h
an
ce
m
en
t
o
f
tr
an
s
latio
n
p
r
ec
is
io
n
an
d
ass
ess
m
en
t
s
tan
d
ar
d
s
is
cr
u
cial
to
im
p
r
o
v
in
g
th
e
o
v
er
all
ca
lib
r
e
o
f
m
ac
h
in
e
tr
an
s
latio
n
s
.
Ad
d
itio
n
ally
,
au
to
m
atic
ev
alu
atio
n
s
y
s
tem
s
,
s
u
ch
as
th
e
B
L
E
U
m
etr
ic
an
d
its
m
o
d
if
ied
v
er
s
io
n
s
,
h
av
e
p
r
o
v
en
b
e
n
ef
icial
in
ev
alu
atin
g
m
ac
h
in
e
tr
an
s
latio
n
q
u
ality
an
d
e
n
s
u
r
in
g
th
at
it
m
ee
ts
th
e
n
ec
ess
ar
y
s
tan
d
a
r
d
s
f
o
r
ef
f
ec
tiv
e
co
m
m
u
n
icatio
n
i
n
v
ar
i
o
u
s
lan
g
u
ag
e
p
air
s
an
d
d
o
m
ain
s
.
3.
M
E
T
H
O
DO
L
O
G
Y
T
h
e
m
eth
o
d
o
l
o
g
y
u
s
ed
in
th
is
s
tu
d
y
is
cr
u
cial
f
o
r
e
n
s
u
r
in
g
t
h
e
v
alid
ity
an
d
r
eliab
ilit
y
o
f
t
h
e
r
esear
ch
f
in
d
in
g
s
.
T
h
is
s
ec
tio
n
p
r
esen
t
s
th
e
s
y
s
tem
atic
ap
p
r
o
ac
h
to
d
ev
elo
p
i
n
g
th
e
m
ac
h
in
e
tr
an
s
latio
n
s
y
s
tem
f
o
r
co
n
v
er
tin
g
Kr
eo
l
Mo
r
is
ien
to
E
n
g
lis
h
an
d
v
ice
v
e
r
s
a.
T
h
e
m
eth
o
d
o
l
o
g
y
e
m
p
lo
y
e
d
in
th
i
s
s
tu
d
y
in
v
o
lv
e
d
a
co
m
b
in
atio
n
o
f
q
u
an
titativ
e
an
d
q
u
alitativ
e
r
esear
ch
m
eth
o
d
s
.
3
.
1
.
Da
t
a
s
et
a
nd
da
t
a
co
llect
io
n
T
h
e
s
tu
d
y
u
tili
s
es
a
d
ataset
o
f
5
0
,
0
0
0
E
n
g
lis
h
s
en
ten
ce
s
p
ai
r
ed
with
th
eir
tr
an
s
lated
co
u
n
t
er
p
ar
ts
in
KR
M
.
T
h
e
d
ataset
in
clu
d
es
all
th
e
wo
r
d
s
f
r
o
m
th
e
th
ir
d
ed
it
io
n
o
f
Dik
s
io
n
e
r
Mo
r
is
ien
.
T
h
e
d
ataset
f
ea
tu
r
es
a
lar
g
e
v
ar
iety
o
f
s
h
o
r
t
s
en
ten
ce
s
(
ar
o
u
n
d
4
w
o
r
d
s
)
,
m
ed
iu
m
-
s
ized
s
en
ten
ce
s
(
b
etwe
en
5
to
1
0
wo
r
d
s
)
,
an
d
l
o
n
g
s
en
ten
ce
s
(
ab
o
v
e
1
0
wo
r
d
s
)
.
T
h
ese
s
en
ten
ce
s
p
r
im
ar
ily
r
ef
le
ct
ev
er
y
d
ay
u
s
e
f
o
r
g
e
n
er
al
p
u
r
p
o
s
es.
T
h
e
d
ataset
was
co
llected
th
r
o
u
g
h
v
ar
io
u
s
s
o
u
r
ce
s
,
in
clu
d
in
g
o
n
lin
e
r
eso
u
r
ce
s
,
b
o
o
k
s
,
an
d
co
n
v
er
s
atio
n
s
with
n
ativ
e
s
p
ea
k
er
s
o
f
KR
M
.
3
.
2
.
Da
t
a
pre
-
pro
ce
s
s
ing
T
h
e
d
ataset
was p
r
ep
ar
ed
f
o
r
a
n
aly
s
is
b
y
im
p
lem
en
tin
g
d
ata
p
r
e
-
p
r
o
ce
s
s
in
g
m
eth
o
d
s
.
T
h
es
e
m
eth
o
d
s
in
v
o
lv
ed
elim
in
atin
g
p
u
n
ctu
at
io
n
,
co
n
v
er
tin
g
all
te
x
t
to
l
o
wer
ca
s
e,
an
d
d
iv
id
in
g
th
e
s
en
ten
ce
s
in
to
s
ep
ar
ate
wo
r
d
s
.
T
h
e
d
ata
p
r
e
-
p
r
o
ce
s
s
in
g
s
tag
e
was
n
ec
ess
ar
y
to
e
n
s
u
r
e
a
clea
n
er
a
n
d
m
o
r
e
m
an
ag
ea
b
le
d
ataset
f
o
r
an
aly
s
is
.
T
h
e
s
tep
s
f
o
r
d
ata
p
r
e
-
p
r
o
ce
s
s
in
g
d
etails ca
n
b
e
d
iv
id
ed
as f
o
llo
ws
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
A
n
eu
r
a
l m
a
ch
in
e
t
r
a
n
s
la
tio
n
s
ystem
fo
r
K
r
eo
l
R
ep
ib
lik
Mo
r
is
a
n
d
E
n
g
lis
h
…
(
S
a
mee
r
ch
a
n
d
P
u
d
a
r
u
th
)
4981
3
.
2
.
1
.
Da
t
a
t
o
k
enis
a
t
io
n
T
h
e
d
ata
p
r
e
-
p
r
o
ce
s
s
in
g
s
tag
e
in
th
is
s
tu
d
y
in
v
o
l
v
es
lo
ad
in
g
th
e
d
ataset
o
f
E
n
g
lis
h
s
en
ten
ce
s
p
air
ed
with
th
eir
tr
an
s
lated
co
u
n
ter
p
a
r
ts
in
th
e
s
y
s
tem
.
T
h
is
s
tu
d
y
e
m
p
lo
y
ed
d
ata
to
k
e
n
is
atio
n
as
a
d
ata
p
r
e
-
p
r
o
ce
s
s
in
g
m
eth
o
d
t
o
co
n
v
er
t
th
e
E
n
g
lis
h
an
d
KR
M
s
en
ten
ce
s
in
to
a
lis
t
o
f
in
d
i
v
id
u
al
wo
r
d
s
o
r
to
k
e
n
s
.
Sev
er
al
to
k
en
is
atio
n
m
eth
o
d
s
ar
e
av
ailab
le
in
th
e
s
y
s
tem
,
in
clu
d
in
g
Su
b
w
o
r
d
T
e
x
tEn
co
d
er
f
o
r
b
r
ea
k
in
g
d
o
wn
te
x
t
d
ata
in
to
s
u
b
wo
r
d
u
n
its
an
d
B
y
teT
ex
tEn
c
o
d
er
f
o
r
d
iv
id
in
g
tex
t d
ata
in
t
o
to
k
e
n
s
at
th
e
b
y
te
le
v
el.
3
.
2
.
2
.
Vo
ca
bu
la
ry
g
ener
a
t
io
n
Vo
ca
b
u
lar
y
g
en
er
atio
n
is
a
cr
u
cial
s
tep
in
d
ata
p
r
e
-
p
r
o
ce
s
s
in
g
.
I
t
in
v
o
lv
es
cr
ea
tin
g
a
co
m
p
r
eh
en
s
iv
e
lis
t
o
f
all
u
n
iq
u
e
wo
r
d
s
in
t
h
e
d
ataset.
T
h
is
lis
t
m
ap
s
wo
r
d
s
to
n
u
m
e
r
ical
r
ep
r
esen
tatio
n
s
(
wo
r
d
em
b
ed
d
i
n
g
s
)
f
o
r
f
u
r
th
er
an
aly
s
is
.
T
h
e
v
o
ca
b
u
lar
y
g
e
n
er
atio
n
s
tep
in
d
ata
p
r
e
-
p
r
o
ce
s
s
in
g
en
s
u
r
es
th
at
ev
er
y
u
n
iq
u
e
wo
r
d
in
th
e
d
ataset
is
ac
co
u
n
ted
f
o
r
an
d
ass
ig
n
ed
a
n
u
m
er
ical
r
e
p
r
ese
n
tatio
n
,
allo
win
g
f
o
r
f
u
r
th
e
r
s
t
u
d
y
a
n
d
m
o
d
ellin
g
.
T
h
is
p
r
o
ce
d
u
r
e
o
f
f
e
r
s
v
ar
io
u
s
m
eth
o
d
s
f
o
r
c
r
ea
tin
g
v
o
c
ab
u
la
r
ies.
I
t
also
id
en
tif
ies
N
-
g
r
am
s
with
in
s
en
ten
ce
s
to
ef
f
ec
tiv
ely
ch
ar
ac
ter
is
e
th
e
ess
en
ce
an
d
s
ig
n
i
f
ican
ce
co
n
v
ey
e
d
b
y
th
e
wo
r
d
s
.
3
.
2
.
3
.
F
ea
t
ure
s
ca
lin
g
Featu
r
e
s
ca
lin
g
is
a
d
ju
s
tin
g
t
h
e
s
ca
le
o
f
f
ea
tu
r
es to
a
c
o
n
s
is
ten
t r
an
g
e
t
o
p
r
ev
en
t
ce
r
tain
f
e
atu
r
es f
r
o
m
d
is
p
r
o
p
o
r
tio
n
ately
im
p
ac
tin
g
th
e
m
o
d
el
’
s
p
er
f
o
r
m
an
ce
.
Var
io
u
s
m
eth
o
d
s
ar
e
av
ailab
le
f
o
r
f
ea
tu
r
e
s
ca
lin
g
,
in
clu
d
in
g
Stan
d
ar
d
ize,
wh
ich
en
ab
les
n
o
r
m
alis
atio
n
o
f
th
e
i
n
p
u
t
d
ata.
N
o
r
m
aliza
tio
n
p
lay
a
s
ig
n
if
ican
t
r
o
le
in
en
s
u
r
in
g
th
e
s
tab
ilit
y
an
d
ef
f
ic
ien
cy
o
f
t
r
ain
in
g
i
n
n
eu
r
al
m
a
ch
in
e
tr
an
s
latio
n
m
o
d
els.
3
.
2
.
4
.
E
nco
din
g
ca
t
eg
o
rica
l v
a
ria
bles
T
h
is
d
ata
p
r
e
-
p
r
o
ce
s
s
in
g
s
tep
i
n
v
o
lv
es
co
n
v
er
tin
g
ca
teg
o
r
ical
v
ar
iab
les
in
to
n
u
m
er
ical
r
ep
r
e
s
en
tatio
n
s
th
at
th
e
al
g
o
r
ith
m
s
ca
n
u
n
d
er
s
tan
d
.
C
ateg
o
r
ical
v
ar
ia
b
les
m
u
s
t
b
e
c
o
n
v
e
r
ted
t
o
n
u
m
er
ical
v
alu
es
b
ef
o
r
e
th
e
y
ca
n
b
e
u
tili
s
ed
in
m
ac
h
in
e
lear
n
in
g
m
o
d
els.
T
h
e
r
e
ar
e
d
if
f
er
en
t
way
s
in
wh
ich
th
is
ca
n
d
o
n
e.
T
h
is
in
clu
d
e
m
eth
o
d
s
s
u
ch
as lab
el
en
c
o
d
in
g
,
o
n
e
-
h
o
t e
n
co
d
in
g
,
b
i
n
ar
y
e
n
co
d
in
g
a
n
d
f
r
eq
u
e
n
cy
e
n
co
d
in
g
.
3
.
3
.
T
o
k
eniza
t
io
n
T
o
k
en
is
atio
n
is
a
f
u
n
d
am
e
n
tal
s
tep
in
d
ata
p
r
ep
r
o
ce
s
s
in
g
.
I
t
in
v
o
lv
es
co
n
v
er
tin
g
th
e
in
p
u
t
tex
t
in
to
a
s
er
ies
o
f
to
k
en
s
(
wo
r
d
s
o
r
s
u
b
wo
r
d
s
)
s
u
itab
le
f
o
r
in
p
u
t
in
t
o
th
e
d
ee
p
n
eu
r
al
m
ac
h
in
e
tr
an
s
latio
n
(
DNM
T
)
m
o
d
el.
Sev
er
al
to
k
en
is
atio
n
ap
p
r
o
ac
h
es
h
av
e
b
ee
n
s
u
g
g
e
s
ted
f
o
r
NM
T
an
d
DNM
T
,
s
u
ch
as
wo
r
d
-
lev
el,
ch
ar
ac
ter
-
lev
el,
an
d
s
u
b
wo
r
d
-
lev
el
to
k
en
is
atio
n
[
2
7
]
.
T
o
k
e
n
is
in
g
wo
r
d
s
at
th
e
lev
el
o
f
in
d
iv
id
u
al
wo
r
d
u
n
its
is
a
s
im
p
le
m
eth
o
d
,
b
u
t
it
m
ay
n
ee
d
h
el
p
wi
th
o
u
t
-
of
-
v
o
ca
b
u
lar
y
(
OOV)
ter
m
s
a
n
d
lan
g
u
a
g
es
with
co
m
p
lex
m
o
r
p
h
o
lo
g
y
.
T
o
k
en
is
in
g
at
th
e
ch
ar
ac
ter
lev
el
ca
n
ef
f
ec
tiv
el
y
p
r
o
ce
s
s
o
u
t
-
of
-
v
o
ca
b
u
lar
y
w
o
r
d
s
an
d
is
r
esil
ien
t
to
v
ar
iatio
n
s
in
s
p
ellin
g
.
Ho
wev
er
,
th
is
m
eth
o
d
p
r
o
d
u
c
es
ex
ten
d
ed
s
eq
u
en
ce
s
th
at
m
ay
p
o
s
e
ch
allen
g
es
f
o
r
th
e
m
o
d
el
d
u
r
in
g
p
r
o
ce
s
s
in
g
.
T
o
ad
d
r
ess
th
e
lim
itatio
n
s
o
f
wo
r
d
-
lev
el
an
d
ch
ar
ac
ter
-
le
v
el
to
k
en
is
atio
n
,
s
u
b
wo
r
d
-
lev
el
to
k
en
is
atio
n
h
as
g
ain
ed
p
o
p
u
lar
ity
.
Su
b
wo
r
d
-
le
v
el
to
k
en
is
ati
o
n
b
r
ea
k
s
d
o
wn
wo
r
d
s
in
to
s
m
aller
u
n
its
o
r
s
u
b
wo
r
d
s
,
wh
ich
allo
ws
f
o
r
b
etter
h
an
d
lin
g
o
f
OOV
ter
m
s
an
d
lan
g
u
ag
es
with
co
m
p
lex
m
o
r
p
h
o
lo
g
y
.
T
o
k
en
is
atio
n
at
th
e
s
u
b
wo
r
d
l
ev
el
u
s
in
g
b
y
te
p
air
e
n
co
d
i
n
g
(
B
PE)
h
as
b
ec
o
m
e
wid
ely
ad
o
p
ted
f
o
r
NM
T
an
d
DNM
T
task
s
b
ec
a
u
s
e
it
o
f
f
e
r
s
a
m
i
d
d
le
g
r
o
u
n
d
b
etwe
en
wo
r
d
-
a
n
d
ch
ar
ac
ter
-
lev
el
t
o
k
en
is
atio
n
.
B
PE
d
iv
i
d
es
th
e
tex
t
in
to
s
u
b
wo
r
d
u
n
its
t
h
at
ar
e
s
h
ar
e
d
ac
r
o
s
s
d
if
f
er
e
n
t
lan
g
u
a
g
es,
th
er
e
b
y
r
e
d
u
ci
n
g
th
e
s
ize
o
f
th
e
v
o
ca
b
u
lar
y
an
d
b
etter
m
a
n
ag
i
n
g
o
u
t
-
of
-
v
o
ca
b
u
lar
y
wo
r
d
s
[
2
7
]
.
3
.
4
.
T
ra
ini
ng
us
ing
t
he
t
ra
ns
f
o
rm
er
mo
del
T
h
e
T
r
a
n
s
f
o
r
m
er
m
o
d
el
is
a
t
y
p
e
o
f
d
ee
p
-
lear
n
i
n
g
m
o
d
el
wh
ich
ca
n
b
e
u
s
ed
f
o
r
m
ac
h
in
e
tr
an
s
latio
n
[
2
8
]
.
T
h
e
co
r
e
c
o
m
p
o
n
en
t
o
f
t
h
is
m
o
d
el
is
th
e
s
elf
-
atten
tio
n
m
ec
h
a
n
is
m
,
wh
ich
allo
ws
th
e
m
o
d
el
to
ev
alu
ate
th
e
v
alu
e
o
f
ea
c
h
wo
r
d
in
th
e
s
en
ten
ce
b
o
th
at
en
co
d
in
g
t
im
e
an
d
at
d
ec
o
d
i
n
g
tim
e.
A
tr
an
s
f
o
r
m
er
m
o
d
el
p
r
o
ce
s
s
es
all
wo
r
d
s
in
a
s
en
ten
ce
at
th
e
s
am
e
tim
e.
T
h
e
en
c
o
d
er
tak
es
th
e
in
p
u
t
s
eq
u
e
n
ce
in
t
h
e
s
o
u
r
ce
la
n
g
u
a
g
e
an
d
p
r
o
ce
s
s
es
it
u
s
in
g
m
u
ltip
le
s
elf
-
atten
tio
n
lay
er
s
an
d
f
ee
d
-
f
o
r
war
d
n
e
u
r
al
n
etwo
r
k
s
.
E
ac
h
lay
er
in
t
h
e
en
co
d
er
r
ef
in
es
th
e
r
e
p
r
esen
t
atio
n
o
f
th
e
i
n
p
u
t
s
eq
u
en
ce
.
T
h
e
d
ec
o
d
er
tak
es
th
e
o
u
tp
u
t
o
f
th
e
en
c
o
d
er
a
n
d
g
en
er
ates
th
e
o
u
tp
u
t
s
eq
u
e
n
c
e
in
th
e
ta
r
g
et
lan
g
u
ag
e
[
2
8
]
.
Similar
to
th
e
en
co
d
er
,
th
e
d
ec
o
d
er
co
n
s
is
ts
o
f
m
u
ltip
le
lay
er
s
o
f
s
elf
-
atten
tio
n
an
d
f
ee
d
-
f
o
r
war
d
n
e
u
r
al
n
et
wo
r
k
s
.
Ho
wev
er
,
th
e
d
ec
o
d
er
also
in
co
r
p
o
r
ates a
n
ad
d
itio
n
al
atten
tio
n
m
ec
h
an
is
m
ca
lled
en
co
d
er
-
d
ec
o
d
er
atte
n
tio
n
,
allo
win
g
it
t
o
f
o
c
u
s
o
n
r
elev
an
t
p
ar
ts
o
f
th
e
in
p
u
t seq
u
en
ce
d
u
r
in
g
d
ec
o
d
in
g
[
2
8
]
.
T
h
e
s
tan
d
ar
d
tr
an
s
f
o
r
m
er
m
o
d
el
f
r
o
m
th
e
T
en
s
o
r
2
T
e
n
s
o
r
lib
r
ar
y
was
u
s
ed
in
th
is
s
tu
d
y
.
T
en
s
o
r
2
T
en
s
o
r
o
f
f
er
s
a
co
m
p
r
eh
en
s
iv
e
f
r
am
ewo
r
k
f
o
r
tr
ain
i
n
g
,
ev
alu
atin
g
,
an
d
d
ep
l
o
y
in
g
m
ac
h
in
e
tr
an
s
latio
n
m
o
d
els
[
2
9
]
.
T
e
n
s
o
r
2
T
en
s
o
r
also
s
u
p
p
o
r
ts
d
is
tr
ib
u
ted
tr
ain
in
g
ac
r
o
s
s
m
u
ltip
le
GPU
s
an
d
/o
r
T
PUs
.
T
h
is
allo
ws
lar
g
er
d
atasets
an
d
m
o
r
e
co
m
p
lex
m
o
d
els
to
b
e
ac
ce
s
s
ed
.
A
d
ataset
o
f
4
8
,
0
0
0
p
ar
allel
s
en
ten
ce
s
was
u
s
ed
f
o
r
tr
ain
in
g
.
A
s
et
o
f
1
,
000
s
en
t
en
ce
s
was
u
s
ed
f
o
r
test
in
g
,
an
d
an
o
t
h
er
s
et
o
f
1
,
0
0
0
s
en
t
en
ce
s
was
u
s
ed
f
o
r
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
13
,
No
.
4
,
Dec
em
b
er
2
0
2
4
:
4
9
7
6
-
4
9
8
7
4982
v
alid
atio
n
.
All
th
e
s
en
ten
ce
s
w
er
e
p
r
e
p
ar
ed
b
y
th
e
r
esear
ch
te
am
.
T
h
e
f
ir
s
t
2
5
,
8
1
0
p
a
r
allel
s
en
ten
ce
s
ca
m
e
f
r
o
m
th
e
Mo
r
is
ia
1
.
0
p
r
o
ject
[
1
1
]
.
T
h
e
tr
ain
in
g
was
r
u
n
f
o
r
1
0
0
k
s
tep
s
o
n
a
W
in
d
o
w
s
1
0
lap
to
p
with
16
GB
R
AM
an
d
a
2
5
6
GB
SS
D.
T
h
e
B
L
E
U
s
co
r
e,
a
v
ailab
le
in
th
e
T
en
s
o
r
2
T
en
s
o
r
lib
r
ar
y
,
was
u
s
ed
to
ev
alu
ate
th
e
q
u
ality
o
f
th
e
tr
an
s
latio
n
s
.
T
en
s
o
r
2
T
en
s
o
r
also
p
r
o
v
id
es
u
tili
ties
f
o
r
lo
ad
in
g
a
n
d
r
u
n
n
in
g
tr
ain
ed
m
o
d
els in
p
r
o
d
u
ctio
n
en
v
i
r
o
n
m
e
n
ts
.
T
h
e
b
est
m
o
d
el
w
as
s
av
ed
an
d
u
p
lo
ad
ed
o
n
Dig
italOce
an
[
3
0
]
,
wh
ich
p
r
o
v
i
d
es
a
clo
u
d
-
b
ased
in
f
r
astru
ctu
r
e
f
o
r
h
o
s
tin
g
s
ca
lab
le
v
i
r
tu
al
m
ac
h
in
es
(
d
r
o
p
lets
)
.
A
web
in
ter
f
ac
e
to
ac
ce
s
s
th
is
tr
an
s
latio
n
s
er
v
ice
was
th
e
n
d
ev
elo
p
e
d
o
n
k
r
eo
lr
ep
i
b
lik
m
o
r
is
.
n
et.
T
h
e
m
o
b
ile
ap
p
also
p
r
o
v
id
es
th
e
s
am
e
s
er
v
ice
f
o
r
tr
a
n
s
latin
g
f
r
o
m
K
R
M
to
E
n
g
lis
h
an
d
v
ice
v
er
s
a.
4.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
I
n
th
is
p
ar
t,
th
e
s
y
s
tem
'
s
f
in
d
in
g
s
an
d
cr
itical
ass
es
s
m
en
t
ar
e
p
r
esen
ted
.
T
h
e
im
p
licatio
n
s
o
f
th
ese
f
in
d
in
g
s
f
o
r
f
u
t
u
r
e
r
esear
ch
ar
e
also
d
is
cu
s
s
ed
.
B
ased
o
n
th
ese
r
esu
lts
,
r
ec
o
m
m
en
d
atio
n
s
f
o
r
p
r
ac
titi
o
n
e
r
s
ar
e
also
p
r
o
v
id
e
d
.
Fig
u
r
e
1
d
ep
icts
th
e
tr
an
s
latio
n
to
o
l
o
n
th
e
Kr
e
o
lR
ep
ib
lik
Mo
r
is
.
n
et
web
s
ite,
wh
ich
en
ab
les
u
s
er
s
to
tr
an
s
late
f
r
o
m
KR
M
to
E
n
g
li
s
h
an
d
v
ice
-
v
er
s
a.
Ad
d
itio
n
ally
,
as
u
s
er
s
in
p
u
t
tex
t
f
o
r
t
r
an
s
latio
n
in
th
e
p
r
o
v
i
d
ed
tex
tb
o
x
,
th
e
to
o
l
o
f
f
er
s
p
o
ten
ti
al
wo
r
d
s
u
g
g
esti
o
n
s
in
KR
M
.
Fig
u
r
e
2
illu
s
tr
ates
a
s
im
ilar
tr
an
s
latio
n
ap
p
r
o
ac
h
f
r
o
m
E
n
g
lis
h
to
KR
M
.
Fig
u
r
e
1
.
T
r
an
s
latio
n
to
o
l t
o
tr
an
s
late
f
r
o
m
KR
M
to
E
n
g
lis
h
Fig
u
r
e
2
.
T
r
an
s
latio
n
to
o
l t
o
tr
an
s
late
f
r
o
m
E
n
g
lis
h
to
KR
M
T
h
e
u
s
er
m
u
s
t
ch
o
o
s
e
t
h
e
s
o
u
r
ce
lan
g
u
ag
e
a
n
d
e
n
ter
th
e
c
o
r
r
ec
t
wo
r
d
,
p
h
r
ase,
o
r
s
en
te
n
ce
b
ef
o
r
e
click
in
g
th
e
T
r
an
s
late
b
u
tto
n
t
o
r
ec
eiv
e
th
e
tr
an
s
lated
tex
t
in
th
e
tar
g
et
lan
g
u
ag
e.
T
h
e
s
o
u
r
ce
p
h
r
ase
in
KR
M
,
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
A
n
eu
r
a
l m
a
ch
in
e
t
r
a
n
s
la
tio
n
s
ystem
fo
r
K
r
eo
l
R
ep
ib
lik
Mo
r
is
a
n
d
E
n
g
lis
h
…
(
S
a
mee
r
ch
a
n
d
P
u
d
a
r
u
th
)
4983
“
Azo
r
d
i
en
n
b
o
n
zo
u
r
p
o
u
al
l
ap
laz
”
,
is
co
n
v
er
ted
in
to
th
e
c
o
r
r
esp
o
n
d
in
g
E
n
g
lis
h
s
en
ten
ce
,
“
T
o
d
ay
is
a
g
o
o
d
d
ay
to
g
o
to
th
e
b
e
ac
h
”
,
as
s
h
o
wn
in
Fig
u
r
e
1
.
W
h
ile
u
n
d
er
g
o
in
g
tr
an
s
latio
n
,
a
m
ess
ag
e
'
T
r
an
s
latio
n
in
p
r
o
g
r
ess
,
p
lease
wait
a
m
o
m
en
t
’
ap
p
ea
r
s
o
n
th
e
s
cr
ee
n
to
n
o
tify
u
s
er
s
ab
o
u
t th
e
o
n
g
o
in
g
tr
a
n
s
latio
n
an
d
ad
v
is
es th
em
to
wait
f
o
r
th
e
r
esu
lt.
A
b
asic
s
p
ellin
g
ch
ec
k
e
r
f
u
n
ctio
n
ality
is
also
av
ailab
le
in
th
e
p
o
r
tal
.
Fig
u
r
e
3
s
h
o
ws
a
s
ce
n
ar
io
wh
er
e
th
e
wo
r
d
s
‘
s
ik
lo
n
n
’
an
d
‘
s
o
u
v
a
n
t’
a
r
e
n
o
t
a
p
p
r
o
p
r
iately
wr
itten
b
u
t
th
e
t
o
o
l
ca
n
p
r
o
v
id
e
th
e
co
r
r
ec
t sp
ellin
g
f
o
r
th
ese
two
wo
r
d
s
.
Fig
u
r
e
3
.
T
r
an
s
latio
n
to
o
l sh
o
win
g
s
p
ellin
g
ch
ec
k
e
r
f
u
n
ctio
n
ality
T
h
e
tr
an
s
latio
n
s
er
v
ice
av
ailab
le
o
n
th
e
web
s
ite
is
a
ls
o
ac
c
ess
ib
le
a
s
a
co
n
v
en
ien
t
an
d
u
s
er
-
f
r
ien
d
ly
An
d
r
o
id
m
o
b
ile
ap
p
licatio
n
,
m
ak
in
g
it
ea
s
ier
f
o
r
u
s
er
s
to
ac
ce
s
s
th
e
lan
g
u
ag
e
tr
a
n
s
latio
n
s
er
v
ice
o
n
th
e
g
o
.
Fig
u
r
e
4
d
e
p
icts
th
e
m
o
b
ile
a
p
p
in
te
r
f
ac
e
d
es
ig
n
e
x
p
licitly
cr
ea
ted
f
o
r
t
r
an
s
latin
g
b
etwe
en
KR
M
an
d
v
ar
i
o
u
s
o
th
er
lan
g
u
ag
es.
T
h
e
m
o
b
ile
a
p
p
licatio
n
allo
ws u
s
er
s
to
s
ele
ct
th
eir
d
esire
d
s
o
u
r
ce
an
d
ta
r
g
et
lan
g
u
ag
es,
in
p
u
t
th
eir
tex
t
f
o
r
tr
an
s
latio
n
,
an
d
r
ec
eiv
e
th
e
tr
an
s
lated
r
esu
lt.
Fig
u
r
e
5
s
h
o
ws
s
o
m
e
s
am
p
le
tr
a
n
s
latio
n
s
r
ec
o
r
d
e
d
in
th
e
tr
an
s
latio
n
h
is
to
r
y
s
ec
ti
o
n
.
T
h
e
tr
an
s
latio
n
s
y
s
tem
h
as
u
n
d
e
r
g
o
n
e
ex
ten
s
iv
e
test
in
g
with
d
if
f
er
en
t
KR
M
s
en
ten
ce
s
an
d
h
as
b
ee
n
co
m
p
a
r
ed
with
th
e
p
r
ev
io
u
s
tr
a
n
s
latio
n
s
y
s
tem
,
Mo
r
is
ia
1
.
0
.
T
h
e
test
in
g
r
esu
lts
r
ev
ea
le
d
th
at
th
e
n
ew
tr
an
s
latio
n
to
o
l o
n
th
e
Kr
eo
lR
ep
ib
lik
Mo
r
is
.
n
et
web
s
ite
o
u
tp
er
f
o
r
m
s
Mo
r
is
ia
1
.
0
.
T
ab
le
1
p
r
esen
ts
th
e
co
n
v
e
r
s
io
n
o
f
s
en
ten
ce
s
f
r
o
m
KR
M
to
E
n
g
lis
h
.
T
h
e
tab
le
also
co
m
p
ar
es
th
e
ea
r
lier
tr
an
s
latio
n
s
y
s
tem
Mo
r
is
ia
1
.
0
an
d
th
e
u
p
d
ated
s
y
s
tem
Mo
r
is
ia
2
.
0
.
T
h
e
ter
m
‘
ak
r
ed
ite’
was
ab
s
en
t
in
th
e
o
r
i
g
in
al
M
o
r
is
ia
1
.
0
d
ataset
an
d
co
u
ld
n
o
t
b
e
ac
cu
r
ately
t
r
an
s
lated
.
As
a
r
esu
lt,
it
h
as
b
ee
n
in
clu
d
ed
in
th
e
n
ew
Mo
r
is
ia
2
.
0
d
ataset,
lead
i
n
g
to
im
p
r
o
v
ed
s
en
ten
ce
tr
an
s
l
atio
n
s
.
Similar
ly
,
alth
o
u
g
h
th
e
wo
r
d
‘
d
ek
atlo
n
’
was
p
ar
t
o
f
th
e
d
ataset,
it
co
u
l
d
n
o
t
b
e
tr
an
s
lated
c
o
r
r
ec
tly
.
A
s
in
g
le
o
cc
u
r
r
e
n
ce
o
f
a
wo
r
d
with
in
a
d
ataset
m
ay
n
o
t
b
e
en
o
u
g
h
f
o
r
th
e
m
o
d
el
to
lear
n
h
o
w
to
u
s
e
it
p
r
o
p
er
ly
.
T
h
e
ter
m
‘
f
iziy
a
d
’
h
as
b
ee
n
in
ter
p
r
eted
as
‘
tr
o
u
b
lem
a
k
er
’
.
W
h
ile
th
is
d
o
es
n
o
t
p
er
f
ec
tly
m
atch
th
e
o
r
ig
i
n
al
s
en
ten
ce
in
E
n
g
lis
h
,
it
p
r
es
en
ts
a
m
o
r
e
r
eliab
le
tr
an
s
latio
n
th
an
Mo
r
is
ia
1
.
0
.
T
h
e
tr
an
s
latio
n
ac
cu
r
ac
y
also
im
p
r
o
v
es
s
ig
n
if
ican
tly
f
o
r
th
e
f
o
u
r
t
h
an
d
f
if
t
h
s
en
ten
ce
s
.
I
n
s
u
m
m
ar
y
,
t
h
er
e
i
s
a
m
ar
k
ed
im
p
r
o
v
e
m
en
t
in
tr
a
n
s
latio
n
ac
cu
r
ac
y
f
r
o
m
Mo
r
is
ia
1
.
0
to
M
o
r
is
ia
2
.
0
wh
en
co
n
v
er
tin
g
KR
M
to
E
n
g
lis
h
.
T
ab
le
2
p
r
esen
ts
t
h
e
tr
a
n
s
latio
n
o
f
s
en
ten
ce
s
f
r
o
m
E
n
g
lis
h
to
KR
M
.
T
h
e
c
o
m
p
a
r
is
o
n
b
etwe
en
th
e
f
o
r
m
er
s
y
s
tem
,
Mo
r
is
ia
1
.
0
,
a
n
d
th
e
p
r
esen
t sy
s
tem
,
Mo
r
is
ia
2
.
0
,
is
illu
s
tr
ated
in
T
ab
le
2
.
I
n
th
e
f
ir
s
t sen
ten
ce
,
th
e
ter
m
‘
g
en
tle’
was
ab
s
en
t
i
n
th
e
Mo
r
is
ia
1
.
0
d
ataset
an
d
co
u
ld
n
o
t
b
e
ac
cu
r
ately
tr
an
s
lated
.
Nev
er
th
eless
,
it
h
as
b
ee
n
in
clu
d
ed
in
t
h
e
M
o
r
i
s
ia
2
.
0
d
ataset,
r
esu
ltin
g
in
an
im
p
r
o
v
ed
tr
an
s
latio
n
f
o
r
th
is
p
ar
ticu
lar
s
en
ten
ce
.
Sen
ten
ce
s
2
,
3
,
an
d
4
in
d
icate
th
at
th
er
e
h
av
e
b
ee
n
en
h
a
n
ce
m
en
ts
in
th
e
q
u
ali
ty
o
f
tr
a
n
s
lated
s
en
ten
ce
s
wh
en
m
o
v
in
g
f
r
o
m
M
o
r
is
ia
1
.
0
to
Mo
r
is
ia
2
.
0
.
T
h
e
ter
m
‘
r
estau
r
an
ts
’
in
t
h
e
f
if
t
h
s
en
ten
ce
w
as
m
is
s
in
g
f
r
o
m
th
e
Mo
r
is
ia
1
.
0
d
ataset
an
d
was
n
o
t
tr
an
s
lated
ac
cu
r
ately
.
I
ts
ad
d
itio
n
to
th
e
Mo
r
is
ia
2
.
0
d
ataset
h
as
im
p
r
o
v
ed
t
h
e
o
u
tp
u
t
q
u
ality
.
Ov
er
all,
th
er
e
is
a
n
o
ticea
b
le
im
p
r
o
v
em
en
t
in
tr
an
s
latio
n
q
u
ality
f
r
o
m
Mo
r
is
ia
1
.
0
t
o
Mo
r
is
ia
2
.
0
w
h
en
tr
an
s
latin
g
f
r
o
m
E
n
g
lis
h
to
KR
M
.
T
h
u
s
,
in
cr
ea
s
in
g
th
e
s
ize
o
f
t
h
e
d
ataset
b
y
ab
o
u
t
2
4
,
0
0
0
n
ew
s
en
ten
ce
s
p
o
s
itiv
ely
im
p
ac
ts
th
e
tr
an
s
latio
n
q
u
al
ity
.
Ho
we
v
er
,
th
e
B
L
E
U
s
co
r
e
in
cr
ea
s
e
d
o
n
ly
f
r
o
m
3
0
.
3
0
(
Mo
r
is
ia
1
.
0
)
to
3
1
.
4
6
(
Mo
r
is
ia
2
.
0
)
f
o
r
tr
an
s
latio
n
f
r
o
m
KR
M
to
E
n
g
lis
h
.
Fo
r
tr
an
s
latio
n
f
r
o
m
E
n
g
lis
h
to
KR
M,
th
e
B
L
E
U
s
co
r
e
in
cr
ea
s
ed
f
r
o
m
2
6
.
3
4
(
Mo
r
is
ia
1
.
0
)
t
o
2
8
.
1
5
(
M
o
r
is
ia
2
.
0
)
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
13
,
No
.
4
,
Dec
em
b
er
2
0
2
4
:
4
9
7
6
-
4
9
8
7
4984
Fig
u
r
e
4
.
T
r
an
s
latio
n
Ap
p
Fig
u
r
e
5
.
T
r
an
s
latio
n
h
is
to
r
y
T
ab
le
1
.
T
r
a
n
s
latio
n
f
r
o
m
KR
M
to
E
n
g
lis
h
#
K
r
e
o
l
R
e
p
i
b
l
i
k
M
o
r
i
s
En
g
l
i
sh
t
r
a
n
sl
a
t
i
o
n
i
n
M
o
r
i
si
a
1
.
0
En
g
l
i
sh
t
r
a
n
sl
a
t
i
o
n
i
n
M
o
r
i
si
a
2
.
0
1
P
a
t
o
u
b
a
n
n
l
i
n
i
v
e
r
si
t
e
k
i
a
k
r
e
d
i
t
e
.
N
o
t
a
l
l
u
n
i
v
e
r
si
t
i
e
s wh
o
a
r
e
p
r
e
c
t
e
d
.
N
o
t
a
l
l
t
h
e
u
n
i
v
e
r
si
t
i
e
s t
h
a
t
a
r
e
a
c
c
r
e
d
i
t
e
d
.
2
Li
f
i
n
n
s
o
r
t
i
p
r
e
mi
e
d
a
n
d
e
k
a
t
l
o
n
.
S
h
e
c
a
m
e
o
u
t
f
i
r
s
t
i
n
t
h
e
f
i
r
st
s
t
a
g
e
.
S
h
e
c
a
m
e
o
u
t
f
i
r
s
t
i
n
t
h
e
me
d
i
t
a
t
i
o
n
.
3
F
i
n
n
e
n
a
a
n
k
o
r
e
n
n
f
i
z
i
y
a
d
d
a
n
l
a
meri
k
h
i
e
r
.
Th
e
r
e
w
a
s
o
n
c
e
m
o
r
e
o
f
f
i
l
d
i
n
A
meric
a
.
Th
e
r
e
w
a
s
o
n
e
mo
r
e
t
r
o
u
b
l
e
m
a
k
e
r
i
n
A
meric
a
.
4
To
m
i
n
n
k
o
k
i
n
p
l
i
n
l
a
r
z
a
n
d
e
p
i
M
a
r
y
.
To
m
h
a
s fo
o
l
e
d
m
o
n
e
y
f
r
o
m
M
a
r
y
.
To
m
c
o
u
l
d
h
a
v
e
s
t
o
l
e
n
t
h
e
m
o
n
e
y
f
r
o
m
M
a
r
y
.
5
To
b
i
z
i
n
e
v
i
t
f
e
r
b
a
n
n
e
r
e
r
k
o
u
msa.
Y
o
u
mu
s
t
a
v
o
i
d
ma
k
i
n
g
su
c
h
a
mi
st
a
k
e
.
Y
o
u
sh
o
u
l
d
a
v
o
i
d
m
a
k
i
n
g
mi
s
t
a
k
e
s
l
i
k
e
t
h
a
t
.
T
ab
le
2
.
T
r
a
n
s
latio
n
f
r
o
m
E
n
g
lis
h
to
KR
M
#
En
g
l
i
sh
K
R
M
t
r
a
n
sl
a
t
i
o
n
i
n
M
o
r
i
si
a
1
.
0
K
R
M
t
r
a
n
sl
a
t
i
o
n
i
n
M
o
r
i
si
a
2
.
0
1
H
e
w
a
s
a
s
g
e
n
t
l
e
a
m
a
n
a
s
e
v
e
r
l
i
v
e
d
.
Li
t
i
k
o
u
ma
e
n
n
m
i
si
e
k
i
z
a
me
v
i
v
.
Li
t
i
o
s
i
d
o
u
k
i
e
n
n
d
i
m
o
u
n
n
k
a
p
a
v
e
x
i
st
e
.
2
S
h
e
m
a
d
e
t
h
e
s
a
me
mi
s
t
a
k
e
a
g
a
i
n
.
Li
f
i
n
n
f
e
r
m
e
m
e
r
e
r
.
Li
f
i
n
n
r
e
f
e
r
mem
e
r
e
r
.
3
I
l
i
st
e
n
e
d
t
o
t
h
e
mu
s
i
c
o
f
b
i
r
d
s.
M
o
t
i
e
k
o
u
t
l
a
m
i
z
i
k
so
b
a
n
n
sw
a
z
o
.
M
o
t
i
e
k
o
u
t
l
a
m
i
z
i
k
b
a
n
n
z
w
a
z
o
.
4
S
h
e
'l
l
b
e
u
p
a
r
o
u
n
d
b
y
t
h
i
s
a
f
t
e
r
n
o
o
n
.
N
o
u
b
i
z
i
n
f
e
r
p
r
e
p
o
u
s
a
l
a
p
r
e
m
i
d
i
-
l
a
.
Li
p
o
u
l
e
v
e
d
a
n
l
a
p
r
e
mi
d
i
.
5
Th
e
c
i
t
y
h
a
s
a
n
a
b
u
n
d
a
n
c
e
o
f
f
i
n
e
r
e
st
a
u
r
a
n
t
s
.
La
v
i
l
e
n
a
e
n
n
a
b
o
n
d
a
n
s
k
o
r
e
k
.
La
v
i
l
e
n
a
e
n
n
a
b
o
n
d
a
n
s
b
a
n
n
b
o
n
r
e
st
o
r
a
n
.
4.
CO
NCLU
SI
O
N
T
h
e
Kr
eo
l
Mo
r
is
ien
o
r
KR
M
lan
g
u
ag
e
h
as
b
ec
o
m
e
in
cr
ea
s
i
n
g
ly
p
o
p
u
lar
,
cr
ea
tin
g
a
d
em
a
n
d
f
o
r
an
ac
ce
s
s
ib
le
tech
n
o
lo
g
y
p
latf
o
r
m
f
ac
ilit
atin
g
its
lear
n
in
g
.
T
o
a
d
d
r
ess
th
is
n
ee
d
,
an
o
n
lin
e
p
lat
f
o
r
m
was
d
ev
elo
p
e
d
to
f
ac
ilit
ate
th
e
tr
an
s
latio
n
o
f
KR
M
to
E
n
g
lis
h
an
d
v
ice
v
er
s
a.
T
h
is
s
y
s
tem
is
ca
p
ab
le
o
f
tr
an
s
latin
g
b
o
th
in
d
iv
id
u
al
wo
r
d
s
an
d
co
m
p
let
e
s
en
ten
ce
s
.
Mo
r
is
ia
2
.
0
h
as
b
ee
n
d
ev
elo
p
e
d
as
an
An
d
r
o
id
ap
p
licatio
n
an
d
will
b
e
ac
ce
s
s
ib
le
o
n
Play
Sto
r
e.
T
h
e
tr
an
s
latio
n
q
u
ality
f
r
o
m
K
R
M
to
E
n
g
lis
h
an
d
v
ice
v
e
r
s
a
is
co
m
p
ar
ab
le
b
ased
o
n
th
e
B
L
E
U
s
co
r
e
ev
alu
atio
n
.
Mo
r
is
ia
2
.
0
ac
h
iev
ed
a
B
L
E
U
s
co
r
e
o
f
3
1
.
4
6
f
o
r
tr
an
s
latin
g
KR
M
in
to
E
n
g
lis
h
.
T
h
e
ea
r
lier
ev
alu
atio
n
f
o
r
Mo
r
is
ia
1
.
0
y
ield
ed
a
s
co
r
e
o
f
3
0
.
3
0
.
Mo
r
is
ia
2
.
0
ac
h
iev
e
d
a
B
L
E
U
s
co
r
e
o
f
2
8
.
1
5
f
o
r
tr
an
s
latin
g
E
n
g
lis
h
in
to
KR
M,
r
ep
r
esen
tin
g
an
im
p
r
o
v
e
m
en
t
f
r
o
m
th
e
p
r
ev
i
o
u
s
s
co
r
e
o
f
2
6
.
3
4
in
Mo
r
is
ia
1
.
0
,
in
d
icatin
g
th
at
th
e
e
x
p
an
d
ed
d
ataset
h
as
n
o
tab
l
y
en
h
a
n
ce
d
tr
an
s
latio
n
q
u
ality
.
Am
o
n
g
o
th
er
c
o
m
p
ar
a
b
le
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
A
n
eu
r
a
l m
a
ch
in
e
t
r
a
n
s
la
tio
n
s
ystem
fo
r
K
r
eo
l
R
ep
ib
lik
Mo
r
is
a
n
d
E
n
g
lis
h
…
(
S
a
mee
r
ch
a
n
d
P
u
d
a
r
u
th
)
4985
Kr
eo
l
lan
g
u
ag
e
tr
an
s
latio
n
p
latf
o
r
m
s
,
Mo
r
is
ia
2
.
0
h
as
d
em
o
n
s
tr
ated
it
i
s
ca
p
ab
ilit
y
to
d
e
liv
er
h
ig
h
er
-
q
u
ality
tr
an
s
latio
n
s
.
ACK
NO
WL
E
DG
E
M
E
NT
S
T
h
is
p
ap
er
is
b
ased
o
n
a
w
o
r
k
s
u
p
p
o
r
te
d
b
y
th
e
H
ig
h
e
r
E
d
u
ca
tio
n
C
o
m
m
is
s
io
n
(
HE
C
)
o
f
Ma
u
r
itiu
s
u
n
d
er
t
h
e
awa
r
d
n
u
m
b
er
I
NT
-
2022
-
03.
RE
F
E
R
E
NC
E
S
[
1
]
“
Ec
o
n
o
m
i
c
a
n
d
s
o
c
i
a
l
i
n
d
i
c
a
t
o
r
s,”
S
t
a
t
i
st
i
c
s
M
a
u
r
i
t
i
u
s
.
A
c
c
e
sse
d
:
F
e
b
.
2
1
,
2
0
2
4
.
[
O
n
l
i
n
e
]
.
A
v
a
i
l
a
b
l
e
:
h
t
t
p
s
:
/
/
st
a
t
sm
a
u
r
i
t
i
u
s
.
g
o
v
m
u
.
o
r
g
/
D
o
c
u
me
n
t
s
/
S
t
a
t
i
st
i
c
s
/
ESI
/
2
0
2
2
/
EI
1
6
8
7
/
2
0
2
2
P
o
p
u
l
a
t
i
o
n
C
e
n
su
s
_
M
a
i
n
R
e
su
l
t
s
_
1
8
1
1
2
0
2
2
.
p
d
f
.
[
2
]
“
I
n
t
e
r
v
i
e
w
:
D
e
r
n
i
e
d
i
a
l
o
g
a
v
e
k
D
e
v
V
i
r
a
h
s
a
w
my
(
1
)
,
”
L
e
M
a
u
ri
c
i
e
n
.
A
c
c
e
ss
e
d
:
F
e
b
.
2
1
,
2
0
2
4
.
[
O
n
l
i
n
e
]
.
A
v
a
i
l
a
b
l
e
:
h
t
t
p
s
:
/
/
w
w
w
.
l
e
ma
u
r
i
c
i
e
n
.
c
o
m/
w
e
e
k
-
e
n
d
/
i
n
t
e
r
v
i
e
w
-
d
e
r
n
i
e
-
d
i
a
l
o
g
-
a
v
e
k
-
d
e
v
-
v
i
r
a
h
sawm
y
-
1
/
6
0
7
4
6
5
/
[
3
]
Z.
B
o
o
d
e
e
a
a
n
d
S
.
P
u
d
a
r
u
t
h
,
“
K
r
e
o
l
M
o
r
i
si
e
n
t
o
En
g
l
i
s
h
a
n
d
En
g
l
i
s
h
t
o
K
r
e
o
l
M
o
r
i
s
i
e
n
t
r
a
n
sl
a
t
i
o
n
s
y
s
t
e
m
u
s
i
n
g
a
t
t
e
n
t
i
o
n
a
n
d
t
r
a
n
sf
o
r
mer
mo
d
e
l
,
”
I
n
t
e
rn
a
t
i
o
n
a
l
J
o
u
rn
a
l
o
f
C
o
m
p
u
t
i
n
g
a
n
d
D
i
g
i
t
a
l
S
y
s
t
e
m
s
,
v
o
l
.
9
,
n
o
.
6
,
p
p
.
1
1
4
3
–
1
1
5
3
,
N
o
v
.
2
0
2
0
,
d
o
i
:
1
0
.
1
2
7
8
5
/
i
j
c
d
s/
0
9
0
6
0
1
2
.
[
4
]
S
.
P
a
o
l
i
a
n
d
H
.
D
a
v
i
d
s
o
n
,
“
P
r
a
g
ma
t
i
c
mark
e
r
s
a
n
d
v
e
r
b
a
d
i
c
e
n
d
i
:
A
n
i
n
v
e
s
t
i
g
a
t
i
o
n
o
f
M
a
u
r
i
t
i
a
n
C
r
e
o
l
e
,
”
J
o
u
r
n
a
l
o
f
P
ra
g
m
a
t
i
c
s
,
v
o
l
.
2
1
4
,
p
p
.
1
0
7
–
1
2
6
,
S
e
p
.
2
0
2
3
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
p
r
a
g
ma.
2
0
2
3
.
0
6
.
0
1
1
.
[
5
]
Y
.
W
u
a
n
d
Y
.
Q
i
n
,
“
M
a
c
h
i
n
e
t
r
a
n
sl
a
t
i
o
n
o
f
E
n
g
l
i
s
h
sp
e
e
c
h
:
C
o
mp
a
r
i
s
o
n
o
f
mu
l
t
i
p
l
e
a
l
g
o
r
i
t
h
ms,”
J
o
u
r
n
a
l
o
f
I
n
t
e
l
l
i
g
e
n
t
S
y
st
e
m
s
,
v
o
l
.
3
1
,
n
o
.
1
,
p
p
.
1
5
9
–
1
6
7
,
J
a
n
.
2
0
2
2
,
d
o
i
:
1
0
.
1
5
1
5
/
j
i
s
y
s
-
2
0
2
2
-
0
0
0
5
.
[
6
]
J.
-
X
.
H
u
a
n
g
,
K
.
-
S
.
Le
e
,
a
n
d
Y
.
-
K
.
K
i
m
,
“
H
y
b
r
i
d
t
r
a
n
s
l
a
t
i
o
n
w
i
t
h
c
l
a
ssi
f
i
c
a
t
i
o
n
:
r
e
v
i
si
t
i
n
g
r
u
l
e
-
b
a
s
e
d
a
n
d
n
e
u
r
a
l
m
a
c
h
i
n
e
t
r
a
n
s
l
a
t
i
o
n
,
”
El
e
c
t
ro
n
i
c
s
,
v
o
l
.
9
,
n
o
.
2
,
Jan
.
2
0
2
0
,
d
o
i
:
1
0
.
3
3
9
0
/
e
l
e
c
t
r
o
n
i
c
s
9
0
2
0
2
0
1
.
[
7
]
P
.
W
i
l
l
i
a
ms
,
R
.
S
e
n
n
r
i
c
h
,
M
.
P
o
s
t
,
a
n
d
P
.
K
o
e
h
n
,
“
S
y
n
t
a
x
-
b
a
se
d
s
t
a
t
i
s
t
i
c
a
l
mac
h
i
n
e
t
r
a
n
sl
a
t
i
o
n
,
”
i
n
S
y
n
t
h
e
s
i
s
L
e
c
t
u
res
o
n
H
u
m
a
n
L
a
n
g
u
a
g
e
T
e
c
h
n
o
l
o
g
i
e
s
,
C
h
a
m:
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
,
2
0
1
6
,
d
o
i
:
1
0
.
1
0
0
7
/
9
7
8
-
3
-
0
3
1
-
0
2
1
6
4
-
0.
[
8
]
Z.
T
a
n
e
t
a
l
.
,
“
N
e
u
r
a
l
m
a
c
h
i
n
e
t
r
a
n
s
l
a
t
i
o
n
:
A
r
e
v
i
e
w
o
f
m
e
t
h
o
d
s,
r
e
s
o
u
r
c
e
s,
a
n
d
t
o
o
l
s
,
”
AI
O
p
e
n
,
v
o
l
.
1
,
p
p
.
5
–
2
1
,
2
0
2
0
.
[
9
]
J.
Z
h
a
n
g
a
n
d
C
.
Z
o
n
g
,
“
N
e
u
r
a
l
m
a
c
h
i
n
e
t
r
a
n
sl
a
t
i
o
n
:
C
h
a
l
l
e
n
g
e
s
,
p
r
o
g
r
e
ss a
n
d
f
u
t
u
r
e
,
”
S
c
i
e
n
c
e
C
h
i
n
a
T
e
c
h
n
o
l
o
g
i
c
a
l
S
c
i
e
n
c
e
s
,
v
o
l
.
6
3
,
n
o
.
1
0
,
p
p
.
2
0
2
8
–
2
0
5
0
,
O
c
t
.
2
0
2
0
,
d
o
i
:
1
0
.
1
0
0
7
/
s1
1
4
3
1
-
0
2
0
-
1
6
3
2
-
x.
[
1
0
]
V
.
R
a
g
n
i
a
n
d
L.
N
.
V
i
e
i
r
a
,
“
W
h
a
t
h
a
s
c
h
a
n
g
e
d
w
i
t
h
n
e
u
r
a
l
mac
h
i
n
e
t
r
a
n
s
l
a
t
i
o
n
?
A
c
r
i
t
i
c
a
l
r
e
v
i
e
w
o
f
h
u
m
a
n
f
a
c
t
o
r
s,”
P
e
rsp
e
c
t
i
v
e
s
,
v
o
l
.
3
0
,
n
o
.
1
,
p
p
.
1
3
7
–
1
5
8
,
J
a
n
.
2
0
2
2
,
d
o
i
:
1
0
.
1
0
8
0
/
0
9
0
7
6
7
6
X
.
2
0
2
1
.
1
8
8
9
0
0
5
.
[
1
1
]
S
.
P
u
d
a
r
u
t
h
e
t
a
l
.
,
“
M
o
r
i
s
i
a
:
a
n
e
u
r
a
l
mac
h
i
n
e
t
r
a
n
sl
a
t
i
o
n
s
y
s
t
e
m
t
o
t
r
a
n
s
l
a
t
e
b
e
t
w
e
e
n
K
r
e
o
l
M
o
r
i
si
e
n
a
n
d
En
g
l
i
sh
,
”
i
n
T
R
Al
i
n
e
a
,
v
o
l
.
2
3
,
2
0
2
1
.
[
1
2
]
B
.
N
a
t
h
,
C
.
K
u
m
b
h
a
r
,
a
n
d
B
.
T.
K
h
o
a
,
“
A
s
t
u
d
y
o
n
a
p
p
r
o
a
c
h
e
s
t
o
n
e
u
r
a
l
mac
h
i
n
e
t
r
a
n
s
l
a
t
i
o
n
,
”
J
o
u
rn
a
l
o
f
L
o
g
i
st
i
c
s
,
I
n
f
o
rm
a
t
i
c
s
a
n
d
S
e
r
v
i
c
e
S
c
i
e
n
c
e
,
v
o
l
.
9
,
n
o
.
1
,
p
p
.
2
7
1
–
2
8
3
,
A
u
g
.
2
0
2
2
,
d
o
i
:
1
0
.
3
3
1
6
8
/
L
I
S
S
.
2
0
2
2
.
0
3
1
9
.
[
1
3
]
L.
B
e
n
k
o
v
a
a
n
d
L.
B
e
n
k
o
,
“
N
e
u
r
a
l
mac
h
i
n
e
t
r
a
n
s
l
a
t
i
o
n
a
s
a
n
o
v
e
l
a
p
p
r
o
a
c
h
t
o
ma
c
h
i
n
e
t
r
a
n
sl
a
t
i
o
n
,
”
i
n
D
i
v
a
i
2
0
2
0
:
1
3
T
h
I
n
t
e
r
n
a
t
i
o
n
a
l
S
c
i
e
n
t
i
f
i
c
C
o
n
f
e
re
n
c
e
o
n
D
i
s
t
a
n
c
e
L
e
a
r
n
i
n
g
i
n
A
p
p
l
i
e
d
I
n
f
o
r
m
a
t
i
c
s
,
p
p
.
4
9
9
–
5
0
8
,
2
0
2
0
.
[
1
4
]
H
.
P
h
a
n
a
n
d
A
.
Ja
n
n
e
s
a
r
i
,
“
S
t
a
t
i
st
i
c
a
l
mac
h
i
n
e
t
r
a
n
sl
a
t
i
o
n
o
u
t
p
e
r
f
o
r
ms
n
e
u
r
a
l
m
a
c
h
i
n
e
t
r
a
n
s
l
a
t
i
o
n
i
n
s
o
f
t
w
a
r
e
e
n
g
i
n
e
e
r
i
n
g
:
w
h
y
a
n
d
h
o
w
,
”
i
n
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
1
st
AC
M
S
I
G
S
O
FT
I
n
t
e
r
n
a
t
i
o
n
a
l
W
o
rks
h
o
p
o
n
Re
p
r
e
se
n
t
a
t
i
o
n
L
e
a
r
n
i
n
g
f
o
r
S
o
f
t
w
a
r
e
En
g
i
n
e
e
r
i
n
g
a
n
d
Pro
g
ra
m
L
a
n
g
u
a
g
e
s
,
N
e
w
Y
o
r
k
,
N
Y
,
U
S
A
:
A
C
M
,
N
o
v
.
2
0
2
0
,
p
p
.
3
–
12
,
d
o
i
:
1
0
.
1
1
4
5
/
3
4
1
6
5
0
6
.
3
4
2
3
5
7
6
.
[
1
5
]
H.
-
H
.
P
h
a
n
-
V
u
,
V
.
T.
Tr
a
n
,
V
.
N
.
N
g
u
y
e
n
,
H
.
V
.
D
a
n
g
,
a
n
d
P
.
T.
D
o
,
“
N
e
u
r
a
l
ma
c
h
i
n
e
t
r
a
n
sl
a
t
i
o
n
b
e
t
w
e
e
n
V
i
e
t
n
a
mese
a
n
d
En
g
l
i
s
h
:
a
n
e
mp
i
r
i
c
a
l
st
u
d
y
,
”
J
o
u
rn
a
l
o
f
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
a
n
d
C
y
b
e
r
n
e
t
i
c
s
,
v
o
l
.
3
5
,
n
o
.
2
,
p
p
.
1
4
7
–
1
6
6
,
J
u
n
.
2
0
1
9
,
d
o
i
:
1
0
.
1
5
6
2
5
/
1
8
1
3
-
9
6
6
3
/
3
5
/
2
/
1
3
2
3
3
.
[
1
6
]
R
.
K
r
ü
g
e
r
,
“
E
x
p
l
i
c
i
t
a
t
i
o
n
i
n
n
e
u
r
a
l
m
a
c
h
i
n
e
t
r
a
n
s
l
a
t
i
o
n
,
”
A
c
r
o
ss
L
a
n
g
u
a
g
e
s
a
n
d
C
u
l
t
u
res
,
v
o
l
.
2
1
,
n
o
.
2
,
p
p
.
1
9
5
–
2
1
6
,
D
e
c
.
2
0
2
0
,
d
o
i
:
1
0
.
1
5
5
6
/
0
8
4
.
2
0
2
0
.
0
0
0
1
2
.
[
1
7
]
B.
-
E.
B
a
t
s
u
k
h
,
“
S
e
n
t
e
n
c
e
s
t
r
u
c
t
u
r
e
a
n
d
b
o
u
n
d
a
r
y
f
o
r
d
e
e
p
n
e
u
r
a
l
m
a
c
h
i
n
e
t
r
a
n
sl
a
t
i
o
n
a
l
i
g
n
m
e
n
t
m
o
d
e
l
,
”
i
n
P
ro
c
e
e
d
i
n
g
s
o
f
t
h
e
Fu
t
u
r
e
T
e
c
h
n
o
l
o
g
i
e
s
C
o
n
f
e
r
e
n
c
e
,
2
0
2
3
,
p
p
.
5
0
8
–
5
2
0
,
d
o
i
:
1
0
.
1
0
0
7
/
9
7
8
-
3
-
0
3
1
-
1
8
3
4
4
-
7
_
3
6
.
[
1
8
]
J.
Z
h
a
n
g
,
H
.
L
u
a
n
,
M
.
S
u
n
,
F
.
Z
h
a
i
,
J.
X
u
,
a
n
d
Y
.
Li
u
,
“
N
e
u
r
a
l
m
a
c
h
i
n
e
t
r
a
n
sl
a
t
i
o
n
w
i
t
h
e
x
p
l
i
c
i
t
p
h
r
a
se
a
l
i
g
n
m
e
n
t
,
”
I
EEE
/
A
C
M
T
ra
n
s
a
c
t
i
o
n
s
o
n
A
u
d
i
o
,
S
p
e
e
c
h
,
a
n
d
L
a
n
g
u
a
g
e
Pr
o
c
e
ssi
n
g
,
v
o
l
.
2
9
,
p
p
.
1
0
0
1
–
1
0
1
0
,
2
0
2
1
,
d
o
i
:
1
0
.
1
1
0
9
/
TA
S
LP.
2
0
2
1
.
3
0
5
7
8
3
1
.
[
1
9
]
A
.
Emn
a
,
S
.
K
c
h
a
o
u
,
a
n
d
R
.
B
o
u
j
e
l
b
a
n
,
“
N
e
u
r
a
l
ma
c
h
i
n
e
t
r
a
n
sl
a
t
i
o
n
o
f
l
o
w
r
e
so
u
r
c
e
l
a
n
g
u
a
g
e
s:
a
p
p
l
i
c
a
t
i
o
n
t
o
t
r
a
n
s
c
r
i
p
t
i
o
n
s
o
f
Tu
n
i
s
i
a
n
D
i
a
l
e
c
t
,
”
i
n
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
I
n
t
e
l
l
i
g
e
n
t
S
y
st
e
m
s
a
n
d
P
a
t
t
e
rn
Re
c
o
g
n
i
t
i
o
n
,
2
0
2
2
,
p
p
.
2
3
4
–
2
4
7
,
d
o
i
:
1
0
.
1
0
0
7
/
9
7
8
-
3
-
0
3
1
-
0
8
2
7
7
-
1
_
2
0
.
[
2
0
]
K
.
M
.
M
.
A
d
l
a
o
n
a
n
d
N
.
M
a
r
c
o
s,
“
B
u
i
l
d
i
n
g
t
h
e
l
a
n
g
u
a
g
e
r
e
s
o
u
r
c
e
f
o
r
a
C
e
b
u
a
n
o
-
F
i
l
i
p
i
n
o
n
e
u
r
a
l
ma
c
h
i
n
e
t
r
a
n
s
l
a
t
i
o
n
s
y
st
e
m,”
i
n
Pro
c
e
e
d
i
n
g
s
o
f
t
h
e
2
0
1
9
3
rd
I
n
t
e
rn
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
N
a
t
u
r
a
l
L
a
n
g
u
a
g
e
Pr
o
c
e
ss
i
n
g
a
n
d
I
n
f
o
rm
a
t
i
o
n
Re
t
ri
e
v
a
l
,
N
e
w
Y
o
r
k
,
U
S
A
:
A
C
M
,
J
u
n
.
2
0
1
9
,
p
p
.
1
2
7
–
1
3
2
,
d
o
i
:
1
0
.
1
1
4
5
/
3
3
4
2
8
2
7
.
3
3
4
2
8
3
3
.
[
2
1
]
A
.
V
i
l
l
a
n
u
e
v
a
e
t
a
l
.
,
“
M
o
b
i
l
e
-
b
a
se
d
t
r
a
n
s
l
a
t
i
o
n
s
y
st
e
m
f
o
r
C
e
b
u
a
n
o
l
a
n
g
u
a
g
e
w
i
t
h
o
b
j
e
c
t
d
e
t
e
c
t
i
o
n
f
o
r
t
r
a
v
e
l
a
ssi
st
a
n
c
e
u
s
i
n
g
n
e
u
r
a
l
mac
h
i
n
e
t
r
a
n
s
l
a
t
i
o
n
,
”
i
n
2
0
1
9
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
I
n
f
o
rm
a
t
i
o
n
a
n
d
C
o
m
m
u
n
i
c
a
t
i
o
n
s
T
e
c
h
n
o
l
o
g
y
(
I
C
O
I
AC
T
)
,
I
EEE,
J
u
l
.
2
0
1
9
,
p
p
.
5
2
3
–
5
2
8
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
O
I
A
C
T4
6
7
0
4
.
2
0
1
9
.
8
9
3
8
5
6
5
.
[
2
2
]
K
.
P
a
p
i
n
e
n
i
,
S
.
R
o
u
k
o
s
,
T.
W
a
r
d
,
a
n
d
W
.
-
J.
Z
h
u
,
“
B
l
e
u
:
A
me
t
h
o
d
f
o
r
a
u
t
o
m
a
t
i
c
e
v
a
l
u
a
t
i
o
n
o
f
m
a
c
h
i
n
e
t
r
a
n
s
l
a
t
i
o
n
,
”
i
n
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
4
0
t
h
A
n
n
u
a
l
M
e
e
t
i
n
g
o
n
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
-
A
C
L
’
0
2
,
M
o
r
r
i
st
o
w
n
,
N
J,
U
S
A
:
A
sso
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
0
1
,
d
o
i
:
1
0
.
3
1
1
5
/
1
0
7
3
0
8
3
.
1
0
7
3
1
3
5
.
[
2
3
]
S
.
L
y
o
n
s
,
“
Q
u
a
l
i
t
y
o
f
Th
a
i
t
o
En
g
l
i
s
h
m
a
c
h
i
n
e
t
r
a
n
s
l
a
t
i
o
n
,
”
i
n
K
n
o
w
l
e
d
g
e
M
a
n
a
g
e
m
e
n
t
a
n
d
A
c
q
u
i
si
t
i
o
n
f
o
r
I
n
t
e
l
l
i
g
e
n
t
S
y
s
t
e
m
s
:
1
4
t
h
Pa
c
i
f
i
c
R
i
m
K
n
o
w
l
e
d
g
e
Ac
q
u
i
si
t
i
o
n
W
o
rks
h
o
p
,
P
K
AW
2
0
1
6
,
Ph
u
k
e
t
,
T
h
a
i
l
a
n
d
,
A
u
g
u
st
2
2
-
2
3
,
2
0
1
6
,
Pro
c
e
e
d
i
n
g
s
1
4
,
2
0
1
6
,
p
p
.
2
6
1
–
270
,
d
o
i
:
1
0
.
1
0
0
7
/
9
7
8
-
3
-
3
1
9
-
4
2
7
0
6
-
5
_
2
0
.
[
2
4
]
S
.
B
a
n
e
r
j
e
e
a
n
d
A
.
L
a
v
i
e
,
“
M
ETEO
R
:
A
n
a
u
t
o
m
a
t
i
c
me
t
r
i
c
f
o
r
mt
e
v
a
l
u
a
t
i
o
n
w
i
t
h
i
m
p
r
o
v
e
d
c
o
r
r
e
l
a
t
i
o
n
w
i
t
h
h
u
man
j
u
d
g
m
e
n
t
s,
”
i
n
I
n
t
ri
n
si
c
a
n
d
Ex
t
ri
n
si
c
E
v
a
l
u
a
t
i
o
n
Me
a
su
r
e
s
f
o
r
M
a
c
h
i
n
e
T
ra
n
s
l
a
t
i
o
n
a
n
d
/
o
r
S
u
m
m
a
ri
za
t
i
o
n
,
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
W
o
rks
h
o
p
AC
L
2
0
0
5
,
2
0
0
5
,
p
p
.
6
5
–
7
2
.
[
2
5
]
A
.
A
g
a
r
w
a
l
a
n
d
A
.
La
v
i
e
,
“
M
e
t
e
o
r
,
m
-
b
l
e
u
a
n
d
m
-
t
e
r
:
E
v
a
l
u
a
t
i
o
n
m
e
t
r
i
c
s
f
o
r
h
i
g
h
-
c
o
r
r
e
l
a
t
i
o
n
w
i
t
h
h
u
ma
n
r
a
n
k
i
n
g
s
o
f
mac
h
i
n
e
t
r
a
n
s
l
a
t
i
o
n
o
u
t
p
u
t
,
”
i
n
3
r
d
Wo
rks
h
o
p
o
n
S
t
a
t
i
st
i
c
a
l
Ma
c
h
i
n
e
T
ra
n
sl
a
t
i
o
n
,
W
MT
2
0
0
8
a
t
t
h
e
A
n
n
u
a
l
M
e
e
t
i
n
g
o
f
t
h
e
As
so
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,
AC
L
2
0
0
8
,
2
0
0
8
,
p
p
.
1
1
5
–
1
1
8
.
[
2
6
]
“
B
r
o
a
d
o
p
e
r
a
t
i
o
n
a
l
l
a
n
g
u
a
g
e
t
r
a
n
sl
a
t
i
o
n
(
B
O
LT),
”
N
a
t
i
o
n
a
l
I
n
st
i
t
u
t
e
o
f
S
t
a
n
d
a
r
d
s
a
n
d
T
e
c
h
n
o
l
o
g
y
.
A
c
c
e
sse
d
:
M
a
r
.
1
2
,
2
0
2
4
.
[
O
n
l
i
n
e
]
.
A
v
a
i
l
a
b
l
e
:
h
t
t
p
s:
/
/
w
w
w
.
n
i
s
t
.
g
o
v
/
i
t
l
/
i
a
d
/
mi
g
/
b
r
o
a
d
-
o
p
e
r
a
t
i
o
n
a
l
-
l
a
n
g
u
a
g
e
-
t
r
a
n
sl
a
t
i
o
n
-
b
o
l
t
.
Evaluation Warning : The document was created with Spire.PDF for Python.