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.
1
5
,
No
.
2
,
A
p
r
il 2
0
2
6
,
p
p
.
1
8
0
4
~
1
8
1
3
I
SS
N:
2
2
5
2
-
8
9
3
8
,
DOI
: 1
0
.
1
1
5
9
1
/ijai.v
15
.i
2
.
p
p
1
8
0
4
-
1
8
1
3
1804
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
a
i
.
ia
esco
r
e.
co
m
Indo
BE
RT for
ed
uca
tiona
l as
ses
sm
ent:
co
mpa
ra
tive
a
na
ly
sis
of
trans
former mo
d
els
in Indo
nesia
n
questio
n genera
t
i
o
n
H
a
nd
a
ru
J
a
t
i
1
,
Yuni
a
r
I
nd
ri
ha
ps
a
ri
1
,
P
ra
da
na
Set
ia
la
na
1
,
Da
na
ng
Wij
a
y
a
2
,
Sa
t
y
a
Adhi
y
a
k
s
a
Ardy
3
,
Dhis
t
a
Dwi
Nur
Ar
dia
ns
y
a
h
1
1
D
e
p
a
r
t
me
n
t
o
f
El
e
c
t
r
o
n
i
c
s a
n
d
I
n
f
o
r
mat
i
c
s
En
g
i
n
e
e
r
i
n
g
E
d
u
c
a
t
i
o
n
,
U
n
i
v
e
r
si
t
a
s
N
e
g
e
r
i
Y
o
g
y
a
k
a
r
t
a
,
S
l
e
m
a
n
,
I
n
d
o
n
e
si
a
2
D
e
p
a
r
t
me
n
t
o
f
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
a
n
d
I
n
f
o
r
ma
t
i
o
n
E
n
g
i
n
e
e
r
i
n
g
,
N
a
t
i
o
n
a
l
C
e
n
t
r
a
l
U
n
i
v
e
r
si
t
y
,
Ta
o
y
u
a
n
,
Ta
i
w
a
n
3
D
e
p
a
r
t
me
n
t
o
f
El
e
c
t
r
o
n
i
c
a
n
d
C
o
m
p
u
t
e
r
E
n
g
i
n
e
e
r
i
n
g
,
N
a
t
i
o
n
a
l
Ta
i
w
a
n
U
n
i
v
e
r
s
i
t
y
o
f
S
c
i
e
n
c
e
a
n
d
T
e
c
h
n
o
l
o
g
y
,
Ta
i
p
e
i
,
Ta
i
w
a
n
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Sep
2
6
,
2
0
2
5
R
ev
is
ed
J
an
5
,
2
0
2
6
Acc
ep
ted
J
an
2
2
,
2
0
2
6
Th
is
stu
d
y
a
sk
s
wh
e
th
e
r
a
m
o
n
o
l
in
g
u
a
l
e
n
c
o
d
e
r
c
a
n
re
a
li
stica
ll
y
o
u
tp
e
rfo
r
m
m
u
lt
il
in
g
u
a
l
a
n
d
lar
g
e
r
tran
sf
o
rm
e
r
m
o
d
e
ls
f
o
r
In
d
o
n
e
sia
n
a
u
to
m
a
ti
c
q
u
e
stio
n
g
e
n
e
ra
ti
o
n
(AQ
G
)
wh
e
n
a
ll
m
o
d
e
ls
sh
a
re
th
e
sa
m
e
train
in
g
b
u
d
g
e
t.
We
c
o
m
p
a
re
In
d
o
n
e
sia
n
b
id
i
re
c
ti
o
n
a
l
e
n
c
o
d
e
r
re
p
re
se
n
tati
o
n
s
fro
m
tran
sfo
rm
e
rs
(
In
d
o
BER
T
)
,
m
u
lt
il
in
g
u
a
l
BER
T
(m
BERT
)
,
a
n
d
B
ERT
-
larg
e
u
sin
g
a
si
n
g
le
f
in
e
-
tu
n
i
n
g
p
i
p
e
li
n
e
with
a
n
sw
e
r
h
i
g
h
li
g
h
ti
n
g
,
a
p
p
li
e
d
to
a
n
In
d
o
n
e
sia
n
v
e
rsio
n
o
f
T
y
DiQA
-
G
o
ld
P
a
n
d
a
2
0
,
0
0
0
tra
n
sla
ted
su
b
se
t
o
f
S
Qu
AD
2
.
0
.
We
e
v
a
l
u
a
te
m
o
d
e
l
q
u
a
li
t
y
u
si
n
g
b
il
i
n
g
u
a
l
e
v
a
lu
a
ti
o
n
u
n
d
e
rst
u
d
y
sc
o
re
n
-
g
ra
m
4
(
BL
EU
-
4
)
,
m
e
tri
c
fo
r
e
v
a
lu
a
ti
o
n
o
f
tran
sla
ti
o
n
with
e
x
p
li
c
it
o
rd
e
ri
n
g
(
M
ET
E
OR
)
,
a
n
d
ROU
G
E
-
Li
n
c
o
l
n
(
R
OU
G
E
-
L
)
.
In
d
o
BERT
c
o
n
siste
n
t
ly
a
c
h
iev
e
s
th
e
b
e
st
sc
o
re
s
o
n
b
o
th
d
a
ta
se
ts
(e
.
g
.
,
BLE
U
-
4
o
f
1
9
.
6
9
o
n
T
y
DiQA
-
G
o
ld
P
a
n
d
3
.
7
9
o
n
th
e
S
Q
u
AD
2
.
0
su
b
se
t
)
wh
il
e
re
q
u
iri
n
g
les
s
c
o
m
p
u
tatio
n
th
a
n
m
BERT
a
n
d
BERT
-
larg
e
.
O
u
r
re
su
lt
s
sh
o
w
th
a
t
lan
g
u
a
g
e
-
s
p
e
c
ifi
c
p
re
train
in
g
g
i
v
e
s
c
lea
r
a
d
v
a
n
tag
e
s
fo
r
In
d
o
n
e
sia
n
AQ
G
,
y
ield
in
g
h
i
g
h
e
r
a
c
c
u
ra
c
y
a
t
lo
we
r
c
o
m
p
u
tatio
n
a
l
c
o
st
th
a
n
m
u
lt
il
in
g
u
a
l
o
r
larg
e
r
e
n
c
o
d
e
rs.
Th
e
wo
r
k
c
lo
se
s
a
g
a
p
in
I
n
d
o
n
e
sia
n
AQ
G
b
e
n
c
h
m
a
rk
in
g
b
y
p
r
o
v
id
i
n
g
th
e
f
i
rst
h
e
a
d
-
to
-
h
e
a
d
c
o
m
p
a
riso
n
o
f
In
d
o
BERT
,
m
BERT
,
a
n
d
BERT
-
larg
e
u
n
d
e
r
a
sh
a
re
d
fin
e
-
t
u
n
i
n
g
a
n
d
e
v
a
lu
a
ti
o
n
p
ro
t
o
c
o
l.
F
o
r
e
d
u
c
a
ti
o
n
a
l
a
ss
e
ss
m
e
n
t,
th
e
fi
n
d
i
n
g
s
o
ffe
r
a
p
ra
c
ti
c
a
l
re
c
ip
e
fo
r
b
u
il
d
in
g
d
e
p
lo
y
a
b
le
AQ
G
sy
ste
m
s
o
n
m
i
d
-
r
a
n
g
e
G
P
Us
th
a
t
g
e
n
e
ra
te
h
i
g
h
e
r
-
q
u
a
li
t
y
q
u
e
stio
n
s with
o
u
t
p
r
o
h
ib
it
iv
e
train
i
n
g
o
r
i
n
fe
re
n
c
e
b
u
d
g
e
ts.
K
ey
w
o
r
d
s
:
Au
to
m
atic
q
u
esti
o
n
g
e
n
er
atio
n
B
E
R
T
-
lar
g
e
E
d
u
ca
tio
n
al
ass
ess
m
en
t
I
n
d
o
B
E
R
T
I
n
d
o
n
esian
NL
P
Mo
n
o
lin
g
u
al
p
r
etr
ain
i
n
g
Mu
ltil
in
g
u
al
B
E
R
T
T
h
is i
s
a
n
o
p
e
n
a
c
c
e
ss
a
rticle
u
n
d
e
r th
e
CC B
Y
-
SA
li
c
e
n
se
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
Han
d
ar
u
J
ati
Dep
ar
tm
en
t o
f
E
lectr
o
n
ics an
d
I
n
f
o
r
m
atics E
n
g
in
ee
r
i
n
g
E
d
u
ca
tio
n
,
Un
iv
er
s
itas
Neg
er
i Y
o
g
y
ak
ar
ta
Dep
o
k
,
Slem
an
,
Y
o
g
y
a
k
ar
ta
5
5
2
8
1
,
I
n
d
o
n
esia
E
m
ail: h
an
d
ar
u
@
u
n
y
.
ac
.
id
1.
I
NT
RO
D
UCT
I
O
N
A
s
tan
d
ar
d
m
eth
o
d
o
f
ass
ess
i
n
g
s
tu
d
en
t
u
n
d
er
s
tan
d
in
g
th
r
o
u
g
h
tar
g
eted
q
u
esti
o
n
in
g
h
as
r
em
ain
ed
a
co
r
e
co
n
ce
p
t
o
f
ed
u
ca
tio
n
al
ass
es
s
m
en
t
an
d
a
v
alu
ab
le
to
o
l
f
o
r
u
n
d
er
s
tan
d
i
n
g
,
p
r
o
v
i
d
e
d
th
e
q
u
esti
o
n
s
ar
e
alig
n
ed
with
a
p
p
r
o
p
r
iate
lea
r
n
in
g
o
b
jectiv
es
[
1
]
.
I
n
th
is
co
n
t
ex
t,
cr
ea
tin
g
h
ig
h
-
q
u
ality
q
u
esti
o
n
s
f
o
r
ass
ess
in
g
s
tu
d
en
t
u
n
d
er
s
tan
d
in
g
is
a
d
em
an
d
in
g
task
,
ev
en
f
o
r
ex
p
e
r
ien
ce
d
teac
h
er
s
.
T
h
e
item
s
m
u
s
t
b
e
v
alid
,
co
h
er
en
t,
an
d
v
ar
ie
d
en
o
u
g
h
to
elicit
m
o
r
e
th
an
s
im
p
le
r
ec
all
[
2
]
–
[
6
]
.
Au
to
m
atic
q
u
esti
o
n
g
en
er
atio
n
(
AQG)
clo
s
es
th
i
s
g
ap
b
y
au
to
m
atica
lly
g
en
er
atin
g
item
s
f
r
o
m
tex
t
an
d
/o
r
d
atab
ases
,
th
er
eb
y
im
p
r
o
v
in
g
co
v
er
ag
e
an
d
v
ar
iet
y
with
less
au
th
o
r
in
g
tim
e
[
7
]
,
[
8
]
.
C
ar
ef
u
lly
d
esig
n
ed
AQG
h
as
t
h
e
p
o
ten
t
ial
to
im
p
r
o
v
e
ass
ess
m
en
t
ef
f
icien
cy
an
d
r
ef
o
cu
s
atten
tio
n
f
r
o
m
r
em
em
b
e
r
in
g
an
s
wer
s
to
r
ea
s
o
n
in
g
an
d
u
n
d
er
s
tan
d
in
g
.
T
h
e
ev
o
lu
tio
n
o
f
AQG
ca
n
b
e
tr
ac
ed
b
ac
k
to
its
o
r
ig
in
s
in
1
9
7
6
[
9
]
.
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
I
n
d
o
B
E
R
T fo
r
ed
u
ca
tio
n
a
l a
s
s
ess
men
t:
co
mp
a
r
a
tive
a
n
a
lysi
s
o
f tra
n
s
fo
r
mer m
o
d
els
… (
Ha
n
d
a
r
u
Ja
ti
)
1805
Fro
m
a
m
eth
o
d
o
lo
g
ical
p
er
s
p
ec
tiv
e,
th
e
ev
o
lu
tio
n
o
f
AQG
m
eth
o
d
s
b
eg
an
with
r
u
le
-
b
ased
p
ip
elin
es
th
at
in
clu
d
ed
s
em
an
tic/s
y
n
tactic
m
atch
in
g
an
d
h
an
d
wr
itten
tem
p
lates.
AQG
h
as
ex
p
an
d
ed
b
ey
o
n
d
th
e
class
r
o
o
m
an
d
f
o
u
n
d
ap
p
licatio
n
s
i
n
c
o
m
m
er
cial
s
ec
to
r
s
s
u
ch
as
c
h
atb
o
t
d
e
v
elo
p
m
en
t
[
7
]
an
d
h
ea
lt
h
ca
r
e
s
er
v
ices
[
1
0
]
.
I
t
p
r
o
g
r
ess
ed
to
n
e
u
r
al
m
o
d
els
th
at
lear
n
q
u
esti
o
n
-
a
n
s
wer
in
g
d
ir
ec
tly
f
r
o
m
d
ata
[
1
1
]
–
[
1
5
]
.
T
h
e
p
r
o
ce
s
s
r
eq
u
ir
es
s
y
s
tem
s
to
ac
cu
r
ately
in
ter
p
r
et
tex
t
an
d
g
en
er
ate
well
-
s
tr
u
ctu
r
ed
,
m
ea
n
in
g
f
u
l
q
u
esti
o
n
s
[
1
6
]
–
[
1
8
]
.
E
n
co
d
er
-
d
ec
o
d
er
m
o
d
els
a
n
d
cr
o
s
s
-
lin
g
u
al
tr
a
n
s
f
er
h
av
e
m
a
d
e
th
em
ap
p
licab
le
in
lo
w
-
r
es
o
u
r
ce
en
v
ir
o
n
m
en
ts
as
well
[
1
9
]
,
[
2
0
]
,
a
n
d
p
r
e
-
tr
ain
in
g
g
o
als
s
u
ch
as
m
ask
ed
lan
g
u
ag
e
m
o
d
elin
g
(
ML
M)
h
av
e
en
h
an
ce
d
lin
g
u
is
tic
ca
p
ab
ilit
ies
[
2
1
]
,
[
2
2
]
.
T
h
e
r
ec
u
r
r
en
t
n
eu
r
al
n
etwo
r
k
s
(
R
NNs)
en
ab
le
n
atu
r
al
lan
g
u
ag
e
p
r
o
ce
s
s
in
g
(
NL
P)
task
s
b
y
co
n
s
id
er
in
g
th
e
s
eq
u
en
tial
co
n
tex
t.
Ho
wev
e
r
,
R
NNs
f
ac
e
ch
allen
g
es
lik
e
v
an
is
h
in
g
g
r
a
d
ien
ts
an
d
lo
n
g
d
ep
e
n
d
en
cies.
T
h
e
lo
n
g
s
h
o
r
t
-
te
r
m
m
e
m
o
r
y
(
L
STM
)
n
etwo
r
k
s
s
o
l
v
e
th
e
ab
o
v
e
c
h
allen
g
es
b
y
in
co
r
p
o
r
atin
g
m
e
m
o
r
y
ce
lls
to
s
to
r
e
d
ata
f
o
r
lo
n
g
s
eq
u
en
ce
s
[
2
3
]
–
[
2
6
]
.
T
h
e
u
s
e
o
f
tr
an
s
f
o
r
m
er
m
o
d
e
ls
h
as
f
u
r
t
h
er
a
d
v
an
ce
d
t
h
e
f
i
eld
o
f
AQG
.
T
h
ey
ef
f
icien
tly
ex
p
lo
r
in
g
lo
n
g
-
r
a
n
g
e
d
e
p
en
d
e
n
cies
.
C
u
r
r
en
tly
,
NL
P
is
wid
ely
ap
p
ly
in
g
m
o
d
els
lik
e
b
i
d
ir
ec
tio
n
al
en
co
d
er
r
ep
r
esen
tatio
n
s
f
r
o
m
t
r
an
s
f
o
r
m
er
s
(
B
E
R
T
)
,
g
en
er
ativ
e
p
r
e
-
tr
ain
ed
tr
a
n
s
f
o
r
m
er
(
GPT)
,
an
d
tex
t
-
to
-
tex
t
tr
an
s
f
er
tr
an
s
f
o
r
m
er
(
T
5
)
[
2
7
]
–
[
4
0
]
.
Fo
r
I
n
d
o
n
esian
,
I
n
d
o
n
esian
B
E
R
T
(
I
n
d
o
B
E
R
T
)
p
r
o
v
id
es
a
m
o
n
o
lin
g
u
al
en
co
d
er
p
r
e
-
tr
ain
ed
o
n
I
n
d
o
n
esian
tex
t
[
4
1
]
.
Pre
v
io
u
s
s
tu
d
ies
h
av
e
in
v
esti
g
ated
m
o
n
o
lin
g
u
al
en
c
o
d
er
s
f
o
r
I
n
d
o
n
esian
AQGs
u
s
in
g
s
eq
u
en
ce
-
to
-
s
eq
u
en
ce
ar
c
h
itectu
r
es
(
o
p
en
n
eu
r
al
m
ac
h
in
e
t
r
an
s
latio
n
(
Op
en
NM
T
)
u
s
in
g
b
id
ir
ec
tio
n
al
g
ated
r
ec
u
r
r
en
t
u
n
it
(
B
iGR
U)
,
b
id
ir
ec
tio
n
al
lo
n
g
s
h
o
r
t
-
ter
m
m
e
m
o
r
y
(
B
i
-
L
STM
)
,
o
r
tr
an
s
f
o
r
m
er
)
th
at
ac
h
iev
e
co
m
p
etitiv
e
p
er
f
o
r
m
a
n
ce
o
n
SQu
AD
2
.
0
tr
an
s
latio
n
an
d
T
y
DiQA
-
s
ty
le
d
atasets
[
4
1
]
–
[
4
4
]
.
Ho
wev
er
,
t
h
er
e
h
as
b
ee
n
n
o
c
o
m
p
r
e
h
en
s
iv
e
e
v
alu
atio
n
th
at
s
im
u
ltan
eo
u
s
ly
co
n
s
id
er
s
th
e
ef
f
ec
tiv
en
ess
an
d
co
s
t
ef
f
icien
c
y
o
f
en
co
d
er
s
f
o
r
g
en
er
al
AQGs.
I
n
th
is
p
a
p
er
,
I
n
d
o
B
E
R
T
is
co
m
p
ar
ed
f
o
r
th
e
f
ir
s
t
ti
m
e
to
m
B
E
R
T
an
d
B
E
R
T
-
lar
g
e
in
ter
m
s
o
f
th
eir
p
er
f
o
r
m
a
n
ce
o
n
id
en
tical
d
o
wn
s
tr
ea
m
d
atasets
,
th
e
I
n
d
o
n
esian
SQu
AD
2
.
0
(
2
0
,
0
0
0
s
u
b
s
et)
a
n
d
T
y
DiQA
-
Go
ld
P,
u
s
in
g
b
ilin
g
u
al
ev
alu
at
io
n
u
n
d
er
s
tu
d
y
s
co
r
e
n
-
g
r
am
4
(
B
L
E
U
-
4
)
,
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
)
,
an
d
R
OUGE
-
L
in
co
ln
(
R
OUGE
-
L
)
s
co
r
es,
r
esp
ec
tiv
ely
,
as e
v
alu
atio
n
cr
iter
ia
f
o
r
AQG
m
o
d
els.
T
h
er
e
ar
e
th
r
ee
m
ain
co
n
tr
ib
u
tio
n
s
f
r
o
m
th
is
s
tu
d
y
.
First,
to
th
e
b
est
o
f
o
u
r
k
n
o
wled
g
e,
th
is
s
tu
d
y
o
f
f
er
s
th
e
f
ir
s
t
co
m
p
r
eh
e
n
s
iv
e
co
m
p
a
r
is
o
n
o
f
I
n
d
o
B
E
R
T
,
m
u
ltil
in
g
u
al
B
E
R
T
(
m
B
E
R
T
)
,
a
n
d
B
E
R
T
-
lar
g
e
f
o
r
th
e
I
n
d
o
n
esian
AQG
task
u
n
d
er
e
q
u
al
t
r
ain
in
g
an
d
test
in
g
co
n
d
itio
n
s
.
T
h
e
s
ec
o
n
d
co
n
tr
ib
u
tio
n
is
a
co
m
p
r
eh
e
n
s
iv
e
f
in
e
-
t
u
n
in
g
p
ip
elin
e
th
at
lev
er
a
g
es
th
e
an
s
wer
-
h
ig
h
lig
h
tin
g
m
ec
h
a
n
is
m
v
ia
th
e
tag
s
`
[
HL
]
…[
/HL
]
`
to
en
s
u
r
e
th
e
m
o
d
el
m
ai
n
tain
s
atten
tio
n
o
n
th
e
r
elev
an
t
a
n
s
wer
p
o
r
ti
o
n
an
d
en
a
b
les
th
e
g
en
er
atio
n
o
f
g
en
u
in
e
q
u
esti
o
n
s
.
T
h
e
th
ir
d
co
n
tr
ib
u
tio
n
is
an
an
aly
s
is
o
f
ef
f
icien
cy
asp
ec
ts
,
in
clu
d
in
g
th
e
n
u
m
b
er
o
f
p
ar
am
eter
s
,
th
e
ti
m
e
p
er
ep
o
ch
,
GPU
m
em
o
r
y
co
n
s
u
m
p
tio
n
,
an
d
p
r
o
c
ess
in
g
r
ate,
in
a
p
r
ac
tical
s
ettin
g
o
n
m
id
-
r
a
n
g
e
g
r
ap
h
i
cs
p
r
o
ce
s
s
in
g
u
n
it
(
GPU
)
m
o
d
els.
T
o
g
eth
er
,
th
ese
co
n
tr
i
b
u
tio
n
s
s
h
o
w
th
at
lan
g
u
ag
e
-
s
p
ec
if
ic
p
r
etr
ain
in
g
ca
n
o
f
f
er
a
p
r
ac
tical
ad
v
a
n
tag
e
o
v
er
m
u
ltil
in
g
u
al
an
d
l
ar
g
er
en
co
d
er
s
f
o
r
I
n
d
o
n
esian
AQG.
B
ey
o
n
d
th
is
task
,
th
e
s
am
e
in
s
ig
h
t
is
r
ele
v
an
t
f
o
r
o
th
er
I
n
d
o
n
esian
NL
P
ap
p
licatio
n
s
,
s
u
ch
as
s
u
m
m
ar
izatio
n
,
tr
an
s
latio
n
,
an
d
ad
ap
tiv
e
lear
n
in
g
s
y
s
tem
s
,
wh
er
e
r
eso
u
r
ce
-
ef
f
icien
t
y
et
ac
cu
r
ate
m
o
d
els
ar
e
ess
en
tial
[
4
1
]
–
[
4
4
]
.
2.
M
E
T
H
O
D
W
e
ad
o
p
t
a
r
esear
ch
an
d
d
ev
e
lo
p
m
en
t
m
eth
o
d
o
l
o
g
y
to
b
u
ild
an
d
e
v
alu
ate
I
n
d
o
n
esian
AQG
m
o
d
els
u
s
in
g
tr
an
s
f
o
r
m
er
-
b
ased
tech
n
iq
u
es.
E
ac
h
o
f
th
e
t
h
r
ee
p
r
e
-
tr
ain
ed
B
E
R
T
v
ar
ian
ts
(
I
n
d
o
B
E
R
T
,
m
B
E
R
T
,
an
d
B
E
R
T
-
lar
g
e)
was
f
in
e
-
tu
n
ed
o
n
o
u
r
I
n
d
o
n
esian
q
u
esti
o
n
an
s
wer
in
g
(
QA)
d
atasets
u
n
d
er
a
co
n
s
is
ten
t
tr
ain
in
g
r
eg
im
en
[
4
5
]
,
[
4
6
]
.
Fin
e
-
tu
n
i
n
g
allo
ws
th
e
m
o
d
els
to
ad
ap
t
th
eir
p
r
e
-
tr
ai
n
ed
lan
g
u
ag
e
u
n
d
er
s
tan
d
in
g
to
th
e
s
p
ec
if
ic
task
o
f
q
u
esti
o
n
g
en
er
atio
n
.
T
o
e
n
s
u
r
e
a
f
air
co
m
p
ar
is
o
n
,
we
m
ain
tain
e
d
th
e
s
am
e
d
ata
p
r
ep
r
o
ce
s
s
in
g
,
tr
ain
in
g
s
ch
e
d
u
le,
an
d
o
p
tim
izatio
n
p
ar
am
eter
s
f
o
r
all
m
o
d
els,
ad
ju
s
tin
g
o
n
ly
th
e
m
o
d
el
-
s
p
ec
if
ic
co
m
p
o
n
en
ts
s
u
ch
as
th
e
to
k
en
izer
an
d
p
r
e
-
tr
ain
ed
weig
h
ts
.
All
ex
p
er
im
e
n
ts
wer
e
im
p
lem
en
te
d
i
n
Py
th
o
n
u
s
in
g
th
e
Py
T
o
r
c
h
d
ee
p
lear
n
in
g
f
r
am
ewo
r
k
,
lev
er
ag
in
g
its
f
ac
ilit
ies
f
o
r
tr
an
s
f
o
r
m
er
m
o
d
els
an
d
s
eq
u
en
ce
g
en
er
atio
n
.
T
h
is
u
n
i
f
o
r
m
m
eth
o
d
o
lo
g
y
e
n
ab
les
a
d
ir
ec
t
p
er
f
o
r
m
a
n
ce
c
o
m
p
ar
is
o
n
to
d
eter
m
in
e
th
e
o
p
tim
al
m
o
d
el
f
o
r
I
n
d
o
n
esian
AQG.
2
.
1
.
P
r
o
ce
du
re
T
h
e
o
v
e
r
all
p
r
o
ce
d
u
r
e
f
o
r
d
e
v
elo
p
in
g
th
e
I
n
d
o
n
esian
q
u
est
io
n
g
en
e
r
atio
n
m
o
d
els
was
id
en
tical
f
o
r
I
n
d
o
B
E
R
T
,
m
B
E
R
T
,
an
d
B
E
R
T
-
lar
g
e.
W
e
f
o
llo
wed
th
r
ee
m
ain
s
tag
es
f
o
r
ea
ch
m
o
d
el:
d
ataset
p
r
ep
ar
atio
n
,
m
o
d
el
f
in
e
-
tu
n
in
g
(
tr
ain
in
g
)
,
an
d
ev
alu
atio
n
.
T
h
is
s
u
p
er
v
is
ed
lear
n
in
g
p
ip
elin
e
ad
h
er
es
to
s
tan
d
ar
d
p
r
ac
tices
in
m
ac
h
in
e
lear
n
in
g
m
o
d
el
d
e
v
elo
p
m
en
t,
en
s
u
r
in
g
th
at
ea
ch
m
o
d
el
u
n
d
er
g
o
es
th
e
s
am
e
s
e
q
u
en
ce
o
f
s
tep
s
.
B
y
k
ee
p
in
g
th
e
p
r
o
ce
d
u
r
e
co
n
s
is
ten
t,
we
ca
n
attr
ib
u
te
p
er
f
o
r
m
an
ce
d
if
f
er
en
ce
s
to
th
e
m
o
d
el
s
th
em
s
elv
es
r
ath
er
th
an
to
a
n
y
v
ar
iatio
n
in
p
r
o
c
ess
in
g
.
T
h
e
o
r
ch
estra
tio
n
o
f
th
ese
p
r
o
ce
s
s
es
is
f
ac
ilit
ated
th
r
o
u
g
h
a
Py
th
o
n
en
v
ir
o
n
m
en
t a
n
d
u
s
in
g
Py
T
o
r
ch
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2252
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
15
,
No
.
2
,
Ap
r
il
20
26
:
1
8
0
4
-
1
8
1
3
1806
2
.
2
.
Da
t
a
prepa
ra
t
i
o
n
W
e
u
s
e
SQu
AD
2
.
0
[
4
7
]
a
n
d
T
y
DiQA
-
Go
ld
P
[
4
8
]
p
ar
al
lel
s
o
u
r
ce
s
o
f
co
n
te
x
t
-
an
s
wer
-
q
u
esti
o
n
tr
ip
les.
W
e
f
ir
s
t
tr
an
s
late
b
o
th
co
r
p
o
r
a
in
to
I
n
d
o
n
esian
an
d
th
en
s
p
lit
th
em
in
to
t
r
ain
in
g
,
v
alid
atio
n
,
an
d
test
s
ets.
B
ec
au
s
e
SQu
AD
2
.
0
is
s
u
b
s
tan
tially
lar
g
er
th
an
T
y
Di
QA
-
Go
ld
P,
we
s
am
p
le
2
0
,
0
0
0
QA
p
air
s
f
r
o
m
th
e
SQu
AD
2
.
0
tr
ai
n
in
g
s
p
lit.
T
h
is
ch
o
ice
k
ee
p
s
tr
ain
in
g
ti
m
e
m
an
a
g
ea
b
le
a
n
d
y
ield
s
a
tr
ain
in
g
s
et
th
at
is
co
m
p
ar
ab
le
i
n
s
ize
to
T
y
DiQA
-
Go
ld
P.
All
th
r
ee
m
o
d
els
ar
e
tr
ain
ed
o
n
th
e
s
am
e
s
et
o
f
i
n
s
tan
ce
s
to
en
s
u
r
e
a
f
air
co
m
p
ar
is
o
n
.
T
h
e
r
esu
ltin
g
I
n
d
o
n
esian
p
ar
a
g
r
ap
h
s
,
an
s
wer
s
,
an
d
r
ef
er
e
n
ce
q
u
esti
o
n
s
p
r
o
v
id
e
a
s
h
ar
e
d
s
u
p
er
v
is
io
n
s
ig
n
al
ac
r
o
s
s
e
n
co
d
er
s
.
T
o
b
alan
ce
t
r
ain
in
g
tim
e
an
d
m
ain
tain
a
c
o
m
p
a
r
ab
le
co
m
p
u
tatio
n
al
b
u
d
g
et
o
n
a
s
in
g
le
1
6
GB
GPU,
we
ca
p
th
e
SQu
AD
2
.
0
p
o
r
tio
n
at
2
0
,
0
0
0
in
s
tan
ce
s
wh
il
e
r
etain
in
g
t
h
e
f
u
ll
T
y
DiQA
-
Go
ld
P
s
p
lit.
As
a
r
o
b
u
s
tn
ess
ch
ec
k
,
we
r
etr
ain
I
n
d
o
B
E
R
T
o
n
d
is
jo
in
t
2
0
,
0
0
0
S
Qu
AD
s
am
p
les
an
d
o
b
s
er
v
e
n
o
q
u
alitativ
e
c
h
an
g
e
in
th
e
m
o
d
el
r
an
k
i
n
g
o
r
co
n
clu
s
io
n
s
.
2
.
3
.
M
o
del f
ine
-
t
un
ing
W
e
f
r
am
e
th
e
I
n
d
o
n
esian
an
s
wer
-
awa
r
e
AQG
as
a
co
n
d
itio
n
al
s
eq
u
en
ce
g
en
er
atio
n
with
tr
an
s
f
o
r
m
er
en
co
d
er
s
.
Du
r
i
n
g
tr
ain
in
g
,
ea
ch
m
o
d
el
r
ec
eiv
es
a
p
ass
ag
e
in
wh
ich
th
e
tar
g
et
an
s
wer
s
p
an
is
m
ar
k
ed
with
[
HL
]
…[
/HL
]
an
d
lear
n
s
to
p
r
ed
ict
th
e
n
ex
t
q
u
esti
o
n
to
k
en
at
[
MA
SK]
g
iv
en
th
e
h
ig
h
lig
h
ted
co
n
tex
t
an
d
th
e
p
r
ev
io
u
s
ly
g
e
n
er
ated
to
k
en
s
(
cf
.
B
E
R
T
-
b
ased
q
u
esti
o
n
g
en
er
atio
n
(
QG
)
[
4
9
]
–
[
5
1
]
a
n
d
th
e
I
n
d
o
B
E
R
T
s
ettin
g
[
4
1
]
)
.
T
h
e
[
HL
]
tag
s
a
ct
as
a
s
o
f
t
p
o
i
n
ter
to
th
e
an
s
wer
s
p
an
,
h
elp
in
g
th
e
m
o
d
el
f
o
cu
s
o
n
th
e
in
ten
d
ed
co
n
ten
t
an
d
r
ed
u
cin
g
o
f
f
-
tar
g
e
t
q
u
esti
o
n
s
.
W
e
ap
p
ly
th
e
s
am
e
f
in
e
-
tu
n
in
g
r
ec
ip
e
to
I
n
d
o
B
E
R
T
,
m
B
E
R
T
,
an
d
B
E
R
T
-
lar
g
e
an
d
k
ee
p
th
e
co
r
e
tr
ain
in
g
h
y
p
er
p
a
r
am
eter
s
f
ix
ed
to
p
r
eser
v
e
p
r
o
to
co
l
p
a
r
ity
ac
r
o
s
s
m
o
d
els.
T
ab
le
1
s
u
m
m
ar
izes
th
e
lear
n
in
g
r
ate,
b
atch
c
o
n
f
i
g
u
r
atio
n
,
s
eq
u
en
ce
len
g
th
,
o
p
tim
izer
,
p
r
ec
is
io
n
,
an
d
ea
r
ly
-
s
to
p
p
in
g
s
ettin
g
s
f
o
r
ea
ch
en
c
o
d
er
.
T
ab
le
1
.
Hy
p
er
p
ar
a
m
eter
s
u
s
ed
ac
r
o
s
s
m
o
d
els
M
o
d
e
l
Le
a
r
n
i
n
g
r
a
t
e
B
a
t
c
h
si
z
e
G
r
a
d
i
e
n
t
a
c
c
u
m
u
l
a
t
i
o
n
Ef
f
e
c
t
i
v
e
b
a
t
c
h
M
a
x
s
e
q
l
e
n
g
t
h
O
p
t
i
mi
z
e
r
P
r
e
c
i
s
i
o
n
Ea
r
l
y
st
o
p
p
i
n
g
Ep
o
c
h
I
n
d
o
B
ER
T
5
×
1
0
-
5
8
4
32
1
2
8
A
d
a
mW
16
Y
3
mBE
R
T
5
×
1
0
-
5
8
4
32
1
2
8
A
d
a
mW
16
Y
3
B
ER
T
-
l
a
r
g
e
5
×
1
0
-
5
8
8
64
96
A
d
a
mW
16
Y
3
W
e
tr
ain
all
th
r
ee
en
co
d
er
s
w
ith
Ad
am
W
at
a
f
ix
ed
lear
n
in
g
r
ate
o
f
5
×1
0
-
5
a
n
d
a
p
e
r
-
d
e
v
ice
b
atch
s
ize
o
f
8
,
u
s
in
g
FP
1
6
m
ix
ed
p
r
ec
is
io
n
.
T
o
ac
h
iev
e
lar
g
er
e
f
f
ec
tiv
e
b
atch
s
izes
with
o
u
t
e
x
ce
ed
in
g
1
6
GB
o
f
GPU
m
em
o
r
y
,
we
u
s
e
g
r
ad
ie
n
t
ac
cu
m
u
latio
n
:
I
n
d
o
B
E
R
T
an
d
m
B
E
R
T
ac
cu
m
u
late
4
s
tep
s
(
ef
f
ec
tiv
e
b
atch
s
ize
3
2
)
,
wh
er
ea
s
B
E
R
T
-
lar
g
e
ac
cu
m
u
lates 8
s
tep
s
(
ef
f
ec
tiv
e
b
atch
s
ize
6
4
)
to
s
tab
ilize
tr
ain
in
g
f
o
r
t
h
e
d
ee
p
er
24
-
lay
er
ar
ch
itectu
r
e
.
W
e
s
et
th
e
m
ax
im
u
m
s
eq
u
e
n
ce
len
g
th
to
1
2
8
to
k
e
n
s
f
o
r
I
n
d
o
B
E
R
T
an
d
m
B
E
R
T
,
an
d
9
6
to
k
en
s
f
o
r
B
E
R
T
-
lar
g
e,
to
b
alan
ce
c
o
n
tex
tu
al
c
o
v
er
a
g
e
with
m
em
o
r
y
u
s
ag
e.
E
ac
h
m
o
d
el
tr
ain
s
f
o
r
u
p
to
3
ep
o
ch
s
with
ea
r
ly
s
to
p
p
in
g
b
ased
o
n
v
alid
atio
n
lo
s
s
,
p
r
o
v
id
in
g
a
f
air
,
c
o
m
p
u
tatio
n
ally
co
m
p
a
r
ab
le
s
etu
p
ac
r
o
s
s
en
co
d
er
s
.
2
.
4
.
E
v
a
lua
t
i
o
n
W
e
ev
alu
ate
th
e
g
en
er
ated
q
u
esti
o
n
s
with
B
L
E
U
-
4
[
5
2
]
,
M
E
T
E
OR
[
5
3
]
,
an
d
R
OUGE
-
L
[
5
4
]
,
with
ad
d
itio
n
al
b
ac
k
g
r
o
u
n
d
in
te
x
t
s
u
m
m
ar
izatio
n
[
5
5
]
.
B
L
E
U
-
4
m
ea
s
u
r
es
n
-
g
r
am
p
r
ec
is
io
n
with
a
b
r
e
v
ity
p
en
alty
,
ME
T
E
OR
em
p
h
asize
s
r
ec
all
u
s
in
g
s
y
n
o
n
y
m
an
d
s
tem
m
atch
in
g
,
an
d
R
OUGE
-
L
co
m
p
u
tes
o
v
er
lap
v
ia
th
e
lo
n
g
est
co
m
m
o
n
s
u
b
s
eq
u
en
ce
.
W
e
u
s
e
r
e
f
er
en
c
e
im
p
lem
en
tatio
n
s
with
d
e
f
au
lt
s
ettin
g
s
.
W
e
ac
k
n
o
wled
g
e
t
h
at
th
ese
n
-
g
r
a
m
o
v
er
lap
m
etr
ics d
o
n
o
t f
u
lly
r
ef
lect
p
ed
ag
o
g
ical
q
u
ality
o
r
s
em
an
tic
ad
eq
u
ac
y
;
we
r
ev
is
it th
eir
lim
itatio
n
s
in
th
e
d
is
cu
s
s
io
n
an
d
p
r
o
v
i
d
e
co
m
p
lete
f
o
r
m
u
las an
d
n
o
tatio
n
.
2
.
5
.
M
o
del a
rc
hite
ct
ure
Fig
u
r
e
1
illu
s
tr
ates
th
e
I
n
d
o
B
E
R
T
-
AQG
p
ip
elin
e.
W
e
f
ir
s
t
m
ar
k
th
e
a
n
s
wer
s
p
an
in
th
e
p
ass
ag
e
with
[
HL
]
…[
/HL
]
,
to
o
b
tain
th
e
h
i
g
h
lig
h
ted
co
n
tex
t
C
’
.
T
h
e
h
ig
h
lig
h
ted
co
n
te
x
t
an
d
th
e
p
ar
ti
al
q
u
esti
o
n
to
k
en
s
ar
e
th
en
to
k
e
n
ized
,
a
n
d
en
co
d
ed
with
I
n
d
o
B
E
R
T
.
At
ea
c
h
d
ec
o
d
in
g
s
tep
i
,
th
e
m
o
d
el
p
r
ed
icts
th
e
n
ex
t
q
u
esti
o
n
to
k
e
n
at
[
MA
SK]
g
i
v
en
C
’
an
d
th
e
p
r
e
v
io
u
s
ly
g
e
n
er
ated
to
k
e
n
s
q
̂
1
:
−
1
;
th
e
n
ew
to
k
e
n
is
ap
p
en
d
e
d
u
n
til
th
e
f
u
ll
q
u
esti
o
n
(
q
)
.
Fo
r
m
o
tiv
atio
n
a
n
d
p
r
io
r
w
o
r
k
o
n
[
HL
]
tag
g
in
g
i
n
a
n
s
wer
-
awa
r
e
q
u
esti
o
n
g
en
er
atio
n
,
we
r
ef
er
th
e
r
ea
d
e
r
to
s
u
b
s
ec
tio
n
2
.
5
an
d
in
s
tu
d
i
es
[
4
1
]
,
[
4
9
]
,
[
5
0
]
.
Fo
r
m
ally
,
we
f
o
llo
w
th
e
B
E
R
-
b
ased
h
ier
ar
ch
ical
lab
el
-
awa
r
e
s
en
ten
ce
q
u
esti
o
n
g
en
er
atio
n
(
B
E
R
T
-
HL
SQG
)
f
o
r
m
u
latio
n
[
4
4
]
.
I
n
d
o
B
E
R
T
h
as
th
e
B
E
R
T
-
b
ase
ar
ch
itectu
r
e
b
u
t
is
p
r
e
-
tr
ain
ed
o
n
I
n
d
o
n
esian
.
At
d
ec
o
d
i
n
g
s
tep
i
,
th
e
m
o
d
el
r
ec
eiv
es th
e
in
p
u
t
s
eq
u
en
ce
in
(
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
I
n
d
o
B
E
R
T fo
r
ed
u
ca
tio
n
a
l a
s
s
ess
men
t:
co
mp
a
r
a
tive
a
n
a
lysi
s
o
f tra
n
s
fo
r
mer m
o
d
els
… (
Ha
n
d
a
r
u
Ja
ti
)
1807
=
(
[
C
L
S
]
,
,
[
SEP
]
,
q
̂
1
,
…
,
q
̂
,
[
M
A
SK
]
)
(
1
)
Her
e,
q
̂
1
=
−
1
ar
e
th
e
p
r
ev
io
u
s
ly
g
e
n
e
r
ated
q
u
esti
o
n
to
k
en
s
,
an
d
[
M
A
SK
]
m
ar
k
s
th
e
p
o
s
itio
n
to
p
r
ed
ict
n
ex
t.
T
h
e
n
ex
t
-
t
o
k
en
d
is
tr
ib
u
tio
n
is
p
r
o
d
u
ce
d
f
r
o
m
th
e
f
in
al
h
i
d
d
e
n
s
tate
at
[
M
A
SK
]
v
ia
an
af
f
i
n
e
lay
er
a
n
d
s
o
f
tm
ax
in
(
2
)
,
an
d
t
h
e
to
k
e
n
ch
o
ice
is
m
ad
e
b
y
ar
g
m
a
x
in
(
3
)
.
Pr
(
|
)
=
(
ℎ
[
m
as
k
]
.
+
)
(
2
)
q
̂
=
Pr
(
|
)
(
3
)
T
o
g
eth
er
,
(
1
)
to
(
3
)
f
o
r
m
alize
th
e
lo
o
p
d
e
p
icted
in
Fig
u
r
e
1
:
h
ig
h
lig
h
t
th
e
a
n
s
wer
s
p
an
,
en
co
d
e
t
h
e
co
n
tex
t
with
I
n
d
o
B
E
R
T
,
p
r
ed
ict
t
h
e
n
ex
t to
k
en
at
[
MA
SK]
,
an
d
a
p
p
en
d
it to
th
e
p
ar
tial q
u
esti
o
n
u
n
til [
SEP]
.
Fig
u
r
e
1
.
I
n
d
o
B
E
R
T
-
AQG
p
ip
elin
e
2
.
6
.
Da
t
a
a
na
ly
s
is
t
ec
hn
iqu
e
s
Data
an
aly
s
is
en
co
m
p
ass
ed
th
e
ev
alu
atio
n
o
f
ea
c
h
m
o
d
el
co
n
f
ig
u
r
atio
n
,
wh
ich
u
n
d
er
wen
t
th
e
tr
ain
in
g
s
tag
e
u
tili
zin
g
s
tan
d
ar
d
au
to
m
atic
ev
alu
atio
n
m
etr
ic
s
to
co
m
p
a
r
e
p
e
r
f
o
r
m
an
ce
r
esu
lts
.
T
h
ese
m
etr
ics
ass
es
s
ed
an
d
d
eter
m
in
ed
th
e
m
o
d
el
co
n
f
ig
u
r
atio
n
with
th
e
m
o
s
t
o
p
tim
al
p
er
f
o
r
m
an
ce
d
u
r
in
g
th
e
tr
ain
in
g
p
r
o
ce
s
s
.
Gr
ap
h
an
d
p
lo
ttin
g
t
o
o
ls
wer
e
ad
ep
tly
u
tili
ze
d
to
v
is
u
ally
in
ter
p
r
et
a
n
d
c
o
m
p
r
e
h
e
n
d
th
e
p
er
f
o
r
m
an
ce
tr
en
d
s
in
an
aly
zi
n
g
th
e
n
ewly
d
ev
elo
p
e
d
QG
m
o
d
el.
3.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
3
.
1
.
Da
t
a
prepa
ra
t
i
o
n
W
e
s
elec
ted
th
e
tr
an
s
lated
SQ
u
AD
2
.
0
an
d
T
y
DiQA
-
Go
ld
P
as
o
u
r
d
atasets
.
W
e
f
o
llo
w
th
e
o
r
ig
in
al
s
p
lits
f
r
o
m
b
o
th
SQu
AD
2
.
0
an
d
T
y
DiQA
-
Go
ld
P.
Fo
r
tr
ain
in
g
,
we
u
s
e
o
n
ly
2
0
,
0
0
0
in
s
tan
ce
s
f
r
o
m
th
e
SQu
AD
2
.
0
tr
ain
in
g
s
et,
wh
ile
we
u
s
e
th
e
en
tire
T
y
DiQA
-
G
o
ld
P d
ataset.
3
.
2
.
E
x
perim
ent
a
l set
t
ing
W
e
ap
p
ly
th
e
s
am
e
f
in
e
-
tu
n
i
n
g
s
etu
p
to
I
n
d
o
B
E
R
T
,
m
B
E
R
T
,
an
d
B
E
R
T
-
lar
g
e.
I
n
d
o
B
E
R
T
an
d
m
B
E
R
T
f
o
llo
w
th
e
1
2
-
lay
e
r
B
E
R
T
-
b
ase
co
n
f
ig
u
r
atio
n
,
wh
er
ea
s
B
E
R
T
-
lar
g
e
u
s
es
th
e
2
4
-
lay
er
v
ar
ian
t.
All
ex
p
er
im
en
ts
r
u
n
o
n
a
s
in
g
le
NVI
DI
A
R
T
X
4
0
6
0
T
i
GPU
(
1
6
GB
VR
AM
)
.
W
e
u
s
e
Ad
a
m
W
with
a
lear
n
in
g
r
ate
o
f
5
×1
0
-
5
,
th
e
m
ax
im
u
m
s
eq
u
en
ce
len
g
t
h
s
p
ec
if
ied
in
T
a
b
le
1
,
FP
1
6
m
ix
ed
p
r
ec
is
io
n
,
a
n
d
ea
r
ly
s
to
p
p
i
n
g
.
Fo
r
I
n
d
o
B
E
R
T
an
d
m
B
E
R
T
,
we
u
s
e
an
ad
eq
u
ate
b
atch
s
ize
o
f
3
2
,
a
n
d
f
o
r
B
E
R
T
-
lar
g
e,
we
r
ely
o
n
a
s
m
aller
p
er
-
s
tep
b
atch
with
g
r
ad
ien
t
ac
cu
m
u
latio
n
to
f
it
with
in
1
6
GB
o
f
VR
AM
;
th
e
s
m
all
er
b
atch
also
ad
d
s
r
eg
u
lar
izatio
n
u
n
d
er
lim
ited
d
ata.
W
e
m
atch
th
e
to
tal
n
u
m
b
er
o
f
o
p
tim
izatio
n
s
tep
s
an
d
th
e
lear
n
in
g
r
ate
s
ch
ed
u
le
ac
r
o
s
s
m
o
d
els
s
o
th
at
o
b
s
er
v
ed
d
if
f
er
e
n
ce
s
r
ef
lec
t
ar
ch
itectu
r
al
an
d
p
r
e
-
tr
ain
in
g
ef
f
ec
ts
r
ath
er
th
an
th
e
tr
ain
in
g
p
r
o
t
o
co
l.
T
o
m
ain
tain
co
m
p
u
tatio
n
al
-
b
u
d
g
et
p
ar
ity
in
t
h
is
s
in
g
le
-
GPU
s
ettin
g
,
we
ca
p
th
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2252
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
15
,
No
.
2
,
Ap
r
il
20
26
:
1
8
0
4
-
1
8
1
3
1808
SQu
AD
2
.
0
tr
ain
in
g
p
o
r
tio
n
at
2
0
,
0
0
0
in
s
tan
ce
s
(
with
a
r
o
b
u
s
tn
ess
r
e
-
s
am
p
lin
g
ch
ec
k
y
ield
in
g
th
e
s
am
e
q
u
alitativ
e
r
an
k
i
n
g
)
wh
ile
r
etain
in
g
th
e
f
u
ll T
y
DiQA
-
Go
ld
P
s
p
lit
.
3
.
3
.
E
v
a
lua
t
i
o
n
T
h
e
p
er
f
o
r
m
a
n
ce
o
f
I
n
d
o
B
E
R
T
,
m
B
E
R
T
,
an
d
B
E
R
T
-
lar
g
e
o
n
th
e
two
I
n
d
o
n
esian
QA
d
atasets
is
s
u
m
m
ar
ized
in
T
ab
le
s
2
a
n
d
3
.
I
n
d
o
B
E
R
T
ac
h
iev
ed
th
e
h
i
g
h
est
s
co
r
es
ac
r
o
s
s
all
ev
alu
atio
n
m
etr
ics
o
n
b
o
th
d
atasets
,
wh
ile
m
B
E
R
T
s
h
o
w
ed
th
e
s
ec
o
n
d
-
b
est
p
er
f
o
r
m
a
n
ce
an
d
B
E
R
T
-
lar
g
e
th
e
lo
west.
T
h
is
r
an
k
in
g
is
co
n
s
is
ten
t
f
o
r
ea
c
h
m
etr
ic
(
B
L
E
U
-
1
th
r
o
u
g
h
B
L
E
U
-
4
,
ME
T
E
OR
,
an
d
R
OUGE
-
L
)
,
r
ef
le
ctin
g
th
e
a
d
v
an
ta
g
es
o
f
a
lan
g
u
ag
e
-
s
p
ec
if
ic
m
o
d
el
o
v
er
a
m
u
ltil
in
g
u
al
m
o
d
el
an
d
th
e
d
r
awb
ac
k
s
o
f
u
s
in
g
a
lar
g
e
m
o
d
el
n
o
t
p
r
e
-
tr
ain
ed
in
th
e
tar
g
et
lan
g
u
ag
e.
T
ab
le
2
.
Per
f
o
r
m
an
ce
o
f
th
e
th
r
ee
m
o
d
els o
n
th
e
T
y
DiQA
-
G
o
ld
P d
ataset
M
o
d
e
l
B
LEU
-
1
B
LEU
-
2
B
LEU
-
3
B
LEU
-
4
M
ET
EO
R
R
O
U
G
E
-
L
I
n
d
o
B
ER
T
5
4
.
8
1
3
8
.
6
8
2
6
.
8
6
1
9
.
6
9
3
1
.
6
2
5
8
.
8
3
mBE
R
T
4
7
.
6
7
3
0
.
8
3
1
9
.
8
2
1
3
.
6
9
2
5
.
6
4
5
2
.
1
5
B
ER
T
-
l
a
r
g
e
4
6
.
1
5
2
7
.
7
8
1
7
.
2
5
1
1
.
1
2
2
4
.
3
4
5
0
.
6
4
On
T
y
DiQA
-
Go
ld
P,
I
n
d
o
B
E
R
T
’
s
B
L
E
U
-
4
s
co
r
e
o
f
1
9
.
6
9
is
ab
o
u
t
6
p
o
in
ts
h
ig
h
er
t
h
an
m
B
E
R
T
’
s
1
3
.
6
9
an
d
8
.
5
p
o
i
n
ts
h
ig
h
er
th
an
B
E
R
T
-
lar
g
e
’
s
1
1
.
1
2
.
L
ik
e
wis
e,
I
n
d
o
B
E
R
T
lead
s
s
u
b
s
tan
tially
in
ME
T
E
OR
an
d
R
OUGE
-
L
,
in
d
icatin
g
it
g
en
er
ates
q
u
esti
o
n
s
th
at
n
o
t
o
n
ly
m
atch
th
e
r
e
f
er
en
ce
wo
r
d
i
n
g
m
o
r
e
clo
s
ely
b
u
t
also
ca
p
tu
r
e
m
o
r
e
o
f
th
e
r
ef
er
en
ce
co
n
te
n
t.
m
B
E
R
T
’
s
s
c
o
r
es,
wh
ile
lo
wer
th
an
I
n
d
o
B
E
R
T
’
s
,
ar
e
clea
r
ly
ab
o
v
e
th
o
s
e
o
f
B
E
R
T
-
lar
g
e
.
No
tab
ly
,
m
B
E
R
T
o
u
tp
er
f
o
r
m
s
B
E
R
T
-
lar
g
e
b
y
ar
o
u
n
d
2
-
3
p
o
in
ts
o
n
m
o
s
t
m
etr
ics,
d
em
o
n
s
tr
atin
g
th
e
b
e
n
ef
it
o
f
m
u
ltil
in
g
u
al
p
r
e
-
tr
ain
i
n
g
th
at
in
clu
d
es
I
n
d
o
n
esian
:
e
v
en
th
o
u
g
h
m
B
E
R
T
is
a
s
m
aller
m
o
d
el
th
a
n
B
E
R
T
-
lar
g
e
,
its
f
am
iliar
ity
w
ith
I
n
d
o
n
esian
g
i
v
es
it
an
ed
g
e.
B
E
R
T
-
lar
g
e
’
s
u
n
d
er
p
er
f
o
r
m
an
ce
o
n
T
y
DiQ
A
-
Go
ld
P
s
u
g
g
ests
th
at
its
la
r
g
e
ca
p
ac
ity
r
em
ai
n
s
u
n
d
er
u
tili
ze
d
d
u
e
t
o
th
e
m
o
d
el’
s
lack
o
f
p
r
io
r
I
n
d
o
n
esi
an
k
n
o
wled
g
e
a
n
d
th
e
lim
ited
f
in
e
-
tu
n
in
g
d
ata
av
ailab
le.
I
n
p
r
ac
tical
ter
m
s
,
th
e
I
n
d
o
B
E
R
T
m
o
d
el
s
h
o
ws
s
tr
en
g
th
in
h
a
n
d
lin
g
t
h
e
T
y
DiQA
-
Go
ld
P
m
ater
ial,
lik
ely
lev
er
a
g
in
g
its
p
r
e
-
tr
ain
e
d
u
n
d
er
s
tan
d
i
n
g
o
f
I
n
d
o
n
esian
n
u
an
ce
s
to
p
r
o
d
u
ce
m
o
r
e
ac
c
u
r
ate
an
d
f
lu
e
n
t q
u
esti
o
n
s
.
T
ab
le
3
.
Per
f
o
r
m
an
ce
o
f
th
e
th
r
ee
m
o
d
els o
n
th
e
SQu
AD
2
.
0
(
2
0
,
0
0
0
s
u
b
s
et)
d
ataset
M
o
d
e
l
B
LEU
-
1
B
LEU
-
2
B
LEU
-
3
B
LEU
-
4
M
ET
EO
R
R
O
U
G
E
-
L
I
n
d
o
B
ER
T
3
3
.
7
5
1
6
.
2
4
7
.
3
2
3
.
7
9
1
6
.
2
5
3
7
.
4
5
mBE
R
T
2
8
.
2
3
1
0
.
3
1
3
.
6
4
1
.
5
1
1
2
.
6
3
3
2
.
7
7
B
ER
T
-
l
a
r
ge
2
5
.
3
0
8
.
7
2
3
.
1
9
1
.
4
6
1
1
.
2
2
3
0
.
3
2
On
th
e
SQu
AD
2
.
0
s
u
b
s
et,
th
e
o
v
er
all
s
co
r
es
ar
e
l
o
wer
f
o
r
all
m
o
d
els,
r
ef
lectin
g
th
e
in
cr
ea
s
ed
d
if
f
icu
lty
o
f
th
is
d
ataset
an
d
th
e
s
m
aller
tr
ain
in
g
s
am
p
le.
I
n
d
o
B
E
R
T
s
t
ill h
o
ld
s
th
e
h
ig
h
est
s
co
r
es b
y
a
n
o
tab
le
m
ar
g
in
.
Fo
r
in
s
tan
ce
,
I
n
d
o
B
E
R
T
’
s
B
L
E
U
-
4
(
3
.
7
9
)
is
m
o
r
e
t
h
an
d
o
u
b
le
th
at
o
f
m
B
E
R
T
(
1
.
5
1
)
o
r
B
E
R
T
-
lar
g
e
(
1
.
4
6
)
.
T
h
is
d
r
a
m
atic
g
a
p
u
n
d
er
s
co
r
es
I
n
d
o
B
E
R
T
’
s
ef
f
icien
cy
in
lear
n
in
g
f
r
o
m
a
lim
ited
d
ataset.
m
B
E
R
T
an
d
B
E
R
T
-
lar
g
e
b
o
th
s
tr
u
g
g
le
o
n
th
is
d
ataset,
b
u
t
m
B
E
R
T
m
ain
tain
s
a
s
lig
h
t
ad
v
an
tag
e
o
v
er
B
E
R
T
-
lar
g
e
o
n
ev
er
y
m
etr
ic.
T
h
e
d
if
f
er
en
ce
s
b
etwe
en
m
B
E
R
T
an
d
B
E
R
T
-
lar
g
e
,
th
o
u
g
h
s
m
all
in
ab
s
o
lu
te
ter
m
s
h
er
e,
co
n
s
is
ten
tly
f
av
o
r
m
B
E
R
T
(
R
OUGE
-
L
o
f
3
2
.
7
7
v
s
.
3
0
.
3
2
)
,
r
ein
f
o
r
cin
g
th
at
k
n
o
win
g
th
e
lan
g
u
ag
e
is
cr
u
cial
f
o
r
p
er
f
o
r
m
a
n
ce
.
T
h
e
B
E
R
T
-
lar
g
e
m
o
d
el,
d
esp
ite
h
a
v
in
g
o
v
er
th
r
ee
tim
es
th
e
p
ar
am
eter
s
o
f
th
e
o
th
er
s
,
f
ails
to
o
u
t
p
er
f
o
r
m
t
h
e
b
ase
m
o
d
els
in
th
is
lo
w
-
r
eso
u
r
ce
s
ce
n
ar
i
o
.
W
e
attr
ib
u
te
th
is
to
its
in
a
b
ilit
y
to
g
en
er
alize
f
r
o
m
s
u
ch
a
lim
ited
f
in
e
-
tu
n
i
n
g
s
et,
its
lar
g
e
ca
p
ac
ity
ca
n
n
o
t
b
e
ef
f
ec
tiv
ely
u
s
ed
with
o
u
t
f
ar
m
o
r
e
d
ata.
I
n
d
o
B
E
R
T
,
b
y
co
n
t
r
ast,
av
o
i
d
s
th
is
p
itfa
ll
th
an
k
s
to
its
p
r
io
r
I
n
d
o
n
esian
p
r
e
-
tr
ain
in
g
,
wh
ich
allo
ws
it
to
g
en
er
alize
b
etter
f
r
o
m
t
h
e
s
am
e
s
m
all
s
am
p
le.
T
h
e
r
esu
lts
s
h
o
w
th
at
I
n
d
o
B
E
R
T
is
th
e
b
est
-
p
er
f
o
r
m
in
g
m
o
d
el
f
o
r
I
n
d
o
n
esian
q
u
esti
o
n
g
e
n
er
atio
n
i
n
o
u
r
ex
p
er
im
en
ts
,
ex
ce
llin
g
e
s
p
ec
ially
in
s
ce
n
ar
io
s
with
lim
ited
tr
ain
in
g
d
ata.
T
h
e
m
u
ltil
in
g
u
al
m
B
E
R
T
p
r
o
v
id
es d
ec
en
t
p
er
f
o
r
m
a
n
ce
an
d
ca
n
b
e
co
n
s
id
er
ed
a
s
tr
o
n
g
b
aselin
e
f
o
r
I
n
d
o
n
esian
AQ
G,
b
u
t
it
co
n
s
is
ten
tly
lag
s
b
eh
in
d
I
n
d
o
B
E
R
T
.
T
h
e
l
ar
g
e
-
ca
p
ac
ity
B
E
R
T
-
lar
g
e
m
o
d
el
d
id
n
o
t
y
iel
d
an
y
p
er
f
o
r
m
an
ce
b
en
e
f
it
in
th
is
co
n
tex
t;
o
n
th
e
c
o
n
tr
ar
y
,
it
u
n
d
er
p
e
r
f
o
r
m
ed
ev
e
n
th
e
s
m
aller
m
B
E
R
T
.
Fro
m
a
p
r
ac
tical
p
er
s
p
ec
tiv
e,
th
ese
f
in
d
in
g
s
s
u
g
g
est
th
at
f
o
r
I
n
d
o
n
esian
-
lan
g
u
a
g
e
AQG
task
s
,
o
n
e
s
h
o
u
ld
f
av
o
r
a
m
o
d
el
th
at
h
as
b
ee
n
p
r
e
-
tr
ain
ed
o
n
I
n
d
o
n
esian
r
ath
er
t
h
an
s
i
m
p
ly
o
p
tin
g
f
o
r
a
m
o
d
el
wi
th
g
r
ea
ter
s
ize
o
r
g
en
e
r
al
m
u
ltil
in
g
u
al
tr
ain
in
g
.
Ad
d
itio
n
ally
,
I
n
d
o
B
E
R
T
’
s
s
u
p
er
io
r
p
e
r
f
o
r
m
an
ce
,
c
o
u
p
le
d
with
its
r
elativ
ely
s
m
aller
s
i
ze
an
d
th
u
s
f
aster
in
f
er
en
ce
a
n
d
tr
ain
in
g
,
m
ak
es
it
an
attr
ac
tiv
e
c
h
o
ice
f
o
r
r
ea
l
-
w
o
r
ld
a
p
p
licatio
n
s
w
h
er
e
co
m
p
u
tatio
n
a
l
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
I
n
d
o
B
E
R
T fo
r
ed
u
ca
tio
n
a
l a
s
s
ess
men
t:
co
mp
a
r
a
tive
a
n
a
lysi
s
o
f tra
n
s
fo
r
mer m
o
d
els
… (
Ha
n
d
a
r
u
Ja
ti
)
1809
r
eso
u
r
ce
s
an
d
tr
ai
n
in
g
d
ata
m
ay
b
e
lim
ited
.
Me
a
n
wh
ile,
m
B
E
R
T
’
s
r
esu
lt
s
in
d
icate
th
at
if
a
m
u
lti
-
lan
g
u
ag
e
s
o
lu
tio
n
is
r
e
q
u
ir
ed
,
it
ca
n
h
an
d
le
I
n
d
o
n
esian
QG
r
ea
s
o
n
a
b
ly
well,
th
o
u
g
h
with
s
o
m
e
lo
s
s
i
n
q
u
esti
o
n
q
u
ality
.
B
E
R
T
-
lar
g
e
,
g
iv
en
its
r
eso
u
r
c
e
d
em
an
d
s
an
d
lo
w
p
a
y
o
f
f
h
e
r
e,
wo
u
ld
lik
ely
o
n
ly
b
e
ju
s
tifie
d
if
s
ig
n
if
ican
tly
m
o
r
e
I
n
d
o
n
esian
tr
ain
in
g
d
ata
wer
e
av
ailab
le
o
r
if
an
I
n
d
o
n
e
s
ian
-
s
p
ec
if
ic
lar
g
e
m
o
d
el
we
r
e
p
r
e
-
tr
ain
e
d
.
3
.
4
.
E
f
f
iciency
a
nd
re
s
o
urce
us
a
g
e
W
e
c
o
m
p
a
r
e
m
o
d
e
l
s
i
z
e
,
t
im
e
p
e
r
e
p
o
c
h
,
p
e
a
k
V
R
A
M
u
s
a
g
e
,
a
n
d
i
n
f
e
r
e
n
c
e
t
h
r
o
u
g
h
p
u
t
,
a
n
d
s
u
m
m
a
r
i
z
e
t
h
e
r
es
u
lt
s
i
n
T
a
b
le
4
.
A
s
s
h
o
w
n
i
n
T
a
b
l
e
4
,
I
n
d
o
B
E
R
T
i
s
t
h
e
m
o
s
t
e
f
f
i
c
i
e
n
t
m
o
d
e
l
:
i
t
u
s
e
s
f
ew
e
r
p
a
r
a
m
e
t
e
r
s
(
1
2
4
M
)
,
t
r
a
i
n
s
f
a
s
te
r
(
1
h
/
e
p
o
c
h
)
,
r
e
q
u
i
r
e
s
l
e
s
s
VR
A
M
(
4
.
8
G
B
)
,
a
n
d
r
e
a
c
h
e
s
t
h
e
h
i
g
h
e
s
t
t
h
r
o
u
g
h
p
u
t
(
5
,
2
0
0
t
o
k
e
n
s
/
s
)
.
m
B
E
R
T
s
i
ts
in
t
h
e
m
i
d
d
l
e
(
1
7
9
M
;
3
h
/
e
p
o
c
h
;
1
1
G
B
;
3
,
2
0
0
t
o
k
e
n
s
/
s
)
,
w
h
e
r
e
a
s
B
E
R
T
-
l
a
r
g
e
is
t
h
e
m
o
s
t
r
e
s
o
u
r
c
e
-
d
e
m
a
n
d
i
n
g
w
i
t
h
t
h
e
l
o
w
es
t
t
h
r
o
u
g
h
p
u
t
(
3
4
0
M
;
4
h
/
e
p
o
c
h
;
1
1
GB
;
1
,
6
0
0
t
o
k
e
n
s
/
s
)
.
T
a
k
e
n
t
o
g
e
t
h
e
r
,
T
a
b
l
e
4
s
h
o
ws
t
h
a
t
I
n
d
o
B
E
R
T
o
f
f
e
r
s
t
h
e
b
es
t
t
r
a
d
e
-
o
f
f
b
e
t
w
e
e
n
ac
c
u
r
a
c
y
a
n
d
e
f
f
ic
i
en
c
y
f
o
r
I
n
d
o
n
e
s
i
a
n
A
Q
G
u
n
d
e
r
r
e
a
li
s
ti
c
r
es
o
u
r
c
e
b
u
d
g
e
t
s
.
T
ab
le
4
.
E
f
f
icien
cy
s
u
m
m
ar
y
o
f
I
n
d
o
B
E
R
T
,
m
B
E
R
T
,
an
d
B
E
R
T
-
lar
g
e
(
p
ar
am
ete
r
s
(
M)
,
ti
m
e
p
er
ep
o
ch
(
h
)
,
p
ea
k
GPU
m
em
o
r
y
(
GB
)
,
th
r
o
u
g
h
p
u
t
(
to
k
e
n
s
/s
)
)
M
o
d
e
l
P
a
r
a
me
t
e
r
(
M
)
Ti
me
p
e
r
e
p
o
c
h
(
h
)
P
e
a
k
G
P
U
mem
o
r
y
(
G
B
)
Th
r
o
u
g
h
p
u
t
(
t
o
k
e
n
/
s)
I
n
d
o
B
ER
T
1
2
4
1
4
.
8
5
,
2
0
0
mBE
R
T
1
7
9
3
11
3
,
2
0
0
B
ER
T
-
l
a
r
ge
3
4
0
4
11
1
,
6
0
0
T
h
ese
ef
f
icien
cy
r
esu
lts
ca
r
r
y
d
ir
ec
t
s
y
s
tem
im
p
licatio
n
s
.
I
n
d
o
B
E
R
T
is
a
p
r
ac
tical
b
a
ck
b
o
n
e
f
o
r
I
n
d
o
n
esian
AQG
o
n
m
id
-
r
a
n
g
e
GPUs
(
≈
5
GB
VR
AM
)
,
en
ab
lin
g
f
ast
r
etr
ain
in
g
(
≈
1
h
/ep
o
ch
)
an
d
h
ig
h
th
r
o
u
g
h
p
u
t
(
5
,
2
0
0
t
o
k
en
s
/s
)
.
I
f
m
u
ltil
in
g
u
al
s
u
p
p
o
r
t
is
r
e
q
u
i
r
ed
,
m
B
E
R
T
is
a
r
ea
s
o
n
ab
le
f
allb
ac
k
at
a
h
ig
h
er
co
s
t,
wh
er
ea
s
B
E
R
T
-
lar
g
e
is
u
n
lik
ely
to
b
e
ju
s
tifie
d
with
o
u
t
s
u
b
s
tan
tially
m
o
r
e
I
n
d
o
n
esian
d
ata
an
d
c
o
m
p
u
te.
3
.
5
.
L
im
it
a
t
io
ns
a
nd
f
uture
wo
rk
s
Ou
r
ev
alu
atio
n
r
elies
o
n
au
t
o
m
atic
o
v
er
lap
m
etr
ics
(
B
L
E
U
-
4
,
ME
T
E
OR
,
an
d
R
OUGE
-
L
)
.
Ho
wev
er
,
th
ese
s
co
r
es
co
r
r
elate
o
n
ly
im
p
er
f
ec
tly
with
p
ed
a
g
o
g
ical
u
s
ef
u
ln
ess
an
d
p
er
ce
iv
e
d
q
u
esti
o
n
q
u
ality
,
a
n
d
we
d
id
n
o
t
in
clu
d
e
h
u
m
an
-
r
ater
s
tu
d
ies
with
teac
h
er
s
o
r
s
u
b
ject
-
m
atter
ex
p
er
ts
.
T
h
e
I
n
d
o
n
esian
SQu
AD
2
.
0
p
o
r
tio
n
u
s
es
a
2
0
,
0
0
0
tr
an
s
lated
s
u
b
s
et,
wh
ich
m
ay
co
n
tain
tr
an
s
latio
n
ar
tifa
cts
an
d
th
er
ef
o
r
e
d
o
es
n
o
t
f
u
lly
m
ir
r
o
r
class
r
o
o
m
d
is
co
u
r
s
e.
At
th
e
s
am
e
tim
e
,
T
y
DiQA
-
Go
ld
P
f
o
c
u
s
es
o
n
in
f
o
r
m
atio
n
-
s
ee
k
in
g
q
u
esti
o
n
s
r
ath
er
t
h
an
cu
r
r
ic
u
lu
m
-
alig
n
e
d
m
ater
ials
.
W
o
r
k
in
g
o
n
a
s
in
g
l
e
1
6
GB
GPU,
we
u
s
ed
s
m
alle
r
ef
f
ec
tiv
e
b
atc
h
es
f
o
r
B
E
R
T
-
lar
g
e
an
d
c
o
u
ld
n
o
t
ex
p
lo
r
e
a
v
ast
h
y
p
er
p
ar
am
ete
r
s
p
ac
e;
ev
en
th
o
u
g
h
we
m
atc
h
ed
th
e
n
u
m
b
er
o
f
o
p
tim
izatio
n
s
tep
s
ac
r
o
s
s
m
o
d
els
an
d
r
a
n
a
r
e
-
s
am
p
lin
g
r
o
b
u
s
tn
ess
ch
ec
k
o
n
SQu
AD
2
.
0
,
s
o
m
e
co
n
f
o
u
n
d
i
n
g
f
ac
to
r
s
m
ay
p
er
s
is
t.
W
e
also
lim
it
o
u
r
co
m
p
ar
is
o
n
t
o
en
co
d
er
-
o
n
ly
t
r
an
s
f
o
r
m
e
r
s
an
d
d
o
n
o
t
b
e
n
ch
m
ar
k
d
ec
o
d
er
o
r
in
s
tr
u
ctio
n
-
tu
n
e
d
lar
g
e
lan
g
u
ag
e
m
o
d
els.
I
n
f
u
tu
r
e
wo
r
k
,
we
p
la
n
to
ad
d
e
x
p
er
t
a
n
d
teac
h
er
r
atin
g
s
,
cu
r
r
ic
u
lu
m
-
s
p
ec
if
ic
I
n
d
o
n
esian
d
atasets
,
an
d
a
b
r
o
a
d
er
s
et
o
f
m
o
d
el
f
am
ilies
.
4.
CO
NCLU
SI
O
N
T
h
is
s
t
u
d
y
ev
al
u
at
e
d
I
n
d
o
B
E
R
T
,
m
B
E
R
T
,
a
n
d
B
E
R
T
-
l
ar
g
e
f
o
r
I
n
d
o
n
esia
n
AQ
G
an
d
f
o
u
n
d
t
h
at
I
n
d
o
B
E
R
T
c
o
n
s
is
t
e
n
tl
y
p
r
o
d
u
ce
d
t
h
e
h
ig
h
est
-
q
u
a
lit
y
q
u
est
io
n
s
w
h
i
le
u
s
i
n
g
r
es
o
u
r
ce
s
m
o
s
t
ef
f
ic
ie
n
tl
y
.
m
B
E
R
T
r
e
m
ai
n
s
a
v
ia
b
l
e
m
u
l
tili
n
g
u
al
alte
r
n
at
iv
e,
al
b
e
it
w
it
h
a
m
o
d
e
s
t
ac
cu
r
a
c
y
t
r
a
d
e
-
o
f
f
,
w
h
e
r
e
as
B
E
R
T
-
la
r
g
e
o
f
f
er
s
n
o
cl
ea
r
a
d
v
a
n
t
ag
e
i
n
t
h
is
s
e
tti
n
g
a
n
d
r
e
q
u
i
r
es
a
g
r
ea
t
er
c
o
m
p
u
t
ati
o
n
al
b
u
d
g
et
.
W
e
t
h
e
r
e
f
o
r
e
r
ec
o
m
m
en
d
I
n
d
o
B
E
R
T
as
th
e
d
ef
au
lt
b
a
ck
b
o
n
e
f
o
r
I
n
d
o
n
es
ia
n
AQG
i
n
e
d
u
ca
t
io
n
al
a
p
p
lic
ati
o
n
s
an
d
d
e
p
l
o
y
m
e
n
ts
o
n
m
i
d
-
r
a
n
g
e
GPUs
.
Fo
r
d
e
v
el
o
p
e
r
s
a
n
d
e
d
u
c
at
o
r
s
,
a
p
r
a
cti
ca
l
a
p
p
r
o
ac
h
is
to
f
i
n
e
-
t
u
n
e
I
n
d
o
B
E
R
T
w
it
h
[
H
L
]
tag
g
i
n
g
at
a
le
ar
n
i
n
g
r
a
te
o
f
5
×1
0
-
5
(
b
at
ch
s
iz
e
1
6
,
o
r
8
f
o
r
lar
g
er
m
o
d
els
)
a
n
d
u
til
iz
e
FP
1
6
/m
ix
e
d
p
r
ec
is
i
o
n
wit
h
e
a
r
l
y
s
t
o
p
p
in
g
.
W
e
r
ec
o
m
m
e
n
d
e
v
al
u
a
ti
n
g
s
y
s
te
m
s
u
s
i
n
g
B
L
E
U,
ME
T
E
OR
,
a
n
d
R
OU
GE
-
L
.
F
o
r
d
e
p
l
o
y
m
en
t,
e
n
a
b
le
b
atc
h
i
n
g
/
c
ac
h
i
n
g
a
n
d
c
o
n
s
id
e
r
FP
1
6
o
r
8
-
b
it i
n
f
e
r
e
n
c
e
t
o
m
e
et
t
y
p
ic
al
l
ate
n
c
y
an
d
m
e
m
o
r
y
b
u
d
g
e
ts
o
n
m
i
d
-
r
an
g
e
(
~
5
GB
VR
AM
)
GPUs
.
ACK
NO
WL
E
DG
M
E
N
T
S
W
e
u
s
ed
AI
to
o
ls
s
o
lely
to
im
p
r
o
v
e
th
e
clar
ity
,
g
r
am
m
ar
,
an
d
co
n
s
is
ten
cy
o
f
th
e
m
an
u
s
cr
ip
t
tex
t.
All
tech
n
ical
co
n
ten
t,
a
n
aly
s
es,
f
i
g
u
r
es,
a
n
d
co
n
clu
s
io
n
s
wer
e
p
r
o
d
u
ce
d
a
n
d
v
er
if
ied
b
y
t
h
e
au
th
o
r
s
.
No
d
ata
wer
e
g
en
er
ated
,
m
o
d
if
ie
d
,
o
r
s
y
n
th
e
s
ized
u
s
in
g
AI
to
o
ls
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2252
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
15
,
No
.
2
,
Ap
r
il
20
26
:
1
8
0
4
-
1
8
1
3
1810
F
UNDING
I
NF
O
R
M
A
T
I
O
N
T
h
is
r
esear
ch
was
s
u
p
p
o
r
te
d
b
y
Un
iv
er
s
itas
Neg
er
i
Y
o
g
y
ak
a
r
ta
(
UNY)
u
n
d
er
C
o
n
tr
ac
t
No
.
T
/1
0
.
1
2
/UN3
4
.
1
5
/PT.
0
1
.
0
2
/2
0
2
3
.
T
h
e
f
u
n
d
e
r
h
a
d
n
o
r
o
l
e
in
s
tu
d
y
d
esig
n
,
d
ata
co
ll
ec
tio
n
an
d
an
aly
s
is
,
d
ec
is
io
n
to
p
u
b
lis
h
,
o
r
p
r
ep
a
r
a
tio
n
o
f
th
e
m
an
u
s
cr
ip
t.
AUTHO
R
CO
NT
RI
B
UT
I
O
NS ST
A
T
E
M
E
N
T
T
h
is
jo
u
r
n
al
u
s
es
th
e
C
o
n
tr
ib
u
to
r
R
o
les
T
ax
o
n
o
m
y
(
C
R
ed
iT)
to
r
ec
o
g
n
ize
in
d
iv
id
u
al
au
th
o
r
co
n
tr
ib
u
tio
n
s
,
r
ed
u
ce
au
th
o
r
s
h
ip
d
is
p
u
tes,
an
d
f
ac
ilit
ate
co
llab
o
r
atio
n
.
Na
m
e
o
f
Aut
ho
r
C
M
So
Va
Fo
I
R
D
O
E
Vi
Su
P
Fu
Han
d
ar
u
J
ati
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
Yu
n
iar
I
n
d
r
ih
ap
s
ar
i
✓
✓
✓
✓
✓
✓
✓
✓
Pra
d
an
a
Setialan
a
✓
✓
✓
✓
✓
✓
✓
Dan
an
g
W
ijay
a
✓
✓
✓
✓
✓
✓
✓
Saty
a
Ad
h
iy
ak
s
a
Ar
d
y
✓
✓
✓
✓
✓
✓
✓
Dh
is
ta
Dwi
Nu
r
Ar
d
ian
s
y
ah
✓
✓
✓
✓
✓
✓
✓
C
:
C
o
n
c
e
p
t
u
a
l
i
z
a
t
i
o
n
M
:
M
e
t
h
o
d
o
l
o
g
y
So
:
So
f
t
w
a
r
e
Va
:
Va
l
i
d
a
t
i
o
n
Fo
:
Fo
r
mal
a
n
a
l
y
s
i
s
I
:
I
n
v
e
s
t
i
g
a
t
i
o
n
R
:
R
e
so
u
r
c
e
s
D
:
D
a
t
a
C
u
r
a
t
i
o
n
O
:
W
r
i
t
i
n
g
-
O
r
i
g
i
n
a
l
D
r
a
f
t
E
:
W
r
i
t
i
n
g
-
R
e
v
i
e
w
&
E
d
i
t
i
n
g
Vi
:
Vi
su
a
l
i
z
a
t
i
o
n
Su
:
Su
p
e
r
v
i
s
i
o
n
P
:
P
r
o
j
e
c
t
a
d
mi
n
i
st
r
a
t
i
o
n
Fu
:
Fu
n
d
i
n
g
a
c
q
u
i
si
t
i
o
n
CO
NF
L
I
C
T
O
F
I
N
T
E
R
E
S
T
ST
A
T
E
M
E
NT
T
h
e
au
t
h
o
r
d
ec
lar
es
t
h
at
th
e
r
e
ar
e
n
o
k
n
o
wn
co
n
f
licts
o
f
in
ter
est
ass
o
ciate
d
with
th
is
p
u
b
licatio
n
.
T
h
er
e
ar
e
n
o
f
in
a
n
cial
o
r
p
e
r
s
o
n
al
r
elatio
n
s
h
ip
s
th
at
co
u
ld
in
ap
p
r
o
p
r
iately
in
f
l
u
en
ce
o
r
b
ias
th
e
co
n
ten
t
o
f
th
is
wo
r
k
.
I
NF
O
RM
E
D
CO
NS
E
N
T
No
t
ap
p
licab
le.
T
h
is
s
tu
d
y
d
id
n
o
t
in
v
o
l
v
e
h
u
m
an
p
ar
tici
p
an
ts
,
h
u
m
an
d
ata,
o
r
a
n
y
p
er
s
o
n
ally
id
en
tifia
b
le
in
f
o
r
m
atio
n
.
All
d
ata
u
s
ed
wer
e
eith
er
p
u
b
licl
y
av
ailab
le,
f
u
lly
an
o
n
y
m
ize
d
,
o
r
d
er
i
v
ed
f
r
o
m
n
o
n
‑
h
u
m
an
s
o
u
r
ce
s
,
an
d
th
er
e
f
o
r
e
n
o
in
f
o
r
m
ed
co
n
s
en
t w
as
r
eq
u
ir
ed
f
r
o
m
in
d
iv
i
d
u
als.
E
T
H
I
CAL AP
P
RO
V
AL
No
t
ap
p
licab
le.
T
h
is
r
esear
ch
d
id
n
o
t
in
v
o
lv
e
h
u
m
an
s
u
b
jects,
h
u
m
an
b
io
lo
g
ical
m
ater
ials
,
o
r
ex
p
er
im
en
tal
p
r
o
ce
d
u
r
es o
n
a
n
im
als.
DATA AV
AI
L
AB
I
L
I
T
Y
T
h
e
d
ata
th
at
s
u
p
p
o
r
t
th
e
f
in
d
in
g
s
o
f
t
h
is
s
tu
d
y
ar
e
o
p
en
ly
a
v
ailab
le
in
th
e
G
ith
u
b
.
co
m
at
h
ttp
s
://g
ith
u
b
.
co
m
/g
o
o
g
le
-
r
esear
ch
-
d
atasets
/ty
d
iq
a.
RE
F
E
R
E
NC
E
S
[
1
]
A
.
K
.
S
a
l
mo
n
a
n
d
M
.
X
.
B
a
r
r
e
r
a
,
“
I
n
t
e
n
t
i
o
n
a
l
q
u
e
st
i
o
n
i
n
g
t
o
p
r
o
m
o
t
e
t
h
i
n
k
i
n
g
a
n
d
l
e
a
r
n
i
n
g
,
”
T
h
i
n
k
i
n
g
S
k
i
l
l
s
a
n
d
C
re
a
t
i
v
i
t
y
,
v
o
l
.
4
0
,
Ju
n
.
2
0
2
1
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
t
sc.
2
0
2
1
.
1
0
0
8
2
2
.
[
2
]
R
.
G
.
W
.
M
u
e
l
l
e
r
,
“
Ex
a
mi
n
i
n
g
t
e
a
c
h
e
r
s’
d
e
v
e
l
o
p
me
n
t
a
n
d
i
m
p
l
e
me
n
t
a
t
i
o
n
o
f
c
o
mp
e
l
l
i
n
g
q
u
e
s
t
i
o
n
s,
”
S
o
c
i
a
l
S
t
u
d
i
e
s
R
e
se
a
rc
h
a
n
d
Pra
c
t
i
c
e
,
v
o
l
.
1
3
,
n
o
.
1
,
p
p
.
1
–
1
5
,
M
a
y
2
0
1
8
,
d
o
i
:
1
0
.
1
1
0
8
/
S
S
R
P
-
08
-
2
0
1
7
-
0
0
4
2
.
[
3
]
S
.
S
a
a
r
i
n
e
n
,
S
.
K
r
i
s
h
n
a
m
u
r
t
h
i
,
K
.
F
i
s
l
e
r
,
a
n
d
P
.
T.
W
i
l
so
n
,
“
H
a
r
n
e
ssi
n
g
t
h
e
w
i
sd
o
m
o
f
t
h
e
c
l
a
sses
,
”
i
n
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
5
0
t
h
AC
M
T
e
c
h
n
i
c
a
l
S
y
m
p
o
s
i
u
m
o
n
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
E
d
u
c
a
t
i
o
n
,
F
e
b
.
2
0
1
9
,
p
p
.
6
0
6
–
6
1
2
,
d
o
i
:
1
0
.
1
1
4
5
/
3
2
8
7
3
2
4
.
3
2
8
7
5
0
4
.
[
4
]
T.
To
f
a
d
e
,
J.
E
l
sn
e
r
,
a
n
d
S
.
T
.
H
a
i
n
e
s,
“
B
e
st
p
r
a
c
t
i
c
e
st
r
a
t
e
g
i
e
s
f
o
r
e
f
f
e
c
t
i
v
e
u
se
o
f
q
u
e
s
t
i
o
n
s
a
s
a
t
e
a
c
h
i
n
g
t
o
o
l
,
”
Am
e
r
i
c
a
n
J
o
u
rn
a
l
o
f
Ph
a
rm
a
c
e
u
t
i
c
a
l
E
d
u
c
a
t
i
o
n
,
v
o
l
.
7
7
,
n
o
.
7
,
S
e
p
.
2
0
1
3
,
d
o
i
:
1
0
.
5
6
8
8
/
a
j
p
e
7
7
7
1
5
5
.
[
5
]
C
.
C
h
i
n
a
n
d
D
.
E.
B
r
o
w
n
,
“
S
t
u
d
e
n
t
-
g
e
n
e
r
a
t
e
d
q
u
e
s
t
i
o
n
s:
a
m
e
a
n
i
n
g
f
u
l
a
s
p
e
c
t
o
f
l
e
a
r
n
i
n
g
i
n
sc
i
e
n
c
e
,
”
I
n
t
e
rn
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
S
c
i
e
n
c
e
E
d
u
c
a
t
i
o
n
,
v
o
l
.
2
4
,
n
o
.
5
,
p
p
.
5
2
1
–
5
4
9
,
M
a
y
2
0
0
2
,
d
o
i
:
1
0
.
1
0
8
0
/
0
9
5
0
0
6
9
0
1
1
0
0
9
5
2
4
9
.
[
6
]
M
.
G
o
t
t
l
i
e
b
,
J.
B
a
i
l
i
t
z
,
M
.
F
i
x
,
E.
S
h
a
p
p
e
l
l
,
a
n
d
M
.
J.
W
a
g
n
e
r
,
“
E
d
u
c
a
t
o
r
’
s
b
l
u
e
p
r
i
n
t
:
a
h
o
w
‐
t
o
g
u
i
d
e
f
o
r
d
e
v
e
l
o
p
i
n
g
h
i
g
h
‐
q
u
a
l
i
t
y
mu
l
t
i
p
l
e
‐
c
h
o
i
c
e
q
u
e
st
i
o
n
s,
”
A
EM
Ed
u
c
a
t
i
o
n
a
n
d
T
r
a
i
n
i
n
g
,
v
o
l
.
7
,
n
o
.
1
,
F
e
b
.
2
0
2
3
,
d
o
i
:
1
0
.
1
0
0
2
/
a
e
t
2
.
1
0
8
3
6
.
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
I
n
d
o
B
E
R
T fo
r
ed
u
ca
tio
n
a
l a
s
s
ess
men
t:
co
mp
a
r
a
tive
a
n
a
lysi
s
o
f tra
n
s
fo
r
mer m
o
d
els
… (
Ha
n
d
a
r
u
Ja
ti
)
1811
[
7
]
S
.
S
o
n
i
,
P
.
K
u
mar
,
a
n
d
A
.
S
a
h
a
,
“
A
u
t
o
m
a
t
i
c
q
u
e
s
t
i
o
n
g
e
n
e
r
a
t
i
o
n
:
a
sy
st
e
ma
t
i
c
r
e
v
i
e
w
,
”
S
S
RN
E
l
e
c
t
r
o
n
i
c
J
o
u
rn
a
l
,
2
0
1
9
,
d
o
i
:
1
0
.
2
1
3
9
/
s
sr
n
.
3
4
0
3
9
2
6
.
[
8
]
K
.
P
.
K
u
m
a
r
e
t
a
l
.
,
“
A
u
t
o
m
a
t
e
d
q
u
e
st
i
o
n
p
a
p
e
r
g
e
n
e
r
a
t
o
r
u
s
i
n
g
LL
M
,
”
I
n
t
e
rn
a
t
i
o
n
a
l
J
o
u
rn
a
l
o
f
R
e
se
a
rc
h
a
n
d
I
n
n
o
v
a
t
i
o
n
i
n
Ap
p
l
i
e
d
S
c
i
e
n
c
e
,
v
o
l
.
X
,
n
o
.
I
V
,
p
p
.
2
6
6
–
2
7
5
,
2
0
2
5
,
d
o
i
:
1
0
.
5
1
5
8
4
/
I
JR
I
A
S
.
2
0
2
5
.
1
0
0
4
0
0
2
0
.
[
9
]
J.
H
.
W
o
l
f
e
,
“
A
u
t
o
m
a
t
i
c
q
u
e
s
t
i
o
n
g
e
n
e
r
a
t
i
o
n
f
r
o
m
t
e
x
t
-
a
n
a
i
d
t
o
i
n
d
e
p
e
n
d
e
n
t
s
t
u
d
y
,
”
i
n
P
ro
c
e
e
d
i
n
g
s
o
f
t
h
e
AC
M
S
I
G
C
S
E
-
S
I
G
C
U
E
t
e
c
h
n
i
c
a
l
s
y
m
p
o
si
u
m
o
n
C
o
m
p
u
t
e
r
sci
e
n
c
e
a
n
d
e
d
u
c
a
t
i
o
n
,
1
9
7
6
,
p
p
.
1
0
4
–
1
1
2
,
d
o
i
:
1
0
.
1
1
4
5
/
8
0
0
1
0
7
.
8
0
3
4
5
9
.
[
1
0
]
P
.
Ta
n
w
a
r
,
K
.
B
a
n
s
a
l
,
A
.
S
h
a
r
ma
,
a
n
d
N
.
D
a
g
a
r
,
“
A
I
b
a
se
d
c
h
a
t
b
o
t
f
o
r
h
e
a
l
t
h
c
a
r
e
u
s
i
n
g
ma
c
h
i
n
e
l
e
a
r
n
i
n
g
,
”
i
n
2
0
2
3
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
Ad
v
a
n
c
e
s
i
n
C
o
m
p
u
t
a
t
i
o
n
,
C
o
m
m
u
n
i
c
a
t
i
o
n
a
n
d
I
n
f
o
rm
a
t
i
o
n
T
e
c
h
n
o
l
o
g
y
(
I
C
AI
C
C
I
T
)
,
N
o
v
.
2
0
2
3
,
p
p
.
7
5
1
–
7
5
5
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
A
I
C
C
I
T6
0
2
5
5
.
2
0
2
3
.
1
0
4
6
5
7
2
8
.
[
1
1
]
M
.
F
l
o
r
a
n
d
B
.
R
i
o
r
d
a
n
,
“
A
sem
a
n
t
i
c
r
o
l
e
-
b
a
se
d
a
p
p
r
o
a
c
h
t
o
o
p
e
n
-
d
o
m
a
i
n
a
u
t
o
ma
t
i
c
q
u
e
st
i
o
n
g
e
n
e
r
a
t
i
o
n
,
”
i
n
P
ro
c
e
e
d
i
n
g
s
o
f
t
h
e
T
h
i
r
t
e
e
n
t
h
W
o
rks
h
o
p
o
n
I
n
n
o
v
a
t
i
v
e
U
se
o
f
N
L
P
f
o
r
Bu
i
l
d
i
n
g
Ed
u
c
a
t
i
o
n
a
l
A
p
p
l
i
c
a
t
i
o
n
s
,
2
0
1
8
,
p
p
.
2
5
4
–
2
6
3
,
d
o
i
:
1
0
.
1
8
6
5
3
/
v
1
/
W
1
8
-
0
5
3
0
.
[
1
2
]
A
.
G
a
s
h
k
o
v
a
n
d
M
.
E
l
t
s
o
v
a
,
“
A
u
t
o
mat
i
z
a
t
i
o
n
o
f
a
n
sw
e
r
s
t
o
q
u
e
s
t
i
o
n
s
b
y
m
a
t
c
h
i
n
g
sy
n
t
a
c
t
i
c
g
r
a
p
h
s
,
”
i
n
Pro
c
e
e
d
i
n
g
s
o
f
t
h
e
X
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
“
W
o
rd
,
U
t
t
e
r
a
n
c
e
,
T
e
x
t
:
C
o
g
n
i
t
i
v
e
,
P
ra
g
m
a
t
i
c
a
n
d
C
u
l
t
u
r
a
l
As
p
e
c
t
s”
(
WUT
2
0
2
0
)
,
A
u
g
.
2
0
2
0
,
p
p
.
4
1
9
–
4
2
5
,
d
o
i
:
1
0
.
1
5
4
0
5
/
e
p
sb
s
.
2
0
2
0
.
0
8
.
4
9
.
[
1
3
]
E.
S
n
e
i
d
e
r
s,
“
A
u
t
o
mat
e
d
q
u
e
s
t
i
o
n
a
n
sw
e
r
i
n
g
u
s
i
n
g
q
u
e
st
i
o
n
t
e
m
p
l
a
t
e
s
t
h
a
t
c
o
v
e
r
t
h
e
c
o
n
c
e
p
t
u
a
l
m
o
d
e
l
o
f
t
h
e
d
a
t
a
b
a
s
e
,
”
i
n
N
a
t
u
r
a
l
L
a
n
g
u
a
g
e
Pro
c
e
ss
i
n
g
a
n
d
I
n
f
o
rm
a
t
i
o
n
S
y
st
e
m
s (NLD
B
2
0
0
2
)
,
2
0
0
2
,
p
p
.
2
3
5
–
2
3
9
,
d
o
i
:
1
0
.
1
0
0
7
/
3
-
5
4
0
-
3
6
2
7
1
-
1
_
2
4
.
[
1
4
]
G
.
K
u
r
d
i
,
J.
L
e
o
,
B
.
P
a
r
si
a
,
U
.
S
a
t
t
l
e
r
,
a
n
d
S
.
A
l
-
E
mari
,
“
A
sy
s
t
e
mat
i
c
r
e
v
i
e
w
o
f
a
u
t
o
m
a
t
i
c
q
u
e
st
i
o
n
g
e
n
e
r
a
t
i
o
n
f
o
r
e
d
u
c
a
t
i
o
n
a
l
p
u
r
p
o
ses,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
Art
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
i
n
Ed
u
c
a
t
i
o
n
,
v
o
l
.
3
0
,
n
o
.
1
,
p
p
.
1
2
1
–
2
0
4
,
M
a
r
.
2
0
2
0
,
d
o
i
:
1
0
.
1
0
0
7
/
s
4
0
5
9
3
-
019
-
0
0
1
8
6
-
y.
[
1
5
]
M
.
La
s
t
a
n
d
G
.
D
a
n
o
n
,
“
A
u
t
o
ma
t
i
c
q
u
e
st
i
o
n
g
e
n
e
r
a
t
i
o
n
,
”
WIREs
D
a
t
a
M
i
n
i
n
g
a
n
d
K
n
o
w
l
e
d
g
e
D
i
s
c
o
v
e
r
y
,
v
o
l
.
1
0
,
n
o
.
6
,
N
o
v
.
2
0
2
0
,
d
o
i
:
1
0
.
1
0
0
2
/
w
i
d
m.
1
3
8
2
.
[
1
6
]
M
.
Ess
a
m,
M
.
A
.
D
e
i
f
,
a
n
d
R
.
E
l
g
o
h
a
r
y
,
“
D
e
c
i
p
h
e
r
i
n
g
A
r
a
b
i
c
q
u
e
st
i
o
n
:
a
d
e
d
i
c
a
t
e
d
su
r
v
e
y
o
n
A
r
a
b
i
c
q
u
e
s
t
i
o
n
a
n
a
l
y
s
i
s
me
t
h
o
d
s,
c
h
a
l
l
e
n
g
e
s,
l
i
mi
t
a
t
i
o
n
s
a
n
d
f
u
t
u
r
e
p
a
t
h
w
a
y
s,”
Ar
t
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
Re
v
i
e
w
,
v
o
l
.
5
7
,
n
o
.
9
,
A
u
g
.
2
0
2
4
,
d
o
i
:
1
0
.
1
0
0
7
/
s
1
0
4
6
2
-
0
2
4
-
1
0
8
8
0
-
6.
[
1
7
]
I
.
E
.
F
a
t
t
o
h
,
A
.
E
.
A
b
o
u
t
a
b
l
,
a
n
d
M
.
H
.
H
a
g
g
a
g
,
“
T
a
p
p
i
n
g
i
n
t
o
t
h
e
p
o
w
e
r
o
f
a
u
t
o
ma
t
i
c
q
u
e
st
i
o
n
g
e
n
e
r
a
t
i
o
n
,
”
I
n
t
e
rn
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
C
o
m
p
u
t
e
r A
p
p
l
i
c
a
t
i
o
n
s
,
v
o
l
.
1
0
3
,
n
o
.
1
,
p
p
.
1
–
6
,
O
c
t
.
2
0
1
4
,
d
o
i
:
1
0
.
5
1
2
0
/
1
8
0
3
5
-
7
6
3
6
.
[
1
8
]
A
.
A
sa
d
i
a
n
d
R
.
S
a
f
a
b
a
k
h
s
h
,
“
T
h
e
e
n
c
o
d
e
r
-
d
e
c
o
d
e
r
f
r
a
m
e
w
o
r
k
a
n
d
i
t
s
a
p
p
l
i
c
a
t
i
o
n
s
,
”
i
n
D
e
e
p
L
e
a
r
n
i
n
g
:
C
o
n
c
e
p
t
s
a
n
d
Arc
h
i
t
e
c
t
u
r
e
s
,
S
p
r
i
n
g
e
r
,
C
h
a
m
,
2
0
2
0
,
p
p
.
1
3
3
–
1
6
7
.
d
o
i
:
1
0
.
1
0
0
7
/
9
7
8
-
3
-
0
3
0
-
3
1
7
5
6
-
0
_
5
.
[
1
9
]
C
.
Es
c
o
l
a
n
o
,
M
.
R
.
C
.
-
j
u
ss
à
,
a
n
d
J
.
A
.
R
.
F
o
n
o
l
l
o
s
a
,
“
M
u
l
t
i
l
i
n
g
u
a
l
ma
c
h
i
n
e
t
r
a
n
s
l
a
t
i
o
n
:
d
e
e
p
a
n
a
l
y
s
i
s
o
f
l
a
n
g
u
a
g
e
-
sp
e
c
i
f
i
c
e
n
c
o
d
e
r
-
d
e
c
o
d
e
r
s,
”
J
o
u
rn
a
l
o
f
Art
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
R
e
se
a
rc
h
,
v
o
l
.
7
3
,
p
p
.
1
5
3
5
–
1
5
5
2
,
A
p
r
.
2
0
2
2
,
d
o
i
:
1
0
.
1
6
1
3
/
j
a
i
r
.
1
.
1
2
6
9
9
.
[
2
0
]
J.
Y
u
,
S
.
W
a
n
g
,
a
n
d
J.
Y
i
n
,
“
A
d
a
p
t
i
v
e
c
r
o
ss
-
l
i
n
g
u
a
l
q
u
e
s
t
i
o
n
g
e
n
e
r
a
t
i
o
n
w
i
t
h
mi
n
i
m
a
l
r
e
so
u
r
c
e
s,
”
T
h
e
C
o
m
p
u
t
e
r
J
o
u
r
n
a
l
,
v
o
l
.
6
4
,
n
o
.
7
,
p
p
.
1
0
5
6
–
1
0
6
8
,
A
u
g
.
2
0
2
1
,
d
o
i
:
1
0
.
1
0
9
3
/
c
o
m
j
n
l
/
b
x
a
b
1
0
6
.
[
2
1
]
D
.
Lu
i
t
e
l
,
S
.
H
a
ssa
n
i
,
a
n
d
M
.
S
a
b
e
t
z
a
d
e
h
,
“
I
mp
r
o
v
i
n
g
r
e
q
u
i
r
e
me
n
t
s
c
o
mp
l
e
t
e
n
e
ss:
a
u
t
o
ma
t
e
d
a
ss
i
st
a
n
c
e
t
h
r
o
u
g
h
l
a
r
g
e
l
a
n
g
u
a
g
e
mo
d
e
l
s
,
”
Re
q
u
i
rem
e
n
t
s E
n
g
i
n
e
e
ri
n
g
,
v
o
l
.
2
9
,
n
o
.
1
,
p
p
.
7
3
–
9
5
,
M
a
r
.
2
0
2
4
,
d
o
i
:
1
0
.
1
0
0
7
/
s
0
0
7
6
6
-
024
-
0
0
4
1
6
-
3.
[
2
2
]
Y
.
Zh
a
o
,
R
.
C
a
o
,
J.
B
a
i
,
W
.
M
a
,
a
n
d
H
.
S
h
i
n
n
o
u
,
“
D
e
t
e
r
m
i
n
i
n
g
t
h
e
l
o
g
i
c
a
l
r
e
l
a
t
i
o
n
b
e
t
w
e
e
n
t
w
o
s
e
n
t
e
n
c
e
s
b
y
u
si
n
g
t
h
e
mas
k
e
d
l
a
n
g
u
a
g
e
m
o
d
e
l
o
f
B
E
R
T
,
”
i
n
2
0
2
0
I
n
t
e
rn
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
T
e
c
h
n
o
l
o
g
i
e
s
a
n
d
A
p
p
l
i
c
a
t
i
o
n
s
o
f
Art
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
(
T
AAI)
,
D
e
c
.
2
0
2
0
,
p
p
.
2
2
8
–
2
3
1
,
d
o
i
:
1
0
.
1
1
0
9
/
TA
A
I
5
1
4
1
0
.
2
0
2
0
.
0
0
0
4
9
.
[
2
3
]
K
.
M
.
T
a
r
w
a
n
i
a
n
d
S
.
E
d
e
m,
“
S
u
r
v
e
y
o
n
r
e
c
u
r
r
e
n
t
n
e
u
r
a
l
n
e
t
w
o
r
k
i
n
n
a
t
u
r
a
l
l
a
n
g
u
a
g
e
p
r
o
c
e
ss
i
n
g
,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
En
g
i
n
e
e
ri
n
g
T
re
n
d
s
a
n
d
T
e
c
h
n
o
l
o
g
y
,
v
o
l
.
4
8
,
n
o
.
6
,
p
p
.
3
0
1
–
3
0
4
,
J
u
n
.
2
0
1
7
,
d
o
i
:
1
0
.
1
4
4
4
5
/
2
2
3
1
5
3
8
1
/
I
JETT
-
V
4
8
P
2
5
3
.
[
2
4
]
A
.
G
r
a
v
e
s,
“
L
o
n
g
sh
o
r
t
-
t
e
r
m
mem
o
r
y
,
”
i
n
S
u
p
e
r
v
i
s
e
d
S
e
q
u
e
n
c
e
L
a
b
e
l
l
i
n
g
w
i
t
h
Re
c
u
rr
e
n
t
N
e
u
r
a
l
N
e
t
w
o
rks
,
S
p
r
i
n
g
e
r
,
B
e
r
l
i
n
,
H
e
i
d
e
l
b
e
r
g
,
2
0
1
2
,
p
p
.
3
7
–
45
,
d
o
i
:
1
0
.
1
0
0
7
/
9
7
8
-
3
-
6
4
2
-
2
4
7
9
7
-
2
_
4
.
[
2
5
]
N
.
Z
u
c
c
h
e
t
a
n
d
A
.
O
r
v
i
e
t
o
,
“
R
e
c
u
r
r
e
n
t
n
e
u
r
a
l
n
e
t
w
o
r
k
s:
v
a
n
i
sh
i
n
g
a
n
d
e
x
p
l
o
d
i
n
g
g
r
a
d
i
e
n
t
s
a
r
e
n
o
t
t
h
e
e
n
d
o
f
t
h
e
s
t
o
r
y
,
”
i
n
Pro
c
e
e
d
i
n
g
s
o
f
t
h
e
3
8
t
h
I
n
t
e
rn
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
N
e
u
r
a
l
I
n
f
o
rm
a
t
i
o
n
Pr
o
c
e
ssi
n
g
S
y
st
e
m
,
2
0
2
4
,
p
p
.
1
3
9
4
0
2
–
1
3
9
4
4
3
.
[
2
6
]
P
.
N
e
r
e
l
l
a
,
D
.
P
i
t
t
u
,
S
.
U
n
d
r
a
k
o
n
d
a
,
S
.
C
h
e
n
n
a
ms
e
t
t
y
,
V
.
P
.
K
u
m
a
r
S
,
a
n
d
V
.
K
.
K
i
s
h
o
r
e
K
,
“
A
n
e
f
f
i
c
i
e
n
t
S
e
q
2
S
e
q
m
o
d
e
l
t
o
p
r
e
d
i
c
t
q
u
e
st
i
o
n
a
n
d
a
n
sw
e
r
r
e
sp
o
n
se
sy
st
e
m,
”
i
n
2
0
2
4
S
e
c
o
n
d
I
n
t
e
rn
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
Ad
v
a
n
c
e
s
i
n
I
n
f
o
rm
a
t
i
o
n
T
e
c
h
n
o
l
o
g
y
(
I
C
AI
T
)
,
Ju
l
.
2
0
2
4
,
p
p
.
1
–
6
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
A
I
T6
1
6
3
8
.
2
0
2
4
.
1
0
6
9
0
3
4
3
.
[
2
7
]
Y
.
S
u
n
,
“
Th
e
e
v
o
l
u
t
i
o
n
o
f
t
r
a
n
sf
o
r
mer
mo
d
e
l
s
f
r
o
m
u
n
i
d
i
r
e
c
t
i
o
n
a
l
t
o
b
i
d
i
r
e
c
t
i
o
n
a
l
i
n
n
a
t
u
r
a
l
l
a
n
g
u
a
g
e
p
r
o
c
e
s
si
n
g
,
”
A
p
p
l
i
e
d
a
n
d
C
o
m
p
u
t
a
t
i
o
n
a
l
En
g
i
n
e
e
ri
n
g
,
v
o
l
.
4
2
,
n
o
.
1
,
p
p
.
2
8
1
–
2
8
9
,
F
e
b
.
2
0
2
4
,
d
o
i
:
1
0
.
5
4
2
5
4
/
2
7
5
5
-
2
7
2
1
/
4
2
/
2
0
2
3
0
7
9
4
.
[
2
8
]
N
.
P
a
t
w
a
r
d
h
a
n
,
S
.
M
a
r
r
o
n
e
,
a
n
d
C
.
S
a
n
s
o
n
e
,
“
Tr
a
n
sf
o
r
m
e
r
s
i
n
t
h
e
r
e
a
l
w
o
r
l
d
:
a
s
u
r
v
e
y
o
n
N
LP
a
p
p
l
i
c
a
t
i
o
n
s,”
I
n
f
o
rm
a
t
i
o
n
,
v
o
l
.
1
4
,
n
o
.
4
,
A
p
r
.
2
0
2
3
,
d
o
i
:
1
0
.
3
3
9
0
/
i
n
f
o
1
4
0
4
0
2
4
2
.
[
2
9
]
M
.
A
r
sa
l
a
n
,
“
Tr
a
n
sf
o
r
mers
i
n
n
a
t
u
r
a
l
l
a
n
g
u
a
g
e
p
r
o
c
e
ssi
n
g
:
a
c
o
m
p
r
e
h
e
n
si
v
e
r
e
v
i
e
w
,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
f
o
r
Re
s
e
a
r
c
h
i
n
Ap
p
l
i
e
d
S
c
i
e
n
c
e
a
n
d
E
n
g
i
n
e
e
ri
n
g
T
e
c
h
n
o
l
o
g
y
,
v
o
l
.
1
2
,
n
o
.
5
,
p
p
.
5
5
9
1
–
5
5
9
7
,
M
a
y
2
0
2
4
,
d
o
i
:
1
0
.
2
2
2
1
4
/
i
j
r
a
se
t
.
2
0
2
4
.
6
2
8
6
3
.
[
3
0
]
J.
P
a
r
k
,
M
.
R
.
B
a
b
a
e
i
,
S
.
A
.
M
u
n
o
z
,
A
.
N
.
V
e
n
k
a
t
,
a
n
d
J.
D
.
H
e
d
e
n
g
r
e
n
,
“
S
i
mu
l
t
a
n
e
o
u
s
mu
l
t
i
st
e
p
t
r
a
n
sf
o
r
mer
a
r
c
h
i
t
e
c
t
u
r
e
f
o
r
mo
d
e
l
p
r
e
d
i
c
t
i
v
e
c
o
n
t
r
o
l
,
”
C
o
m
p
u
t
e
rs
& C
h
e
m
i
c
a
l
E
n
g
i
n
e
e
r
i
n
g
,
v
o
l
.
1
7
8
,
O
c
t
.
2
0
2
3
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
c
o
m
p
c
h
e
me
n
g
.
2
0
2
3
.
1
0
8
3
9
6
.
[
3
1
]
M
.
Za
h
e
e
r
e
t
a
l
.
,
“
B
i
g
b
i
r
d
:
t
r
a
n
sf
o
r
mers
f
o
r
l
o
n
g
e
r
se
q
u
e
n
c
e
s
,
”
i
n
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
3
4
t
h
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
N
e
u
r
a
l
I
n
f
o
rm
a
t
i
o
n
Pr
o
c
e
ssi
n
g
S
y
st
e
,
2
0
2
0
,
p
p
.
1
7
2
8
3
–
1
7
2
9
7
.
[
3
2
]
B
.
Z
h
u
a
n
g
,
J
.
Li
u
,
Z.
P
a
n
,
H
.
H
e
,
Y
.
W
e
n
g
,
a
n
d
C
.
S
h
e
n
,
“
A
su
r
v
e
y
o
n
e
f
f
i
c
i
e
n
t
t
r
a
i
n
i
n
g
o
f
t
r
a
n
sf
o
r
mers
,
”
i
n
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
T
h
i
r
t
y
-
S
e
c
o
n
d
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
i
n
t
C
o
n
f
e
re
n
c
e
o
n
Art
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
,
A
u
g
.
2
0
2
3
,
p
p
.
6
8
2
3
–
6
8
3
1
,
d
o
i
:
1
0
.
2
4
9
6
3
/
i
j
c
a
i
.
2
0
2
3
/
7
6
4
.
[
3
3
]
A
.
B
h
a
t
t
a
c
h
a
r
y
y
a
,
“
R
e
v
o
l
u
t
i
o
n
i
z
i
n
g
k
n
o
w
l
e
d
g
e
r
e
t
r
i
e
v
a
l
:
a
c
o
m
p
r
e
h
e
n
si
v
e
st
u
d
y
o
f
q
u
e
s
t
i
o
n
a
n
sw
e
r
i
n
g
sy
s
t
e
m,”
S
S
RN
El
e
c
t
r
o
n
i
c
J
o
u
rn
a
l
,
2
0
2
3
,
d
o
i
:
1
0
.
2
1
3
9
/
ssr
n
.
4
6
4
9
5
8
5
.
[
3
4
]
J.
D
e
v
l
i
n
,
M
.
-
W
.
C
h
a
n
g
,
K
.
L
e
e
,
a
n
d
K
.
T
o
u
t
a
n
o
v
a
,
“
B
ER
T
:
p
r
e
-
t
r
a
i
n
i
n
g
o
f
d
e
e
p
b
i
d
i
r
e
c
t
i
o
n
a
l
t
r
a
n
sf
o
r
m
e
r
s
f
o
r
l
a
n
g
u
a
g
e
u
n
d
e
r
s
t
a
n
d
i
n
g
,
”
i
n
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
2
0
1
9
C
o
n
f
e
r
e
n
c
e
o
f
t
h
e
N
o
rt
h
,
2
0
1
9
,
p
p
.
4
1
7
1
–
4
1
8
6
,
d
o
i
:
1
0
.
1
8
6
5
3
/
v
1
/
N
1
9
-
1
4
2
3
.
[
3
5
]
M
.
S
u
g
u
n
a
a
n
d
K
.
S
.
S
.
P
r
a
b
h
a
,
“
R
e
c
i
p
r
o
c
a
t
i
n
g
e
n
c
o
d
e
r
p
o
r
t
r
a
y
a
l
f
r
o
m
r
e
l
i
a
b
l
e
t
r
a
n
sf
o
r
m
e
r
d
e
p
e
n
d
e
n
t
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
mem
o
r
y
f
o
r
q
u
e
s
t
i
o
n
a
n
d
a
n
sw
e
r
i
n
g
t
e
x
t
c
l
a
ss
i
f
i
c
a
t
i
o
n
,
”
I
E
EE
Ac
c
e
ss
,
v
o
l
.
1
2
,
p
p
.
1
1
7
8
0
0
–
1
1
7
8
1
1
,
2
0
2
4
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
4
.
3
4
2
6
6
0
4
.
[
3
6
]
N
.
F
a
t
i
m
a
,
S
.
M
.
D
a
u
d
p
o
t
a
,
Z
.
K
a
st
r
a
t
i
,
A
.
S
.
I
mr
a
n
,
S
.
H
a
ssa
n
,
a
n
d
N
.
S
.
El
mi
t
w
a
l
l
y
,
“
I
mp
r
o
v
i
n
g
n
e
w
s
h
e
a
d
l
i
n
e
t
e
x
t
g
e
n
e
r
a
t
i
o
n
q
u
a
l
i
t
y
t
h
r
o
u
g
h
f
r
e
q
u
e
n
t
P
O
S
-
T
a
g
p
a
t
t
e
r
n
s
a
n
a
l
y
si
s
,
”
En
g
i
n
e
e
ri
n
g
A
p
p
l
i
c
a
t
i
o
n
s
o
f
Ar
t
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
,
v
o
l
.
1
2
5
,
O
c
t
.
2
0
2
3
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
e
n
g
a
p
p
a
i
.
2
0
2
3
.
1
0
6
7
1
8
.
[
3
7
]
A
.
K
a
r
a
k
,
K
.
K
u
n
a
l
,
N
.
D
a
r
a
p
a
n
e
n
i
,
a
n
d
A
.
R
.
P
a
d
u
r
i
,
“
I
mp
l
e
me
n
t
a
t
i
o
n
o
f
G
P
T
mo
d
e
l
s
f
o
r
t
e
x
t
g
e
n
e
r
a
t
i
o
n
i
n
h
e
a
l
t
h
c
a
r
e
d
o
ma
i
n
,
”
EAI E
n
d
o
rse
d
T
r
a
n
sa
c
t
i
o
n
s
o
n
AI
a
n
d
Ro
b
o
t
i
c
s
,
v
o
l
.
3
,
A
p
r
.
2
0
2
4
,
d
o
i
:
1
0
.
4
1
0
8
/
a
i
r
o
.
4
0
8
2
.
[
3
8
]
D
.
J
y
o
t
i
,
J.
S
r
i
v
a
s
t
a
v
a
,
a
n
d
D
.
P
.
M
a
h
a
t
o
,
“
I
mp
l
e
m
e
n
t
i
n
g
T
5
f
o
r
t
e
x
t
s
u
mm
a
r
i
z
a
t
i
o
n
:
a
n
a
l
g
o
r
i
t
h
mi
c
a
p
p
r
o
a
c
h
,
”
i
n
2
0
2
5
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
ren
c
e
o
n
I
n
f
o
rm
a
t
i
o
n
N
e
t
w
o
rki
n
g
(
I
C
O
I
N
)
,
Ja
n
.
2
0
2
5
,
p
p
.
6
4
8
–
6
5
2
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
O
I
N
6
3
8
6
5
.
2
0
2
5
.
1
0
9
9
2
7
6
6
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2252
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
15
,
No
.
2
,
Ap
r
il
20
26
:
1
8
0
4
-
1
8
1
3
1812
[
3
9
]
J.
R
a
n
g
a
n
a
t
h
a
n
a
n
d
G
.
A
b
u
k
a
,
“
Te
x
t
su
mm
a
r
i
z
a
t
i
o
n
u
si
n
g
t
r
a
n
sf
o
r
mer
m
o
d
e
l
,
”
i
n
2
0
2
2
N
i
n
t
h
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
S
o
c
i
a
l
N
e
t
w
o
rks
An
a
l
y
si
s,
Ma
n
a
g
e
m
e
n
t
a
n
d
S
e
c
u
r
i
t
y
(
S
N
A
MS
)
,
N
o
v
.
2
0
2
2
,
p
p
.
1
–
5
,
d
o
i
:
1
0
.
1
1
0
9
/
S
N
A
M
S
5
8
0
7
1
.
2
0
2
2
.
1
0
0
6
2
6
9
8
.
[
4
0
]
T.
T,
M
.
B
o
r
a
h
,
P
.
D
a
d
u
r
e
,
a
n
d
P
.
P
a
k
r
a
y
,
“
C
o
m
p
a
r
a
t
i
v
e
a
n
a
l
y
s
i
s
o
f
T
5
mo
d
e
l
f
o
r
a
b
s
t
r
a
c
t
i
v
e
t
e
x
t
su
mm
a
r
i
z
a
t
i
o
n
o
n
d
i
f
f
e
r
e
n
t
d
a
t
a
se
t
s,
”
S
S
RN
El
e
c
t
ro
n
i
c
J
o
u
r
n
a
l
,
2
0
2
2
,
d
o
i
:
1
0
.
2
1
3
9
/
ssr
n
.
4
0
9
6
4
1
3
.
[
4
1
]
F
.
K
o
t
o
,
A
.
R
a
h
i
mi
,
J.
H
.
La
u
,
a
n
d
T.
B
a
l
d
w
i
n
,
“
I
n
d
o
LE
M
a
n
d
I
n
d
o
B
ER
T:
a
b
e
n
c
h
m
a
r
k
d
a
t
a
s
e
t
a
n
d
p
r
e
-
t
r
a
i
n
e
d
l
a
n
g
u
a
g
e
m
o
d
e
l
f
o
r
I
n
d
o
n
e
s
i
a
n
N
LP
,
”
i
n
Pro
c
e
e
d
i
n
g
s
o
f
t
h
e
2
8
t
h
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
C
o
m
p
u
t
a
t
i
o
n
a
l
L
i
n
g
u
i
st
i
c
s
,
2
0
2
0
,
p
p
.
7
5
7
–
7
7
0
,
d
o
i
:
1
0
.
1
8
6
5
3
/
v
1
/
2
0
2
0
.
c
o
l
i
n
g
-
m
a
i
n
.
6
6
.
[
4
2
]
K
.
V
i
n
c
e
n
t
i
o
a
n
d
D
.
S
u
h
a
r
t
o
n
o
,
“
A
u
t
o
m
a
t
i
c
q
u
e
s
t
i
o
n
g
e
n
e
r
a
t
i
o
n
m
o
n
o
l
i
n
g
u
a
l
mu
l
t
i
l
i
n
g
u
a
l
p
r
e
-
t
r
a
i
n
e
d
m
o
d
e
l
s
u
s
i
n
g
R
N
N
a
n
d
t
r
a
n
sf
o
r
mer
i
n
l
o
w
r
e
s
o
u
r
c
e
I
n
d
o
n
e
si
a
n
l
a
n
g
u
a
g
e
,
”
I
n
f
o
rm
a
t
i
c
a
,
v
o
l
.
4
6
,
n
o
.
7
,
N
o
v
.
2
0
2
2
,
d
o
i
:
1
0
.
3
1
4
4
9
/
i
n
f
.
v
4
6
i
7
.
4
2
3
6
.
[
4
3
]
F
.
J.
M
u
i
s
a
n
d
A
.
P
u
r
w
a
r
i
a
n
t
i
,
“
S
e
q
u
e
n
c
e
-
to
-
se
q
u
e
n
c
e
l
e
a
r
n
i
n
g
f
o
r
I
n
d
o
n
e
si
a
n
a
u
t
o
ma
t
i
c
q
u
e
st
i
o
n
g
e
n
e
r
a
t
o
r
,
”
i
n
2
0
2
0
7
t
h
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
A
d
v
a
n
c
e
I
n
f
o
rm
a
t
i
c
s:
C
o
n
c
e
p
t
s,
T
h
e
o
ry
a
n
d
Ap
p
l
i
c
a
t
i
o
n
s
(
I
C
AI
C
T
A)
,
S
e
p
.
2
0
2
0
,
p
p
.
1
–
6
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
A
I
C
TA
4
9
8
6
1
.
2
0
2
0
.
9
4
2
9
0
3
2
.
[
4
4
]
M
.
M
.
H
e
n
r
y
,
G
.
N
.
E
l
w
i
r
e
h
a
r
d
j
a
,
a
n
d
B
.
P
a
r
d
a
m
e
a
n
,
“
A
u
t
o
ma
t
i
c
q
u
e
st
i
o
n
g
e
n
e
r
a
t
i
o
n
f
o
r
b
a
h
a
sa
I
n
d
o
n
e
s
i
a
e
x
a
m
i
n
a
t
i
o
n
u
s
i
n
g
c
o
p
y
n
e
t
,
”
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
4
5
,
p
p
.
9
5
3
–
9
6
2
,
2
0
2
4
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
p
r
o
c
s.
2
0
2
4
.
1
0
.
3
2
3
.
[
4
5
]
Y.
-
H
.
C
h
a
n
a
n
d
Y
.
-
C
.
F
a
n
,
“
B
E
R
T
f
o
r
q
u
e
st
i
o
n
g
e
n
e
r
a
t
i
o
n
,
”
i
n
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
1
2
t
h
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
ren
c
e
o
n
N
a
t
u
r
a
l
L
a
n
g
u
a
g
e
G
e
n
e
r
a
t
i
o
n
,
2
0
1
9
,
p
p
.
1
7
3
–
1
7
7
.
d
o
i
:
1
0
.
1
8
6
5
3
/
v
1
/
W
1
9
-
8
6
2
4
.
[
4
6
]
Y.
-
H
.
C
h
a
n
a
n
d
Y
.
-
C
.
F
a
n
,
“
A
r
e
c
u
r
r
e
n
t
B
E
R
T
-
b
a
se
d
m
o
d
e
l
f
o
r
q
u
e
s
t
i
o
n
g
e
n
e
r
a
t
i
o
n
,
”
i
n
Pro
c
e
e
d
i
n
g
s
o
f
t
h
e
2
n
d
Wo
r
k
s
h
o
p
o
n
Ma
c
h
i
n
e
R
e
a
d
i
n
g
f
o
r
Q
u
e
st
i
o
n
An
s
w
e
ri
n
g
,
2
0
1
9
,
p
p
.
1
5
4
–
1
6
2
.
d
o
i
:
1
0
.
1
8
6
5
3
/
v
1
/
D
1
9
-
5
8
2
1
.
[
4
7
]
P
.
R
a
j
p
u
r
k
a
r
,
R
.
J
i
a
,
a
n
d
P
.
L
i
a
n
g
,
“
K
n
o
w
w
h
a
t
y
o
u
d
o
n
’
t
k
n
o
w
:
u
n
a
n
sw
e
r
a
b
l
e
q
u
e
s
t
i
o
n
s
f
o
r
S
Q
u
A
D
,
”
i
n
Pr
o
c
e
e
d
i
n
g
s o
f
t
h
e
5
6
t
h
An
n
u
a
l
M
e
e
t
i
n
g
o
f
t
h
e
Asso
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
st
i
c
s
(
Vo
l
u
m
e
2
:
S
h
o
rt
P
a
p
e
rs)
,
2
0
1
8
,
p
p
.
7
8
4
–
7
8
9
,
d
o
i
:
1
0
.
1
8
6
5
3
/
v
1
/
P
1
8
-
2
1
2
4
.
[
4
8
]
J.
H
.
C
l
a
r
k
e
t
a
l
.
,
“
Ty
D
i
Q
A
:
a
b
e
n
c
h
mar
k
f
o
r
i
n
f
o
r
m
a
t
i
o
n
-
se
e
k
i
n
g
q
u
e
st
i
o
n
a
n
sw
e
r
i
n
g
i
n
t
y
p
o
l
o
g
i
c
a
l
l
y
d
i
v
e
r
se
l
a
n
g
u
a
g
e
s,
”
T
ra
n
s
a
c
t
i
o
n
s
o
f
t
h
e
Asso
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
st
i
c
s
,
v
o
l
.
8
,
p
p
.
4
5
4
–
4
7
0
,
D
e
c
.
2
0
2
0
,
d
o
i
:
1
0
.
1
1
6
2
/
t
a
c
l
_
a
_
0
0
3
1
7
.
[
4
9
]
D
.
Q
.
N
g
u
y
e
n
a
n
d
A
.
T
.
N
g
u
y
e
n
,
“
P
h
o
B
E
R
T:
p
r
e
-
t
r
a
i
n
e
d
l
a
n
g
u
a
g
e
mo
d
e
l
s
f
o
r
V
i
e
t
n
a
m
e
se,
”
i
n
Fi
n
d
i
n
g
s
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:
EM
N
L
P
2
0
2
0
,
2
0
2
0
,
p
p
.
1
0
3
7
–
1
0
4
2
,
d
o
i
:
1
0
.
1
8
6
5
3
/
v
1
/
2
0
2
0
.
f
i
n
d
i
n
g
s
-
e
m
n
l
p
.
9
2
.
[
5
0
]
H.
-
G
.
Zh
a
o
,
X
.
-
Z.
L
i
,
a
n
d
X
.
K
a
n
g
,
“
D
e
v
e
l
o
p
me
n
t
o
f
a
n
a
r
t
i
f
i
c
i
a
l
i
n
t
e
l
l
i
g
e
n
c
e
c
u
r
r
i
c
u
l
u
m
d
e
s
i
g
n
f
o
r
c
h
i
l
d
r
e
n
i
n
Ta
i
w
a
n
a
n
d
i
t
s
i
mp
a
c
t
o
n
l
e
a
r
n
i
n
g
o
u
t
c
o
mes,
”
H
u
m
a
n
i
t
i
e
s
a
n
d
S
o
c
i
a
l
S
c
i
e
n
c
e
s
C
o
m
m
u
n
i
c
a
t
i
o
n
s
,
v
o
l
.
1
1
,
n
o
.
1
,
O
c
t
.
2
0
2
4
,
d
o
i
:
1
0
.
1
0
5
7
/
s
4
1
5
9
9
-
0
2
4
-
0
3
8
3
9
-
z.
[
5
1
]
W
.
Z
h
a
n
g
,
X
.
Li
,
Y
.
Y
a
n
g
,
a
n
d
R
.
D
o
n
g
,
“
P
r
e
-
t
r
a
i
n
i
n
g
o
n
m
i
x
e
d
d
a
t
a
f
o
r
l
o
w
-
r
e
s
o
u
r
c
e
n
e
u
r
a
l
ma
c
h
i
n
e
t
r
a
n
sl
a
t
i
o
n
,
”
I
n
f
o
rm
a
t
i
o
n
,
v
o
l
.
1
2
,
n
o
.
3
,
M
a
r
.
2
0
2
1
,
d
o
i
:
1
0
.
3
3
9
0
/
i
n
f
o
1
2
0
3
0
1
3
3
.
[
5
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
LEU
:
a
me
t
h
o
d
f
o
r
a
u
t
o
mat
i
c
e
v
a
l
u
a
t
i
o
n
o
f
ma
c
h
i
n
e
t
r
a
n
sl
a
t
i
o
n
,
”
i
n
Pro
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
,
2
0
0
1
,
p
p
.
3
1
1
–
318
,
d
o
i
:
1
0
.
3
1
1
5
/
1
0
7
3
0
8
3
.
1
0
7
3
1
3
5
.
[
5
3
]
S
.
B
a
n
e
r
j
e
e
a
n
d
A
.
La
v
i
e
,
“
M
ETEO
R
:
a
n
a
u
t
o
m
a
t
i
c
m
e
t
r
i
c
f
o
r
M
T
e
v
a
l
u
a
t
i
o
n
w
i
t
h
i
mp
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
me
n
t
s,”
i
n
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
AC
L
Wo
r
k
s
h
o
p
o
n
I
n
t
ri
n
s
i
c
a
n
d
Ex
t
ri
n
si
c
Ev
a
l
u
a
t
i
o
n
Me
a
s
u
res
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
z
a
t
i
o
n
,
2
0
0
5
,
p
p
.
6
5
–
7
2
.
[
5
4
]
C.
-
Y
.
L
i
n
,
“
R
O
U
G
E:
a
p
a
c
k
a
g
e
f
o
r
a
u
t
o
m
a
t
i
c
e
v
a
l
u
a
t
i
o
n
o
f
s
u
mm
a
r
i
e
s
,
”
i
n
T
e
x
t
S
u
m
m
a
ri
z
a
t
i
o
n
Bra
n
c
h
e
s
O
u
t
,
B
a
r
c
e
l
o
n
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
st
i
c
s
,
2
0
0
4
,
p
p
.
7
4
–
8
1
.
[
5
5
]
D
.
S
u
l
e
i
m
a
n
a
n
d
A
.
A
w
a
j
a
n
,
“
D
e
e
p
l
e
a
r
n
i
n
g
b
a
se
d
a
b
st
r
a
c
t
i
v
e
t
e
x
t
s
u
mm
a
r
i
z
a
t
i
o
n
:
a
p
p
r
o
a
c
h
e
s,
d
a
t
a
se
t
s,
e
v
a
l
u
a
t
i
o
n
me
a
su
r
e
s,
a
n
d
c
h
a
l
l
e
n
g
e
s,”
M
a
t
h
e
m
a
t
i
c
a
l
Pr
o
b
l
e
m
s
i
n
E
n
g
i
n
e
e
ri
n
g
,
v
o
l
.
2
0
2
0
,
p
p
.
1
–
2
9
,
A
u
g
.
2
0
2
0
,
d
o
i
:
1
0
.
1
1
5
5
/
2
0
2
0
/
9
3
6
5
3
4
0
.
B
I
O
G
RAP
H
I
E
S O
F
AUTH
O
RS
H
a
n
d
a
r
u
J
a
ti
re
c
e
iv
e
d
h
is
P
h
.
D.
d
e
g
re
e
in
C
o
m
p
u
ter
a
n
d
In
f
o
rm
a
ti
o
n
S
c
ien
c
e
fro
m
Un
iv
e
rsiti
Tek
n
o
l
o
g
i
P
e
tro
n
a
s,
M
a
lay
sia
.
He
is
c
u
rre
n
tl
y
a
n
a
ss
o
c
iate
p
ro
fe
ss
o
r
a
t
th
e
De
p
a
rtme
n
t
o
f
El
e
c
tro
n
ics
a
n
d
In
fo
rm
a
ti
c
s
En
g
i
n
e
e
rin
g
Ed
u
c
a
ti
o
n
,
U
n
iv
e
rsitas
Ne
g
e
r
i
Yo
g
y
a
k
a
rta,
In
d
o
n
e
sia
.
His
re
se
a
rc
h
in
tere
sts
in
c
l
u
d
e
m
a
c
h
i
n
e
lea
rn
in
g
,
a
rti
ficia
l
in
telli
g
e
n
c
e
,
d
e
c
isio
n
s
u
p
p
o
rt
sy
ste
m
s,
d
a
ta
m
in
in
g
,
so
ftwa
re
d
e
v
e
lo
p
m
e
n
t,
a
n
d
v
o
c
a
ti
o
n
a
l
e
d
u
c
a
ti
o
n
.
He
h
a
s
p
u
b
li
sh
e
d
n
u
m
e
ro
u
s
p
a
p
e
rs
in
in
tern
a
ti
o
n
a
l
jo
u
rn
a
ls
a
n
d
c
o
n
fe
re
n
c
e
s,
p
a
rti
c
u
larly
i
n
th
e
a
re
a
s
o
f
AI
a
p
p
li
c
a
ti
o
n
s
f
o
r
e
d
u
c
a
ti
o
n
a
n
d
i
n
telli
g
e
n
t
sy
ste
m
s.
He
h
a
s
a
lso
b
e
e
n
i
n
v
o
lv
e
d
in
se
v
e
ra
l
c
o
ll
a
b
o
ra
ti
v
e
re
se
a
rc
h
p
ro
jec
ts
a
n
d
c
o
m
m
u
n
it
y
se
rv
ice
a
c
ti
v
it
ies
fo
c
u
si
n
g
o
n
d
ig
it
a
l
lea
rn
in
g
in
n
o
v
a
t
io
n
.
In
th
is
p
a
p
e
r,
h
e
c
o
n
tri
b
u
ted
t
o
c
o
n
c
e
p
tu
a
li
z
a
ti
o
n
,
m
e
th
o
d
o
l
o
g
y
d
e
sig
n
,
a
n
d
su
p
e
rv
isi
o
n
.
He
c
a
n
b
e
c
o
n
ta
c
ted
a
t
e
m
a
il
:
h
a
n
d
a
r
u
@u
n
y
.
a
c
.
i
d
.
Yun
ia
r
In
d
r
ih
a
p
sa
r
i
is
a
n
a
ss
istan
t
p
r
o
fe
ss
o
r
a
t
U
n
iv
e
rsitas
N
e
g
e
ri
Yo
g
y
a
k
a
rta
(UN
Y),
In
d
o
n
e
sia
,
a
n
d
a
P
h
.
D
.
stu
d
e
n
t
a
t
Na
ti
o
n
a
l
Taiwa
n
Un
iv
e
rsity
o
f
S
c
ien
c
e
a
n
d
Tec
h
n
o
l
o
g
y
,
Taiwa
n
.
S
h
e
re
c
e
iv
e
d
h
e
r
Ba
c
h
e
lo
r’s
d
e
g
re
e
in
El
e
c
tri
c
a
l
En
g
in
e
e
ri
n
g
a
n
d
M
a
ste
r’s
d
e
g
re
e
i
n
I
n
fo
rm
a
ti
o
n
Tec
h
n
o
lo
g
y
fr
o
m
Un
i
v
e
rsitas
G
a
d
jah
M
a
d
a
,
In
d
o
n
e
sia
.
He
r
re
se
a
rc
h
in
tere
sts
in
c
lu
d
e
so
c
ial
n
e
two
rk
a
n
a
ly
sis,
e
-
l
e
a
rn
in
g
,
h
u
m
a
n
-
c
o
m
p
u
ter
in
tera
c
ti
o
n
,
a
n
d
e
d
u
c
a
ti
o
n
a
l
tec
h
n
o
l
o
g
y
.
In
th
is
p
a
p
e
r,
sh
e
c
o
n
tri
b
u
ted
to
c
o
n
c
e
p
tu
a
li
z
a
ti
o
n
,
writi
n
g
–
o
ri
g
i
n
a
l
d
ra
ft
p
re
p
a
ra
t
io
n
,
a
n
d
f
o
rm
a
l
a
n
a
ly
sis.
S
h
e
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
y
u
n
iar@u
n
y
.
a
c
.
id
.
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
I
n
d
o
B
E
R
T fo
r
ed
u
ca
tio
n
a
l a
s
s
ess
men
t:
co
mp
a
r
a
tive
a
n
a
lysi
s
o
f tra
n
s
fo
r
mer m
o
d
els
… (
Ha
n
d
a
r
u
Ja
ti
)
1813
Pra
d
a
n
a
S
e
tia
l
a
n
a
is
a
n
a
ss
ist
a
n
t
p
r
o
fe
ss
o
r
in
th
e
De
p
a
rtme
n
t
o
f
El
e
c
tro
n
ics
a
n
d
In
fo
rm
a
ti
c
s
En
g
in
e
e
ri
n
g
Ed
u
c
a
ti
o
n
a
t
Un
i
v
e
rsitas
Ne
g
e
ri
Yo
g
y
a
k
a
rta,
In
d
o
n
e
sia
.
He
re
c
e
iv
e
d
h
is
Ba
c
h
e
lo
r
d
e
g
re
e
i
n
I
n
fo
r
m
a
ti
c
s
En
g
i
n
e
e
rin
g
E
d
u
c
a
ti
o
n
fro
m
Un
i
v
e
rsitas
Ne
g
e
ri
Yo
g
y
a
k
a
rta
a
n
d
h
is
M
a
ste
r
d
e
g
re
e
in
I
n
fo
rm
a
ti
o
n
Tec
h
n
o
l
o
g
y
fr
o
m
Un
iv
e
rsitas
G
a
d
jah
M
a
d
a
,
In
d
o
n
e
sia
.
His
re
se
a
rc
h
in
tere
sts
in
c
l
u
d
e
s
o
ftwa
re
e
n
g
in
e
e
rin
g
,
n
a
tu
ra
l
lan
g
u
a
g
e
p
r
o
c
e
ss
in
g
,
m
o
b
il
e
a
n
d
c
l
o
u
d
c
o
m
p
u
ti
n
g
,
a
n
d
d
a
tab
a
se
sy
ste
m
s.
In
th
is
p
a
p
e
r,
h
e
c
o
n
tri
b
u
ted
t
o
m
e
th
o
d
o
lo
g
y
,
so
ftwa
re
d
e
v
e
l
o
p
m
e
n
t,
a
n
d
d
a
ta
c
u
ra
ti
o
n
.
He
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
p
ra
d
a
n
a
.
se
ti
a
lan
a
@u
n
y
.
a
c
.
i
d
.
Da
n
a
n
g
Wija
y
a
re
c
e
iv
e
d
h
is
Ba
c
h
e
lo
r
d
e
g
re
e
in
In
f
o
rm
a
ti
c
s
En
g
i
n
e
e
rin
g
Ed
u
c
a
ti
o
n
fro
m
U
n
iv
e
rsitas
Ne
g
e
ri
Yo
g
y
a
k
a
rta,
In
d
o
n
e
sia
,
a
n
d
h
is
M
a
ste
r
d
e
g
re
e
fr
o
m
t
h
e
In
tern
a
ti
o
n
a
l
M
a
ste
r’s
p
ro
g
ra
m
in
Artifi
c
ial
In
telli
g
e
n
c
e
,
Na
ti
o
n
a
l
Ce
n
tral
Un
i
v
e
rsity
,
Taiwa
n
.
His
re
se
a
rc
h
in
tere
sts
i
n
c
lu
d
e
a
rt
ifi
c
ial
in
telli
g
e
n
c
e
a
n
d
n
a
t
u
ra
l
la
n
g
u
a
g
e
p
ro
c
e
ss
in
g
.
I
n
t
h
is
p
a
p
e
r,
h
e
c
o
n
tri
b
u
ted
to
i
n
v
e
stig
a
ti
o
n
,
v
a
li
d
a
ti
o
n
,
a
n
d
v
is
u
a
li
z
a
ti
o
n
.
He
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
d
a
n
a
n
g
wija
y
a
7
5
0
@g
m
a
il
.
c
o
m
.
S
a
ty
a
Adh
i
y
a
k
sa
Ar
d
y
re
c
e
iv
e
d
h
is
Ba
c
h
e
lo
r
d
e
g
re
e
in
In
f
o
rm
a
ti
o
n
Tec
h
n
o
l
o
g
y
fro
m
Un
iv
e
rsitas
Ne
g
e
ri
Yo
g
y
a
k
a
rta,
In
d
o
n
e
sia
.
He
is
c
u
rre
n
tl
y
a
M
a
ste
r’s
stu
d
e
n
t
in
t
h
e
De
p
a
rtme
n
t
o
f
El
e
c
tro
n
ic
a
n
d
Co
m
p
u
ter
En
g
i
n
e
e
rin
g
a
t
Na
ti
o
n
a
l
Taiwa
n
Un
i
v
e
rsity
o
f
S
c
ien
c
e
a
n
d
Tec
h
n
o
l
o
g
y
,
Taiwa
n
.
His
re
se
a
r
c
h
in
tere
sts
in
c
lu
d
e
a
rti
ficia
l
in
telli
g
e
n
c
e
,
n
a
tu
ra
l
lan
g
u
a
g
e
p
r
o
c
e
ss
in
g
,
a
n
d
e
d
u
c
a
ti
o
n
a
l
tec
h
n
o
lo
g
y
.
I
n
th
is
p
a
p
e
r,
h
e
c
o
n
tri
b
u
ted
to
writi
n
g
–
re
v
iew
a
n
d
e
d
it
in
g
,
re
so
u
rc
e
s,
a
n
d
p
ro
jec
t
a
d
m
in
istrati
o
n
.
He
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
sa
ty
a
a
d
h
iy
a
k
sa
@g
m
a
il
.
c
o
m
.
Dhi
sta
Dw
i
Nur
Ar
d
i
a
n
sy
a
h
h
o
l
d
s
a
Ba
c
h
e
lo
r’s
d
e
g
re
e
in
In
f
o
rm
a
ti
c
s
En
g
i
n
e
e
rin
g
E
d
u
c
a
ti
o
n
fro
m
U
n
i
v
e
rsitas
Ne
g
e
ri
Yo
g
y
a
k
a
rta
(UN
Y),
In
d
o
n
e
sia
.
He
is
p
a
rt
o
f
th
e
Dig
it
a
l
Tran
sfo
rm
a
ti
o
n
Dire
c
to
ra
te
a
t
Un
i
v
e
rsity
Na
h
d
a
tu
l
Ul
a
m
a
Yo
g
y
a
k
a
rta,
wh
e
re
h
e
su
p
p
o
rts
i
n
stit
u
ti
o
n
a
l
d
i
g
it
a
li
z
a
t
io
n
i
n
it
iati
v
e
s.
His
re
se
a
rc
h
in
tere
sts
fo
c
u
s
o
n
m
a
c
h
in
e
lea
rn
in
g
a
n
d
it
s
a
p
p
li
c
a
ti
o
n
s.
In
th
is
p
a
p
e
r,
h
e
c
o
n
tri
b
u
te
d
to
i
n
v
e
stig
a
ti
o
n
,
d
a
ta
c
u
ra
ti
o
n
,
so
ftwa
re
/p
ro
c
e
ss
in
g
sc
rip
ts,
v
is
u
a
li
z
a
ti
o
n
,
a
n
d
wri
ti
n
g
,
re
v
iew
,
a
n
d
e
d
it
in
g
.
He
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
d
h
istad
n
a
@g
m
a
il
.
c
o
m
.
Evaluation Warning : The document was created with Spire.PDF for Python.