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
.
1
,
Feb
r
u
ar
y
2
0
2
6
,
p
p
.
929
~
940
I
SS
N:
2
2
5
2
-
8
9
3
8
,
DOI
: 1
0
.
1
1
5
9
1
/ijai.v
15
.i
1
.
p
p
9
2
9
-
9
4
0
929
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
a
i
.
ia
esco
r
e.
co
m
Text
summ
a
riza
ti
o
n:
BART
,
RF
,
a
nd hy
brid B
ART
-
RF
a
lg
o
rithm comp
a
riso
n
M
uh
a
m
m
a
d Adib
Z
a
m
za
m
,
Ag
us
B
uo
no
,
T
o
t
o
H
a
ry
a
nto
S
c
h
o
o
l
o
f
D
a
t
a
S
c
i
e
n
c
e
,
M
a
t
h
e
mat
i
c
s
a
n
d
I
n
f
o
r
ma
t
i
c
s,
I
P
B
U
n
i
v
e
r
si
t
y
,
B
o
g
o
r
,
I
n
d
o
n
e
s
i
a
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Oct
2
5
,
2
0
2
3
R
ev
is
ed
No
v
1
6
,
2
0
2
5
Acc
ep
ted
Dec
1
5
,
2
0
2
5
Da
ta
a
n
d
in
fo
rm
a
ti
o
n
a
c
c
u
m
u
late
q
u
a
n
t
it
a
ti
v
e
l
y
a
n
d
q
u
a
li
tativ
e
l
y
.
Ab
u
n
d
a
n
t
tex
t
d
a
ta
a
re
p
o
ste
d
o
n
t
h
e
in
ter
n
e
t.
Th
e
n
u
m
b
e
r
c
o
rre
late
s
to
th
e
c
o
m
p
lex
it
y
o
f
th
e
su
m
m
a
riza
ti
o
n
.
A
u
to
m
a
ti
c
t
e
x
t
su
m
m
a
riza
ti
o
n
(AT
S
)
is
o
n
e
o
f
th
e
m
o
st
c
h
a
ll
e
n
g
i
n
g
tas
k
s
i
n
n
a
tu
ra
l
lan
g
u
a
g
e
p
ro
c
e
ss
in
g
(NL
P
).
AT
S
a
p
p
r
o
a
c
h
e
d
i
n
th
re
e
wa
y
s:
e
x
trac
ti
v
e
,
a
b
stra
c
ti
v
e
,
a
n
d
h
y
b
ri
d
.
Hy
b
rid
a
p
p
ro
a
c
h
c
o
m
b
in
e
s
b
o
t
h
e
x
trac
ti
v
e
a
n
d
a
b
stra
c
ti
v
e
.
T
h
is
re
se
a
rc
h
tes
ts
a
n
d
c
o
m
p
a
re
s
p
e
rfo
rm
a
n
c
e
o
f
b
i
d
irec
ti
o
n
a
l
a
u
t
o
-
re
g
re
ss
iv
e
tr
a
n
sfo
rm
e
r
(
BART
)
a
n
d
r
a
n
d
o
m
f
o
re
st
(RF
)
in
d
i
v
id
u
a
ll
y
a
n
d
th
e
p
e
rfo
rm
a
n
c
e
c
o
m
b
in
a
ti
o
n
o
f
h
y
b
ri
d
BART
a
n
d
R
F
in
ATS
.
Th
e
re
se
a
rc
h
sh
o
ws
th
a
t
i
n
d
i
v
id
u
a
ll
y
,
BART
a
n
d
RF
re
c
a
ll
-
o
rien
te
d
u
n
d
e
rst
u
d
y
f
o
r
g
isti
n
g
e
v
a
l
u
a
t
io
n
(ROU
G
E)
sc
o
re
s
a
re
h
a
v
in
g
q
u
it
e
d
iffere
n
c
e
s.
Co
n
se
c
u
ti
v
e
l
y
,
ROU
G
E
RF
sc
o
re
s
in
R1
,
R
2
,
a
n
d
RL
a
re
5
1
.
4
5
,
4
5
.
5
2
,
a
n
d
5
4
.
5
8
re
sp
e
c
ti
v
e
ly
.
M
e
a
n
wh
il
e
,
ROU
G
E
BART
sc
o
re
s
a
re
3
2
.
7
8
,
1
6
.
1
7
,
a
n
d
3
2
.
1
9
.
Co
n
se
c
u
ti
v
e
ly
,
a
v
e
ra
g
e
ROU
G
E
RF
,
BART
,
a
n
d
RF
×
BART
F
-
m
e
a
su
re
a
re
4
5
.
7
3
,
2
1
.
3
8
,
a
n
d
3
1
.
3
1
.
R
F
h
a
s
th
e
h
i
g
h
e
st
a
v
e
ra
g
e
sc
o
re
.
ATS
h
y
b
ri
d
R
F
×
BART
is
sh
o
wn
t
o
b
e
p
e
rfo
rm
e
d
b
e
tt
e
r
t
h
a
n
t
h
e
d
e
fa
u
lt
BART.
Th
e
a
v
e
ra
g
e
ROU
G
E
F
-
m
e
a
su
re
s
fo
r
RF
×
BART
o
b
tain
m
o
d
e
ra
te
sc
o
re
a
t
3
1
.
3
1
.
Th
is
sc
o
re
is
b
e
tt
e
r
th
a
n
t
h
e
d
e
fa
u
lt
BART’s
ROU
G
E
sc
o
re
.
RF
×
BART
c
a
n
b
e
a
n
a
lt
e
rn
a
t
iv
e
t
o
th
e
e
ffe
c
ti
v
e
h
y
b
rid
a
p
p
r
o
a
c
h
.
K
ey
w
o
r
d
s
:
Au
to
m
atic
tex
t su
m
m
ar
izatio
n
B
AR
T
-
RF
Hy
b
r
id
R
an
d
o
m
f
o
r
est
R
OUGE
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
:
T
o
to
Har
y
a
n
to
Sch
o
o
l o
f
Data
Scien
ce
,
Ma
t
h
em
atics a
n
d
I
n
f
o
r
m
atics,
I
PB
Un
iv
er
s
ity
Dr
am
ag
a
B
o
g
o
r
,
W
est J
av
a,
1
6
6
8
0
,
I
n
d
o
n
esia
E
m
ail: to
to
h
ar
y
a
n
to
@
ap
p
s
.
ip
b
.
ac
.
id
1.
I
NT
RO
D
UCT
I
O
N
Data
an
d
in
f
o
r
m
atio
n
d
e
v
elo
p
ed
q
u
an
titativ
ely
an
d
q
u
alitati
v
ely
.
I
n
th
e
last
d
ec
ad
e,
u
s
e
o
f
th
e
i
n
ter
n
et
h
as
g
r
o
wn
ex
p
o
n
en
tially
a
n
d
h
as
b
ec
o
m
e
a
n
in
teg
r
al
p
ar
t
o
f
d
aily
life
[
1
]
.
Alm
o
s
t
ev
er
y
o
n
e
f
r
o
m
c
h
ild
r
en
,
ad
o
lescen
t,
ad
u
lt
,
an
d
o
l
d
p
e
o
p
le
u
s
e
i
n
ter
n
et
a
n
d
g
en
er
a
te
m
ass
iv
e
am
o
u
n
ts
o
f
d
ata.
T
h
e
r
esu
lt
o
f
th
is
in
cr
ea
s
in
g
n
u
m
b
er
o
f
u
s
er
s
is
th
e
s
ig
n
if
ican
t
in
cr
ea
s
e
in
d
a
ta
wh
ich
co
m
p
licates
in
f
o
r
m
a
tio
n
r
etr
iev
al
task
s
.
T
h
er
ef
o
r
e,
u
tili
zin
g
m
o
d
er
n
in
f
o
r
m
atio
n
r
etr
iev
al
tech
n
iq
u
es
b
ec
o
m
es e
v
en
m
o
r
e
im
p
o
r
ta
n
t.
Au
to
m
atic
tex
t
s
u
m
m
ar
izatio
n
(
AT
S)
r
esear
ch
h
as
b
ee
n
co
n
d
u
cted
u
s
in
g
v
a
r
io
u
s
m
eth
o
d
s
.
A
s
u
r
v
ey
s
tu
d
y
co
n
d
u
cte
d
wh
ic
h
s
tated
th
at
th
er
e
ar
e
two
k
in
d
s
o
f
a
b
s
tr
ac
tiv
e
ap
p
r
o
ac
h
es,
n
am
ely
s
tr
u
ctu
r
e
-
b
ased
an
d
s
em
an
tic
-
b
ased
[
2
]
.
T
h
e
s
tr
u
c
tu
r
al
ap
p
r
o
ac
h
ca
n
h
av
e
g
r
a
m
m
ar
p
r
o
b
lem
s
b
ec
a
u
s
e
it
d
o
es
n
o
t
co
n
s
id
er
th
e
s
em
an
tic
r
ep
r
esen
tatio
n
o
f
t
h
e
d
o
cu
m
en
t.
T
h
e
s
em
an
tic
a
p
p
r
o
ac
h
h
as
b
etter
lin
g
u
is
tic
q
u
ality
b
ec
au
s
e
it
in
v
o
lv
es
s
em
an
tic
r
ep
r
esen
tatio
n
s
an
d
s
em
an
tic
r
elatio
n
s
o
f
tex
t
d
o
cu
m
en
ts
.
T
h
e
s
em
an
tic
m
eth
o
d
ad
d
r
ess
es
th
e
is
s
u
e
o
f
a
s
tr
u
ctu
r
al
a
p
p
r
o
ac
h
th
at
r
e
d
u
ce
s
r
e
d
u
n
d
an
cies
in
th
e
s
u
m
m
ar
y
.
T
h
e
s
u
m
m
ar
y
cr
ea
ted
p
r
o
d
u
ce
s
b
etter
an
d
d
en
s
er
co
h
esio
n
.
R
esear
ch
o
n
AT
S sp
ec
if
ically
o
n
I
n
d
o
n
esian
is
s
till
r
elativ
ely
s
m
all
[
3
]
.
A
v
ar
iatio
n
o
f
th
e
tr
a
n
s
f
o
r
m
er
n
eu
r
al
n
etwo
r
k
in
tr
o
d
u
ce
d
,
ca
lled
b
i
d
ir
ec
tio
n
al
au
t
o
-
r
eg
r
ess
iv
e
tr
an
s
f
o
r
m
er
s
(
B
AR
T
)
,
wh
ich
i
s
a
v
ar
iatio
n
o
f
b
id
ir
ec
tio
n
al
e
n
co
d
er
tr
a
n
s
f
o
r
m
e
r
s
(
B
E
R
T
)
[
4
]
.
B
AR
T
u
s
ed
b
o
th
p
r
etr
ain
in
g
an
d
f
in
e
-
tu
n
in
g
lea
r
n
in
g
ap
p
r
o
ac
h
.
Pre
tr
ain
in
g
is
d
o
n
e
s
o
th
at
th
e
m
o
d
el
h
as
a
g
en
er
al
u
n
d
er
s
tan
d
in
g
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.
1
5
,
No
.
1
,
Feb
r
u
ar
y
2
0
2
6
:
929
-
9
4
0
930
o
f
th
e
in
p
u
t
d
o
cu
m
e
n
ts
.
Fin
e
tu
n
in
g
p
r
o
ce
s
s
ed
to
s
er
v
e
s
p
ec
if
ic
p
u
r
p
o
s
es
in
a
n
atu
r
al
l
an
g
u
ag
e
p
r
o
ce
s
s
in
g
(
NL
P)
co
n
te
x
t.
B
AR
T
ca
n
ac
c
o
m
p
lis
h
s
p
ec
if
ic
g
o
als s
u
c
h
as
g
en
e
r
atin
g
ab
s
tr
ac
t d
ialo
g
u
e,
q
u
esti
o
n
a
n
s
wer
in
g
an
d
s
u
m
m
ar
izatio
n
[
4
]
.
I
n
th
e
o
th
e
r
h
a
n
d
,
r
a
n
d
o
m
f
o
r
e
s
t
(
R
F)
was u
s
ed
in
th
e
AT
S i
n
a
s
tu
d
y
an
d
h
a
d
g
o
o
d
s
u
m
m
ar
y
q
u
alit
y
r
esu
lts
in
ter
m
s
o
f
r
elev
an
ce
an
d
n
o
v
elty
[
5
]
.
R
F
class
if
ie
s
wh
eth
er
a
s
en
ten
ce
s
h
o
u
ld
b
e
in
clu
d
ed
in
th
e
s
u
m
m
ar
y
o
r
n
o
t.
T
h
e
class
if
ier
is
tr
ain
ed
u
s
in
g
th
e
s
co
r
e
f
ea
tu
r
es
an
d
s
u
m
m
ar
y
in
f
o
r
m
atio
n
f
r
o
m
ea
ch
s
en
ten
ce
f
r
o
m
th
e
d
o
cu
m
en
t
s
et
p
r
ev
io
u
s
ly
cr
ea
ted
.
T
h
e
r
ef
o
r
e
,
R
F
ca
n
b
e
co
n
s
id
er
ed
as
th
e
o
n
e
a
p
p
r
o
ac
h
in
o
u
r
n
e
x
t
s
tu
d
y
.
Au
to
m
atica
lly
g
en
e
r
atin
g
in
f
o
r
m
ativ
e
s
u
m
m
ar
ies
r
e
m
ain
s
a
co
m
p
lex
task
.
Pre
v
io
u
s
p
ap
er
s
tated
th
at
h
y
b
r
id
ap
p
r
o
ac
h
is
o
n
e
o
f
g
o
o
d
ap
p
r
o
ac
h
t
o
g
et
o
f
b
o
th
e
x
tr
ac
tiv
e
an
d
a
b
s
tr
ac
tiv
e
s
u
m
m
ar
izatio
n
ad
v
an
ta
g
es,
b
u
t th
er
e
is
n
o
t e
n
o
u
g
h
p
r
o
o
f
o
f
r
esear
c
h
co
n
d
u
cte
d
to
s
u
p
p
o
r
t th
is
ar
g
u
m
en
t.
T
h
is
is
p
ar
ti
cu
lar
ly
tr
u
e
f
o
r
AT
S
r
esear
ch
in
th
e
I
n
d
o
n
esian
lan
g
u
ag
e,
wh
e
r
e
th
e
ex
p
lo
r
atio
n
o
f
h
y
b
r
id
a
p
p
r
o
ac
h
es
is
s
till
li
m
ited
.
T
o
ad
d
r
ess
th
is
g
ap
,
we
p
r
o
p
o
s
e
a
n
o
v
el
h
y
b
r
id
s
u
m
m
ar
izatio
n
ap
p
r
o
a
c
h
f
o
r
I
n
d
o
n
esian
tex
t.
Ou
r
m
o
d
el
lev
er
ag
es
th
e
s
tr
en
g
th
s
o
f
b
o
t
h
e
x
tr
ac
tiv
e
an
d
a
b
s
tr
ac
tiv
e
tech
n
i
q
u
es.
T
h
e
ex
tr
ac
tiv
e
co
m
p
o
n
en
t,
p
o
wer
ed
b
y
th
e
R
F,
id
en
tifie
s
th
e
m
o
s
t
r
elev
an
t
s
en
ten
ce
s
with
in
th
e
d
o
c
u
m
en
t.
T
h
e
ab
s
tr
ac
tiv
e
c
o
m
p
o
n
en
t,
u
t
ilizin
g
th
e
B
AR
T
,
th
en
g
en
e
r
ates a
co
n
cise su
m
m
ar
y
b
y
r
ef
o
r
m
u
latin
g
an
d
co
m
b
in
in
g
t
h
e
k
ey
p
o
in
ts
ex
tr
ac
t
ed
in
th
e
f
i
r
s
t stag
e.
2.
RE
L
AT
E
D
WO
RK
S
2
.
1
.
Aut
o
m
a
t
ic
t
ex
t
s
um
m
a
r
iza
t
io
n
AT
S
is
th
e
p
r
o
ce
s
s
o
f
co
n
d
en
s
in
g
in
f
o
r
m
atio
n
a
n
d
id
ea
s
in
a
tex
t
in
to
s
h
o
r
ter
tex
ts
[
2
]
.
Sy
s
tem
atic
liter
atu
r
e
r
ev
iew
(
SLR)
s
tu
d
i
es
o
n
AT
S
h
av
e
b
ee
n
co
n
d
u
cted
.
Gen
er
al
p
r
o
ce
s
s
o
f
ac
tiv
ities
in
clu
d
ed
ar
e:
r
ep
r
esen
tatio
n
,
s
co
r
i
n
g
,
a
n
d
te
x
t selec
tio
n
f
o
r
s
u
m
m
ar
ies.
T
h
e
f
ir
s
t
s
tu
d
y
o
n
AT
S
was
c
o
n
d
u
cte
d
in
1
9
5
8
.
E
x
p
er
im
en
ts
ar
e
co
n
d
u
cted
o
n
e
x
tr
ac
tin
g
tech
n
ical
an
d
m
a
g
az
in
e
ar
ticles
[
6
]
.
Statis
tical
in
f
o
r
m
atio
n
s
u
ch
as
th
e
n
u
m
b
er
o
f
wo
r
d
s
in
ar
ticles an
d
th
eir
d
is
tr
ib
u
tio
n
ar
e
u
s
ed
.
SLR
co
n
d
u
cte
d
with
a
s
im
ilar
d
is
cu
s
s
io
n
,
n
am
ely
d
is
cu
s
s
in
g
ap
p
r
o
ac
h
es
to
to
p
ic
r
ep
r
esen
tatio
n
,
co
n
tex
tu
al
in
f
lu
e
n
ce
s
o
n
s
u
m
m
ar
ies,
in
d
icato
r
-
b
ased
r
e
p
r
e
s
en
ta
tio
n
ap
p
r
o
ac
h
es
an
d
th
e
s
en
ten
ce
s
elec
tio
n
p
r
o
ce
s
s
in
s
u
m
m
a
r
ies
[
7
]
,
[
8
]
.
A
s
u
r
v
e
y
s
tu
d
y
th
e
f
ea
tu
r
es
o
f
ex
tr
ac
tiv
e
AT
S,
s
u
p
e
r
v
is
ed
an
d
u
n
s
u
p
e
r
v
is
ed
m
eth
o
d
s
[
8
]
.
T
h
e
ex
tr
ac
tiv
e
a
p
p
r
o
ac
h
is
class
if
ied
as
co
h
er
en
t
in
te
r
m
s
o
f
th
e
r
esu
lts
o
f
th
e
s
u
m
m
ar
y
.
T
h
e
s
u
p
er
v
is
ed
m
eth
o
d
r
eq
u
ir
es
m
ap
p
ed
d
ata
f
r
o
m
h
u
m
an
s
th
at
m
ar
k
s
s
en
ten
ce
s
o
r
p
a
r
ts
o
f
te
x
t
th
at
ar
e
in
clu
d
ed
in
th
e
s
u
m
m
ar
y
.
T
h
e
u
n
s
u
p
e
r
v
is
ed
ap
p
r
o
ac
h
is
s
im
p
ler
th
a
n
th
e
s
u
p
er
v
is
ed
,
as
th
e
lab
el
is
n
o
t
r
eq
u
i
r
ed
in
s
u
m
m
ar
y
g
e
n
er
atio
n
.
T
h
e
AT
S
ar
ch
itectu
r
e
in
g
e
n
e
r
al
in
clu
d
es
a
s
er
ies
o
f
p
r
e
p
r
o
ce
s
s
in
g
,
p
r
o
ce
s
s
in
g
,
a
n
d
p
o
s
t
p
r
o
ce
s
s
in
g
[
2
]
.
Mo
s
t
AT
S
s
y
s
tem
s
th
at
p
er
f
o
r
m
well
ar
e
f
o
c
u
s
ed
o
n
o
n
e
g
o
al,
o
n
e
m
eth
o
d
f
o
r
a
s
p
e
cif
ic
ap
p
r
o
ac
h
,
o
n
e
d
o
m
ain
s
p
ec
if
ic
t
o
th
e
AT
S.
T
h
e
p
r
e
p
r
o
ce
s
s
in
g
s
tag
e
in
A
T
S
is
to
cr
ea
te
a
s
tr
u
ctu
r
ed
r
ep
r
esen
tatio
n
o
f
th
e
o
r
ig
in
al
tex
t
u
s
in
g
v
ar
io
u
s
tech
n
iq
u
es
s
u
ch
as
s
eg
m
en
tatio
n
,
wo
r
d
s
to
k
en
izatio
n
,
r
em
o
v
al
o
f
s
to
p
-
wo
r
d
s
,
p
ar
t
-
of
-
s
p
ee
ch
tag
g
in
g
,
s
tem
m
in
g
an
d
th
e
lik
e.
T
h
e
p
r
o
ce
s
s
in
g
s
tag
e
is
th
e
a
p
p
licatio
n
o
f
tech
n
iq
u
es
f
o
r
m
a
k
in
g
a
s
u
m
m
ar
y
.
T
h
e
p
o
s
t
-
p
r
o
ce
s
s
in
g
s
tag
e
is
s
o
lv
in
g
th
e
r
em
ain
in
g
p
r
o
b
le
m
s
is
in
th
e
s
u
m
m
ar
y
r
esu
lts
s
u
ch
as
s
en
ten
ce
s
eq
u
en
cin
g
.
AT
S
is
v
er
y
o
f
ten
u
s
ed
in
te
x
t
m
in
in
g
an
d
an
aly
tical
ap
p
licatio
n
s
s
u
ch
as
in
f
o
r
m
atio
n
r
etr
iev
al,
in
f
o
r
m
atio
n
ex
tr
ac
tio
n
,
q
u
esti
o
n
an
s
wer
in
g
.
AT
S
h
as
u
s
ed
in
f
r
ar
e
d
r
ad
iatio
n
(
IR
)
tec
h
n
iq
u
es
to
s
tr
en
g
th
e
n
s
ea
r
ch
en
g
in
e
ca
p
a
b
ilit
ies
[
9
]
.
Fu
r
th
er
a
p
p
licatio
n
s
in
cl
u
d
e
m
ed
ia
s
u
ch
as
:
n
ews,
o
p
in
io
n
/
s
en
tim
en
t,
Mic
r
o
b
lo
g
/Tw
ee
t,
b
o
o
k
s
,
s
to
r
ies/
n
o
v
els
,
em
ail,
b
io
m
ed
ic
al
d
o
cu
m
en
ts
,
leg
al
d
o
cu
m
e
n
ts
,
an
d
s
cien
tific
jo
u
r
n
als
.
T
ex
t
s
u
m
m
ar
izatio
n
is
also
ap
p
licab
le
f
o
r
s
o
f
twar
e
s
u
m
m
ar
izatio
n
,
as
th
e
ar
tifa
cts
o
r
d
o
cu
m
en
tatio
n
an
d
s
o
u
r
ce
co
d
e
ar
e
b
asically
tex
t
[
1
0
]
.
So
f
twar
e
s
u
m
m
ar
i
za
tio
n
is
a
f
ield
th
at
is
s
til
l
u
n
d
er
d
e
v
elo
p
m
e
n
t.
So
f
twar
e
s
u
m
m
ar
izatio
n
d
is
cu
s
s
es
th
e
p
r
o
ce
s
s
o
f
g
en
er
atin
g
r
ep
r
esen
tatio
n
s
o
f
o
n
e
o
r
m
o
r
e
s
o
f
twar
e
ar
tifa
cts
th
at
co
n
tain
in
f
o
r
m
atio
n
th
at
s
tak
eh
o
ld
er
s
n
ee
d
to
p
er
f
o
r
m
s
p
ec
if
ic
task
s
in
s
o
f
twar
e
en
g
in
ee
r
in
g
.
2
.
2
.
Ra
nd
o
m
f
o
re
s
t
R
F
is
an
en
s
em
b
le
lear
n
in
g
-
b
ased
alg
o
r
ith
m
,
wh
ich
o
p
er
ate
s
b
y
b
u
ild
in
g
d
ec
is
io
n
tr
ee
s
(
DT
)
.
E
ac
h
tr
ee
b
u
ilt
f
r
o
m
p
ar
t
o
f
th
e
d
ata
s
et
u
s
in
g
th
e
b
o
o
ts
tr
ap
ag
g
r
eg
atin
g
(
B
ag
g
in
g
)
m
eth
o
d
.
T
h
e
m
ain
ad
v
an
tag
es
o
f
R
F
ar
e
s
im
p
licity
an
d
ef
f
ec
tiv
en
ess
.
T
h
e
d
r
awb
ac
k
o
f
R
F
is
lo
w
in
ter
p
r
eta
b
ilit
y
.
T
h
e
o
u
tp
u
t
class
is
th
e
m
o
d
e
o
r
av
er
ag
e
(
m
ea
n
)
o
f
all
tr
ee
r
esu
lts
th
at
wer
e
b
u
ilt
p
r
ev
io
u
s
ly
[
1
1
]
.
R
F
is
o
f
te
n
u
s
ed
in
NL
P
ca
s
es,
s
u
ch
as
lan
g
u
ag
e
m
o
d
elin
g
[
1
2
]
,
s
en
ti
m
en
t a
n
aly
s
is
[
1
3
]
,
d
etec
tio
n
o
f
s
ar
ca
s
m
[
1
1
]
,
an
d
s
u
m
m
ar
i
za
tio
n
[
3
]
.
2
.
3
.
B
idi
re
ct
io
na
l a
uto
-
re
g
r
ess
iv
e
t
ra
ns
f
o
rm
er
s
B
AR
T
i
s
a
tr
an
s
f
o
r
m
er
m
o
d
e
l,
a
d
en
o
is
in
g
au
to
e
n
co
d
e
r
th
at
ca
n
b
e
u
s
ed
in
a
v
er
y
wid
e
r
an
g
e
o
f
task
s
[
4
]
.
T
r
an
s
f
o
r
m
er
s
ca
p
a
b
le
o
f
h
a
n
d
lin
g
s
eq
u
e
n
tial
d
ata
u
s
in
g
f
ewe
r
r
eso
u
r
ce
s
t
h
an
r
ec
u
r
r
en
t
n
e
u
r
al
n
etwo
r
k
s
(
R
NN)
o
r
co
n
v
o
lu
ti
o
n
al
n
eu
r
al
n
etwo
r
k
s
(
C
NN)
[
1
4
]
.
Few
o
f
B
AR
T
'
s
p
r
ed
ec
ess
o
r
s
ar
e
th
e
B
E
R
T
an
d
GPT
m
o
d
els
[
1
5
]
,
[
1
6
]
.
B
AR
T
i
s
p
ar
ticu
lar
ly
ef
f
ec
tiv
e
wh
en
f
in
e
-
tu
n
ed
f
o
r
tex
t
g
e
n
er
atio
n
.
Fin
etu
n
e
d
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
Text
s
u
mma
r
iz
a
tio
n
:
B
A
R
T,
R
F
,
a
n
d
h
yb
r
id
B
A
R
T
-
R
F
a
lg
o
r
ith
m
co
mp
a
r
is
o
n
(
Mu
h
a
mma
d
A
d
ib
Z
a
mza
m)
931
B
AR
T
m
o
d
el
ca
n
p
er
f
o
r
m
m
a
n
y
NL
P
ac
tiv
ities
s
u
ch
as seq
u
en
ce
class
if
icatio
n
,
to
k
en
class
if
icatio
n
,
s
eq
u
e
n
ce
g
en
er
atio
n
a
n
d
m
ac
h
in
e
tr
a
n
s
latio
n
.
B
AR
T
u
s
es
th
e
s
tan
d
ar
d
s
eq
2
s
eq
tr
an
s
f
o
r
m
er
ar
ch
itectu
r
e
,
u
tili
zin
g
GeL
Us
as
ac
tiv
atio
n
f
u
n
ctio
n
[
4
]
.
T
h
e
b
asic
m
o
d
el
u
s
ed
is
6
lay
e
r
s
o
f
th
e
en
c
o
d
er
a
n
d
d
ec
o
d
er
.
T
h
e
lar
g
e
m
o
d
el
h
as
1
2
lay
er
s
o
n
th
e
en
c
o
d
er
a
n
d
d
ec
o
d
er
.
T
h
e
ar
c
h
itectu
r
e
b
eh
av
es
lik
e
B
E
R
T
with
th
e
d
if
f
er
en
ce
s
o
f
:
i
)
ea
ch
lay
e
r
o
n
th
e
d
ec
o
d
er
p
er
f
o
r
m
s
cr
o
s
s
-
atten
tio
n
o
n
th
e
last
h
id
d
en
lay
er
o
n
th
e
en
c
o
d
er
,
a
n
d
ii
)
B
E
R
T
u
s
e
s
an
ad
d
itio
n
al
f
e
ed
-
f
o
r
war
d
n
etwo
r
k
b
ef
o
r
e
wo
r
d
p
r
ed
ictio
n
,
wh
er
e
B
AR
T
is
n
o
t.
B
AR
T
co
n
tai
n
s
1
0
%
m
o
r
e
p
ar
am
eter
s
th
a
n
B
E
R
T
.
T
h
e
b
r
ief
co
m
p
ar
is
o
n
o
f
B
AR
T
an
d
B
E
R
T
is
s
h
o
wn
at
Fig
u
r
e
1
.
S
tan
d
ar
d
tr
an
s
f
o
r
m
er
ar
c
h
itectu
r
e
i
s
s
h
o
wn
in
Fig
u
r
e
1
(
a)
.
A
g
e
n
er
al
co
m
p
ar
is
o
n
o
f
th
e
B
AR
T
ar
ch
itectu
r
e
with
GPT
an
d
B
E
R
T
is
s
h
o
wn
in
Fig
u
r
e
1
(
b
)
.
(
a)
(
b
)
Fig
u
r
e
1
.
C
o
m
p
a
r
is
o
n
o
f
tr
an
s
f
o
r
m
er
-
b
ased
m
o
d
els f
o
r
NL
P
,
h
ig
h
lig
h
tin
g
th
eir
a
r
ch
itectu
r
a
l d
if
f
er
en
ces
:
(
a)
s
tan
d
ar
d
tr
an
s
f
o
r
m
er
ar
c
h
itectu
r
e
[
1
4
]
an
d
(
b
)
ar
ch
itectu
r
e
co
m
p
ar
is
o
n
o
f
GPT,
B
E
R
T
,
an
d
B
AR
T
[
4
]
3.
M
E
T
H
O
D
T
h
e
s
u
m
m
ar
y
s
y
s
tem
p
r
o
ce
s
s
co
n
s
is
ts
o
f
s
ev
er
al
p
r
o
ce
s
s
es
:
i
)
in
p
u
t
d
ata
in
th
e
f
o
r
m
o
f
tex
t,
ii
)
p
r
e
-
p
r
o
ce
s
s
,
iii
)
p
r
o
ce
s
s
ea
c
h
alg
o
r
ith
m
to
p
r
o
d
u
ce
an
a
u
to
m
atic
s
u
m
m
ar
y
,
an
d
iv
)
ev
alu
atio
n
.
T
h
e
tex
t
d
ata
ar
e
n
ews
ar
ticles
f
r
o
m
th
e
lip
u
tan
6
d
ataset
[
1
7
]
.
T
h
e
p
r
e
p
r
o
ce
s
s
co
n
s
is
ts
o
f
ca
s
e
f
o
ld
i
n
g
,
lab
el
m
ap
p
in
g
,
to
k
en
izatio
n
an
d
to
k
e
n
r
em
o
v
al.
T
h
er
e
ar
e
3
s
ce
n
ar
i
o
s
o
f
th
e
alg
o
r
ith
m
th
at
will
b
e
u
s
ed
,
B
AR
T
,
R
F
,
an
d
a
h
y
b
r
id
co
m
b
i
n
a
tio
n
o
f
R
F
an
d
B
AR
T
.
T
h
en
,
r
ec
all
-
o
r
ien
ted
u
n
d
er
s
tu
d
y
f
o
r
g
is
tin
g
ev
alu
atio
n
(
R
OUGE
)
m
etr
ic
is
th
en
ca
lcu
lated
o
n
t
h
e
s
u
m
m
ar
ies
o
b
tain
ed
.
T
h
e
s
u
m
m
ar
i
za
tio
n
s
y
s
tem
p
r
o
p
o
s
ed
is
s
h
o
wn
in
Fig
u
r
e
2
.
T
h
e
d
ata
to
b
e
u
s
ed
is
L
ip
u
tan
6
I
n
d
o
n
esian
s
u
m
m
ar
izatio
n
d
atas
et
[
1
7
]
.
T
h
e
p
r
o
p
o
s
ed
R
F
×
B
AR
T
h
y
b
r
id
m
o
d
el
is
d
e
f
in
ed
b
y
a
two
-
s
tag
e
p
r
o
ce
s
s
o
p
er
atin
g
in
an
ex
tr
ac
tiv
e
-
ab
s
tr
ac
tiv
e
f
ash
io
n
.
C
r
u
cially
,
th
e
R
F
co
m
p
o
n
en
t
f
u
n
ctio
n
s
as
th
e
f
ir
s
t
s
tag
e:
an
in
p
u
t
s
elec
tio
n
lay
er
.
I
t
r
an
k
s
an
d
s
elec
ts
th
e
m
o
s
t
s
alien
t
s
en
ten
ce
s
f
r
o
m
th
e
s
o
u
r
ce
d
o
c
u
m
en
t
b
ased
o
n
a
s
et
o
f
lin
g
u
is
tic
an
d
p
o
s
itio
n
al
f
ea
tu
r
es.
On
ly
th
ese
h
ig
h
-
q
u
al
ity
,
f
ilter
ed
s
en
ten
ce
s
ar
e
p
ass
ed
to
th
e
B
AR
T
m
o
d
el,
wh
ich
th
en
g
en
er
ates
th
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.
1
5
,
No
.
1
,
Feb
r
u
ar
y
2
0
2
6
:
929
-
9
4
0
932
f
in
al
ab
s
tr
ac
tiv
e
s
u
m
m
ar
y
.
T
h
is
ar
ch
itectu
r
e
is
d
esig
n
ed
t
o
o
p
tim
ize
th
e
B
AR
T
in
p
u
t,
th
er
eb
y
m
itig
atin
g
r
ed
u
n
d
an
cy
a
n
d
im
p
r
o
v
in
g
th
e
o
v
er
all
co
n
ten
t f
id
elity
o
f
th
e
g
en
er
ated
s
u
m
m
ar
y
.
Fig
u
r
e
2
.
Pro
p
o
s
ed
s
u
m
m
a
r
iza
tio
n
s
y
s
tem
3
.
1
.
Da
t
a
prepro
ce
s
s
T
h
e
p
r
ep
r
o
ce
s
s
co
n
s
is
ts
o
f
ca
s
e
f
o
ld
in
g
,
lab
el
m
ap
p
in
g
,
to
k
en
izatio
n
an
d
to
k
en
r
e
m
o
v
al.
B
y
co
n
v
er
tin
g
all
ch
a
r
ac
ter
s
to
l
o
wer
ca
s
e
(
ca
s
e
f
o
ld
in
g
)
,
wo
r
d
s
d
if
f
er
in
g
o
n
ly
in
ca
p
italiz
atio
n
ar
e
tr
ea
te
d
as
id
en
tical.
L
ab
el
m
ap
p
in
g
is
d
o
n
e
o
n
t
h
e
o
r
ig
in
al
ar
ticle
d
ata
s
o
th
at
in
d
ex
o
f
ea
ch
r
e
f
er
en
ce
s
en
ten
ce
is
p
r
eser
v
ed
.
T
o
k
en
izatio
n
is
th
e
p
r
o
ce
s
s
o
f
m
ak
in
g
th
e
s
m
alle
s
t
elem
en
ts
o
f
tex
t
in
to
f
o
r
m
o
f
in
teg
er
v
ec
to
r
s
.
T
o
k
en
r
em
o
v
al
r
em
o
v
es
u
n
n
e
ce
s
s
ar
y
wo
r
d
s
/elem
en
ts
.
T
o
k
e
n
r
em
o
v
al
is
d
o
n
e
s
o
th
at
th
e
d
ata
th
at
is
p
r
o
ce
s
s
ed
is
s
u
b
s
tan
tial.
3
.
2
.
RF
e
x
t
ra
c
t
iv
e
s
um
ma
riza
t
io
n
Fo
r
ea
ch
d
o
c
u
m
en
t,
we
h
a
v
e
to
e
x
tr
ac
t
th
e
in
f
o
r
m
atio
n
as
th
e
f
ea
tu
r
es.
I
n
t
h
is
s
tu
d
y
,
we
u
s
ed
8
attr
ib
u
tes f
r
o
m
s
en
ten
ce
s
in
o
n
e
d
o
cu
m
e
n
t r
ef
er
s
to
th
e
[
3
]
,
[
1
8
]
as we
ll
:
i)
T
F/IDF
: t
h
e
v
alu
e
o
f
th
is
attr
ib
u
te
is
ca
lcu
lated
b
ased
o
n
th
e
n
u
m
b
er
o
f
ter
m
s
th
at
ap
p
ea
r
in
th
e
s
en
ten
ce
an
d
th
e
n
u
m
b
e
r
in
all
s
en
ten
c
es
in
th
e
d
o
cu
m
en
t.
T
F/IDF
in
th
is
co
n
tex
t
ca
n
b
e
r
e
f
er
r
ed
to
as
(
T
F/ISF).
T
h
e
f
o
r
m
u
la
is
wr
itten
in
(
1
)
.
(
)
=
(
(
)
×
(
)
)
(
1
)
is
f
is
th
e
n
u
m
b
er
o
f
o
cc
u
r
r
en
c
es
o
f
th
e
ith
wo
r
d
in
th
e
e
n
tir
e
s
en
ten
ce
.
tf
is
th
e
o
cc
u
r
r
en
c
e
o
f
th
e
wo
r
d
(
ter
m
f
r
e
q
u
en
c
y
)
in
th
e
ith
s
en
ten
ce
.
len
is
th
e
n
u
m
b
er
o
f
wo
r
d
s
in
th
e
s
en
ten
ce
.
ii)
U
p
p
er
c
a
s
e
:
t
h
i
s
a
t
t
r
ib
u
t
e
w
i
l
l
h
a
v
e
h
i
g
h
e
r
w
e
ig
h
t
in
g
v
a
l
u
e
i
n
s
e
n
t
e
n
c
e
s
t
h
a
t
h
a
v
e
o
n
e
o
r
m
o
r
e
c
a
p
i
ta
l
l
e
t
t
e
r
s
s
u
c
h
a
s
t
h
e
n
a
m
e
o
f
t
h
e
p
er
s
o
n
's
s
u
b
j
e
c
t,
p
l
ac
e
o
r
o
b
je
c
t
n
am
e
.
V
a
lu
e
s
a
r
e
o
b
t
a
i
n
ed
u
s
in
g
(
2
)
a
n
d
(
3
)
.
(
)
=
(
2
)
2
(
)
=
(
)
(
(
)
)
(
3
)
iii)
Pro
p
er
n
o
u
n
: t
h
e
n
u
m
b
er
o
f
n
o
u
n
s
in
a
s
en
ten
ce
p
o
s
itiv
ely
co
r
r
elate
s
with
th
e
v
alu
e
o
f
th
is
attr
ib
u
te.
I
t is
d
ef
in
ed
as r
atio
o
f
n
o
u
n
wo
r
d
o
v
er
th
e
le
n
g
th
o
f
a
s
en
ten
ce
.
T
h
e
v
alu
e
is
o
b
tain
ed
u
s
in
g
(
4
)
.
3
(
)
=
(
4
)
iv
)
C
u
e
p
h
r
ases
:
i
n
g
en
er
al,
m
an
y
s
en
ten
ce
s
b
eg
in
with
p
h
r
ases
s
u
ch
as
s
o
,
in
v
esti
g
atio
n
an
d
th
e
lik
e
.
E
m
p
h
asis
wo
r
d
s
s
u
ch
as
“
terb
a
ik
”,
“
terp
en
tin
g
”,
“
p
a
lin
g
p
en
tin
g
”
wh
ich
e
q
u
iv
alen
t
to
“
b
est”
o
r
“m
o
s
t
im
p
o
r
tan
t”
in
E
n
g
lis
h
an
d
o
th
er
p
h
r
ases
also
s
h
o
w
th
e
g
o
o
d
ch
ar
ac
ter
is
tic
th
at
a
s
en
ten
ce
is
in
clu
d
ed
in
th
e
s
u
m
m
ar
y
.
I
t
is
d
ef
in
ed
as
r
atio
o
f
p
h
r
ase
co
u
n
t
o
v
er
wo
r
d
co
u
n
t
in
a
s
en
ten
ce
.
Attr
ib
u
te
v
alu
es
ar
e
o
b
tain
ed
u
s
in
g
(
5
)
.
4
(
)
=
ℎ
(
5
)
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
Text
s
u
mma
r
iz
a
tio
n
:
B
A
R
T,
R
F
,
a
n
d
h
yb
r
id
B
A
R
T
-
R
F
a
lg
o
r
ith
m
co
mp
a
r
is
o
n
(
Mu
h
a
mma
d
A
d
ib
Z
a
mza
m)
933
v)
Nu
m
er
ical
d
ata
:
n
u
m
er
ical
d
ata/n
u
m
b
er
s
in
d
icate
th
at
a
s
en
ten
ce
m
ay
co
n
tain
im
p
o
r
tan
t
in
f
o
r
m
atio
n
.
T
h
e
v
alu
e
o
f
t
h
is
attr
ib
u
te
is
o
b
tain
ed
u
s
in
g
(
6
)
.
5
(
)
=
/
(
6
)
v
i)
Sen
ten
ce
len
g
th
:
l
o
n
g
s
en
ten
c
es
h
av
e
a
h
ig
h
er
weig
h
t.
I
t
is
d
ef
in
ed
as
r
atio
o
f
wo
r
d
co
u
n
t
in
a
s
en
ten
ce
o
v
er
lo
n
g
est s
en
ten
ce
in
ter
m
o
f
wo
r
d
co
u
n
t.
Valu
es o
b
tain
e
d
b
y
(
7
)
.
6
(
)
=
(
7
)
v
ii)
Sen
ten
ce
p
o
s
itio
n
:
t
h
e
p
o
s
itio
n
o
f
th
e
s
en
ten
ce
in
a
p
ar
a
g
r
ap
h
d
eter
m
in
es
wh
eth
er
th
e
s
en
ten
ce
i
s
co
n
s
id
er
ed
im
p
o
r
tan
t.
I
t
is
ass
u
m
ed
th
at
th
e
f
ir
s
t
s
en
ten
ce
an
d
th
e
last
s
en
ten
ce
ar
e
im
p
o
r
tan
t
s
en
ten
ce
s
.
Valu
es o
b
tain
ed
b
y
(
8
)
.
7
(
)
=
{
1
,
−
,
ℎ
(
8
)
v
iii)
Similar
ity
to
titl
e
:
t
h
is
attr
ib
u
te
ass
ess
e
s
th
e
lev
el
o
f
s
im
ilar
ity
in
th
e
titl
e.
I
t
is
d
ef
in
ed
as
r
atio
o
f
in
ter
s
ec
ted
wo
r
d
.
Valu
es o
b
tain
ed
b
y
(
9
)
.
8
(
)
=
∩
∪
(
9
)
R
F
will
b
e
tr
ain
ed
to
p
er
f
o
r
m
b
in
ar
y
class
if
icatio
n
,
wh
ich
is
to
d
eter
m
in
e
wh
eth
er
a
s
en
t
en
ce
in
a
n
ar
ticle
is
in
clu
d
ed
in
th
e
s
u
m
m
ar
y
.
L
a
b
els
0
an
d
1
ar
e
u
s
ed
in
th
e
d
ataset
to
in
d
icate
wh
e
th
er
a
s
en
ten
ce
is
a
s
u
m
m
ar
y
r
e
f
er
en
ce
.
I
f
th
e
la
b
e
l is 0
th
en
th
e
s
en
ten
ce
d
o
es n
o
t in
clu
d
e
a
s
u
m
m
ar
y
,
if
1
th
e
n
it is
a
s
u
m
m
ar
y
.
3
.
3
.
B
ART
a
bs
t
ra
ct
iv
e
s
um
m
a
riza
t
i
o
n
Ab
s
tr
ac
tiv
e
s
u
m
m
ar
izatio
n
is
o
n
e
o
f
th
e
tex
t
g
e
n
er
atio
n
task
s
.
Su
m
m
ar
izin
g
is
d
o
n
e
b
y
tr
ain
in
g
B
AR
T
with
s
tag
es
o
f
p
r
e
-
tr
ain
i
n
g
a
n
d
f
in
e
-
tu
n
in
g
.
C
o
m
m
o
n
p
r
e
-
tr
ain
in
g
a
p
p
r
o
ac
h
r
ef
i
n
es
its
u
n
d
er
s
tan
d
in
g
b
y
r
ec
o
n
s
tr
u
ctin
g
co
r
r
u
p
te
d
tex
t.
E
ac
h
m
eth
o
d
an
d
p
r
e
-
tr
ain
in
g
ex
am
p
le
ar
e
s
h
o
wn
in
Fig
u
r
e
3
.
I
n
th
e
ex
am
p
le
in
Fig
u
r
e
3
,
wo
r
d
s
o
r
to
k
en
s
ar
e
s
y
m
b
o
lized
b
y
th
e
letter
s
A
th
r
o
u
g
h
E
.
B
AR
T
f
in
e
-
tu
n
in
g
in
th
e
ca
s
e
o
f
s
u
m
m
ar
izatio
n
is
d
o
n
e
in
a
s
p
ec
if
ic
way
ca
lled
s
eq
u
en
ce
g
en
er
atio
n
.
Seq
u
en
ce
g
e
n
er
ati
o
n
is
p
r
o
ce
s
s
ed
b
y
m
an
ip
u
latin
g
t
h
e
tex
t o
f
th
e
in
p
u
t so
th
at
th
e
r
esu
lts
h
av
e
s
im
ilar
m
ea
n
in
g
with
s
h
o
r
ter
le
n
g
th
.
Fig
u
r
e
3
.
B
AR
T
Pre
-
tr
ain
in
g
e
x
am
p
le
[
4
]
3
.
4
.
H
y
brid s
um
m
a
riza
t
io
n
Hy
b
r
id
s
u
m
m
a
r
izatio
n
is
d
o
n
e
b
y
co
m
b
in
in
g
t
h
e
R
F
an
d
B
AR
T
alg
o
r
ith
m
s
.
R
F
is
u
tili
ze
d
at
ex
tr
ac
tiv
e
s
u
m
m
ar
izatio
n
.
T
h
e
s
u
m
m
ar
y
r
esu
lts
o
f
th
e
R
F
ar
e
th
en
p
r
o
ce
s
s
ed
o
n
th
e
B
AR
T
f
o
r
s
u
m
m
ar
y
r
e
g
en
er
atio
n
.
I
llu
s
tr
atio
n
h
y
b
r
id
s
u
m
m
a
r
izatio
n
s
ce
n
ar
io
is
s
h
o
wn
in
Fig
u
r
e
4
.
Fig
u
r
e
4
.
Hy
b
r
id
s
u
m
m
ar
izatio
n
s
ce
n
ar
io
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.
1
5
,
No
.
1
,
Feb
r
u
ar
y
2
0
2
6
:
929
-
9
4
0
934
3
.
5
.
Rec
a
ll
-
o
rient
ed
un
derst
ud
y
f
o
r
g
is
t
ing
ev
a
lua
t
io
n
R
O
UG
E
is
o
n
e
o
f
t
h
e
a
u
t
o
m
a
ti
c
m
e
as
u
r
e
m
e
n
t
m
e
t
h
o
d
s
.
I
n
t
h
i
s
m
et
h
o
d
,
t
h
e
s
u
m
m
a
r
y
o
f
t
h
e
r
e
s
u
l
ts
o
f
a
u
t
o
m
a
t
i
o
n
is
c
o
m
p
a
r
e
d
t
o
t
h
e
s
u
m
m
a
r
y
m
a
d
e
b
y
h
u
m
a
n
s
[
1
9
]
.
T
h
e
m
e
as
u
r
e
m
e
n
t
i
s
c
o
n
d
u
c
te
d
b
y
c
a
l
c
u
l
a
t
i
n
g
t
h
e
m
a
t
c
h
i
n
g
u
n
i
ts
s
u
c
h
as
n
-
g
r
a
m
s
(
n
-
w
o
r
d
s
)
,
w
o
r
d
o
r
d
e
r
,
a
n
d
w
o
r
d
p
a
i
r
s
b
et
w
e
e
n
t
h
e
s
u
m
m
a
r
y
o
f
t
h
e
a
u
t
o
m
at
e
d
r
e
s
u
l
ts
c
o
m
p
a
r
e
d
t
o
t
h
e
r
e
f
e
r
e
n
c
e
s
u
m
m
a
r
y
.
A
h
i
g
h
e
r
R
O
U
G
E
s
c
o
r
e
i
n
d
i
c
a
t
e
s
b
e
tt
e
r
s
u
m
m
a
r
iz
a
t
i
o
n
p
e
r
f
o
r
m
a
n
c
e
,
r
e
f
l
e
c
t
i
n
g
a
g
r
e
a
te
r
d
e
g
r
e
e
o
f
o
v
e
r
l
a
p
b
e
t
w
e
e
n
t
h
e
g
e
n
e
r
at
e
d
s
u
m
m
a
r
y
a
n
d
t
h
e
r
e
f
e
r
e
n
c
e
s
u
m
m
a
r
y
.
4.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
4
.
1
.
P
re
pro
ce
s
s
ed
da
t
a
Pre
p
r
o
ce
s
s
in
g
co
n
s
is
ts
o
f
ca
s
e
f
o
ld
in
g
,
lab
el
m
ap
p
in
g
,
to
k
en
i
za
tio
n
an
d
to
k
en
r
em
o
v
al.
Pre
-
p
r
o
ce
s
s
in
g
f
o
r
R
F
m
o
d
el
tr
ain
in
g
d
ata
in
clu
d
es
ca
s
e
f
o
ld
in
g
,
lab
el
m
ap
p
i
n
g
,
an
d
to
k
en
r
em
o
v
al.
L
ab
el
m
ap
p
in
g
is
in
clu
d
e
d
in
f
ea
tu
r
e
en
g
in
ee
r
in
g
.
T
o
k
en
i
za
tio
n
is
o
n
ly
p
r
ep
ar
e
d
f
o
r
t
h
e
d
ev
elo
p
m
e
n
t
o
f
th
e
B
AR
T
m
o
d
el.
C
ase
f
o
ld
in
g
is
n
o
t
u
s
ed
b
ec
au
s
e
it
co
n
f
licts
with
th
e
u
p
p
er
ca
s
e
f
ea
tu
r
e.
T
h
e
u
p
p
e
r
ca
s
e
f
ea
tu
r
e
will
alwa
y
s
h
av
e
a
v
alu
e
o
f
0
if
th
e
ca
s
e
f
o
l
d
in
g
p
r
o
ce
s
s
is
in
clu
d
ed
.
On
th
is
o
cc
asio
n
,
co
m
p
ar
is
o
n
o
f
th
e
ac
cu
r
ac
y
o
f
th
e
r
esu
lts
o
f
th
e
R
F
m
o
d
el
was
ca
r
r
ied
o
u
t
o
n
two
d
ata
tr
ea
tm
en
ts
,
n
am
el
y
with
p
r
e
-
p
r
o
ce
s
s
in
g
a
n
d
with
o
u
t
p
r
e
-
p
r
o
ce
s
s
in
g
.
C
o
m
p
a
r
is
o
n
s
wer
e
m
ad
e
with
r
an
d
o
m
d
ata
o
f
1
,
0
0
0
an
d
5
,
0
0
0
p
o
in
ts
.
Acc
u
r
ac
y
c
o
m
p
ar
is
o
n
o
f
t
h
e
two
R
F m
o
d
els is
s
h
o
wn
in
T
ab
le
1
.
T
ab
le
1
.
Acc
u
r
ac
y
co
m
p
ar
is
o
n
o
f
R
F m
o
d
el
with
an
d
with
o
u
t p
r
ep
r
o
ce
s
s
in
g
No
Tr
a
i
n
d
a
t
a
c
o
u
n
t
Te
st
d
a
t
a
c
o
u
n
t
Te
st
sc
h
e
m
e
A
c
c
u
r
a
c
y
w
i
t
h
o
u
t
p
r
e
p
r
o
c
e
ss
(
%)
A
c
c
u
r
a
c
y
w
i
t
h
r
e
p
r
o
c
e
ss
(
%)
1
1
,
0
0
0
1
0
,
9
7
2
t
r
a
i
n
7
6
.
5
5
7
3
.
3
2
2
1
,
0
0
0
1
0
,
9
7
2
v
a
l
i
d
a
t
i
o
n
7
8
.
4
7
7
4
.
0
6
3
5
,
0
0
0
1
0
,
9
7
2
t
r
a
i
n
7
4
.
7
0
6
7
.
8
1
4
5
,
0
0
0
1
0
,
9
7
2
v
a
l
i
d
a
t
i
o
n
7
5
.
0
6
7
4
.
6
1
C
o
m
p
ar
is
o
n
s
h
o
ws th
e
m
o
d
el
th
at
u
s
es d
ata
with
o
u
t p
r
ep
r
o
c
ess
in
g
h
as h
ig
h
er
ac
cu
r
ac
y
,
th
er
ef
o
r
e
R
F
m
o
d
el
with
o
u
t
p
r
e
p
r
o
ce
s
s
in
g
will
b
e
u
s
ed
.
Featu
r
e
en
g
in
ee
r
in
g
p
r
o
ce
s
s
f
r
o
m
th
e
r
aw
d
ata
in
clu
d
es
th
e
8
f
ea
tu
r
es
d
escr
ib
ed
in
t
h
e
p
r
ev
i
o
u
s
s
ec
tio
n
.
Featu
r
e
en
g
i
n
ee
r
i
n
g
is
p
e
r
f
o
r
m
ed
o
n
all
s
p
lits
.
Sp
lit
tr
ain
with
a
to
tal
o
f
1
9
3
,
8
8
3
d
ata
p
o
in
ts
p
r
o
d
u
ce
s
2
,
6
4
0
,
5
9
0
n
ew
d
ata
p
o
in
t
s
wh
ich
ar
e
a
class
if
icatio
n
d
ataset
o
f
a
s
en
ten
ce
wh
eth
er
it
is
in
clu
d
ed
in
th
e
s
u
m
m
ar
y
o
r
n
o
t.
E
ac
h
o
f
t
h
ese
n
ew
d
ata
p
o
in
ts
is
a
n
u
m
er
i
c
al
r
ep
r
esen
tatio
n
o
f
ea
ch
s
en
ten
ce
.
Data
s
n
ip
p
ets a
r
e
s
h
o
wn
in
T
a
b
le
2
.
C
o
m
p
a
r
i
s
o
n
o
f
t
h
e
a
m
o
u
n
t
o
f
d
a
t
a
b
a
s
e
d
o
n
t
h
e
l
a
b
e
ls
i
n
T
a
b
l
e
3
s
h
o
w
s
t
h
a
t
t
h
e
d
a
ta
s
et
h
a
s
i
m
b
a
l
a
n
ce
c
o
n
d
i
t
i
o
n
s
.
T
h
e
c
o
n
d
i
t
i
o
n
c
a
n
c
a
u
s
e
a
n
o
v
e
r
f
i
t
m
o
d
e
l
t
h
a
t
m
a
k
e
s
t
h
e
c
l
as
s
i
f
i
e
r
m
o
d
e
l
l
e
a
n
t
o
wa
r
d
s
o
n
l
y
o
n
e
c
l
a
s
s
.
R
es
a
m
p
l
i
n
g
t
h
e
d
at
a
w
it
h
s
p
e
cif
i
c
t
e
c
h
n
i
q
u
es
a
l
l
o
w
s
u
s
t
o
t
r
ai
n
a
m
o
d
e
l
t
h
a
t
is
n
o
t
b
i
as
e
d
b
y
t
h
e
i
n
i
t
i
al
i
m
b
a
la
n
c
e
.
T
ab
le
2
.
Sn
ip
p
et
o
f
n
ew
d
ataset
f1
f2
f3
f4
f5
f6
f7
f8
i
d
_
t
e
x
t
l
a
b
e
l
3
.
7
1
0
.
3
5
1
0
0
0
.
6
1
1
0
.
1
2
2
6
4
0
8
1
2
.
5
6
0
.
1
4
0
.
2
9
0
0
0
.
2
5
0
.
3
3
0
2
6
4
0
8
1
4
.
5
0
.
2
5
0
.
5
6
0
0
1
0
.
2
2
0
2
6
4
0
8
0
4
.
4
7
0
.
1
9
1
0
0
0
.
4
9
1
0
2
6
4
1
0
0
3
.
7
6
0
.
1
2
0
.
5
7
0
0
.
1
2
0
.
1
9
0
.
8
7
0
.
1
2
2
6
4
1
0
1
3
.
8
3
0
.
0
9
0
.
7
0
0
.
0
9
0
.
2
6
0
.
6
7
0
2
6
4
1
0
0
T
ab
le
3
.
L
a
b
el
co
u
n
t o
n
n
ew
d
ataset
La
b
e
l
C
o
u
n
t
1
3
7
2
,
4
0
3
0
2
,
2
6
8
,
1
8
7
T
o
k
en
izatio
n
is
d
o
n
e
u
s
in
g
th
e
to
k
en
izer
f
r
o
m
th
e
I
n
d
o
b
ar
t h
u
g
g
in
g
f
ac
e
m
o
d
el.
T
o
k
en
izat
io
n
cr
ea
tes
th
e
ar
ticle
r
ep
r
esen
tatio
n
to
i
n
teg
er
v
ec
to
r
s
tr
u
ctu
r
e.
T
r
an
s
f
o
r
m
er
is
a
m
o
d
el
th
at
p
r
o
ce
s
s
es
in
teg
er
v
ec
to
r
s
tr
u
ctu
r
e
in
p
u
t
an
d
p
r
o
d
u
ce
s
v
ec
to
r
o
u
t
p
u
t
as
we
ll.
T
h
e
r
ev
er
s
e
p
r
o
ce
s
s
o
f
to
k
en
izatio
n
is
d
e
to
k
en
izatio
n
,
wh
ic
h
is
th
e
p
r
o
ce
s
s
o
f
c
o
n
v
e
r
tin
g
i
n
teg
er
v
ec
to
r
s
to
th
eir
o
r
ig
in
al
v
alu
es.
Deto
k
en
izatio
n
is
ca
r
r
i
ed
o
u
t
at
th
e
o
u
tp
u
t
o
f
th
e
tr
an
s
f
o
r
m
er
m
o
d
el.
T
o
k
en
izatio
n
is
d
o
n
e
w
h
en
B
AR
T
is
ab
o
u
t
t
o
b
e
tr
ai
n
ed
.
T
o
k
e
n
izatio
n
is
d
o
n
e
o
n
f
ea
tu
r
e
d
ata
a
n
d
lab
els.
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
Text
s
u
mma
r
iz
a
tio
n
:
B
A
R
T,
R
F
,
a
n
d
h
yb
r
id
B
A
R
T
-
R
F
a
lg
o
r
ith
m
co
mp
a
r
is
o
n
(
Mu
h
a
mma
d
A
d
ib
Z
a
mza
m)
935
I
n
itial
B
A
R
T
tr
ain
in
g
tr
ials
wer
e
co
n
d
u
cted
to
d
eter
m
in
e
th
e
r
eso
u
r
ce
s
to
b
e
u
s
ed
.
T
h
e
tr
ain
in
g
i
s
co
n
d
u
cte
d
o
n
in
f
r
astru
ctu
r
e
as
a
s
er
v
ice
(
I
aa
S)
wh
ic
h
allo
ws
u
s
er
s
to
p
er
f
o
r
m
c
o
m
p
u
tin
g
p
r
o
ce
s
s
es
u
s
in
g
th
e
latest G
PUs
.
T
h
e
m
o
r
e
tr
ain
in
g
d
ata
th
at
is
p
r
o
ce
s
s
ed
,
th
e
h
ig
h
er
th
e
tr
ain
in
g
co
s
ts
.
Data
s
elec
tio
n
was c
ar
r
ied
o
u
t
f
r
o
m
1
9
3
,
8
8
3
d
ata
t
o
m
in
i
m
ize
m
o
d
el
d
ev
elo
p
m
en
t
c
o
s
ts
.
Data
s
elec
tio
n
was
ca
r
r
ied
o
u
t
b
y
u
tili
zin
g
th
e
r
esu
lts
o
f
e
n
g
in
ee
r
i
n
g
f
ea
tu
r
es
f
o
r
p
r
ev
i
o
u
s
R
F
m
o
d
els.
T
h
e
c
o
n
d
itio
n
s
u
s
ed
f
o
r
th
e
s
elec
tio
n
ar
e
an
ar
ticle
m
u
s
t
h
av
e
f
ea
tu
r
e
v
al
u
es
th
at
ar
e
eq
u
al
to
o
r
m
o
r
e
th
a
n
th
e
av
er
ag
e
o
f
ea
ch
f
ea
tu
r
e.
1
0
9
,
5
2
2
d
ata
p
o
in
ts
wer
e
o
b
tain
ed
f
r
o
m
th
is
p
r
o
ce
s
s
.
4
.
2
.
M
o
del c
o
m
pa
riso
n
R
F
h
as
a
wea
k
n
ess
in
th
e
clas
s
if
icatio
n
o
f
im
b
alan
ce
d
d
ata.
T
h
e
class
_
weig
h
t=b
alan
ce
d
p
ar
am
eter
is
u
s
ed
at
th
e
in
itializatio
n
o
f
th
e
R
F
m
o
d
el
u
s
in
g
s
k
lear
n
p
y
th
o
n
m
o
d
u
le,
s
o
th
at
th
e
class
if
ier
m
o
d
el
p
er
f
o
r
m
s
tr
ain
in
g
with
b
alan
ce
d
d
ata
au
to
m
atica
lly
b
y
u
s
in
g
in
v
er
s
ely
p
r
o
p
o
r
tio
n
al
class
weig
h
t.
W
e
u
s
ed
h
u
g
g
in
g
f
ac
e
f
r
am
ewo
r
k
with
B
AR
T
m
o
d
e
l
ca
lled
I
n
d
o
b
a
r
t
[
2
0
]
.
T
h
e
d
ef
a
u
lt
co
n
f
ig
u
r
atio
n
is
:
1
6
b
atch
e
s
,
1
2
h
i
d
d
en
lay
er
s
(
6
en
co
d
er
s
an
d
6
d
ec
o
d
e
r
s
)
.
B
AR
T
f
in
etu
n
ed
in
7
0
ep
o
ch
s
an
d
1
1
5
0
0
0
s
tep
s
o
n
1
×
NVI
DI
A
R
T
X
A6
0
0
0
4
8
GB
.
B
ea
m
s
ea
r
ch
s
ize
o
f
5
is
u
s
ed
in
s
u
m
m
ar
y
g
en
e
r
atio
n
.
Ar
ea
u
n
d
er
r
ec
eiv
er
o
p
er
ato
r
ch
ar
ac
ter
is
tic
(
AURO
C
/AU
C
R
OC
)
is
u
s
ed
to
m
ea
s
u
r
e
clas
s
if
i
er
q
u
ality
[
2
1
]
.
T
h
e
R
OC
g
r
ap
h
is
s
h
o
w
n
in
Fig
u
r
e
5
.
A
r
a
n
d
o
m
class
if
ier
h
as
a
v
alu
e
o
f
0
.
5
a
n
d
a
p
er
f
ec
t
class
if
ier
h
as
a
v
alu
e
o
f
1
.
A
class
if
ier
is
c
o
n
s
id
er
ed
ef
f
ec
tiv
e
if
th
e
AU
R
OC
h
as
a
v
alu
e
o
f
m
o
r
e
th
a
n
0
.
5
a
n
d
p
r
e
f
er
ab
l
y
clo
s
e
to
1
.
T
h
e
AUC R
F v
alu
e
is
s
h
o
wn
to
h
a
v
e
a
v
alu
e
>
0
.
5
,
s
o
it is
co
n
s
id
er
e
d
ef
f
ec
tiv
e
as a
class
if
ier
,
ev
en
th
o
u
g
h
it is
v
er
y
clo
s
e
to
0
.
5
.
Fig
u
r
e
5
.
AUROC
ch
ar
t f
o
r
R
F m
o
d
el
T
h
e
s
u
m
m
ar
y
r
esu
lts
o
f
th
e
co
m
p
ar
is
o
n
o
f
th
e
th
r
ee
s
ce
n
ar
i
o
s
ar
e
s
h
o
wn
in
T
ab
le
4
.
B
AR
T
h
as
r
esu
lt
s
o
f
ab
s
tr
ac
tiv
e
s
u
m
m
ar
y
.
R
F
s
u
cc
ee
d
ed
in
s
u
m
m
a
r
izin
g
ex
tr
a
ctiv
e
.
R
F
×
B
A
R
T
g
iv
es
in
ter
m
ed
iate
r
esu
lts
,
b
etter
th
an
B
AR
T
b
u
t
lo
wer
i
n
s
co
r
es
th
an
R
F.
T
h
e
r
esu
lts
o
f
th
e
B
AR
T
s
u
m
m
ar
y
ca
n
ca
p
tu
r
e
th
e
o
u
tlin
e
o
f
th
e
s
u
b
ject
in
s
en
ten
ce
s
,
s
u
ch
as
"
Kep
o
lis
ian
Dae
r
ah
R
iau
"
b
ec
o
m
es
"Ka
p
o
ld
a
R
iau
".
B
A
R
T
g
en
er
ates
f
air
ly
g
o
o
d
ab
s
tr
ac
t
iv
e
s
u
m
m
ar
ies,
b
u
t
s
till
g
en
er
ates
ch
a
r
ac
ter
s
o
r
to
k
e
n
s
an
d
wo
r
d
s
th
at
is
n
o
t
in
cl
u
d
e
d
in
g
r
ea
t
d
ictio
n
ar
y
o
f
th
e
I
n
d
o
n
esian
lan
g
u
a
g
e
o
r
k
am
u
s
b
esar
b
ah
asa
I
n
d
o
n
e
s
ia
(
KB
B
I
)
ter
m
s
u
ch
as
to
u
w
et
.
At
th
e
en
d
o
f
a
s
en
ten
ce
B
AR
T
s
till
ev
o
k
es
m
ea
n
in
g
less
s
y
m
b
o
ls
s
u
ch
as
p
e
r
io
d
s
,
ex
clam
atio
n
p
o
in
ts
an
d
q
u
esti
o
n
m
ar
k
s
.
T
h
e
r
esu
lts
o
f
th
e
s
u
m
m
a
r
y
o
f
th
e
h
y
b
r
id
R
F
×
B
A
R
T
h
av
e
s
im
ilar
ch
ar
ac
ter
is
tics
to
B
AR
T
.
I
n
g
e
n
er
al,
th
e
s
tatis
tical
co
m
p
ar
is
o
n
o
f
th
e
c
o
u
n
t
o
f
wo
r
d
s
in
th
e
s
u
m
m
a
r
y
r
esu
lts
is
s
h
o
wn
in
T
a
b
le
5
.
4
.
3
.
E
v
a
lua
t
i
o
n
R
OUGE
ev
alu
atio
n
u
s
es R
OUGE
n
-
g
r
am
,
R
OUGE
lo
n
g
est c
o
m
m
o
n
s
u
b
s
eq
u
e
n
ce
(
R
L
)
an
d
R
OUGE
weig
h
ted
L
C
S
(
R
W
)
.
R
OUG
E
r
ec
all
n
-
g
r
am
s
ar
e
d
is
p
lay
e
d
with
th
e
m
etr
ics
R
1
,
R
2
to
R
n
.
R
L
co
n
s
id
er
s
th
e
lo
n
g
est
s
im
ilar
s
u
b
s
eq
u
en
ce
o
f
two
tex
ts
.
R
W
also
u
s
e
s
L
C
S
with
co
n
s
id
er
atio
n
o
f
th
e
weig
h
t
in
ea
ch
L
C
S.
E
ac
h
m
etr
ic
h
as
r
ec
all,
p
r
ec
is
io
n
,
an
d
f
-
m
ea
s
u
r
e
m
ea
s
u
r
em
en
ts
.
R
ec
all
r
ep
r
esen
ts
r
a
tio
b
etwe
en
c
o
u
n
t
o
v
er
lap
p
i
n
g
wo
r
d
o
f
r
ef
e
r
en
c
e
an
d
ca
n
d
id
ate
s
u
m
m
ar
y
o
v
er
r
ef
er
en
ce
s
u
m
m
a
r
y
.
Pre
cisi
o
n
r
ep
r
esen
ts
r
atio
b
etwe
en
co
u
n
t
o
v
e
r
lap
p
in
g
wo
r
d
o
f
r
e
f
er
en
ce
an
d
ca
n
d
id
at
e
s
u
m
m
ar
y
o
v
er
ca
n
d
id
ate
s
u
m
m
ar
y
.
F
-
m
ea
s
u
r
e
r
ep
r
esen
ts
h
ar
m
o
n
ic
r
atio
b
etwe
e
n
r
ec
all
an
d
p
r
ec
is
io
n
.
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.
1
5
,
No
.
1
,
Feb
r
u
ar
y
2
0
2
6
:
929
-
9
4
0
936
T
ab
le
4
.
Sam
p
le
s
u
m
m
ar
izatio
n
r
esu
lts
P
a
r
t
Te
x
t
(
i
n
I
n
d
o
n
e
s
i
a
n
)
Te
x
t
(
i
n
En
g
l
i
s
h
)
R
a
w
t
e
x
t
Li
p
u
t
a
n
6
.
c
o
m
,
J
a
k
a
r
t
a
:
K
e
p
o
l
i
si
a
n
D
a
e
r
a
h
Ri
a
u
b
e
r
t
e
k
a
d
m
e
m
b
e
r
a
n
t
a
s
p
e
l
a
k
u
p
e
n
y
e
l
u
n
d
u
p
a
n
k
a
y
u
y
a
n
g
k
e
r
a
p
t
e
rj
a
d
i
d
i
R
i
a
u
.
S
e
l
a
i
n
i
t
u
,
Po
l
d
a
s
e
t
e
m
p
a
t
j
u
g
a
a
k
a
n
m
e
m
b
e
ra
n
g
u
s
m
e
n
i
p
u
l
a
si
d
a
n
a
r
e
b
o
i
sa
s
i
d
a
n
i
u
r
a
n
h
a
si
l
h
u
t
a
n
.
D
e
m
i
k
i
a
n
d
i
t
e
g
a
sk
a
n
K
e
p
a
l
a
Po
l
d
a
Ri
a
u
Br
i
g
a
d
i
r
J
e
n
d
e
ra
l
P
o
l
i
si
J
o
h
n
y
Y
o
d
j
a
n
a
,
seu
s
a
i
d
i
l
a
n
t
i
k
m
e
n
j
a
d
i
K
a
p
o
l
d
a
Ri
a
u
o
l
e
h
K
e
p
a
l
a
Po
l
r
i
J
e
n
d
e
r
a
l
Po
l
i
s
i
S
u
r
o
y
o
Bi
m
a
n
t
o
ro
,
d
i
J
a
k
a
rt
a
,
b
a
r
u
-
b
a
r
u
i
n
i
.
Me
n
u
ru
t
J
o
h
n
y
,
p
e
l
a
k
u
t
i
n
d
a
k
k
r
i
m
i
n
a
l
y
a
n
g
k
e
ra
p
m
e
n
j
a
r
a
h
k
a
y
u
d
i
Ri
a
u
a
k
a
n
d
i
t
i
n
d
a
k
t
e
g
a
s
.
"
S
a
y
a
t
a
k
a
k
a
n
p
a
n
d
a
n
g
b
u
l
u
,
"
j
a
n
j
i
J
o
h
n
y
.
S
e
l
a
i
n
i
t
u
,
i
a
b
e
rt
e
k
a
d
m
e
n
y
e
l
i
d
i
k
i
d
u
g
a
a
n
m
a
n
i
p
u
l
a
si
d
a
n
a
r
e
b
o
i
s
a
s
i
d
a
n
i
u
ra
n
h
a
si
l
h
u
t
a
n
se
b
e
s
a
r
R
p
6
8
0
m
i
l
i
a
r
y
a
n
g
d
i
l
a
k
u
k
a
n
se
b
u
a
h
p
e
r
u
sa
h
a
a
n
k
a
y
u
d
i
Ri
a
u
.
S
e
m
e
n
t
a
r
a
i
t
u
,
se
l
a
i
n
m
e
l
a
n
t
i
k
J
o
h
n
y
Y
o
d
y
a
n
a
,
K
a
p
o
l
r
i
j
u
g
a
m
e
l
a
n
t
i
k
I
n
s
p
e
k
t
u
r
J
e
n
d
e
ra
l
P
o
l
i
si
Fi
rm
a
n
G
a
n
i
m
e
n
j
a
d
i
K
a
p
o
l
d
a
S
u
l
a
w
e
si
S
e
l
a
t
a
n
d
a
n
Bri
g
j
e
n
P
o
l
.
Ed
d
y
D
a
r
n
a
d
i
m
e
n
j
a
d
i
K
a
p
o
l
d
a
M
a
l
u
k
u
.
S
e
l
a
i
n
i
t
u
,
Bi
m
a
n
t
o
r
o
j
u
g
a
m
e
l
a
n
t
i
k
K
o
m
i
sa
r
i
s
Be
s
a
r
Po
l
.
T
o
t
o
k
S
o
e
n
a
rj
o
m
e
n
j
a
d
i
K
a
p
o
l
d
a
J
a
m
b
i
,
Br
i
g
j
e
n
Po
l
.
S
u
g
i
r
i
m
e
n
j
a
d
i
K
a
p
o
l
d
a
L
a
m
p
u
n
g
,
d
a
n
Br
i
g
j
e
n
Po
l
.
D
w
i
P
u
rw
a
n
t
o
m
e
n
j
a
d
i
K
a
p
o
l
d
a
Be
n
g
k
u
l
u
.
(
I
C
H
/
Ed
i
P
ri
y
o
n
o
d
a
n
A
n
d
i
Azri
l
)
.
Li
p
u
t
a
n
6
.
c
o
m,
Ja
k
a
r
t
a
:
Th
e
R
i
a
u
R
e
g
i
o
n
a
l
P
o
l
i
c
e
a
r
e
d
e
t
e
r
mi
n
e
d
t
o
c
r
a
c
k
d
o
w
n
o
n
t
i
m
b
e
r
smu
g
g
l
e
r
s
w
h
o
o
f
t
e
n
o
p
e
r
a
t
e
i
n
R
i
a
u
.
I
n
a
d
d
i
t
i
o
n
,
t
h
e
l
o
c
a
l
p
o
l
i
c
e
w
i
l
l
a
l
s
o
c
r
a
c
k
d
o
w
n
o
n
t
h
e
m
a
n
i
p
u
l
a
t
i
o
n
o
f
r
e
f
o
r
e
st
a
t
i
o
n
f
u
n
d
s
a
n
d
f
o
r
e
st
p
r
o
d
u
c
t
l
e
v
i
e
s
.
T
h
i
s
w
a
s
e
m
p
h
a
si
z
e
d
b
y
R
i
a
u
R
e
g
i
o
n
a
l
P
o
l
i
c
e
C
h
i
e
f
B
r
i
g
a
d
i
e
r
G
e
n
e
r
a
l
Jo
h
n
y
Y
o
d
j
a
n
a
a
f
t
e
r
b
e
i
n
g
i
n
a
u
g
u
r
a
t
e
d
a
s
R
i
a
u
R
e
g
i
o
n
a
l
P
o
l
i
c
e
C
h
i
e
f
b
y
N
a
t
i
o
n
a
l
P
o
l
i
c
e
C
h
i
e
f
G
e
n
e
r
a
l
S
u
r
o
y
o
B
i
ma
n
t
o
r
o
i
n
J
a
k
a
r
t
a
r
e
c
e
n
t
l
y
.
A
c
c
o
r
d
i
n
g
t
o
J
o
h
n
y
,
c
r
i
mi
n
a
l
s
w
h
o
f
r
e
q
u
e
n
t
l
y
l
o
o
t
t
i
m
b
e
r
i
n
R
i
a
u
w
i
l
l
b
e
d
e
a
l
t
w
i
t
h
sev
e
r
e
l
y
.
“
I
w
i
l
l
n
o
t
d
i
s
c
r
i
mi
n
a
t
e
,
”
Jo
h
n
y
p
r
o
mi
s
e
d
.
I
n
a
d
d
i
t
i
o
n
,
h
e
i
s
d
e
t
e
r
mi
n
e
d
t
o
i
n
v
e
s
t
i
g
a
t
e
a
l
l
e
g
a
t
i
o
n
s
o
f
m
a
n
i
p
u
l
a
t
i
o
n
o
f
r
e
f
o
r
e
s
t
a
t
i
o
n
f
u
n
d
s
a
n
d
f
o
r
e
st
p
r
o
d
u
c
t
l
e
v
i
e
s
a
mo
u
n
t
i
n
g
t
o
R
p
6
8
0
b
i
l
l
i
o
n
b
y
a
t
i
mb
e
r
c
o
m
p
a
n
y
i
n
R
i
a
u
.
M
e
a
n
w
h
i
l
e
,
i
n
a
d
d
i
t
i
o
n
t
o
i
n
a
u
g
u
r
a
t
i
n
g
J
o
h
n
y
Y
o
d
y
a
n
a
,
t
h
e
N
a
t
i
o
n
a
l
P
o
l
i
c
e
C
h
i
e
f
a
l
s
o
i
n
a
u
g
u
r
a
t
e
d
I
n
sp
e
c
t
o
r
G
e
n
e
r
a
l
F
i
r
ma
n
G
a
n
i
a
s
S
o
u
t
h
S
u
l
a
w
e
s
i
P
o
l
i
c
e
C
h
i
e
f
a
n
d
B
r
i
g
a
d
i
e
r
G
e
n
e
r
a
l
Ed
d
y
D
a
r
n
a
d
i
a
s
M
a
l
u
k
u
P
o
l
i
c
e
C
h
i
e
f
.
F
u
r
t
h
e
r
m
o
r
e
,
B
i
m
a
n
t
o
r
o
a
l
so
a
p
p
o
i
n
t
e
d
S
e
n
i
o
r
C
o
mm
i
ssi
o
n
e
r
T
o
t
o
k
S
o
e
n
a
r
j
o
a
s
t
h
e
Jamb
i
P
o
l
i
c
e
C
h
i
e
f
,
B
r
i
g
a
d
i
e
r
G
e
n
e
r
a
l
S
u
g
i
r
i
a
s
t
h
e
La
mp
u
n
g
P
o
l
i
c
e
C
h
i
e
f
,
a
n
d
B
r
i
g
a
d
i
e
r
G
e
n
e
r
a
l
D
w
i
P
u
r
w
a
n
t
o
a
s
t
h
e
B
e
n
g
k
u
l
u
P
o
l
i
c
e
C
h
i
e
f
.
(
I
C
H
/
E
d
i
P
r
i
y
o
n
o
a
n
d
A
n
d
i
A
z
r
i
l
)
.
Ex
t
r
e
f
e
r
e
n
c
e
Li
p
u
t
a
n
6
.
c
o
m,
J
a
k
a
r
t
a
:
K
e
p
o
l
i
s
i
a
n
D
a
e
r
a
h
Ri
a
u
b
e
r
t
e
k
a
d
m
e
m
b
e
r
a
n
t
a
s
p
e
l
a
k
u
p
e
n
y
e
l
u
n
d
u
p
a
n
k
a
y
u
y
a
n
g
k
e
ra
p
t
e
r
j
a
d
i
d
i
Ri
a
u
.
“
S
a
y
a
t
a
k
a
k
a
n
p
a
n
d
a
n
g
b
u
l
u
,
”
j
a
n
j
i
J
o
h
n
y
.
T
o
t
o
k
S
o
e
n
a
r
j
o
m
e
n
j
a
d
i
K
a
p
o
l
d
a
J
a
m
b
i
,
Br
i
g
j
e
n
P
o
l
.
Li
p
u
t
a
n
6
.
c
o
m,
Ja
k
a
r
t
a
:
Th
e
R
i
a
u
R
e
g
i
o
n
a
l
P
o
l
i
c
e
a
r
e
d
e
t
e
r
mi
n
e
d
t
o
c
r
a
c
k
d
o
w
n
o
n
t
i
m
b
e
r
smu
g
g
l
e
r
s
w
h
o
o
f
t
e
n
o
p
e
r
a
t
e
i
n
R
i
a
u
.
“
I
w
i
l
l
n
o
t
d
i
s
c
r
i
m
i
n
a
t
e
,
”
p
r
o
m
i
se
d
J
o
h
n
y
.
T
o
t
o
k
S
o
e
n
a
r
j
o
b
e
c
a
me
t
h
e
Ja
mb
i
R
e
g
i
o
n
a
l
P
o
l
i
c
e
C
h
i
e
f
,
B
r
i
g
a
d
i
e
r
G
e
n
e
r
a
l
P
o
l
.
A
b
s refer
e
n
c
e
K
a
p
o
l
d
a
Ri
a
u
b
a
ru
Bri
g
j
e
n
P
o
l
.
J
o
h
n
y
Y
o
d
j
a
n
a
b
e
r
t
e
k
a
d
m
e
m
b
e
r
a
n
t
a
s p
e
l
a
k
u
p
e
n
y
e
l
u
n
d
u
p
a
n
k
a
y
u
d
i
Ri
a
u
.
I
a
b
e
r
j
a
n
j
i
a
k
a
n
m
e
n
i
n
d
a
k
t
e
g
a
s
p
e
l
a
k
u
t
a
n
p
a
p
a
n
d
a
n
g
b
u
l
u
.
Th
e
n
e
w
R
i
a
u
P
o
l
i
c
e
C
h
i
e
f
,
B
r
i
g
a
d
i
e
r
G
e
n
e
r
a
l
J
o
h
n
y
Y
o
d
j
a
n
a
,
i
s
d
e
t
e
r
m
i
n
e
d
t
o
c
r
a
c
k
d
o
w
n
o
n
t
i
mb
e
r
smu
g
g
l
e
r
s
i
n
R
i
a
u
.
H
e
h
a
s
p
r
o
m
i
se
d
t
o
t
a
k
e
f
i
r
m
a
c
t
i
o
n
a
g
a
i
n
st
p
e
r
p
e
t
r
a
t
o
r
s w
i
t
h
o
u
t
e
x
c
e
p
t
i
o
n
.
B
A
R
T
k
a
p
o
l
d
a
r
i
a
u
b
r
i
g
j
e
n
p
o
l
i
s
i
j
o
h
n
y
y
o
d
j
a
n
a
b
e
rj
a
n
j
i
m
e
n
y
e
l
i
d
i
k
i
d
u
g
a
a
n
m
a
n
i
p
u
l
a
s
i
d
a
n
a
re
b
o
i
s
a
si
d
a
n
i
u
r
a
n
h
a
s
i
l
h
u
t
a
n
se
b
e
sa
r
rp
6
8
0
m
i
l
i
a
r.
p
o
l
ri
m
e
m
b
e
r
a
n
g
u
s
m
e
n
i
p
u
l
a
s
i
d
a
n
i
u
ra
n
h
u
t
a
n
.
k
a
s
u
s
i
n
i
a
k
a
n
d
i
l
a
p
o
rk
a
n
k
e
m
a
sy
a
ra
k
a
t
.
j
o
h
n
y
m
e
n
g
a
t
a
k
a
n
t
a
k
a
k
a
n
p
a
n
d
a
n
g
b
u
l
u
.
.
.
k
e
j
a
g
u
n
g
a
d
a
l
a
h
p
e
l
a
k
u
t
i
n
d
a
k
k
ri
m
i
n
a
l
.
p
o
l
ri
.
.
.
.
.
.
r
i
a
u
p
o
l
i
c
e
c
h
i
e
f
b
r
i
g
a
d
i
e
r
g
e
n
e
r
a
l
j
o
h
n
y
y
o
d
j
a
n
a
p
r
o
m
i
se
d
t
o
i
n
v
e
st
i
g
a
t
e
a
l
l
e
g
a
t
i
o
n
s o
f
man
i
p
u
l
a
t
i
o
n
o
f
r
e
f
o
r
e
st
a
t
i
o
n
f
u
n
d
s
a
n
d
f
o
r
e
s
t
r
e
v
e
n
u
e
c
o
n
t
r
i
b
u
t
i
o
n
s
a
mo
u
n
t
i
n
g
t
o
r
p
6
8
0
b
i
l
l
i
o
n
.
t
h
e
n
a
t
i
o
n
a
l
p
o
l
i
c
e
w
i
l
l
c
r
a
c
k
d
o
w
n
o
n
m
a
n
i
p
u
l
a
t
i
o
n
a
n
d
f
o
r
e
s
t
r
e
v
e
n
u
e
c
o
n
t
r
i
b
u
t
i
o
n
s.
t
h
i
s
c
a
s
e
w
i
l
l
b
e
r
e
p
o
r
t
e
d
t
o
t
h
e
p
u
b
l
i
c
.
j
o
h
n
y
s
a
i
d
h
e
w
o
u
l
d
n
o
t
d
i
s
c
r
i
m
i
n
a
t
e
.
.
.
t
h
e
a
t
t
o
r
n
e
y
g
e
n
e
r
a
l
's
o
f
f
i
c
e
i
s
t
h
e
p
e
r
p
e
t
r
a
t
o
r
o
f
c
r
i
mi
n
a
l
a
c
t
s.
t
h
e
n
a
t
i
o
n
a
l
p
o
l
i
c
e
.
.
.
RF
Li
p
u
t
a
n
6
.
c
o
m
,
J
a
k
a
r
t
a
:
K
e
p
o
l
i
si
a
n
D
a
e
r
a
h
R
i
a
u
b
e
r
t
e
k
a
d
m
e
m
b
e
r
a
n
t
a
s
p
e
l
a
k
u
p
e
n
y
e
l
u
n
d
u
p
a
n
k
a
y
u
y
a
n
g
k
e
r
a
p
t
e
rj
a
d
i
d
i
R
i
a
u
.
S
e
l
a
i
n
i
t
u
,
Po
l
d
a
s
e
t
e
m
p
a
t
j
u
g
a
a
k
a
n
m
e
m
b
e
ra
n
g
u
s
m
e
n
i
p
u
l
a
si
d
a
n
a
r
e
b
o
i
sa
s
i
d
a
n
i
u
r
a
n
h
a
s
i
l
h
u
t
a
n
.
M
e
n
u
r
u
t
J
o
h
n
y
,
p
e
l
a
k
u
t
i
n
d
a
k
k
ri
m
i
n
a
l
y
a
n
g
k
e
r
a
p
m
e
n
j
a
ra
h
k
a
y
u
d
i
Ri
a
u
a
k
a
n
d
i
t
i
n
d
a
k
t
e
g
a
s
“
S
a
y
a
t
a
k
a
k
a
n
p
a
n
d
a
n
g
b
u
l
u
,
”
j
a
n
j
i
J
o
h
n
y
Li
p
u
t
a
n
6
.
c
o
m,
Ja
k
a
r
t
a
:
Th
e
R
i
a
u
R
e
g
i
o
n
a
l
P
o
l
i
c
e
a
r
e
d
e
t
e
r
mi
n
e
d
t
o
c
r
a
c
k
d
o
w
n
o
n
t
i
m
b
e
r
smu
g
g
l
e
r
s
w
h
o
f
r
e
q
u
e
n
t
l
y
o
p
e
r
a
t
e
i
n
R
i
a
u
.
I
n
a
d
d
i
t
i
o
n
,
t
h
e
l
o
c
a
l
p
o
l
i
c
e
w
i
l
l
a
l
so
c
r
a
c
k
d
o
w
n
o
n
t
h
e
man
i
p
u
l
a
t
i
o
n
o
f
r
e
f
o
r
e
st
a
t
i
o
n
f
u
n
d
s
a
n
d
f
o
r
e
s
t
p
r
o
d
u
c
t
l
e
v
i
e
s.
A
c
c
o
r
d
i
n
g
t
o
J
o
h
n
y
,
c
r
i
mi
n
a
l
s
w
h
o
f
r
e
q
u
e
n
t
l
y
l
o
o
t
t
i
m
b
e
r
i
n
R
i
a
u
w
i
l
l
b
e
d
e
a
l
t
w
i
t
h
sev
e
r
e
l
y
.
“
I
w
i
l
l
n
o
t
d
i
s
c
r
i
mi
n
a
t
e
,
”
Jo
h
n
y
p
r
o
mi
s
e
d
.
RF
×
B
A
R
T
k
e
p
o
l
i
s
i
a
n
d
a
e
r
a
h
r
i
a
u
b
e
r
t
e
k
a
d
m
e
m
b
e
r
a
n
t
a
s
p
e
l
a
k
u
p
e
n
y
e
l
u
n
d
u
p
a
n
k
a
y
u
y
a
n
g
k
e
r
a
p
t
e
rj
a
d
i
d
i
ri
a
u
.
s
e
l
a
i
n
i
t
u
,
p
o
l
d
a
se
t
e
m
p
a
t
j
u
g
a
a
k
a
n
m
e
m
b
e
ra
n
g
u
s
m
e
n
i
p
u
l
a
si
d
a
n
a
re
b
o
i
s
a
si
d
a
n
i
u
r
a
n
h
a
s
i
l
h
u
t
a
n
.
t
n
i
a
k
a
n
m
e
n
i
n
d
a
k
t
e
g
a
s
t
o
u
w
e
t
-
n
y
a
?
`
'
k
e
t
e
r
a
n
g
a
n
j
o
h
n
y
,
p
e
n
g
e
d
a
r
a
k
a
n
t
h
e
r
i
a
u
r
e
g
i
o
n
a
l
p
o
l
i
c
e
a
r
e
d
e
t
e
r
mi
n
e
d
t
o
c
r
a
c
k
d
o
w
n
o
n
t
i
m
b
e
r
s
mu
g
g
l
e
r
s
w
h
o
f
r
e
q
u
e
n
t
l
y
o
p
e
r
a
t
e
i
n
r
i
a
u
.
i
n
a
d
d
i
t
i
o
n
,
t
h
e
l
o
c
a
l
p
o
l
i
c
e
w
i
l
l
a
l
s
o
c
r
a
c
k
d
o
w
n
o
n
t
h
e
man
i
p
u
l
a
t
i
o
n
o
f
r
e
f
o
r
e
s
t
a
t
i
o
n
f
u
n
d
s
a
n
d
f
o
r
e
st
p
r
o
d
u
c
t
l
e
v
i
e
s
.
w
i
l
l
t
h
e
t
n
i
t
a
k
e
f
i
r
m
a
c
t
i
o
n
a
g
a
i
n
st
t
h
e
smu
g
g
l
e
r
s?
a
c
c
o
r
d
i
n
g
t
o
j
o
h
n
y
,
t
h
e
d
i
st
r
i
b
u
t
o
r
s w
i
l
l
T
ab
le
5
.
Statis
tical
co
m
p
ar
is
o
n
o
f
wo
r
d
co
u
n
t
S
t
a
t
i
st
i
c
B
A
R
T
RF
RF
×
B
A
R
T
Te
st
c
o
u
n
t
1
0
9
7
2
1
0
9
7
2
1
0
9
7
2
M
e
a
n
4
7
,
9
9
6
4
,
7
6
5
0
,
7
3
S
t
d
d
e
v
3
,
9
0
2
3
,
1
3
1
6
,
8
3
M
i
n
32
7
6
2
5
%
45
48
38
5
0
%
48
62
49
7
5
%
51
79
62
M
a
x
61
1
8
3
1
3
2
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
Text
s
u
mma
r
iz
a
tio
n
:
B
A
R
T,
R
F
,
a
n
d
h
yb
r
id
B
A
R
T
-
R
F
a
lg
o
r
ith
m
co
mp
a
r
is
o
n
(
Mu
h
a
mma
d
A
d
ib
Z
a
mza
m)
937
T
h
e
ca
lcu
latio
n
o
f
th
e
R
OUGE
s
co
r
e
i
n
th
is
h
y
b
r
id
m
eth
o
d
u
s
es
v
ar
iab
les
t
h
at
m
atch
th
e
co
n
tex
t
o
f
th
e
m
o
d
el.
T
h
e
R
OUGE
ca
lcu
latio
n
f
o
r
B
AR
T
u
s
es
an
ab
s
tr
ac
t
s
u
m
m
ar
y
o
f
th
e
d
ataset
as
a
r
ef
er
e
n
ce
.
T
h
e
R
OUGE
ca
lcu
latio
n
f
o
r
R
F
u
s
es
th
e
ex
tr
ac
tiv
e
s
u
m
m
ar
y
o
f
th
e
d
ataset
as
a
r
ef
er
en
ce
.
T
h
e
R
OUGE
ca
lcu
latio
n
f
o
r
h
y
b
r
i
d
R
F
×
B
AR
T
u
s
es
th
e
ab
s
tr
ac
t
an
d
e
x
tr
ac
tiv
e
s
u
m
m
a
r
y
o
f
th
e
d
ataset
as
a
r
ef
er
e
n
ce
an
d
th
e
R
F
×
B
A
R
T
r
esu
lts
as
a
h
y
p
o
th
esis
s
u
m
m
a
r
y
.
T
h
e
h
y
b
r
i
d
ca
lcu
latio
n
(
R
F
×
B
AR
T
)
is
th
e
av
er
ag
e
o
f
t
h
e
ab
s
tr
ac
tiv
e
R
OUGE
s
co
r
e
(
R
F
×
B
A
R
T
1
)
an
d
th
e
e
x
tr
ac
tiv
e
s
co
r
e
(
R
F
×
B
AR
T
2
)
.
C
o
m
p
ar
is
o
n
o
f
th
e
R
OUGE
s
co
r
e
ev
alu
atio
n
o
f
th
e
r
esu
lts
o
f
all
m
o
d
els
i
s
s
h
o
wn
in
T
ab
le
6
.
T
h
e
co
m
p
a
r
is
o
n
o
f
th
ese
s
co
r
es
s
h
o
ws
th
at
th
e
R
F
m
o
d
el
is
th
e
m
o
d
el
with
th
e
h
ig
h
est
R
OUGE
s
co
r
e
o
f
t
h
e
m
aj
o
r
ity
o
f
m
etr
ics.
T
h
er
e
is
o
n
e
R
F
×
B
A
R
T
1
s
co
r
e
th
at
ca
n
ex
ce
ed
th
e
R
OUGE
R
F sco
r
e,
n
am
ely
th
e
s
co
r
e
o
n
th
e
R
1
p
r
ec
is
io
n
m
etr
ic
with
a
v
al
u
e
o
f
4
3
.
9
9
.
C
o
m
p
a
r
is
o
n
o
f
th
e
av
e
r
ag
e
s
co
r
es sh
o
wn
in
T
ab
le
7
.
T
ab
le
6
.
R
OUGE
s
co
r
es o
f
all
m
o
d
els
R
O
U
G
E
Ev
a
l
u
a
t
i
o
n
B
A
R
T
RF
RF
×
B
A
R
T
RF
×
B
A
R
T
1
RF
×
B
A
R
T
2
R1
F
-
mea
su
r
e
32
.
78
51
.
45
40
.
81
46
.
61
35
.
02
P
r
e
c
i
s
i
o
n
25
.
37
43
.
39
35
.
94
43
.
99
27
.
89
R
e
c
a
l
l
46
.
97
70
.
06
52
.
54
54
.
05
51
.
02
R2
F
-
mea
su
r
e
16
.
17
45
.
52
28
.
54
38
.
39
18
.
70
P
r
e
c
i
s
i
o
n
12
.
45
38
.
40
25
.
68
36
.
50
14
.
86
R
e
c
a
l
l
23
.
43
62
.
16
35
.
81
44
.
20
27
.
43
R3
F
-
mea
su
r
e
8
.
93
43
.
33
22
.
93
34
.
94
10
.
92
P
r
e
c
i
s
i
o
n
6
.
83
36
.
53
21
.
00
33
.
34
8
.
65
R
e
c
a
l
l
13
.
09
59
.
51
28
.
17
40
.
19
16
.
16
RL
F
-
mea
su
r
e
32
.
19
54
.
48
41
.
50
48
.
64
34
.
36
P
r
e
c
i
s
i
o
n
25
.
81
46
.
77
37
.
15
46
.
11
28
.
19
R
e
c
a
l
l
43
.
17
70
.
04
50
.
60
54
.
68
46
.
51
RW
F
-
mea
su
r
e
16
.
81
33
.
88
22
.
75
27
.
17
18
.
34
P
r
e
c
i
s
i
o
n
17
.
03
40
.
07
28
.
57
37
.
84
19
.
30
R
e
c
a
l
l
16
.
85
31
.
97
20
.
66
22
.
69
18
.
64
T
ab
le
7
.
Av
e
r
ag
e
R
OUGE
s
co
r
es o
f
all
m
o
d
els
Ev
a
l
u
a
t
i
o
n
B
A
R
T
RF
RF
×
B
A
R
T
F
-
mea
su
r
e
2
1
.
3
8
4
5
.
7
3
3
1
.
3
1
P
r
e
c
i
s
i
o
n
1
7
.
5
0
4
1
.
0
3
2
9
.
6
7
R
e
c
a
l
l
2
8
.
7
0
5
8
.
7
5
3
7
.
5
6
T
h
e
h
ig
h
est
R
F
s
co
r
e
lies
in
th
e
R
L
r
ec
all
m
etr
ic
with
a
v
al
u
e
o
f
7
0
.
0
6
.
T
h
e
l
o
west
R
F
s
c
o
r
e
lies
in
th
e
R
W
r
ec
all
m
etr
ic
with
a
v
alu
e
o
f
2
8
.
5
0
.
T
h
e
R
OUGE
R
F
s
co
r
e
h
as
th
e
h
ig
h
est
v
alu
e
b
ec
au
s
e
th
e
m
o
d
el
r
esu
lts
h
av
e
th
e
m
o
s
t
s
im
ilar
it
ies
with
th
e
ex
tr
ac
tiv
e
s
u
m
m
ar
y
tex
t.
T
h
e
R
OUGE
B
AR
T
s
co
r
e
h
as
th
e
lo
west
av
er
ag
e,
m
ax
im
u
m
a
n
d
m
in
im
u
m
v
al
u
es.
T
h
e
h
ig
h
est R
OUGE
B
AR
T
s
co
r
e
lies
in
th
e
R
1
r
ec
all
m
etr
ic
with
a
v
alu
e
o
f
4
6
.
9
7
.
T
h
e
lo
west
R
OUGE
B
AR
T
s
co
r
e
lies
in
th
e
R
3
p
r
ec
is
io
n
m
etr
ic
with
a
v
alu
e
o
f
6
.
8
3
.
T
h
e
R
OUGE
R
F
×
B
A
R
T
s
co
r
e
h
as
s
u
f
f
icien
t
av
e
r
ag
e,
m
ax
im
u
m
an
d
m
in
im
u
m
v
alu
es.
T
h
e
h
ig
h
est
R
OUGE
RF
×
B
AR
T
s
co
r
e
lies
in
th
e
R
1
r
ec
all
m
etr
ic
with
a
v
alu
e
o
f
5
2
.
5
4
.
T
h
e
lo
west
R
OUGE
R
F
×
B
A
R
T
s
co
r
e
lie
s
in
th
e
R
W
r
ec
all
m
etr
ic
with
a
v
alu
e
o
f
2
0
.
6
6
.
Fig
u
r
e
6
p
r
esen
ts
a
co
m
p
a
r
is
o
n
g
r
ap
h
o
f
th
e
a
v
er
ag
e
R
OUGE
F
-
m
ea
s
u
r
e
s
co
r
es
f
o
r
th
r
ee
s
u
m
m
ar
izatio
n
tech
n
iq
u
es:
R
F,
B
AR
T
,
an
d
R
F
×
B
AR
T
.
No
tab
ly
,
R
F
co
n
s
is
ten
tly
ac
h
ie
v
e
s
th
e
h
ig
h
est
s
co
r
es
ac
r
o
s
s
all
m
etr
ics.
T
h
is
ca
n
b
e
attr
ib
u
ted
to
R
F f
o
cu
s
o
n
r
etr
i
ev
in
g
s
en
ten
ce
s
m
o
s
t similar
to
th
e
r
e
f
er
en
ce
te
x
t
th
r
o
u
g
h
a
class
if
icatio
n
p
r
o
ce
s
s
.
W
h
ile
R
F
×
B
A
R
T
p
er
f
o
r
m
s
b
etter
th
an
B
AR
T
,
it
s
till
f
alls
s
h
o
r
t
o
f
R
F
s
tr
o
n
g
s
im
ilar
ity
to
th
e
r
ef
er
e
n
ce
.
A
co
m
p
ar
is
o
n
o
f
th
e
R
F×B
A
R
T
s
co
r
es
with
o
th
er
s
tu
d
ies
s
h
o
wn
at
T
ab
le
8
.
Mu
ltip
le
s
tu
d
ies
h
av
e
s
h
o
wn
th
at
R
F×B
AR
T
o
u
tp
er
f
o
r
m
s
B
AR
T
o
n
all
R
OUGE
m
etr
ics.
So
m
e
s
tu
d
ies h
a
v
e
a
h
i
g
h
er
R
OUGE
s
co
r
e
s
u
ch
as
in
[
4
]
,
[
2
2
]
o
n
th
e
R
1
s
co
r
e.
T
h
e
B
AR
T
m
o
d
el
th
at
s
u
r
p
ass
es
all
R
1
,
R
2
,
an
d
R
L
R
F×B
A
R
T
s
co
r
es
is
th
e
B
AR
T
[
2
3
]
.
T
h
e
R
F×B
AR
T
h
y
b
r
id
m
o
d
el
is
well
-
s
u
ited
f
o
r
d
ep
lo
y
m
en
t
in
p
r
ac
tical
en
v
ir
o
n
m
e
n
ts
r
eq
u
ir
i
n
g
h
ig
h
ly
ac
cu
r
ate
an
d
ef
f
icien
t
s
u
m
m
ar
i
za
tio
n
.
Po
ten
tial
r
ea
l
-
wo
r
ld
ap
p
licatio
n
s
in
clu
d
e
i)
lar
g
e
-
s
ca
le
n
ews
ag
g
r
eg
atio
n
,
wh
er
e
th
e
two
-
s
tag
e
p
r
o
ce
s
s
allo
ws
th
e
R
F
co
m
p
o
n
en
t
to
q
u
ick
ly
f
ilter
k
ey
in
f
o
r
m
atio
n
f
o
r
tim
ely
ab
s
tr
ac
tiv
e
o
u
tp
u
t;
an
d
ii)
s
u
m
m
ar
izatio
n
o
f
co
m
p
lex
leg
al
o
r
m
e
d
ical
d
o
cu
m
e
n
ts
,
wh
er
e
th
e
i
n
h
er
e
n
t
p
r
ec
is
io
n
o
f
th
e
ex
tr
ac
tiv
e
R
F
co
m
p
o
n
en
t
en
s
u
r
es
th
at
cr
itical
clau
s
es
an
d
f
ac
ts
ar
e
p
r
io
r
itized
f
o
r
i
n
clu
s
io
n
in
th
e
f
in
al
s
u
m
m
ar
y
,
m
a
k
in
g
t
h
e
r
esu
ltin
g
d
o
cu
m
en
t b
o
th
co
n
cise a
n
d
l
eg
ally
tr
ac
ea
b
le.
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.
1
5
,
No
.
1
,
Feb
r
u
ar
y
2
0
2
6
:
929
-
9
4
0
938
Fig
u
r
e
6
.
C
o
m
p
a
r
is
o
n
ch
a
r
t o
f
av
er
ag
e
F
-
m
ea
s
u
r
e
R
OUGE
R
F,
B
A
R
T
d
an
R
F
×
B
AR
T
s
c
o
r
es
T
ab
le
8
.
C
o
m
p
a
r
is
o
n
o
f
R
OUGE
s
co
r
es o
f
R
F
×
B
A
R
T
with
p
r
ev
io
u
s
B
AR
T
m
o
d
el
R
e
se
a
r
c
h
e
r
M
e
t
h
o
d
s
R1
R2
RL
U
p
a
d
h
y
a
y
e
t
a
l
.
[
2
4
]
B
A
R
T
3
9
.
2
1
9
.
0
9
3
7
.
9
4
Le
w
i
s
e
t
a
l
.
[
4
]
B
A
R
T
4
4
.
1
6
2
1
.
2
8
4
0
.
9
D
e
Y
o
u
n
g
e
t
a
l
.
[
2
5
]
B
A
R
T
2
7
.
5
6
9
.
4
2
0
.
8
Ed
d
i
n
e
e
t
a
l
.
[
2
2
]
B
A
R
T
4
2
.
4
2
8
.
8
4
0
.
3
K
o
n
d
a
d
a
d
i
e
t
a
l
.
[
2
3
]
B
A
R
T
6
0
.
5
1
4
8
.
1
4
5
7
.
6
5
Th
i
s
w
o
r
k
RF
×
B
A
R
T
4
0
.
8
1
2
8
.
5
4
4
1
.
4
9
5.
CO
NCLU
SI
O
N
R
e
s
e
a
r
c
h
r
e
v
e
a
l
s
a
s
i
g
n
if
i
c
an
t
d
if
f
er
e
n
ce
i
n
R
O
U
G
E
s
co
r
e
b
e
t
we
e
n
th
e
p
e
r
f
o
r
m
an
c
e
o
f
R
F
a
n
d
B
A
R
T
o
n
A
T
S
t
a
s
k
s
.
R
O
U
G
E
F
1
-
s
co
r
e
s
R
1
,
R
2
,
a
n
d
R
L
m
o
d
e
l
o
f
t
h
e
R
F
a
r
e
5
1
.
4
5
,
4
5
.
5
2
,
a
n
d
5
4
.
5
8
.
T
h
e
B
A
R
T
s
c
o
r
e
s
f
o
r
R
1
,
R
2
,
a
n
d
R
L
f
o
r
t
h
e
R
F
m
o
d
e
l
w
er
e
3
2
.
7
8
,
1
6
.
1
7
,
a
n
d
3
2
.
1
9
.
T
h
e
av
e
r
ag
e
R
O
U
G
E
F
-
m
e
a
s
u
r
e
B
A
R
T
s
c
o
r
e
i
s
4
5
.
7
3
.
T
h
e
a
v
er
a
g
e
R
O
U
G
E
F
-
m
e
a
s
u
r
e
B
A
R
T
s
c
o
r
e
i
s
2
1
.
3
8
.
R
F
h
a
s
t
h
e
h
ig
h
e
s
t
a
v
e
r
ag
e
s
co
r
e.
R
F
e
x
c
e
l
s
a
t
e
x
t
r
a
c
t
iv
e
s
u
m
m
a
r
i
z
a
t
io
n
,
wh
e
r
e
a
s
B
A
R
T
ca
n
c
r
e
a
t
e
s
u
m
m
a
r
i
e
s
th
a
t
ex
p
r
e
s
s
i
d
e
a
s
in
n
e
w
w
ay
s
(
a
b
s
t
r
a
c
t
iv
e
s
u
m
m
ar
i
z
a
t
io
n
)
.
Ho
w
e
v
e
r
,
B
A
R
T
-
g
en
e
r
a
t
e
d
t
e
x
t
m
ay
i
n
c
lu
d
e
w
o
r
d
s
o
r
p
h
r
a
s
e
s
n
o
t
f
o
u
n
d
i
n
th
e
K
B
B
I
.
T
h
e
c
o
m
b
in
e
d
R
F
×
B
A
R
T
ap
p
r
o
a
ch
,
wh
i
l
e
o
u
t
s
c
o
r
i
n
g
B
A
R
T
o
n
t
h
e
R
O
U
G
E
m
e
t
r
i
c
,
f
a
l
l
s
s
h
o
r
t
o
f
R
F
p
e
r
f
o
r
m
an
c
e.
T
h
e
R
O
U
G
E
F
-
m
e
a
s
u
r
e
RF
×
B
A
R
T
a
v
e
r
a
g
e
s
c
o
r
e
i
s
3
1
.
3
1
.
RF
×
B
A
R
T
h
a
s
a
m
o
d
er
a
t
e
s
c
o
r
e
.
T
h
i
s
s
c
o
r
e
i
n
c
r
e
a
s
e
s
c
o
m
p
a
r
ed
to
t
h
e
B
A
R
T
s
c
o
r
e
i
n
d
ep
e
n
d
e
n
t
ly
.
RF
×
B
A
R
T
c
a
n
b
e
a
n
e
f
f
e
c
t
i
v
e
a
l
t
er
n
a
t
iv
e
t
o
t
h
e
h
y
b
r
i
d
m
e
th
o
d
ap
p
r
o
ac
h
.
T
h
e
i
n
t
eg
r
a
t
io
n
o
f
s
ev
e
r
a
l
m
e
th
o
d
s
i
n
th
e
c
o
n
c
e
p
t
o
f
h
y
b
r
i
d
s
u
m
m
a
r
i
z
a
t
io
n
i
s
s
t
i
l
l a
t t
h
e
d
e
v
e
lo
p
m
e
n
t s
t
a
g
e.
I
n
f
u
t
u
r
e
s
tu
d
i
es
,
i
t
i
s
e
x
p
e
c
t
ed
t
h
a
t
r
e
s
e
a
r
ch
e
r
s
to
o
f
f
er
m
o
r
e
d
iv
e
r
s
e
h
y
b
r
i
d
ap
p
r
o
a
c
h
/
ar
c
h
it
e
ct
u
r
e
th
a
t
i
s
m
o
r
e
r
o
b
u
s
t
t
h
an
j
u
s
t
r
e
ly
i
n
g
o
n
t
h
e
o
u
t
p
u
t
o
f
e
x
tr
a
c
t
i
v
e
m
e
t
h
o
d
s
.
A
k
ey
b
en
e
f
i
t
o
f
t
h
e
R
F
×
B
A
R
T
a
r
ch
i
t
e
c
tu
r
e
i
s
i
t
s
r
o
b
u
s
t
a
b
i
l
i
ty
t
o
b
a
l
a
n
c
e
i
n
t
e
r
p
r
e
t
ab
i
l
i
t
y
a
n
d
f
lu
e
n
c
y
.
T
h
e
R
F
c
o
m
p
o
n
en
t
o
f
f
er
s
a
h
i
g
h
d
eg
r
ee
o
f
i
n
t
er
p
r
e
t
ab
i
l
i
ty
a
n
d
co
n
tr
o
l
;
i
t
s
f
e
a
t
u
r
e
-
d
r
iv
e
n
d
e
c
i
s
i
o
n
-
m
a
k
in
g
(
e
.
g
.
,
ex
p
l
i
c
i
t
ly
w
e
ig
h
t
in
g
f
e
a
t
u
r
e
s
l
i
k
e
s
en
t
e
n
c
e
p
o
s
i
t
i
o
n
a
n
d
k
ey
p
h
r
a
s
e
s
)
m
a
k
e
s
th
e
e
x
tr
a
c
t
iv
e
s
e
l
ec
t
i
o
n
p
r
o
c
e
s
s
t
r
an
s
p
ar
e
n
t
a
n
d
tr
ac
e
a
b
l
e.
C
o
n
v
er
s
e
l
y
,
t
h
e
B
A
R
T
m
o
d
e
l
u
t
i
l
i
ze
s
i
t
s
p
o
w
er
f
u
l
n
eu
r
a
l
l
a
n
g
u
ag
e
g
en
e
r
a
t
i
o
n
c
ap
a
b
i
l
i
t
i
e
s
t
o
e
n
s
u
r
e
t
h
e
f
in
a
l
s
u
m
m
ar
y
i
s
h
i
g
h
l
y
f
lu
e
n
t
a
n
d
c
o
h
e
r
e
n
t,
r
e
s
u
l
t
i
n
g
in
h
u
m
a
n
-
r
ea
d
ab
l
e
a
n
d
g
r
am
m
a
t
i
ca
l
l
y
c
o
r
r
ec
t
p
r
o
s
e.
T
h
i
s
s
u
c
c
e
s
s
f
u
l
i
n
te
g
r
a
t
i
o
n
m
e
r
g
e
s
th
e
s
t
r
e
n
g
th
s
o
f
s
y
m
b
o
l
i
c
(
R
F
)
a
n
d
n
eu
r
a
l
(
B
A
R
T
)
te
c
h
n
i
q
u
e
s
.
F
UNDING
I
NF
O
R
M
A
T
I
O
N
Fu
n
d
in
g
s
o
u
r
ce
o
f
th
is
r
esear
c
h
h
as
b
ee
n
en
ab
led
b
y
th
e
r
es
ea
r
ch
er
s
th
em
s
elv
es.
Usag
e
o
f
r
eso
u
r
ce
s
th
at
r
eq
u
ir
e
f
u
n
d
ar
e
in
cl
u
d
in
g
v
ir
tu
al
m
ac
h
in
e
an
d
r
em
o
te
G
PU.
Evaluation Warning : The document was created with Spire.PDF for Python.