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.
14
,
No
.
1
,
Feb
r
u
ar
y
2
0
2
5
,
p
p
.
680
~
691
I
SS
N:
2
2
5
2
-
8
9
3
8
,
DOI
: 1
0
.
1
1
5
9
1
/ijai.v
14
.i
1
.
p
p
6
8
0
-
6
9
1
680
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
a
i
.
ia
esco
r
e.
co
m
To
wa
rds eff
ici
ent
knowledg
e e
x
t
ra
ction
:
Na
tural la
n
g
ua
g
e
pro
cess
ing
-
ba
sed
sum
ma
riza
tion o
f
research pa
pe
r
introductio
ns
Nik
it
a
Cha
ud
ha
ri
1
,
Dee
pa
li Vo
ra
1
,
P
a
y
a
l K
a
da
m
2
,
Va
is
ha
li
K
ha
irna
r
3
,
Sh
rut
i P
a
t
il
4
,
K
et
a
n K
o
t
ec
ha
4
1
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
En
g
i
n
e
e
r
i
n
g
,
S
y
mb
i
o
s
i
s I
n
st
i
t
u
t
e
o
f
T
e
c
h
n
o
l
o
g
y
,
S
y
m
b
i
o
s
i
s I
n
t
e
r
n
a
t
i
o
n
a
l
(
D
e
e
med
U
n
i
v
e
r
si
t
y
)
,
P
u
n
e
,
I
n
d
i
a
2
D
e
p
a
r
t
m
e
n
t
o
f
E
l
e
c
t
r
o
n
i
c
s
a
n
d
T
e
l
e
c
o
mm
u
n
i
c
a
t
i
o
n
E
n
g
i
n
e
e
r
i
n
g
,
B
h
a
r
a
t
i
V
i
d
y
a
p
e
e
t
h
(
D
e
e
m
e
d
t
o
b
e
U
n
i
v
e
r
s
i
t
y
)
C
o
l
l
e
g
e
o
f
E
n
g
i
n
e
e
r
i
n
g
,
P
u
n
e
,
I
n
d
i
a
3
D
e
p
a
r
t
me
n
t
o
f
I
n
f
o
r
mat
i
o
n
Te
c
h
n
o
l
o
g
y
,
T
e
r
n
a
C
o
l
l
e
g
e
o
f
E
n
g
i
n
e
e
r
i
n
g
,
N
a
v
i
M
u
m
b
a
i
,
I
n
d
i
a
4
S
y
m
b
i
o
si
s
C
e
n
t
r
e
f
o
r
A
p
p
l
i
e
d
A
r
t
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
(
S
C
A
A
I
)
,
S
y
mb
i
o
s
i
s I
n
st
i
t
u
t
e
o
f
T
e
c
h
n
o
l
o
g
y
P
u
n
e
C
a
mp
u
s,
S
y
mb
i
o
si
s I
n
t
e
r
n
a
t
i
o
n
a
l
(
D
e
e
me
d
U
n
i
v
e
r
si
t
y
)
,
P
u
ne
,
I
n
d
i
a
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
J
an
22
,
2
0
2
4
R
ev
is
ed
J
u
l
10
,
2
0
2
4
Acc
ep
ted
J
u
l
26
,
2
0
2
4
Ac
a
d
e
m
ic
a
n
d
re
se
a
rc
h
p
a
p
e
rs
se
rv
e
a
s
v
a
lu
a
b
le
p
latf
o
rm
s
f
o
r
d
iss
e
m
in
a
ti
n
g
e
x
p
e
rti
se
a
n
d
d
isc
o
v
e
ries
t
o
d
i
v
e
rse
a
u
d
ien
c
e
s.
T
h
e
g
r
o
win
g
v
o
l
u
m
e
o
f
a
c
a
d
e
m
ic
p
a
p
e
rs,
with
n
e
a
rly
7
m
il
li
o
n
n
e
w
p
u
b
li
c
a
ti
o
n
s
a
n
n
u
a
ll
y
,
p
re
se
n
ts
a
fo
rm
id
a
b
le
c
h
a
ll
e
n
g
e
fo
r
stu
d
e
n
t
s
a
n
d
re
se
a
rc
h
e
rs
a
li
k
e
.
Co
n
se
q
u
e
n
tl
y
,
t
h
e
d
e
v
e
lo
p
m
e
n
t
o
f
re
se
a
rc
h
p
a
p
e
r
su
m
m
a
riza
ti
o
n
to
o
ls
h
a
s
b
e
c
o
m
e
c
ru
c
ial
to
d
isti
ll
i
n
g
c
r
u
c
ial
in
si
g
h
ts
e
fficie
n
t
ly
.
T
h
is
stu
d
y
e
x
a
m
in
e
s
th
e
e
ffe
c
ti
v
e
n
e
ss
o
f
p
re
-
train
e
d
m
o
d
e
ls
li
k
e
tex
t
-
to
-
t
e
x
t
tran
sfe
r
tran
sf
o
rm
e
r
(
T5
)
,
b
i
d
irec
ti
o
n
a
l
e
n
c
o
d
e
r
re
p
re
se
n
tatio
n
s
fr
o
m
tra
n
sfo
rm
e
rs
(
BERT
)
,
b
id
irec
ti
o
n
a
l
a
n
d
a
u
to
-
re
g
re
ss
iv
e
tran
sfo
rm
e
r
(
BART
)
,
a
n
d
p
re
-
train
i
n
g
with
e
x
trac
ted
g
a
p
-
se
n
ten
c
e
s
fo
r
a
b
stra
c
ti
v
e
su
m
m
a
riza
ti
o
n
(
P
EG
ASUS
)
o
n
re
se
a
rc
h
p
a
p
e
rs,
i
n
t
ro
d
u
c
in
g
a
n
o
v
e
l
h
y
b
ri
d
m
o
d
e
l
m
e
rg
i
n
g
e
x
trac
ti
v
e
a
n
d
a
b
stra
c
ti
v
e
tec
h
n
iq
u
e
s.
Co
m
p
a
ra
ti
v
e
a
n
a
ly
sis
o
f
su
m
m
a
ries
,
re
c
a
ll
-
o
rien
ted
u
n
d
e
rstu
d
y
f
o
r
g
isti
n
g
e
v
a
lu
a
ti
o
n
(
ROU
G
E
)
a
n
d
b
il
i
n
g
u
a
l
e
v
a
lu
a
ti
o
n
u
n
d
e
rstu
d
y
(
BLE
U
)
sc
o
re
e
v
a
lu
a
ti
o
n
s
a
n
d
a
u
t
h
o
r
e
v
a
lu
a
ti
o
n
h
e
l
p
e
v
a
l
u
a
te
th
e
q
u
a
l
it
y
a
n
d
a
c
c
u
ra
c
y
o
f
th
e
g
e
n
e
ra
ted
su
m
m
a
ries
.
Th
is
a
d
v
a
n
c
e
m
e
n
t
c
o
n
tri
b
u
tes
t
o
e
n
h
a
n
c
in
g
th
e
a
c
c
e
s
sib
il
it
y
a
n
d
e
fficie
n
c
y
o
f
a
ss
imilatin
g
c
o
m
p
le
x
a
c
a
d
e
m
ic
c
o
n
ten
t,
e
m
p
h
a
siz
in
g
t
h
e
imp
o
rta
n
c
e
o
f
a
d
v
a
n
c
e
d
su
m
m
a
riza
ti
o
n
to
o
ls
in
p
ro
m
o
ti
n
g
th
e
a
c
c
e
ss
ib
il
it
y
o
f
a
c
a
d
e
m
ic
k
n
o
wle
d
g
e
.
K
ey
w
o
r
d
s
:
B
AR
T
B
E
R
T
Natu
r
al
lan
g
u
ag
e
p
r
o
ce
s
s
in
g
PEGA
SU
S
T
ex
t su
m
m
ar
izatio
n
T
e
x
t
-
to
-
t
e
x
t
t
r
a
n
s
f
e
r
t
r
a
n
s
f
o
r
m
e
r
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
:
Dee
p
ali
Vo
r
a
Dep
ar
tm
en
t
o
f
C
o
m
p
u
ter
Scie
n
ce
E
n
g
in
ee
r
in
g
,
Sy
m
b
io
s
is
I
n
s
titu
te
o
f
T
ec
h
n
o
lo
g
y
Sy
m
b
io
s
is
I
n
ter
n
atio
n
al
(
Dee
m
ed
Un
iv
er
s
ity
)
Nea
r
L
u
p
in
R
esear
ch
Par
k
,
Gr
am
: L
av
ale,
T
al:
Mu
ls
h
i,
Pu
n
e
,
4
1
2
1
1
5
,
I
n
d
ia
E
m
ail:
d
ee
p
ali.
v
o
r
a@
s
itp
u
n
e.
ed
u
.
in
1.
I
NT
RO
D
UCT
I
O
N
T
h
e
f
ir
s
t
s
et
o
f
s
ch
o
lar
ly
liter
at
u
r
e
d
ates
to
th
e
1
7
th
ce
n
tu
r
y
[
1
]
,
b
eg
in
n
in
g
with
th
e
J
o
u
r
n
al
d
es
s
ça
v
an
s
[
2
]
a
n
d
t
h
e
Ph
ilo
s
o
p
h
ical
T
r
a
n
s
ac
tio
n
s
o
f
t
h
e
R
o
y
al
So
ciet
y
[
3
]
in
1
6
6
5
.
B
y
2
0
1
9
,
Fire
an
d
Gu
estrin
[
4
]
an
aly
ze
d
o
v
e
r
1
2
0
m
illi
o
n
p
ap
er
s
,
esti
m
atin
g
y
ea
r
ly
p
u
b
l
icatio
n
s
at
o
v
er
7
m
illi
o
n
,
in
clu
d
in
g
1
.
8
m
illi
o
n
s
cien
tific
p
ap
er
s
with
m
o
r
e
th
a
n
f
iv
e
r
ef
e
r
en
ce
s
.
R
esear
ch
er
s
f
ac
e
s
ig
n
if
ican
t
ch
allen
g
es
i
n
k
ee
p
in
g
u
p
with
th
e
in
cr
ea
s
in
g
n
u
m
b
er
an
d
c
o
m
p
l
ex
ity
o
f
n
ew
s
ch
o
lar
ly
p
a
p
er
s
,
m
ak
in
g
it
d
if
f
icu
lt
to
s
u
r
v
e
y
e
x
is
tin
g
liter
atu
r
e
o
r
m
o
v
e
t
o
n
ew
r
esear
c
h
ar
ea
s
e
f
f
icien
tly
.
A
u
to
m
atic
s
u
m
m
a
r
i
za
tio
n
,
f
i
r
s
t
p
r
o
p
o
s
ed
b
y
L
u
h
n
in
th
e
1
9
5
0
s
with
th
e
h
eu
r
is
tic
m
eth
o
d
,
id
en
tifi
es
k
ey
s
en
ten
ce
s
b
y
th
eir
h
ig
h
-
f
r
eq
u
en
cy
wo
r
d
s
[
5
]
–
[
7
]
.
A
d
v
an
ce
m
e
n
ts
in
th
e
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
To
w
a
r
d
s
efficien
t k
n
o
w
led
g
e
ex
tr
a
ctio
n
:
N
a
tu
r
a
l la
n
g
u
a
g
e
p
r
o
ce
s
s
in
g
-
b
a
s
ed
.
.
.
(
N
ikita
C
h
a
u
d
h
a
r
i)
681
1
9
8
0
s
an
d
1
9
9
0
s
in
tr
o
d
u
ce
d
n
atu
r
al
lan
g
u
a
g
e
p
r
o
ce
s
s
in
g
tech
n
iq
u
es
to
im
p
r
o
v
e
th
e
u
n
d
er
s
tan
d
i
n
g
an
d
in
ter
p
r
etatio
n
o
f
s
en
ten
ce
s
ig
n
if
ican
ce
in
s
u
m
m
ar
izatio
n
[
8
]
.
L
o
n
g
d
o
cu
m
e
n
ts
,
u
n
lik
e
s
h
o
r
ter
o
n
es,
ar
e
ty
p
ically
s
tr
u
ctu
r
e
d
in
to
s
ec
tio
n
s
to
aid
co
m
p
r
eh
en
s
io
n
[
9
]
,
an
d
as
au
to
m
atic
tex
t
s
u
m
m
ar
izatio
n
g
ain
s
p
o
p
u
lar
ity
,
th
e
r
e
'
s
a
r
i
s
in
g
n
ee
d
to
s
u
m
m
ar
ize
ex
ten
s
iv
e
r
esear
ch
p
ap
er
s
d
u
e
to
th
e
tim
e
-
in
ten
s
iv
e
n
atu
r
e
o
f
p
r
o
ce
s
s
in
g
in
f
o
r
m
atio
n
[
1
0
]
.
Ho
wev
er
,
ch
alle
n
g
es
s
u
ch
as
to
p
ic
id
en
tific
atio
n
an
d
m
is
in
ter
p
r
et
atio
n
p
er
s
is
t
in
lo
n
g
d
o
c
u
m
en
t
s
u
m
m
ar
izatio
n
[
1
1
]
.
Nan
et
a
l
.
[
1
2
]
d
e
m
o
n
s
tr
ated
th
at
ev
alu
atin
g
ca
n
d
i
d
ate
s
u
m
m
ar
ies
f
r
o
m
th
e
c
o
n
v
o
lu
tio
n
al
n
eu
r
al
n
etwo
r
k
s
-
d
aily
m
ail
(
C
NN
-
DM
)
d
ataset
o
n
a
s
in
g
le
NVI
DI
A
V1
0
0
GPU
tak
es
o
v
er
f
o
u
r
d
a
y
s
.
T
h
is
s
tu
d
y
ev
al
u
ates
th
e
ef
f
ec
tiv
en
ess
o
f
p
r
e
-
tr
ain
e
d
tr
an
s
f
o
r
m
er
s
—
T
5
[
1
3
]
,
p
r
e
-
tr
ain
in
g
with
ex
tr
ac
ted
g
ap
-
s
en
ten
ce
s
f
o
r
ab
s
tr
ac
tiv
e
s
u
m
m
ar
i
za
tio
n
(
PEGA
SUS)
[
1
4
]
,
b
id
ir
ec
tio
n
al
en
c
o
d
er
r
ep
r
esen
tatio
n
s
f
r
o
m
tr
a
n
s
f
o
r
m
er
s
(
B
E
R
T
)
[
1
5
]
,
an
d
b
i
d
ir
ec
tio
n
al
an
d
au
to
-
r
eg
r
ess
iv
e
tr
an
s
f
o
r
m
e
r
(
B
AR
T
)
[
1
6
]
—
an
d
ex
p
lo
r
es
a
h
y
b
r
id
ap
p
r
o
ac
h
co
m
b
in
in
g
ex
tr
ac
tiv
e
an
d
ab
s
tr
ac
tiv
e
s
u
m
m
ar
izatio
n
to
en
h
a
n
ce
ac
c
u
r
ac
y
in
s
u
m
m
ar
izin
g
co
m
p
le
x
d
o
c
u
m
en
ts
lik
e
ac
a
d
em
ic
p
a
p
er
s
.
T
h
e
a
p
p
r
o
ac
h
b
len
d
s
tech
n
iq
u
es
f
r
o
m
B
E
R
T
an
d
PEGA
SUS
to
cr
ea
te
a
cu
s
to
m
ized
h
y
b
r
id
m
o
d
el
tailo
r
ed
f
o
r
r
esear
ch
p
a
p
er
in
tr
o
d
u
ctio
n
s
,
aim
in
g
to
s
tr
ik
e
a
b
alan
ce
b
etwe
en
ex
tr
ac
tiv
e
an
d
ab
s
tr
ac
tiv
e
m
eth
o
d
s
.
T
ex
t
s
u
m
m
ar
izatio
n
tech
n
i
q
u
e
s
,
ex
ten
s
iv
ely
s
tu
d
ied
in
n
atu
r
a
l
lan
g
u
ag
e
p
r
o
ce
s
s
in
g
an
d
co
m
p
u
tatio
n
al
lin
g
u
is
tics
,
co
v
er
b
o
th
ex
tr
ac
ti
v
e
an
d
ab
s
tr
ac
tiv
e
m
eth
o
d
s
[
1
7
]
.
Fattah
a
n
d
R
en
[
1
8
]
d
is
cu
s
s
es
ch
allen
g
es
an
d
ad
v
an
ce
m
e
n
ts
in
au
t
o
m
atic
t
ex
t
s
u
m
m
ar
izatio
n
.
Mo
r
e
n
o
[
5
]
em
p
h
asizes
th
e
im
p
o
r
tan
c
e
o
f
s
u
m
m
ar
izatio
n
ac
r
o
s
s
d
iv
er
s
e
d
o
m
ain
s
.
Den
o
m
m
e
et
a
l.
[
1
9
]
ca
te
g
o
r
ize
s
u
m
m
ar
y
ty
p
es a
n
d
p
r
o
v
i
d
e
illu
s
tr
ativ
e
ex
am
p
les.
I
n
s
p
ec
ialized
d
o
m
ain
s
,
r
esea
r
ch
er
s
h
a
v
e
a
d
ap
ted
s
u
m
m
ar
i
za
tio
n
tech
n
iq
u
es
to
s
p
ec
if
ic
co
n
tex
ts
.
Gh
o
s
h
et
a
l.
[
2
0
]
f
o
cu
s
o
n
I
n
d
ian
leg
al
tex
t
s
u
m
m
a
r
izatio
n
u
s
in
g
tex
t
n
o
r
m
aliza
tio
n
a
p
p
r
o
ac
h
es.
J
aa
f
ar
an
d
B
o
u
zo
u
b
aa
[
2
1
]
p
r
o
p
o
s
e
a
h
y
b
r
id
a
p
p
r
o
ac
h
f
o
r
ab
s
tr
ac
tiv
e
s
u
m
m
ar
izatio
n
.
Gu
p
ta
et
a
l
.
[
2
2
]
e
x
p
lo
r
e
tr
an
s
f
o
r
m
er
-
b
ased
m
o
d
els f
o
r
au
to
m
ated
n
ews su
m
m
ar
izatio
n
.
Hy
b
r
id
ap
p
r
o
ac
h
es
co
m
b
in
i
n
g
ex
tr
ac
tiv
e
an
d
a
b
s
tr
ac
tiv
e
m
eth
o
d
s
h
av
e
g
ai
n
ed
p
o
p
u
lar
ity
.
Fab
b
r
izio
et
a
l
.
[
2
3
]
f
o
c
u
s
ed
o
n
o
p
in
i
o
n
ex
tr
ac
tio
n
in
m
u
lti
-
d
o
cu
m
e
n
t
s
u
m
m
ar
izatio
n
,
wh
ile
Fattah
[
2
4
]
p
r
o
p
o
s
ed
a
d
iv
e
r
s
e
m
ac
h
in
e
le
ar
n
in
g
m
o
d
el
f
o
r
e
n
h
an
ce
d
s
u
m
m
ar
izatio
n
.
C
elik
y
ilm
az
an
d
T
u
r
[
2
5
]
in
tr
o
d
u
ce
d
a
h
ier
ar
ch
ical
clu
s
ter
in
g
-
b
ased
h
y
b
r
id
m
o
d
el
f
o
r
o
r
g
an
izin
g
tex
tu
al
co
n
ten
t,
an
d
Ma
n
d
al
et
a
l
.
[
2
6
]
co
m
b
in
ed
s
tr
ateg
ies f
o
r
g
en
er
atin
g
co
n
cise su
m
m
ar
ies.
Desp
ite
th
ese
ad
v
an
ce
m
en
ts
,
co
m
p
r
eh
e
n
s
iv
e
ev
alu
atio
n
s
ar
e
lim
ited
.
Su
r
v
ey
s
in
[
2
7
]
,
[
2
8
]
o
f
f
e
r
in
s
ig
h
ts
in
to
s
u
m
m
ar
izatio
n
t
ec
h
n
iq
u
es,
s
tr
ess
in
g
th
e
n
ee
d
f
o
r
r
ig
o
r
o
u
s
ev
alu
atio
n
.
C
h
r
i
s
ten
s
en
et
a
l.
[
2
9
]
em
p
h
asized
s
ty
le
-
s
p
ec
if
ic
s
u
m
m
ar
izatio
n
f
o
r
b
r
o
ad
ca
s
t
n
ews,
wh
ile
Allah
y
ar
i
et
a
l.
[
3
0
]
p
r
o
v
id
ed
an
o
v
e
r
v
iew
o
f
tex
t
s
u
m
m
ar
izatio
n
m
et
h
o
d
o
lo
g
ies.
Ou
r
s
tu
d
y
in
tr
o
d
u
ce
s
a
n
o
v
el
h
y
b
r
i
d
m
et
h
o
d
in
teg
r
a
tin
g
ex
tr
ac
tiv
e
an
d
ab
s
tr
ac
tiv
e
tech
n
iq
u
es,
aim
in
g
to
ad
v
an
ce
tex
t
s
u
m
m
ar
iz
atio
n
th
r
o
u
g
h
r
ig
o
r
o
u
s
ev
alu
atio
n
an
d
tailo
r
ed
ap
p
r
o
ac
h
es to
ad
d
r
ess
in
f
o
r
m
a
tio
n
o
v
er
l
o
ad
.
2.
M
E
T
H
O
D
Ov
er
th
e
y
ea
r
s
,
m
u
ltip
le
m
et
h
o
d
s
h
av
e
b
ee
n
im
p
lem
en
ted
f
o
r
au
to
m
atic
tex
t
s
u
m
m
ar
i
za
tio
n
b
y
r
esear
ch
er
s
wh
ich
ar
e
p
r
esen
t
ed
in
F
ig
u
r
e
1
.
Un
f
o
r
tu
n
ately
,
n
o
m
eth
o
d
is
ca
p
ab
le
o
f
ac
cu
r
ately
s
u
m
m
ar
izin
g
d
o
cu
m
e
n
ts
o
f
d
if
f
er
e
n
t
d
o
m
ai
n
s
[
3
1
]
–
[
3
6
]
.
B
ased
o
n
t
h
e
o
u
t
p
u
t
an
d
p
u
r
p
o
s
e,
th
e
m
eth
o
d
s
an
d
ap
p
r
o
ac
h
es
f
o
r
s
u
m
m
ar
izatio
n
ca
n
v
ar
y
.
Fig
u
r
e
1
.
T
e
x
t su
m
m
ar
y
m
eth
o
d
o
lo
g
y
[
1
9
]
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.
14
,
No
.
1
,
Feb
r
u
ar
y
2
0
2
5
:
680
-
6
9
1
682
T
o
g
en
er
ate
a
s
u
m
m
ar
y
,
th
e
ex
tr
ac
tiv
e
-
b
ased
ap
p
r
o
ac
h
s
elec
ts
in
f
o
r
m
ativ
e
s
en
ten
ce
s
d
ir
ec
tly
f
r
o
m
th
e
in
p
u
t
d
o
cu
m
en
t
b
ased
o
n
p
r
ed
ef
in
ed
cr
iter
ia,
s
co
r
in
g
ea
ch
s
en
ten
ce
ac
co
r
d
in
g
to
its
f
ea
tu
r
es
to
d
eter
m
in
e
its
r
elev
an
ce
[
1
7
]
.
I
n
co
n
tr
ast,
ab
s
tr
ac
tiv
e
s
u
m
m
ar
izatio
n
r
ew
r
ites
th
e
o
r
ig
in
al
d
o
c
u
m
en
t
u
s
in
g
in
ter
n
al
s
em
an
tic
r
ep
r
esen
tatio
n
s
an
d
n
atu
r
al
la
n
g
u
ag
e
p
r
o
ce
s
s
in
g
(
NL
P)
tech
n
iq
u
es
to
p
r
o
d
u
ce
co
h
er
en
t
an
d
h
u
m
an
-
r
ea
d
a
b
le
s
u
m
m
ar
ies
[
37
]
.
T
h
is
m
eth
o
d
is
m
o
r
e
co
m
p
lex
d
u
e
to
th
e
ch
allen
g
es
o
f
NL
P
an
d
th
e
n
ee
d
f
o
r
e
f
f
ec
tiv
e
co
m
p
r
ess
io
n
tech
n
iq
u
es
[3
8
]
.
Fig
u
r
e
2
illu
s
tr
ates
th
e
b
asic
f
lo
w
o
f
e
x
tr
ac
tiv
e
a
n
d
ab
s
tr
ac
tiv
e
tex
t
s
u
m
m
ar
izatio
n
m
eth
o
d
s
.
Fig
u
r
e
2
.
B
asic e
x
tr
ac
tiv
e
an
d
ab
s
tr
ac
tiv
e
f
lo
wch
ar
t
[3
9
]
Gen
er
ic
tex
t
s
u
m
m
ar
izatio
n
tr
ea
ts
all
in
p
u
ts
u
n
if
o
r
m
ly
,
aim
in
g
to
p
r
o
v
id
e
a
co
m
p
r
eh
en
s
iv
e
u
n
d
er
s
tan
d
i
n
g
wh
ile
m
i
n
im
izin
g
r
ed
u
n
d
a
n
cy
[
4
0
]
.
Ho
wev
e
r
,
with
o
u
t
d
o
m
ain
-
s
p
ec
if
ic
k
n
o
wled
g
e
o
r
q
u
er
ies,
s
u
ch
m
o
d
els
m
ay
o
v
er
lo
o
k
cr
itical
to
p
ics,
esp
ec
ially
ev
id
en
t
wh
en
s
u
m
m
ar
izin
g
co
m
p
lex
d
o
m
ain
s
lik
e
m
ed
ical
r
ep
o
r
ts
[
4
1
]
.
Qu
er
y
-
b
ased
tex
t
s
u
m
m
ar
izatio
n
f
o
c
u
s
es
o
n
ex
tr
ac
tin
g
in
f
o
r
m
atio
n
th
at
d
ir
ec
tl
y
an
s
wer
s
s
p
ec
if
ied
q
u
er
ies
with
in
a
lim
ited
wo
r
d
co
u
n
t,
o
f
te
n
lack
in
g
co
h
er
en
c
e
co
m
p
ar
ed
to
o
th
er
m
eth
o
d
s
[
4
2
]
.
Do
m
ain
-
b
ased
tex
t
s
u
m
m
ar
izatio
n
u
tili
ze
s
d
o
m
ain
-
s
p
ec
if
ic
lear
n
in
g
to
p
r
o
d
u
ce
m
o
r
e
p
r
e
cise
s
u
m
m
ar
ies
tailo
r
ed
to
f
ield
s
s
u
c
h
as
b
io
m
ed
ical
r
esear
ch
a
n
d
le
g
al
d
o
cu
m
en
ts
[
4
3
]
,
h
i
g
h
lig
h
ti
n
g
th
e
im
p
o
r
tan
ce
o
f
i
d
en
tify
i
n
g
d
o
m
ain
-
s
p
ec
if
ic
ter
m
s
.
I
n
d
icativ
e
tex
t
s
u
m
m
ar
izatio
n
ca
p
tu
r
es
k
e
y
to
p
ics
f
r
o
m
th
e
o
r
ig
in
al
tex
t,
u
s
ef
u
l
f
o
r
s
ch
o
lar
ly
a
r
ticles
wh
er
e
ess
en
tial
in
f
o
r
m
atio
n
is
co
n
d
en
s
ed
[
4
4
]
.
I
n
f
o
r
m
ativ
e
tex
t
s
u
m
m
ar
izatio
n
p
r
o
v
id
e
s
m
etad
ata
ab
o
u
t
a
d
o
cu
m
e
n
t'
s
s
co
p
e,
aid
in
g
r
ea
d
er
s
in
d
ec
id
in
g
wh
eth
e
r
to
d
e
lv
e
d
ee
p
er
in
to
t
h
e
co
n
ten
t
[
4
5
]
,
o
f
ten
f
o
u
n
d
o
n
b
o
o
k
c
o
v
er
s
o
r
r
ep
o
r
t
h
ea
d
lin
es.
Aca
d
em
ic
p
ap
er
s
,
s
tr
u
ctu
r
ed
with
s
ec
tio
n
s
lik
e
titl
e,
ab
s
tr
ac
t,
in
tr
o
d
u
ctio
n
,
m
eth
o
d
s
,
r
esu
lts
,
d
is
cu
s
s
io
n
,
a
n
d
r
ef
er
en
ce
s
,
ca
n
b
e
n
ef
it
f
r
o
m
s
eg
m
en
t
-
s
p
ec
if
ic
s
u
m
m
ar
izatio
n
to
ef
f
ec
tiv
el
y
co
v
er
ess
en
tial p
o
in
ts
[
4
6
]
.
Scien
tific
p
ap
er
s
p
o
s
es
a
ch
al
len
g
e
d
u
e
to
th
eir
co
m
p
lex
la
n
g
u
ag
e
.
T
h
e
au
th
o
r
s
ex
p
lo
r
ed
a
h
y
b
r
id
m
eth
o
d
u
s
in
g
b
o
th
ex
t
r
ac
tiv
e
an
d
a
b
s
tr
ac
tiv
e
tech
n
iq
u
es
to
s
u
m
m
ar
ize
p
ap
er
s
an
d
ev
a
lu
ated
tr
an
s
f
o
r
m
er
m
o
d
els'
p
er
f
o
r
m
an
ce
o
n
p
ap
e
r
in
tr
o
d
u
ctio
n
s
.
E
x
tr
ac
tiv
e
te
ch
n
iq
u
es
wer
e
u
s
ed
to
s
elec
t
im
p
o
r
tan
t
s
ec
tio
n
s
o
r
s
en
ten
ce
s
f
r
o
m
th
e
in
p
u
t
d
ata,
in
teg
r
ated
in
to
au
t
o
m
atica
lly
g
en
er
ated
ab
s
tr
ac
tiv
e
s
u
m
m
a
r
ies.
Fab
b
r
izio
et
a
l.
n
o
ted
th
at
h
y
b
r
id
s
u
m
m
ar
ies
ar
e
m
o
r
e
r
ea
d
ab
le
an
d
co
m
p
ac
t
co
m
p
ar
ed
to
p
u
r
ely
ex
tr
ac
tiv
e
m
eth
o
d
s
,
as
d
ep
icted
in
Fig
u
r
e
3
[
4
7
]
–
[
4
9
]
.
T
h
e
h
y
b
r
id
s
u
m
m
a
r
ies
o
f
f
er
b
r
o
ad
co
v
er
ag
e
o
f
in
p
u
t
d
ata,
a
r
e
co
n
cise,
av
o
id
r
ed
u
n
d
an
c
y
,
an
d
m
ain
tain
h
ig
h
lin
g
u
is
tic
q
u
ality
in
r
ea
d
a
b
ilit
y
an
d
co
h
er
en
ce
.
T
h
ey
ar
e
also
tailo
r
ed
d
ir
ec
tly
to
r
ea
d
er
s
'
n
ee
d
s
,
en
h
an
cin
g
u
s
ab
ilit
y
.
T
h
e
s
tu
d
y
test
ed
an
d
ev
alu
ate
d
p
r
e
-
tr
ain
e
d
tr
a
n
s
f
o
r
m
er
s
in
clu
d
in
g
T
5
,
B
E
R
T
,
PE
GASUS
,
an
d
B
AR
T
to
d
ev
elo
p
th
e
h
y
b
r
id
m
o
d
el.
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
To
w
a
r
d
s
efficien
t k
n
o
w
led
g
e
ex
tr
a
ctio
n
:
N
a
tu
r
a
l la
n
g
u
a
g
e
p
r
o
ce
s
s
in
g
-
b
a
s
ed
.
.
.
(
N
ikita
C
h
a
u
d
h
a
r
i)
683
Fig
u
r
e
3
.
A
h
y
b
r
i
d
m
o
d
el
f
o
r
au
to
m
atic
tex
t su
m
m
ar
izatio
n
2
.
1
.
T
ra
ns
f
o
rm
er
mo
dels
o
v
er
v
iew
T
h
is
s
ec
tio
n
p
r
o
v
id
es
a
n
o
v
er
v
iew
o
f
s
ev
er
al
p
r
o
m
in
e
n
t
T
r
an
s
f
o
r
m
er
-
b
ased
m
o
d
els
u
s
ed
in
n
at
u
r
al
lan
g
u
ag
e
p
r
o
ce
s
s
in
g
task
s
.
T
5
,
o
r
te
x
t
-
to
-
tex
t
tr
a
n
s
f
er
tr
a
n
s
f
o
r
m
er
,
s
er
v
es
as
a
v
er
s
atile
ar
ch
itectu
r
e
f
o
r
task
s
lik
e
tr
an
s
latio
n
,
q
u
esti
o
n
an
s
w
er
in
g
,
an
d
class
if
icatio
n
[
1
3
]
a
s
s
h
o
wn
in
Fig
u
r
e
4
.
PEGA
SUS
wh
ich
is
d
ep
icted
u
s
in
g
Fig
u
r
e
5
is
s
p
ec
ialized
f
o
r
ab
s
tr
ac
tiv
e
tex
t
s
u
m
m
a
r
izatio
n
b
y
g
e
n
er
atin
g
s
u
m
m
ar
ies
th
r
o
u
g
h
a
m
ask
in
g
m
ec
h
an
is
m
th
at
cr
ea
tes
an
ex
tr
ac
tiv
e
-
lik
e
s
u
m
m
a
r
y
s
eq
u
en
ce
[
1
4
]
.
B
E
R
T
em
p
lo
y
s
b
id
ir
ec
tio
n
al
co
n
tex
t
to
u
n
d
er
s
tan
d
am
b
ig
u
o
u
s
la
n
g
u
ag
e
in
v
ar
io
u
s
co
n
tex
ts
,
i
n
itially
tr
ain
ed
o
n
W
ik
ip
ed
ia
an
d
later
a
d
ap
ted
f
o
r
Q
an
d
A
task
s
[
1
5
]
as
s
h
o
wn
in
Fig
u
r
e
6
.
B
AR
T
wh
ich
is
s
h
o
wn
in
F
ig
u
r
e
7
f
u
n
ctio
n
s
as
a
d
en
o
is
in
g
au
to
en
co
d
er
d
esig
n
ed
f
o
r
s
eq
u
en
ce
-
to
-
s
eq
u
en
ce
task
s
,
u
tili
zin
g
a
t
r
an
s
f
o
r
m
er
-
b
ased
ar
c
h
itectu
r
e
s
im
ilar
to
B
E
R
T
.
b
u
t
with
a
s
p
ec
if
ic
f
o
cu
s
o
n
te
x
t
r
ea
s
s
em
b
ly
f
r
o
m
c
o
r
r
u
p
ted
i
n
p
u
ts
[
1
6
]
.
Ad
d
itio
n
ally
,
th
e
h
y
b
r
id
m
o
d
el,
th
e
o
n
e
b
ein
g
u
s
ed
h
er
e,
in
t
eg
r
ates
B
E
R
T
f
o
r
ex
tr
ac
tiv
e
s
u
m
m
ar
izatio
n
an
d
PEGA
SUS
f
o
r
ab
s
tr
ac
tiv
e
s
u
m
m
ar
izatio
n
,
c
o
m
b
in
i
n
g
b
o
th
ap
p
r
o
ac
h
es
t
o
en
h
an
ce
s
u
m
m
ar
y
q
u
ality
a
n
d
ac
h
iev
e
h
i
g
h
er
r
ec
all
-
o
r
ien
te
d
u
n
d
er
s
tu
d
y
f
o
r
g
is
tin
g
ev
al
u
atio
n
(
R
OUGE
)
s
co
r
es
[
1
2
]
.
T
o
b
eg
in
,
th
e
r
aw
tex
t w
as p
ass
ed
th
r
o
u
g
h
th
e
B
E
R
T
tr
an
s
f
o
r
m
er
m
o
d
el
to
o
b
tain
a
n
ex
tr
ac
tiv
e
s
u
m
m
ar
y
.
T
h
is
s
u
m
m
ar
y
was
th
en
f
ed
in
t
o
th
e
PEGA
SUS
m
o
d
el,
wh
ich
p
r
o
d
u
ce
d
a
f
in
al
a
b
s
tr
ac
tiv
e
s
u
m
m
ar
y
.
Fig
u
r
e
4
.
T
5
[
1
3
]
Fig
u
r
e
5
.
PEGA
SUS
[
1
4
]
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.
14
,
No
.
1
,
Feb
r
u
ar
y
2
0
2
5
:
680
-
6
9
1
684
Fig
u
r
e
6
.
B
E
R
T
[
1
5
]
Fig
u
r
e
7
.
B
AR
T
[
1
6
]
2
.
2
.
Da
t
a
co
llect
io
n
a
nd
prepro
ce
s
s
ing
Du
r
in
g
d
ata
co
llectio
n
,
in
tr
o
d
u
ctio
n
s
f
r
o
m
s
cien
tific
p
a
p
er
s
au
th
o
r
ed
b
y
Sin
g
h
et
a
l.
[
4
7
]
f
o
r
m
ed
t
h
e
d
ataset.
T
o
e
n
s
u
r
e
u
n
if
o
r
m
ity
,
r
ig
o
r
o
u
s
p
r
ep
r
o
ce
s
s
in
g
tech
n
iq
u
es
wer
e
a
p
p
lied
.
T
h
is
in
cl
u
d
ed
r
em
o
v
in
g
wh
ite
s
p
ac
es,
elim
in
atin
g
s
to
p
wo
r
d
s
,
an
d
clea
r
in
g
s
p
ec
ial
ch
ar
ac
ter
s
to
en
h
an
ce
tex
t q
u
ality
.
All te
x
t w
as c
o
n
v
er
ted
to
lo
wer
ca
s
e
to
s
tan
d
ar
d
ize
ca
s
e,
p
u
n
ctu
atio
n
m
a
r
k
s
wer
e
r
ed
u
ce
d
,
a
n
d
n
u
m
er
ic
ch
ar
ac
t
er
s
wer
e
o
m
itted
.
Op
tio
n
ally
,
s
tem
m
in
g
o
r
lem
m
atiza
tio
n
co
u
ld
f
u
r
th
er
h
o
m
o
g
en
ize
th
e
tex
t.
T
ab
les
1
a
n
d
2
s
h
o
wca
s
e
a
d
ataset
s
am
p
le
f
ea
tu
r
in
g
in
t
r
o
d
u
ctio
n
s
f
r
o
m
p
ap
e
r
s
lik
e
"Sto
c
k
p
r
e
d
ictio
n
u
s
in
g
m
ac
h
in
e
lea
r
n
in
g
:
a
r
ev
iew
p
ap
e
r
,
"
with
co
r
r
esp
o
n
d
in
g
r
ef
er
e
n
ce
s
an
d
g
e
n
er
ated
s
u
m
m
ar
ies.
T
ab
le
1
.
I
n
f
o
r
m
atio
n
o
n
th
e
s
a
m
p
le
A
u
t
h
o
r
s
R
e
f
e
r
e
n
c
e
s
u
mm
a
r
y
S
i
n
g
h
e
t
a
l
.
[
4
7
]
I
n
mac
h
i
n
e
l
e
a
r
n
i
n
g
,
d
a
t
a
i
s
o
b
t
a
i
n
e
d
b
y
k
n
o
w
l
e
d
g
e
e
x
t
r
a
c
t
i
o
n
.
M
a
c
h
i
n
e
l
e
a
r
n
i
n
g
t
e
c
h
n
i
q
u
e
s
l
e
a
r
n
f
r
o
m
p
a
s
t
p
r
e
d
i
c
t
i
o
n
s a
n
d
g
i
v
e
a
n
o
u
t
c
o
m
e
.
M
a
c
h
i
n
e
l
e
a
r
n
i
n
g
i
s
u
se
d
t
o
d
e
scri
b
e
t
h
e
p
r
o
c
e
ss
o
f
e
x
t
r
a
c
t
i
n
g
k
n
o
w
l
e
d
g
e
f
r
o
m
a
c
o
d
e
.
D
a
t
a
i
s
p
r
o
v
i
d
e
d
t
o
t
h
e
g
e
n
e
r
i
c
a
l
g
o
r
i
t
h
ms
a
n
d
l
o
g
i
c
i
s
d
e
v
e
l
o
p
e
d
o
n
t
h
e
b
a
si
s
o
f
t
h
a
t
d
a
t
a
.
M
a
c
h
i
n
e
l
e
a
r
n
i
n
g
c
a
n
b
e
d
e
f
i
n
e
d
a
s
t
h
e
d
a
t
a
w
h
i
c
h
i
s
o
b
t
a
i
n
e
d
b
y
k
n
o
w
l
e
d
g
e
e
x
t
r
a
c
t
i
o
n
.
I
n
st
e
a
d
o
f
w
r
i
t
i
n
g
a
c
o
d
e
f
o
r
e
v
e
r
y
s
p
e
c
i
f
i
c
p
r
o
b
l
e
m
,
d
a
t
a
i
s
p
r
o
v
i
d
e
d
t
o
t
h
e
g
e
n
e
r
i
c
a
l
g
o
r
i
t
h
ms,
a
n
d
l
o
g
i
c
i
s
d
e
v
e
l
o
p
e
d
o
n
t
h
e
b
a
s
i
s
o
f
t
h
a
t
d
a
t
a
T
ab
le
2
.
Gen
e
r
ated
s
u
m
m
ar
ies
f
o
r
th
e
s
am
p
le
A
l
g
o
r
i
t
h
m
G
e
n
e
r
a
t
e
d
s
u
mm
a
r
y
T5
M
a
c
h
i
n
e
l
e
a
r
n
i
n
g
c
a
n
b
e
d
e
f
i
n
e
d
a
s
t
h
e
d
a
t
a
w
h
i
c
h
i
d
o
b
t
a
i
n
e
d
b
y
k
n
o
w
l
e
d
g
e
e
x
t
r
a
c
t
i
o
n
.
T
h
e
t
e
c
h
n
i
q
u
e
f
o
r
mo
st
a
c
c
u
r
a
t
e
p
r
e
d
i
c
t
i
o
n
i
s
b
y
l
e
a
r
n
i
n
g
f
r
o
m
p
a
st
i
n
st
a
n
c
e
s
a
n
d
t
o
ma
k
e
a
p
r
o
g
r
a
m
t
o
d
o
t
h
i
s
i
s
b
e
s
t
p
o
ss
i
b
l
e
w
i
t
h
ma
c
h
i
n
e
l
e
a
r
n
i
n
g
t
e
c
h
n
i
q
u
e
s
P
EG
A
S
U
S
M
a
c
h
i
n
e
l
e
a
r
n
i
n
g
i
s a
b
r
a
n
c
h
o
f
c
o
m
p
u
t
e
r
sci
e
n
c
e
t
h
a
t
d
e
a
l
s
w
i
t
h
t
h
e
e
x
t
r
a
c
t
i
o
n
o
f
d
a
t
a
f
r
o
m
t
h
e
e
n
v
i
r
o
n
me
n
t
a
n
d
i
t
s
s
u
b
se
q
u
e
n
t
u
se
f
o
r
t
h
e
p
u
r
p
o
s
e
o
f
d
e
c
i
s
i
o
n
m
a
k
i
n
g
.
I
t
i
s
a
r
a
p
i
d
l
y
g
r
o
w
i
n
g
f
i
e
l
d
t
h
a
t
i
s
f
i
n
d
i
n
g
i
t
s
w
a
y
i
n
t
o
v
a
r
i
o
u
s
f
i
e
l
d
s
s
u
c
h
a
s
t
h
e
d
a
t
a
t
h
a
t
i
s
o
b
t
a
i
n
e
d
b
y
k
n
o
w
l
e
d
g
e
e
x
t
r
a
c
t
i
o
n
,
i
n
st
e
a
d
o
f
e
x
p
l
i
c
i
t
l
y
p
r
o
g
r
a
mm
e
d
i
n
s
t
e
a
d
t
h
e
y
a
r
e
t
r
a
i
n
e
d
t
o
ma
k
e
d
e
c
i
si
o
n
s
t
h
a
t
a
r
e
d
r
i
v
e
n
b
y
d
a
t
a
.
B
ER
T
M
a
c
h
i
n
e
l
e
a
r
n
i
n
g
c
a
n
b
e
d
e
f
i
n
e
d
a
s
t
h
e
d
a
t
a
w
h
i
c
h
i
s
o
b
t
a
i
n
e
d
b
y
k
n
o
w
l
e
d
g
e
e
x
t
r
a
c
t
i
o
n
,
I
n
st
e
a
d
o
f
w
r
i
t
i
n
g
c
o
d
e
f
o
r
e
v
e
r
y
sp
e
c
i
f
i
c
p
r
o
b
l
e
m,
d
a
t
a
i
s
p
r
o
v
i
d
e
d
t
o
t
h
e
g
e
n
e
r
i
c
a
l
g
o
r
i
t
h
ms
a
n
d
l
o
g
i
c
i
s
d
e
v
e
l
o
p
e
d
o
n
t
h
e
b
a
si
s
o
f
t
h
a
t
d
a
t
a
.
B
A
R
T
M
a
c
h
i
n
e
s d
o
n
t
h
a
v
e
t
o
b
e
p
r
o
g
r
a
mm
e
d
e
x
p
l
i
c
i
t
l
y
i
n
s
t
e
a
d
t
h
e
y
a
r
e
t
r
a
i
n
e
d
t
o
mak
e
d
e
c
i
si
o
n
s
t
h
a
t
a
r
e
d
r
i
v
e
n
b
y
d
a
t
a
.
D
a
t
a
i
s
p
r
o
v
i
d
e
d
t
o
t
h
e
g
e
n
e
r
i
c
a
l
g
o
r
i
t
h
ms
a
n
d
l
o
g
i
c
i
s
d
e
v
e
l
o
p
e
d
o
n
t
h
e
b
a
s
i
s
o
f
t
h
a
t
d
a
t
a
,
W
h
e
n
a
ma
c
h
i
n
e
i
mp
r
o
v
e
s
i
t
s
p
e
r
f
o
r
ma
n
c
e
b
a
s
e
d
o
n
i
t
s
p
a
s
t
e
x
p
e
r
i
e
n
c
e
s i
t
c
a
n
b
e
sai
d
t
h
a
t
t
h
e
ma
c
h
i
n
e
h
a
s
t
r
u
l
y
l
e
a
r
n
t
.
H
y
b
r
i
d
(
B
ER
T
+
P
EG
A
S
U
S
)
Th
e
t
e
r
m
l
e
a
r
n
i
n
g
i
s
u
se
d
t
o
d
e
scri
b
e
t
h
e
p
r
o
c
e
ss
o
f
e
x
t
r
a
c
t
i
n
g
k
n
o
w
l
e
d
g
e
f
r
o
m
a
c
o
d
e
.
Le
a
r
n
i
n
g
c
a
n
b
e
d
e
f
i
n
e
d
a
s
t
h
a
t
d
a
t
a
i
s
o
b
t
a
i
n
e
d
b
y
k
n
o
w
l
e
d
g
e
e
x
t
r
a
c
t
i
o
n
f
r
o
m
a
sp
e
c
i
f
i
c
p
r
o
b
l
e
m,
G
e
b
=
n
e
r
i
c
a
l
g
o
r
i
t
h
m
s
a
r
e
p
r
o
v
i
d
e
d
f
o
r
t
h
e
k
n
o
w
l
e
d
g
e
e
x
t
r
a
c
t
i
o
n
f
r
o
m
a
c
o
d
e
,
w
h
i
c
h
i
s
p
r
o
v
i
d
e
d
o
n
t
h
e
b
a
si
s
o
f
l
o
g
i
c
t
h
a
t
i
s
d
e
v
e
l
o
p
e
d
f
o
r
w
r
i
t
i
n
g
a
sp
e
c
i
f
i
c
p
r
o
b
l
e
m.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
To
w
a
r
d
s
efficien
t k
n
o
w
led
g
e
ex
tr
a
ctio
n
:
N
a
tu
r
a
l la
n
g
u
a
g
e
p
r
o
ce
s
s
in
g
-
b
a
s
ed
.
.
.
(
N
ikita
C
h
a
u
d
h
a
r
i)
685
2
.
3
.
H
y
perpa
ra
m
e
t
er
s
et
t
ing
T
o
en
h
a
n
ce
th
e
s
u
m
m
ar
izatio
n
q
u
ality
,
h
y
p
er
p
ar
a
m
eter
s
f
o
r
ea
c
h
m
eth
o
d
wer
e
t
h
o
u
g
h
tf
u
lly
estab
lis
h
ed
b
ased
o
n
a
co
m
b
i
n
atio
n
o
f
d
ef
au
lt
v
alu
es
a
n
d
d
o
m
ain
ex
p
er
tis
e
as
m
en
tio
n
e
d
in
T
ab
le
3
.
T
h
ese
h
y
p
er
p
ar
am
eter
s
ettin
g
s
ar
e
s
p
ec
if
ically
tailo
r
ed
f
o
r
th
e
h
y
b
r
id
m
o
d
el
.
E
n
s
u
r
in
g
th
at
it
e
f
f
ec
tiv
ely
co
m
b
in
es
ex
tr
ac
tiv
e
an
d
ab
s
tr
ac
tiv
e
s
u
m
m
ar
izatio
n
tech
n
iq
u
es
to
p
r
o
d
u
ce
h
ig
h
-
q
u
ality
s
u
m
m
ar
ies
wh
ile
co
n
s
id
er
in
g
b
o
th
s
o
u
r
ce
co
n
te
n
t a
n
d
f
in
e
-
tu
n
ed
ab
s
tr
ac
tio
n
s
.
T
ab
le
3
.
Hy
p
er
p
ar
a
m
eter
s
ettin
g
s
M
o
d
e
l
Le
a
r
n
i
n
g
r
a
t
e
B
a
t
c
h
si
z
e
N
u
mb
e
r
o
f
e
p
o
c
h
s
W
e
i
g
h
t
d
e
c
a
y
O
p
t
i
mi
z
e
r
t
y
p
e
B
ER
T
2e
-
5
32
3
0
.
0
1
A
d
a
mW
B
A
R
T
3e
-
5
16
4
0
.
0
0
1
A
d
a
m
P
EG
A
S
U
S
1e
-
5
8
3
0
.
0
1
A
d
a
mW
T5
(
TF)
0
.
0
0
1
64
5
0
.
0
0
0
1
A
d
a
m
H
y
b
r
i
d
m
o
d
e
l
1e
-
5
8
4
0
.
0
0
1
A
d
a
m
3.
RE
SU
L
T
S
AND
D
I
SCU
SS
I
O
N
T
h
e
ev
alu
atio
n
o
f
th
e
m
o
d
el
s
u
m
m
ar
ies
f
o
llo
wed
th
e
s
tan
d
ar
d
p
atter
n
f
o
r
ass
ess
in
g
au
to
m
atica
lly
g
en
er
ated
s
u
m
m
ar
ies,
em
p
lo
y
i
n
g
R
OUGE
s
co
r
es,
b
ilin
g
u
al
e
v
alu
atio
n
u
n
d
er
s
tu
d
y
(
B
L
E
U)
s
co
r
es,
an
d
ex
p
er
t
ev
alu
atio
n
.
T
h
e
class
if
icatio
n
o
f
th
ese
ass
ess
m
en
ts
i
s
d
ep
icted
in
Fig
u
r
e
8
.
R
ef
er
en
ce
s
u
m
m
ar
ies
p
r
o
v
id
e
d
b
y
th
e
au
th
o
r
s
o
f
t
h
e
s
am
p
le
p
a
p
er
s
wer
e
u
tili
ze
d
f
o
r
co
m
p
ar
i
s
o
n
an
d
co
m
p
u
tatio
n
o
f
th
e
R
OUGE
an
d
B
L
E
U
s
co
r
es f
o
r
th
e
m
o
d
el
-
g
en
er
ate
d
s
u
m
m
ar
ies.
Fig
u
r
e
8
.
E
v
alu
atio
n
m
eth
o
d
s
T
h
e
R
OUGE
-
1
,
R
OUGE
-
2
,
an
d
R
OUGE
-
L
s
co
r
es
wer
e
co
m
p
u
ted
to
ass
ess
s
u
m
m
ar
izatio
n
p
er
f
o
r
m
an
ce
.
As
d
ep
icted
i
n
T
ab
le
3
,
th
e
R
OUGE
p
r
ec
is
io
n
,
r
ec
all,
an
d
F
-
s
co
r
es
f
o
r
T
5
,
PEGA
SUS,
B
A
R
T
,
B
E
R
T
,
an
d
th
e
h
y
b
r
i
d
(
B
E
R
T
+PE
GAS
US)
m
o
d
els
ar
e
p
r
esen
ted
co
n
ce
r
n
in
g
th
e
s
u
m
m
ar
y
o
f
s
am
p
les.
T
ab
le
4
p
r
o
v
id
es
an
o
v
e
r
v
iew
o
f
th
e
R
OUGE
s
co
r
es
(
1
,
2
,
an
d
L
)
f
o
r
all
f
o
u
r
s
am
p
les.
T
h
e
h
y
b
r
id
,
B
E
R
T
,
an
d
T
5
m
o
d
els
d
em
o
n
s
tr
ated
th
e
h
ig
h
est
s
co
r
es,
wh
ile
B
AR
T
ex
h
ib
ited
th
e
lo
west
p
er
f
o
r
m
an
c
e.
T
h
e
h
y
b
r
id
m
o
d
el
ex
ce
lled
in
th
e
f
ir
s
t
s
am
p
le,
w
h
ile
B
E
R
T
p
er
f
o
r
m
ed
well
in
t
h
e
f
ir
s
t
two
s
am
p
les,
a
n
d
T
5
a
ch
iev
ed
e
x
ce
p
tio
n
al
r
esu
lts
f
o
r
Sam
p
le
1
an
d
Sam
p
le
3
.
I
t
is
ess
en
tial
to
n
o
te
th
at
th
e
R
OUGE
s
co
r
e,
wh
ile
in
f
o
r
m
ativ
e,
d
o
es
n
o
t
f
u
lly
m
ea
s
u
r
e
s
em
an
tic
ac
cu
r
ac
y
in
s
u
m
m
ar
izatio
n
.
T
h
e
r
ef
o
r
e,
tw
o
ad
d
itio
n
al
ev
al
u
atio
n
m
eth
o
d
s
h
av
e
b
ee
n
em
p
lo
y
ed
.
Fig
u
r
e
9
illu
s
tr
ate
t
h
e
av
er
ag
e
R
OUGE
s
co
r
es
f
o
r
ea
ch
m
o
d
el,
o
f
f
er
i
n
g
a
co
n
cise
r
ep
r
esen
tatio
n
o
f
th
eir
co
m
p
a
r
ativ
e
p
er
f
o
r
m
a
n
c
e.
Fig
u
r
e
9
.
Av
e
r
ag
e
R
OUGE
s
co
r
e
f
o
r
ea
ch
m
o
d
el
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.
14
,
No
.
1
,
Feb
r
u
ar
y
2
0
2
5
:
680
-
6
9
1
686
T
ab
le
4
.
R
OUGE
s
co
r
es
S
a
mp
l
e
n
o.
R
e
c
a
l
l
P
r
e
c
i
s
i
o
n
F
-
1
R
O
U
G
E
-
1
R
O
U
G
E
-
2
R
O
U
G
E
-
L
R
O
U
G
E
-
1
R
O
U
G
E
-
2
R
O
U
G
E
-
L
R
O
U
G
E
-
1
R
O
U
G
E
-
2
R
O
U
G
E
-
L
T5
1
0
.
5
7
1
0
.
3
0
0
0
.
5
2
4
0
.
3
5
3
0
.
1
5
0
0
.
3
2
4
0
.
4
3
6
0
.
2
0
0
0
.
4
0
0
2
0
.
1
2
2
0
.
0
0
0
0
.
1
2
2
0
.
1
3
2
0
.
0
0
0
0
.
1
3
2
0
.
1
2
7
0
.
0
0
0
0
.
1
2
7
3
0
.
8
7
0
0
.
6
8
2
0
.
8
7
0
0
.
5
2
6
0
.
3
5
7
0
.
5
2
6
0
.
6
5
6
0
.
4
6
9
0
.
6
5
6
4
0
.
1
3
3
0
.
0
0
0
0
.
1
0
0
0
.
1
0
5
0
.
0
0
0
0
.
0
7
9
0
.
1
1
8
0
.
0
0
0
0
.
0
8
8
P
EG
A
S
U
S
1
0
.
4
7
6
0
.
1
5
0
0
.
4
7
6
0
.
1
7
5
0
.
0
3
8
0
.
1
7
5
0
.
2
5
6
0
.
0
6
1
0
.
2
5
6
2
0
.
3
9
0
0
.
2
2
9
0
.
3
9
0
0
.
5
0
0
0
.
2
5
0
0
.
5
0
0
0
.
4
3
8
0
.
2
3
9
0
.
4
3
8
3
0
.
1
7
4
0
.
0
0
0
0
.
1
3
0
0
.
1
4
8
0
.
0
0
0
0
.
1
1
1
0
.
1
6
0
0
.
0
0
0
0
.
1
2
0
4
0
.
1
6
7
0
.
0
3
3
0
.
1
6
7
0
.
1
8
5
0
.
0
3
1
0
.
1
8
5
0
.
1
7
5
0
.
0
3
2
0
.
1
7
5
B
A
R
T
1
0
.
2
8
6
0
.
0
0
0
0
.
2
3
8
0
.
1
3
6
0
.
0
0
0
0
.
1
1
4
0
.
1
8
5
0
.
0
0
0
0
.
1
5
4
2
0
.
2
4
4
0
.
0
4
2
0
.
1
7
1
0
.
2
5
0
0
.
0
4
7
0
.
1
7
5
0
.
2
4
7
0
.
0
4
4
0
.
1
7
3
3
0
.
1
7
4
0
.
0
0
0
0
.
1
7
4
0
.
0
8
5
0
.
0
0
0
0
.
0
8
5
0
.
1
1
4
0
.
0
0
0
0
.
1
1
4
4
0
.
3
3
3
0
.
0
6
7
0
.
2
6
7
0
.
2
3
8
0
.
0
4
2
0
.
1
9
0
0
.
2
7
8
0
.
0
5
1
0
.
2
2
2
B
ER
T
1
0
.
4
2
9
0
.
3
0
0
0
.
4
2
9
0
.
2
7
3
0
.
1
5
4
0
.
2
7
3
0
.
3
3
3
0
.
2
0
3
0
.
3
3
3
2
0
.
4
3
9
0
.
3
5
4
0
.
4
3
9
0
.
6
2
1
0
.
4
5
9
0
.
6
2
1
0
.
5
1
4
0
.
4
0
0
0
.
5
1
4
3
0
.
2
1
7
0
.
0
0
0
0
.
1
7
4
0
.
0
5
8
0
.
0
0
0
0
.
0
4
7
0
.
0
9
2
0
.
0
0
0
0
.
0
7
3
4
0
.
4
3
3
0
.
2
0
0
0
.
3
6
7
0
.
3
1
7
0
.
1
2
2
0
.
2
6
8
0
.
3
6
6
0
.
1
5
2
0
.
3
1
0
H
y
b
r
i
d
1
0
.
3
8
1
0
.
2
5
0
0
.
3
8
1
0
.
2
1
1
0
.
0
9
4
0
.
2
1
1
0
.
2
7
1
0
.
1
3
7
0
.
2
7
1
2
0
.
1
2
2
0
.
0
0
0
0
.
1
2
2
0
.
2
0
8
0
.
0
0
0
0
.
2
0
8
0
.
1
5
4
0
.
0
0
0
0
.
1
5
4
3
0
.
3
9
1
0
.
0
4
5
0
.
2
1
7
0
.
2
2
0
0
.
0
2
0
0
.
1
2
2
0
.
2
8
1
0
.
0
2
7
0
.
1
5
6
4
0
.
2
0
0
0
.
0
0
0
0
.
1
6
7
0
.
1
7
1
0
.
0
0
0
0
.
1
4
3
0
.
1
8
5
0
.
0
0
0
0
.
1
5
4
T
ab
le
5
illu
s
tr
ates
th
e
s
u
m
m
ar
izatio
n
p
er
f
o
r
m
an
ce
b
ased
o
n
B
L
E
U
s
co
r
es
ac
r
o
s
s
all
m
o
d
el
s
.
No
tab
ly
,
th
e
h
y
b
r
id
an
d
B
E
R
T
m
o
d
els
d
em
o
n
s
tr
ate
ex
ce
p
tio
n
al
p
er
f
o
r
m
an
ce
,
s
u
r
p
ass
in
g
B
AR
T
,
T
5
,
an
d
PEGA
SUS.
T
h
is
p
er
f
o
r
m
an
ce
d
i
f
f
er
en
ce
is
attr
ib
u
ted
to
B
L
E
U
s
co
r
es'
ten
d
en
cy
to
f
av
o
r
ex
tr
ac
tiv
e
s
u
m
m
ar
izatio
n
m
eth
o
d
s
,
alig
n
in
g
with
B
E
R
T
'
s
s
tr
en
g
t
h
s
in
th
is
co
n
tex
t.
Fig
u
r
e
10
v
is
u
ally
d
ep
icts
th
e
m
ea
n
B
L
E
U
s
co
r
es
f
o
r
ea
ch
m
o
d
el,
o
f
f
er
in
g
a
c
o
n
cise
o
v
e
r
v
iew
o
f
th
eir
p
e
r
f
o
r
m
an
ce
s
.
Ho
wev
er
,
it'
s
im
p
o
r
tan
t
to
n
o
te
th
at
wh
ile
B
L
E
U
an
d
R
OUGE
s
co
r
es
a
r
e
v
alu
a
b
le
m
etr
ics,
th
ey
m
ay
n
o
t
f
u
ll
y
ca
p
tu
r
e
t
h
e
s
em
an
tic
ac
cu
r
a
cy
o
f
s
u
m
m
a
r
izatio
n
.
T
h
er
ef
o
r
e,
a
n
au
t
h
o
r
'
s
ev
alu
a
tio
n
is
co
n
d
u
cted
,
as
d
e
p
icted
in
Fig
u
r
e
1
1
,
ass
ess
in
g
s
u
m
m
ar
ies
b
ased
o
n
co
n
cisen
ess
,
ac
cu
r
ac
y
,
an
d
d
etail
p
r
eser
v
atio
n
.
T
h
is
ev
alu
atio
n
is
cr
itical
as
it
p
r
o
v
id
e
s
a
co
m
p
r
eh
e
n
s
iv
e
p
er
s
p
ec
tiv
e
o
n
th
e
q
u
ality
o
f
s
u
m
m
ar
izatio
n
,
c
o
n
s
id
er
in
g
f
ac
to
r
s
b
ey
o
n
d
q
u
an
titativ
e
s
co
r
e
s
.
T
ab
le
5
.
Av
e
r
ag
e
B
L
E
U
Sco
r
es
B
A
R
T
B
ER
T
T5
P
EG
A
S
U
S
H
y
b
r
i
d
S
a
mp
l
e
1
0
.
2
8
6
6
2
1
0
.
2
4
2
2
5
0
.
1
2
0
1
5
0
.
4
1
7
3
9
S
a
mp
l
e
2
0
.
1
6
5
3
2
0
.
8
1
0
4
4
0
.
2
2
8
5
5
0
0
.
1
8
4
0
2
S
a
mp
l
e
3
0
0
.
5
6
9
3
0
0
0
.
0
6
1
3
0
0
.
6
4
4
6
8
S
a
mp
l
e
4
0
.
5
6
2
9
2
0
.
7
9
8
6
3
0
.
0
9
9
4
3
0
0
.
7
2
3
4
3
Fig
u
r
e
1
0
.
Av
er
ag
e
B
L
E
U
s
co
r
e
f
o
r
ea
c
h
m
o
d
el
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
To
w
a
r
d
s
efficien
t k
n
o
w
led
g
e
ex
tr
a
ctio
n
:
N
a
tu
r
a
l la
n
g
u
a
g
e
p
r
o
ce
s
s
in
g
-
b
a
s
ed
.
.
.
(
N
ikita
C
h
a
u
d
h
a
r
i)
687
Fig
u
r
e
1
1
.
Av
er
ag
e
a
u
th
o
r
ev
a
lu
atio
n
s
co
r
es
I
t
is
b
o
th
c
o
m
p
u
tatio
n
ally
a
n
d
th
eo
r
etica
lly
ch
allen
g
in
g
to
ac
h
iev
e
f
lawless
R
OUGE
s
c
o
r
es.
Fo
r
ev
alu
atio
n
,
th
e
m
eth
o
d
in
v
o
lv
ed
ca
lcu
latin
g
th
e
R
OUGE
s
c
o
r
es f
o
r
all
g
en
er
ate
d
s
u
m
m
ar
i
es u
s
in
g
a
r
ef
er
en
ce
s
u
m
m
ar
y
a
p
p
r
o
v
ed
b
y
th
e
au
t
h
o
r
s
o
f
th
e
p
a
p
er
s
.
I
n
t
h
e
ev
al
u
atio
n
,
PEGA
SUS,
B
E
R
T
,
h
y
b
r
id
,
T
5
,
an
d
B
AR
T
wer
e
ass
es
s
ed
ac
r
o
s
s
m
u
ltip
le
s
am
p
les
f
o
r
s
u
m
m
ar
izatio
n
p
e
r
f
o
r
m
a
n
ce
.
I
n
Sam
p
le
1
,
PEG
ASUS,
B
E
R
T
,
an
d
h
y
b
r
id
ac
h
iev
e
d
R
OUGE
1
F1
s
co
r
es
o
f
0
.
2
5
6
,
0
.
3
3
3
,
an
d
0
.
2
7
1
,
r
esp
ec
tiv
ely
,
in
d
i
ca
tin
g
s
atis
f
ac
to
r
y
s
u
m
m
ar
i
za
tio
n
q
u
ality
with
p
o
ten
tial
f
o
r
im
p
r
o
v
em
en
t.
I
n
Sam
p
le
2
,
B
E
R
T
ac
h
iev
ed
an
F1
s
co
r
e
o
f
0
.
5
1
4
,
o
u
tp
er
f
o
r
m
in
g
o
th
er
m
o
d
els,
wh
ile
h
y
b
r
id
f
o
llo
wed
with
0
.
1
8
5
;
T
5
h
ad
m
u
ltip
le
in
co
n
s
is
ten
cies.
Sam
p
le
3
h
ig
h
lig
h
ted
T
5
an
d
B
AR
T
'
s
ef
f
ec
tiv
en
ess
with
R
OU
GE
1
F1
s
co
r
es
o
f
0
.
6
5
6
an
d
0
.
1
1
4
,
r
esp
ec
tiv
ely
,
co
m
p
a
r
ed
to
B
E
R
T
an
d
PEGA
SUS.
Sa
m
p
le
4
s
h
o
wed
B
AR
T
'
s
s
u
p
er
io
r
co
h
er
en
ce
d
esp
ite
lo
wer
R
OUG
E
s
co
r
es
th
an
B
E
R
T
,
wh
ich
ex
ce
lled
in
B
leu
s
co
r
es
alo
n
g
s
id
e
h
y
b
r
id
.
Ov
er
all,
T
5
an
d
B
AR
T
co
n
s
is
ten
tly
p
r
o
v
id
ed
co
m
p
etitiv
e
s
u
m
m
ar
ies,
d
e
m
o
n
s
tr
atin
g
th
eir
ef
f
icac
y
in
d
iv
e
r
s
e
s
u
m
m
ar
izatio
n
task
s
.
R
OUGE
an
d
B
L
E
U
s
co
r
es
ar
e
s
tan
d
ar
d
m
etr
ics
in
n
atu
r
al
lan
g
u
ag
e
p
r
o
ce
s
s
in
g
task
s
,
in
cl
u
d
in
g
tex
t
s
u
m
m
ar
izatio
n
a
n
d
m
ac
h
in
e
t
r
an
s
latio
n
.
Desp
ite
th
eir
u
s
ef
u
ln
ess
,
th
ey
ar
e
lim
ited
in
ass
ess
in
g
s
em
an
tic
an
d
s
tr
u
ctu
r
al
asp
ec
ts
o
f
tex
t,
f
o
cu
s
in
g
m
ain
ly
o
n
s
u
r
f
a
ce
-
lev
el
s
im
ilar
ity
an
d
lack
in
g
co
n
s
id
er
atio
n
f
o
r
s
u
b
jectiv
e
h
u
m
an
u
n
d
er
s
tan
d
in
g
.
Au
th
o
r
ev
alu
atio
n
p
r
o
v
es
to
b
e
m
o
r
e
a
cc
u
r
ate,
as
d
em
o
n
s
tr
ated
in
th
i
s
s
tu
d
y
co
m
p
ar
in
g
th
e
Hy
b
r
id
an
d
B
E
R
T
m
o
d
el
s
.
T
h
e
ev
alu
atio
n
in
v
o
lv
e
d
p
r
o
ce
s
s
in
g
r
an
d
o
m
s
am
p
les
f
r
o
m
th
e
d
ataset
u
s
in
g
th
ese
m
o
d
els,
an
d
ass
ess
in
g
s
u
m
m
ar
ies
b
ased
o
n
cr
iter
ia
s
u
c
h
as
co
n
cisen
ess
,
ac
cu
r
ac
y
,
an
d
d
etail
p
r
eser
v
atio
n
.
Ag
g
r
eg
ated
s
co
r
es
f
r
o
m
R
OUGE
,
B
L
E
U,
an
d
au
th
o
r
ev
al
u
atio
n
s
wer
e
weig
h
ted
an
d
p
r
esen
ted
in
T
ab
le
6
,
with
T
ab
le
7
s
h
o
win
g
th
e
o
v
er
all
a
v
er
ag
e
s
co
r
es a
cr
o
s
s
all
m
o
d
els b
ased
o
n
th
ese
ev
alu
atio
n
m
etr
ics.
T
ab
le
6
.
E
v
alu
atio
n
p
ar
am
eter
d
is
tr
ib
u
tio
n
Ev
a
l
u
a
t
i
o
n
me
t
h
o
d
W
e
i
g
h
t
e
d
p
e
r
c
e
n
t
a
g
e
(
%)
R
O
U
G
E
30
B
LEU
20
A
u
t
h
o
r
e
v
a
l
u
a
t
i
o
n
50
T
ab
le
7
.
T
o
tal
av
er
ag
e
s
co
r
es f
o
r
all
m
o
d
els
S
a
mp
l
e
B
A
R
T
B
ER
T
T5
P
EG
A
S
U
S
H
y
b
r
i
d
S
a
mp
l
e
1
4
.
6
2
7
7
.
4
9
9
6
.
0
2
4
.
8
4
5
.
6
S
a
mp
l
e
2
5
.
0
6
0
7
7
.
6
6
2
4
.
8
3
4
5
.
3
1
5
.
1
6
S
a
mp
l
e
3
3
.
8
3
5
.
6
6
3
7
5
.
9
6
5
3
.
6
6
.
1
3
S
a
mp
l
e
4
5
.
7
0
6
8
6
.
6
9
5
2
4
.
7
9
9
8
3
.
3
6
6
.
2
5
A
v
e
r
a
g
e
4
.
8
6
.
8
8
5
.
4
4
.
2
8
5
.
7
9
B
y
co
m
p
ar
in
g
f
ig
u
r
es
(
Fig
u
r
e
8
,
Fig
u
r
e
9
,
Fig
u
r
e
1
1
,
an
d
Fig
u
r
e
1
2
)
,
it
ca
n
b
e
co
n
clu
d
ed
th
at
th
e
h
ig
h
est
av
er
a
g
e
s
co
r
es
f
o
r
all
s
am
p
les
ar
e
p
r
esen
ted
b
y
B
E
R
T
an
d
th
e
h
y
b
r
id
m
o
d
el;
th
e
latter
p
r
o
v
id
in
g
th
e
b
est
ab
s
tr
ac
tin
g
s
u
m
m
ar
ies
in
t
h
e
m
ajo
r
ity
o
f
ca
s
es,
ac
co
r
d
in
g
to
Fig
u
r
e
1
2
.
Alth
o
u
g
h
t
h
e
h
y
b
r
id
m
o
d
el
g
ar
n
e
r
s
a
s
co
r
e
s
lig
h
tly
lo
wer
th
a
n
o
th
e
r
m
o
d
els
in
b
o
th
class
ic
R
OUG
E
an
d
B
L
E
U
m
etr
ics,
au
th
o
r
s
’
o
v
er
all
ass
ess
m
en
t
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.
14
,
No
.
1
,
Feb
r
u
ar
y
2
0
2
5
:
680
-
6
9
1
688
co
n
f
ir
m
e
d
its
ef
f
e
ctiv
en
ess
in
g
en
er
atin
g
s
u
m
m
ar
y
o
f
r
esear
ch
p
a
p
er
a
b
s
tr
ac
ts
th
at
ar
e
ac
c
u
r
ate
an
d
r
elev
an
t.
T
h
e
co
m
p
ar
is
o
n
i
n
T
ab
le
7
i
n
d
icate
s
th
e
s
u
p
er
io
r
ity
o
f
B
E
R
T
o
n
th
e
co
n
s
id
er
ed
m
etr
ics;
h
o
wev
er
,
it is
cr
itical
to
n
o
te
th
e
s
h
o
r
tco
m
in
g
s
o
f
th
ese
i
n
d
icato
r
s
f
o
r
e
v
alu
atin
g
t
h
e
q
u
ality
o
f
s
u
m
m
ar
ies
i
n
g
e
n
er
al.
So
m
e
o
f
th
e
d
if
f
icu
lties
th
at
wer
e
p
r
esen
t
wer
e
d
ata
h
eter
o
g
en
eity
,
m
e
an
in
g
th
at
th
e
s
ty
les,
co
n
ten
t,
an
d
s
tr
u
ctu
r
es
o
f
r
esear
ch
p
ap
er
s
th
at
wer
e
r
ec
e
iv
ed
wer
e
d
if
f
er
e
n
t,
an
d
ca
u
s
ed
an
is
s
u
e
b
ec
au
s
e
th
ey
co
u
ld
n
o
t b
e
s
tan
d
ar
d
ize
d
to
h
av
e
o
n
e
ef
f
ec
tiv
e
s
u
m
m
ar
i
za
tio
n
m
o
d
el
f
o
r
th
e
m
all.
Fu
r
th
er
m
o
r
e
,
th
e
co
n
ce
p
t
o
f
s
u
b
jectiv
ity
in
ev
alu
atio
n
em
er
g
ed
as a
r
esu
lt o
f
th
e
is
s
u
es with
ev
alu
atio
n
p
r
o
ce
d
u
r
es
th
at
r
ely
o
n
r
elativ
e
ap
p
r
o
ac
h
e
s
,
in
clu
d
in
g
B
L
E
U
an
d
R
OUGE
,
an
d
f
o
r
m
u
las,
as
th
e
latter
ar
e
in
ca
p
ab
le
o
f
co
n
s
id
er
in
g
s
u
ch
q
u
alitativ
e
cr
iter
ia
f
o
r
s
u
m
m
ar
ies
as
co
h
er
en
ce
o
r
in
f
o
r
m
ativ
e
n
ess
d
ee
m
ed
to
b
e
v
ital f
o
r
ev
alu
ati
o
n
an
d
m
ay
p
o
ten
tially
im
p
o
s
e
s
y
s
tem
atic
b
ias.
Fig
u
r
e
1
2
.
T
o
tal
av
er
a
g
e
s
co
r
e
s
f
o
r
ea
ch
s
am
p
le
T
h
e
ev
al
u
atio
n
m
etr
ics,
s
u
ch
a
s
R
OUGE
an
d
B
L
E
U
ar
e
u
s
ed
to
ev
alu
ate
th
e
q
u
ality
o
f
th
e
s
u
m
m
ar
ies
an
d
th
ese
m
etr
ics
ten
d
to
r
ew
ar
d
th
e
m
o
d
els
th
at
f
o
cu
s
m
o
r
e
o
n
th
e
wo
r
d
o
cc
u
r
r
en
ce
,
wh
ich
is
v
er
y
m
u
c
h
in
lin
e
with
ex
tr
ac
tiv
e
m
ec
h
an
is
m
s
,
s
u
ch
as
B
E
R
T
.
Ho
wev
er
,
s
u
m
m
ar
izatio
n
q
u
ality
d
o
es
n
o
t
r
ef
er
to
th
e
b
asic
task
o
f
liter
ally
co
p
y
in
g
co
n
ten
ts
.
T
h
e
p
o
s
t
-
p
r
o
ce
s
s
in
g
co
m
b
in
es
co
n
ten
t
s
elec
tio
n
f
r
o
m
B
E
R
T
with
th
e
s
u
m
m
ar
izatio
n
’
s
ab
s
tr
ac
tiv
e
n
atu
r
e
to
p
r
o
v
id
e
s
h
o
r
t,
lo
g
ical,
an
d
c
o
m
p
r
e
h
en
s
ib
le
s
u
m
m
ar
ies
th
at
co
n
tai
n
e
s
s
en
tial
id
ea
s
as
au
th
o
r
s
d
eter
m
in
e,
as
estab
lis
h
ed
.
T
h
is
b
alan
ce
d
s
ch
em
e
im
p
r
o
v
es
th
e
co
h
er
en
c
y
an
d
th
e
o
v
er
all
s
tr
u
ctu
r
e
o
f
t
h
e
s
u
m
m
a
r
ies,
wh
er
e
Um
B
E
R
T
ev
e
n
o
u
t
B
E
R
T
f
o
r
d
ev
el
o
p
in
g
u
s
er
-
o
r
i
en
tated
s
u
m
m
ar
ies
th
at
f
it
th
e
r
ea
d
er
s
’
r
eq
u
ir
e
m
en
ts
.
T
h
u
s
,
wh
er
e
as
in
th
e
t
r
ad
itio
n
al
m
ea
s
u
r
es
it
is
b
en
ef
icial
f
o
r
B
E
R
T
,
th
e
m
eth
o
d
th
at
is
b
ein
g
d
ev
elo
p
e
d
f
o
r
r
ea
d
er
-
o
r
ien
ted
s
u
m
m
ar
i
za
tio
n
with
th
e
h
elp
o
f
au
th
o
r
ass
es
s
m
en
t
is
m
o
r
e
b
en
ef
icial
f
o
r
th
e
s
am
e
p
u
r
p
o
s
e
o
f
s
u
m
m
ar
izin
g
r
esear
ch
p
ap
er
s
.
3
.
1
.
Appl
ica
t
io
ns
o
f
t
e
x
t
s
um
m
a
riz
a
t
io
n met
ho
ds
a
nd
f
uture
wo
rk
As
f
o
r
th
e
f
u
r
th
e
r
d
ev
el
o
p
m
e
n
t
o
f
th
is
r
esear
ch
,
it
is
n
o
te
wo
r
th
y
to
ex
p
an
d
t
h
e
u
s
e
o
f
th
is
m
o
d
el
b
ey
o
n
d
id
en
tif
y
in
g
th
e
in
tr
o
d
u
ctio
n
s
o
f
r
esear
c
h
p
a
p
er
s
.
T
h
e
C
AST
ap
p
r
o
ac
h
[
5
0
]
ef
f
ec
tiv
ely
class
if
ies
in
tr
a
-
s
ec
tio
n
an
d
cr
o
s
s
-
s
ec
tio
n
r
elatio
n
s
in
s
cien
tific
ar
ticles,
im
p
r
o
v
in
g
m
u
lti
-
ar
ticle
s
u
m
m
ar
iza
tio
n
an
d
ad
d
r
ess
in
g
r
ed
u
n
d
an
cy
an
d
r
ea
d
ab
ilit
y
is
s
u
es.
T
h
u
s
,
th
e
p
r
o
p
o
s
ed
h
y
b
r
id
ap
p
r
o
ac
h
ca
n
b
e
f
u
r
th
er
ex
te
n
d
ed
to
s
y
n
th
esize
s
u
m
m
ar
ies
f
o
r
s
h
o
r
t
tex
t
d
o
c
u
m
en
ts
u
s
in
g
p
r
etr
ai
n
in
g
lan
g
u
ag
e
m
o
d
els
s
u
ch
as
B
E
R
T
[
5
1
]
a
n
d
PEGA
SUS
[
5
2
]
.
T
h
is
ad
ap
tatio
n
p
r
o
v
e
u
s
ef
u
l
wh
en
s
p
ac
e
is
a
co
n
ce
r
n
s
u
ch
as
in
d
iv
id
u
al
twee
ts
,
p
o
s
t
h
ea
d
lin
es
o
r
n
ews
b
r
ief
s
.
Mo
r
eo
v
e
r
,
th
e
m
eth
o
d
’
s
ab
ilit
y
to
g
en
er
ate
ab
s
tr
ac
tiv
e
s
u
m
m
ar
ies
ca
n
b
e
h
elp
f
u
l
in
d
ev
elo
p
in
g
p
r
o
m
p
ts
f
o
r
id
en
tify
i
n
g
b
ac
k
d
o
o
r
attac
k
s
o
n
lan
g
u
ag
e
m
o
d
els.
Fin
ally
,
th
e
s
tr
en
g
th
s
o
f
b
o
th
e
x
tr
ac
tiv
e
an
d
ab
s
tr
ac
tiv
e
elem
en
ts
in
teg
r
ated
in
th
e
ap
p
r
o
ac
h
allo
w
f
o
r
a
d
v
an
cin
g
t
h
e
s
eq
2
s
eq
m
o
d
els
with
t
h
e
b
en
ef
its
o
f
h
ig
h
er
ef
f
ec
tiv
en
ess
an
d
ac
cu
r
ac
y
in
m
u
ltip
le
NL
P
ap
p
licatio
n
s
.
T
h
is
wo
r
k
p
r
o
v
id
es
a
f
o
u
n
d
atio
n
f
o
r
f
u
tu
r
e
wo
r
k
in
th
e
cr
ea
tio
n
o
f
s
tr
o
n
g
er
s
u
m
m
ar
izatio
n
s
y
s
tem
s
as we
ll a
s
th
e
en
h
an
ce
m
e
n
t o
f
t
h
e
NL
P d
is
cip
lin
e
[
5
3
]
-
[
5
5
].
4.
CO
NCLU
SI
O
N
T
h
e
s
tu
d
y
d
ev
elo
p
s
a
m
ac
h
in
e
-
lear
n
in
g
m
o
d
el
s
p
ec
if
ically
f
o
r
s
u
m
m
ar
izin
g
r
esear
ch
an
d
ac
ad
em
ic
p
ap
er
s
.
I
t
a
p
p
lied
s
u
m
m
a
r
izatio
n
m
o
d
els
—
B
AR
T
,
T
5
,
PEGA
SU
S
,
an
d
B
E
R
T
—
to
eig
h
t
s
am
p
les
f
r
o
m
ac
ad
em
ic
wo
r
k
s
'
I
n
tr
o
d
u
ctio
n
s
ec
tio
n
s
.
B
E
R
T
ac
h
iev
ed
th
e
h
ig
h
est
B
L
E
U
s
co
r
es,
d
em
o
n
s
tr
atin
g
s
u
p
e
r
io
r
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
To
w
a
r
d
s
efficien
t k
n
o
w
led
g
e
ex
tr
a
ctio
n
:
N
a
tu
r
a
l la
n
g
u
a
g
e
p
r
o
ce
s
s
in
g
-
b
a
s
ed
.
.
.
(
N
ikita
C
h
a
u
d
h
a
r
i)
689
p
er
f
o
r
m
an
ce
.
A
h
y
b
r
i
d
m
o
d
el
m
er
g
in
g
B
E
R
T
an
d
PEGA
SUS
was
cr
ea
ted
to
g
en
er
ate
s
u
m
m
ar
ies,
co
m
b
in
in
g
ab
s
tr
ac
tiv
e
an
d
ex
tr
ac
tiv
e
tec
h
n
iq
u
es.
T
h
is
h
y
b
r
id
m
o
d
el
s
h
o
wed
ex
ce
llen
t
R
OUGE
an
d
B
L
E
U
s
co
r
es
an
d
r
ec
eiv
ed
p
o
s
itiv
e
f
ee
d
b
ac
k
.
F
u
tu
r
e
e
n
h
an
ce
m
en
ts
co
u
ld
f
u
r
th
er
o
p
tim
ize
th
e
m
o
d
el
f
o
r
p
r
ac
tical
ap
p
licatio
n
s
in
aid
in
g
ac
ad
em
ic
r
esear
ch
.
E
x
p
lo
r
in
g
d
if
f
e
r
en
t
iter
atio
n
s
o
f
h
y
b
r
id
m
o
d
els
in
teg
r
atin
g
d
iv
er
s
e
tech
n
iq
u
es
r
ep
r
esen
ts
a
p
r
o
m
is
in
g
av
en
u
e
f
o
r
f
u
tu
r
e
r
esear
ch
,
in
clu
d
i
n
g
th
e
s
u
m
m
ar
izatio
n
o
f
s
cien
t
if
ic
s
tu
d
y
tex
tb
o
o
k
s
an
d
o
th
e
r
co
m
p
lex
d
o
cu
m
en
ts
.
RE
F
E
R
E
NC
E
S
[
1
]
W
i
k
i
p
e
d
i
a
C
o
n
t
r
i
b
u
t
o
r
s,
“
A
c
a
d
e
mi
c
p
u
b
l
i
sh
i
n
g
,
”
W
i
k
i
p
e
d
i
a
,
2
0
2
0
.
[
O
n
l
i
n
e
]
.
A
v
a
i
l
a
b
l
e
:
h
t
t
p
s
:
/
/
e
n
.
w
i
k
i
p
e
d
i
a
.
o
r
g
/
w
i
k
i
/
A
c
a
d
e
m
i
c
_
p
u
b
l
i
sh
i
n
g
[
2
]
T.
V
o
l
p
e
a
n
d
J.
S
c
h
o
p
f
e
l
,
“
D
i
ssem
i
n
a
t
i
o
n
o
f
k
n
o
w
l
e
d
g
e
a
n
d
c
o
p
y
r
i
g
h
t
:
A
n
h
i
s
t
o
r
i
c
a
l
c
a
se
st
u
d
y
,
”
J
o
u
rn
a
l
o
f
I
n
f
o
rm
a
t
i
o
n
,
C
o
m
m
u
n
i
c
a
t
i
o
n
a
n
d
Et
h
i
c
s
i
n
S
o
c
i
e
t
y
,
v
o
l
.
1
1
,
n
o
.
3
,
p
p
.
1
4
4
–
1
5
5
,
2
0
1
3
,
d
o
i
:
1
0
.
1
1
0
8
/
JI
C
ES
-
06
-
2
0
1
3
-
0
0
1
8
.
[
3
]
L.
P
a
r
t
r
i
d
g
e
,
“
C
e
l
e
b
r
a
t
i
n
g
3
5
0
y
e
a
r
s
o
f
p
h
i
l
o
s
o
p
h
i
c
a
l
t
r
a
n
s
a
c
t
i
o
n
s:
l
i
f
e
sc
i
e
n
c
e
s
p
a
p
e
r
s,
”
P
h
i
l
o
so
p
h
i
c
a
l
T
r
a
n
s
a
c
t
i
o
n
s
o
f
t
h
e
R
o
y
a
l
S
o
c
i
e
t
y
B:
Bi
o
l
o
g
i
c
a
l
S
c
i
e
n
c
e
s
,
v
o
l
.
3
7
0
,
n
o
.
1
6
6
6
,
2
0
1
5
,
d
o
i
:
1
0
.
1
0
9
8
/
r
st
b
.
2
0
1
4
.
0
3
8
0
.
[
4
]
M
.
F
i
r
e
a
n
d
C
.
G
u
e
st
r
i
n
,
“
O
v
e
r
-
o
p
t
i
mi
z
a
t
i
o
n
o
f
a
c
a
d
e
mi
c
p
u
b
l
i
s
h
i
n
g
me
t
r
i
c
s:
o
b
ser
v
i
n
g
g
o
o
d
h
a
r
t
’
s
l
a
w
i
n
a
c
t
i
o
n
,
”
G
i
g
a
S
c
i
e
n
c
e
,
v
o
l
.
8
,
n
o
.
6
,
2
0
1
9
,
d
o
i
:
1
0
.
1
0
9
3
/
g
i
g
a
s
c
i
e
n
c
e
/
g
i
z
0
5
3
.
[
5
]
J.
T
.
‐
M
o
r
e
n
o
,
“
W
h
y
s
u
mm
a
r
i
z
e
t
e
x
t
s
?
,
”
Au
t
o
m
a
t
i
c
T
e
x
t
S
u
m
m
a
r
i
za
t
i
o
n
,
p
p
.
1
–
2
1
,
2
0
1
4
,
d
o
i
:
1
0
.
1
0
0
2
/
9
7
8
1
1
1
9
0
0
4
7
5
2
.
c
h
1
.
[
6
]
A
.
N
e
n
k
o
v
a
a
n
d
K
.
M
c
K
e
o
w
n
,
“
A
u
t
o
mat
i
c
su
m
mari
z
a
t
i
o
n
,
”
F
o
u
n
d
a
t
i
o
n
s
a
n
d
T
re
n
d
s
i
n
I
n
f
o
rm
a
t
i
o
n
Re
t
ri
e
v
a
l
,
v
o
l
.
5
,
n
o
.
2
–
3
,
p
p
.
1
0
3
–
2
3
3
,
2
0
1
1
,
d
o
i
:
1
0
.
1
5
6
1
/
1
5
0
0
0
0
0
0
1
5
.
[
7
]
A
.
V
a
s
h
i
s
h
t
,
“
L
u
h
n
’
s
h
e
u
r
i
s
t
i
c
m
e
t
h
o
d
f
o
r
t
e
x
t
su
mm
a
r
i
z
a
t
i
o
n
,
”
O
p
e
n
G
e
n
u
s
I
Q
:
C
o
m
p
u
t
i
n
g
Ex
p
e
r
t
i
se
&
L
e
g
a
c
y
,
2
0
1
9
.
[
O
n
l
i
n
e
]
.
A
v
a
i
l
a
b
l
e
:
h
t
t
p
s:
/
/
i
q
.
o
p
e
n
g
e
n
u
s.
o
r
g
/
l
u
h
n
s
-
h
e
u
r
i
s
t
i
c
-
met
h
o
d
-
f
o
r
-
t
e
x
t
-
s
u
mm
a
r
i
z
a
t
i
o
n
/
[
8
]
L.
G
o
n
ç
a
l
v
e
s,
“
A
u
t
o
m
a
t
i
c
t
e
x
t
s
u
mm
a
r
i
z
a
t
i
o
n
w
i
t
h
ma
c
h
i
n
e
l
L
e
a
r
n
i
n
g
—
a
n
o
v
e
r
v
i
e
w
,
”
L
u
i
s
f
re
d
g
s
,
2
0
2
0
,
[
O
n
l
i
n
e
]
.
A
v
a
i
l
a
b
l
e
:
h
t
t
p
s
:
/
/
me
d
i
u
m
.
c
o
m/
l
u
i
sf
r
e
d
g
s
/
a
u
t
o
m
a
t
i
c
-
t
e
x
t
-
su
mm
a
r
i
z
a
t
i
o
n
-
w
i
t
h
-
m
a
c
h
i
n
e
-
l
e
a
r
n
i
n
g
-
an
-
o
v
e
r
v
i
e
w
-
6
8
d
e
d
5
7
1
7
a
2
5
[
9
]
A
.
C
o
h
a
n
e
t
a
l
.
,
“
A
d
i
sc
o
u
r
s
e
-
a
w
a
r
e
a
t
t
e
n
t
i
o
n
m
o
d
e
l
f
o
r
a
b
st
r
a
c
t
i
v
e
s
u
mm
a
r
i
z
a
t
i
o
n
o
f
l
o
n
g
d
o
c
u
me
n
t
s,
”
2
0
1
8
C
o
n
f
e
re
n
c
e
o
f
t
h
e
N
o
r
t
h
Am
e
r
i
c
a
n
C
h
a
p
t
e
r
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
st
i
c
s:
H
u
m
a
n
L
a
n
g
u
a
g
e
T
e
c
h
n
o
l
o
g
i
e
s
,
v
o
l
.
2
,
p
p
.
6
1
5
–
6
2
1
,
2
0
1
8
,
d
o
i
:
1
0
.
1
8
6
5
3
/
v
1
/
n
1
8
-
2
0
9
7
.
[
1
0
]
H
.
Y
.
K
o
h
,
J.
J
u
,
M
.
Li
u
,
S
.
P
a
n
,
a
n
d
S
.
P
a
n
,
“
A
n
e
m
p
i
r
i
c
a
l
s
u
r
v
e
y
o
n
l
o
n
g
d
o
c
u
m
e
n
t
s
u
mm
a
r
i
z
a
t
i
o
n
:
d
a
t
a
s
e
t
s
,
m
o
d
e
l
s,
a
n
d
met
r
i
c
s
,
”
AC
M
C
o
m
p
u
t
i
n
g
S
u
rv
e
y
s
,
v
o
l
.
5
5
,
n
o
.
8
,
2
0
2
2
,
d
o
i
:
1
0
.
1
1
4
5
/
3
5
4
5
1
7
6
.
[
1
1
]
U
.
H
a
h
n
a
n
d
I
.
M
a
n
i
,
“
C
h
a
l
l
e
n
g
e
s
o
f
a
u
t
o
ma
t
i
c
su
mm
a
r
i
z
a
t
i
o
n
,
”
C
o
m
p
u
t
e
r
,
v
o
l
.
3
3
,
n
o
.
1
1
,
p
p
.
2
9
–
3
6
,
2
0
0
0
,
d
o
i
:
1
0
.
1
1
0
9
/
2
.
8
8
1
6
9
2
.
[
1
2
]
F
.
N
a
n
e
t
a
l
.
,
“
I
mp
r
o
v
i
n
g
f
a
c
t
u
a
l
c
o
n
si
s
t
e
n
c
y
o
f
a
b
s
t
r
a
c
t
i
v
e
su
mm
a
r
i
z
a
t
i
o
n
v
i
a
q
u
e
s
t
i
o
n
a
n
sw
e
r
i
n
g
,
”
AC
L
-
I
J
C
N
L
P
2
0
2
1
-
5
9
t
h
An
n
u
a
l
Me
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
a
n
d
t
h
e
1
1
t
h
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
N
a
t
u
r
a
l
L
a
n
g
u
a
g
e
Pro
c
e
ssi
n
g
,
Pro
c
e
e
d
i
n
g
s
o
f
t
h
e
C
o
n
f
e
ren
c
e
,
p
p
.
6
8
8
1
–
6
8
9
4
,
2
0
2
1
,
d
o
i
:
1
0
.
1
8
6
5
3
/
v
1
/
2
0
2
1
.
a
c
l
-
l
o
n
g
.
5
3
6
.
[
1
3
]
C
.
R
a
f
f
e
l
e
t
a
l
.
,
“
E
x
p
l
o
r
i
n
g
t
h
e
l
i
mi
t
s
o
f
t
r
a
n
sf
e
r
l
e
a
r
n
i
n
g
w
i
t
h
a
u
n
i
f
i
e
d
t
e
x
t
-
to
-
t
e
x
t
t
r
a
n
sf
o
r
m
e
r
,
”
J
o
u
r
n
a
l
o
f
M
a
c
h
i
n
e
L
e
a
rn
i
n
g
Re
se
a
rc
h
,
v
o
l
.
2
1
,
2
0
2
0
.
[
1
4
]
J.
Z
h
a
n
g
,
Y
.
Zh
a
o
,
M
.
S
a
l
e
h
,
a
n
d
P
.
J.
Li
u
,
“
P
EG
A
S
U
S
:
P
r
e
-
t
r
a
i
n
i
n
g
w
i
t
h
e
x
t
r
a
c
t
e
d
g
a
p
-
s
e
n
t
e
n
c
e
s fo
r
a
b
s
t
r
a
c
t
i
v
e
su
mm
a
r
i
z
a
t
i
o
n
,
”
3
7
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
M
a
c
h
i
n
e
L
e
a
r
n
i
n
g
,
I
C
ML
2
0
2
0
,
p
p
.
1
1
2
6
5
–
1
1
2
7
6
,
2
0
2
0
.
[
1
5
]
J.
D
e
v
l
i
n
,
M
.
W
.
C
h
a
n
g
,
K
.
Le
e
,
a
n
d
K
.
To
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
me
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
,
”
N
AA
C
L
H
L
T
2
0
1
9
-
2
0
1
9
C
o
n
f
e
re
n
c
e
o
f
t
h
e
N
o
r
t
h
Am
e
ri
c
a
n
C
h
a
p
t
e
r
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
st
i
c
s:
H
u
m
a
n
L
a
n
g
u
a
g
e
T
e
c
h
n
o
l
o
g
i
e
s
,
v
o
l
.
1
,
p
p
.
4
1
7
1
–
4
1
8
6
,
2
0
1
9
,
d
o
i
:
1
0
.
1
8
6
5
3
/
v
1
/
N
1
9
-
1
4
2
3
.
[
1
6
]
M
.
L
e
w
i
s
e
t
a
l
.
,
“
B
A
R
T:
d
e
n
o
i
s
i
n
g
s
e
q
u
e
n
c
e
-
to
-
se
q
u
e
n
c
e
p
r
e
-
t
r
a
i
n
i
n
g
f
o
r
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
,
t
r
a
n
sl
a
t
i
o
n
,
a
n
d
c
o
m
p
r
e
h
e
n
si
o
n
,
”
Pr
o
c
e
e
d
i
n
g
s o
f
t
h
e
An
n
u
a
l
Me
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
s
t
i
c
s
,
p
p
.
7
8
7
1
–
7
8
8
0
,
2
0
2
0
,
d
o
i
:
1
0
.
1
8
6
5
3
/
v
1
/
2
0
2
0
.
a
c
l
-
mai
n
.
7
0
3
.
[
1
7
]
N
.
A
n
d
h
a
l
e
a
n
d
L.
A
.
B
e
w
o
o
r
,
“
A
n
o
v
e
r
v
i
e
w
o
f
t
e
x
t
su
mm
a
r
i
z
a
t
i
o
n
t
e
c
h
n
i
q
u
e
s,
”
2
n
d
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
C
o
m
p
u
t
i
n
g
,
C
o
m
m
u
n
i
c
a
t
i
o
n
,
C
o
n
t
r
o
l
a
n
d
A
u
t
o
m
a
t
i
o
n
,
I
C
C
U
BEA
2
0
1
6
,
2
0
1
7
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
C
U
B
EA
.
2
0
1
6
.
7
8
6
0
0
2
4
.
[
1
8
]
M
.
A
.
F
a
t
t
a
h
a
n
d
F
.
R
e
n
,
“
A
u
t
o
m
a
t
i
c
t
e
x
t
s
u
mm
a
r
i
z
a
t
i
o
n
,
”
W
o
rl
d
Ac
a
d
e
m
y
o
f
S
c
i
e
n
c
e
,
E
n
g
i
n
e
e
r
i
n
g
a
n
d
T
e
c
h
n
o
l
o
g
y
,
I
n
t
e
rn
a
t
i
o
n
a
l
J
o
u
rn
a
l
o
f
C
o
m
p
u
t
e
r,
El
e
c
t
r
i
c
a
l
,
A
u
t
o
m
a
t
i
o
n
,
C
o
n
t
r
o
l
a
n
d
I
n
f
o
rm
a
t
i
o
n
En
g
i
n
e
e
ri
n
g
,
v
o
l
.
2
,
n
o
.
1
,
p
p
.
9
0
–
9
3
,
2
0
0
8
.
[
1
9
]
D
.
D
e
n
o
m
me,
N
.
Ta
c
c
o
n
e
,
a
n
d
M
.
F
i
r
e
st
o
n
e
,
“
S
u
mm
a
r
y
d
e
f
i
n
i
t
i
o
n
,
t
y
p
e
s
&
e
x
a
m
p
l
e
s,
”
S
t
u
d
y
L
e
a
rn
L
e
ss
o
n
,
2
0
2
3
.
[
O
n
l
i
n
e
]
.
A
v
a
i
l
a
b
l
e
:
h
t
t
p
s:
/
/
s
t
u
d
y
.
c
o
m
/
l
e
a
r
n
/
l
e
s
so
n
/
w
h
a
t
-
is
-
a
-
su
m
mary
.
h
t
m
l
[
2
0
]
S
.
G
h
o
s
h
,
M
.
D
u
t
t
a
,
a
n
d
T.
D
a
s,
“
I
n
d
i
a
n
l
e
g
a
l
t
e
x
t
s
u
mm
a
r
i
z
a
t
i
o
n
:
a
t
e
x
t
n
o
r
mal
i
z
a
t
i
o
n
-
b
a
se
d
a
p
p
r
o
a
c
h
,
”
I
N
D
I
C
O
N
2
0
2
2
-
2
0
2
2
I
EEE
1
9
t
h
I
n
d
i
a
C
o
u
n
c
i
l
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
,
2
0
2
2
,
d
o
i
:
1
0
.
1
1
0
9
/
I
N
D
I
C
O
N
5
6
1
7
1
.
2
0
2
2
.
1
0
0
3
9
8
9
1
.
[
2
1
]
Y
.
Jaa
f
a
r
a
n
d
K
.
B
o
u
z
o
u
b
a
a
,
“
T
o
w
a
r
d
s
a
n
e
w
h
y
b
r
i
d
a
p
p
r
o
a
c
h
f
o
r
a
b
st
r
a
c
t
i
v
e
su
m
mari
z
a
t
i
o
n
,
”
P
ro
c
e
d
i
a
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
,
v
o
l
.
1
4
2
,
p
p
.
2
8
6
–
2
9
3
,
2
0
1
8
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
p
r
o
c
s
.
2
0
1
8
.
1
0
.
4
9
6
.
[
2
2
]
A
.
G
u
p
t
a
,
D
.
C
h
u
g
h
,
A
n
j
u
m
,
a
n
d
R
.
K
a
t
a
r
y
a
,
“
A
u
t
o
mat
e
d
n
e
w
s
su
m
mariz
a
t
i
o
n
u
si
n
g
t
r
a
n
sf
o
r
m
e
r
s,”
S
u
st
a
i
n
a
b
l
e
A
d
v
a
n
c
e
d
C
o
m
p
u
t
i
n
g
,
v
o
l
.
8
4
0
,
p
p
.
2
4
9
–
2
5
9
,
2
0
2
2
,
d
o
i
:
1
0
.
1
0
0
7
/
9
7
8
-
9
8
1
-
16
-
9
0
1
2
-
9
_
2
1
.
[
2
3
]
G
.
D
.
F
a
b
b
r
i
z
i
o
,
A
.
J
.
S
t
e
n
t
,
a
n
d
R
.
G
a
i
z
a
u
s
k
a
s
,
“
A
h
y
b
r
i
d
a
p
p
r
o
a
c
h
t
o
mu
l
t
i
-
d
o
c
u
me
n
t
s
u
mm
a
r
i
z
a
t
i
o
n
o
f
o
p
i
n
i
o
n
s
i
n
r
e
v
i
e
w
s
,
”
I
N
L
G
2
0
1
4
-
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
8
t
h
I
n
t
e
r
n
a
t
i
o
n
a
l
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
C
o
n
f
e
re
n
c
e
-
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
I
N
L
G
a
n
d
S
I
G
D
I
AL 2
0
1
4
J
o
i
n
t
S
e
ssi
o
n
,
p
p
.
5
4
–
6
3
,
2
0
1
4
,
d
o
i
:
1
0
.
3
1
1
5
/
v
1
/
w
1
4
-
4
4
0
8
.
[
2
4
]
M
.
A
.
F
a
t
t
a
h
,
“
A
h
y
b
r
i
d
m
a
c
h
i
n
e
l
e
a
r
n
i
n
g
m
o
d
e
l
f
o
r
m
u
l
t
i
-
d
o
c
u
m
e
n
t
su
m
m
a
r
i
z
a
t
i
o
n
,
”
A
p
p
l
i
e
d
I
n
t
e
l
l
i
g
e
n
c
e
,
v
o
l
.
4
0
,
n
o
.
4
,
p
p
.
5
9
2
–
6
0
0
,
2
0
1
4
,
d
o
i
:
1
0
.
1
0
0
7
/
s
1
0
4
8
9
-
013
-
0
4
9
0
-
0.
[
2
5
]
A
.
C
e
l
i
k
y
i
l
ma
z
a
n
d
D
.
H
.
-
T
u
r
,
“
A
h
y
b
r
i
d
h
i
e
r
a
r
c
h
i
c
a
l
m
o
d
e
l
f
o
r
m
u
l
t
i
-
d
o
c
u
me
n
t
s
u
mm
a
r
i
z
a
t
i
o
n
,
”
Pro
c
e
e
d
i
n
g
s
o
f
t
h
e
A
n
n
u
a
l
Me
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
,
v
o
l
.
2
0
1
0
,
p
p
.
8
1
5
–
8
2
4
,
2
0
1
0
,
d
o
i
:
1
0
.
2
6
7
1
3
/
j
i
m
s.
v
9
i
3
.
7
6
0
.
[
2
6
]
S
.
M
a
n
d
a
l
,
G
.
K
.
S
i
n
g
h
,
a
n
d
A
.
P
a
l
,
“
A
h
y
b
r
i
d
t
e
x
t
s
u
mm
a
r
i
z
a
t
i
o
n
a
p
p
r
o
a
c
h
,
”
J
o
u
r
n
a
l
o
f
I
n
f
o
rm
a
t
i
c
s
a
n
d
M
a
t
h
e
m
a
t
i
c
a
l
S
c
i
e
n
c
e
s
,
v
o
l
.
9
,
n
o
.
3
,
p
p
.
5
4
7
–
5
5
5
,
2
0
1
7
.
[
2
7
]
M
.
K
i
r
m
a
n
i
,
N
.
M
.
H
a
k
a
k
,
M
.
M
o
h
d
,
a
n
d
M
.
M
o
h
d
,
“
H
y
b
r
i
d
t
e
x
t
s
u
mm
a
r
i
z
a
t
i
o
n
:
a
s
u
r
v
e
y
,
”
Ad
v
a
n
c
e
s
i
n
I
n
t
e
l
l
i
g
e
n
t
S
y
s
t
e
m
s
a
n
d
C
o
m
p
u
t
i
n
g
,
v
o
l
.
7
4
2
,
p
p
.
6
3
–
7
3
,
2
0
1
9
,
d
o
i
:
1
0
.
1
0
0
7
/
9
7
8
-
9
8
1
-
13
-
0
5
8
9
-
4
_
7
.
[
2
8
]
I
.
A
w
a
st
h
i
,
K
.
G
u
p
t
a
,
P
.
S
.
B
h
o
g
a
l
,
S
.
S
.
A
n
a
n
d
,
a
n
d
P
.
K
.
S
o
n
i
,
“
N
a
t
u
r
a
l
l
a
n
g
u
a
g
e
p
r
o
c
e
ssi
n
g
(
N
LP)
b
a
se
d
t
e
x
t
su
mm
a
r
i
z
a
t
i
o
n
-
a
su
r
v
e
y
,
”
T
h
e
6
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
I
n
v
e
n
t
i
v
e
C
o
m
p
u
t
a
t
i
o
n
T
e
c
h
n
o
l
o
g
i
e
s
,
p
p
.
1
3
1
0
–
1
3
1
7
,
2
0
2
1
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
I
C
T5
0
8
1
6
.
2
0
2
1
.
9
3
5
8
7
0
3
.
[
2
9
]
H
.
C
h
r
i
st
e
n
s
e
n
,
B
.
K
o
l
l
u
r
u
,
Y
.
G
o
t
o
h
,
a
n
d
S
.
R
e
n
a
i
s
,
“
F
r
o
m
t
e
x
t
s
u
mm
a
r
i
sa
t
i
o
n
t
o
s
t
y
l
e
-
s
p
e
c
i
f
i
c
su
m
maris
a
t
i
o
n
f
o
r
b
r
o
a
d
c
a
s
t
n
e
w
s
,
”
Ad
v
a
n
c
e
s i
n
I
n
f
o
rm
a
t
i
o
n
R
e
t
r
i
e
v
a
l
(
E
C
I
R
2
0
0
4
)
,
v
o
l
.
2
9
9
7
,
p
p
.
2
2
3
–
2
3
7
,
2
0
0
4
,
d
o
i
:
1
0
.
1
0
0
7
/
9
7
8
-
3
-
5
4
0
-
2
4
7
5
2
-
4
_
1
7
.
Evaluation Warning : The document was created with Spire.PDF for Python.