I
AE
S
I
nte
rna
t
io
na
l J
o
urna
l o
f
Art
if
icia
l In
t
ellig
ence
(
I
J
-
AI
)
Vo
l.
3
,
No
.
3
,
Sep
tem
b
er
201
4
,
p
p
.
112
~
120
I
SS
N:
2252
-
8938
112
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ia
e
s
jo
u
r
n
a
l.c
o
m/o
n
lin
e/in
d
ex
.
p
h
p
/I
J
AI
Effec
t
o
f
Featur
e
Selection o
n
S
m
a
l
l a
nd La
rg
e Docu
m
en
t
Su
m
m
a
ri
z
a
tion
D
.
Y
.
Sa
kh
a
re
*
Ra
j
k
u
m
a
r
**
*
R
e
se
a
rc
h
sc
h
o
lar Bh
a
ra
ti
V
e
e
d
y
a
p
e
e
th
d
e
e
m
e
d
u
n
iv
e
rsity
,
P
u
n
e
,
A
d
d
re
ss
M
a
h
a
ra
sh
tra,
In
d
ia
**
D
RD
O S
c
ien
ti
st ‘
D’,
DIA
T
,
P
u
n
e
M
a
h
a
ra
sh
tra
,
I
n
d
ia
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Ma
y
2
2
,
2
0
1
4
R
ev
i
s
ed
A
u
g
8
,
2
0
1
4
A
cc
ep
ted
A
u
g
2
0
,
2
0
1
4
A
s
th
e
a
m
o
u
n
t
o
f
tex
tu
a
l
In
f
o
rm
a
ti
o
n
i
n
c
re
a
se
s,
w
e
e
x
p
e
rien
c
e
a
n
e
e
d
f
o
r
A
u
to
m
a
ti
c
T
e
x
t
S
u
m
m
a
ri
z
e
rs.
In
A
u
to
m
a
ti
c
su
m
m
a
riza
ti
o
n
a
tex
t
d
o
c
u
m
e
n
t
o
r
a
larg
e
r
c
o
rp
u
s
o
f
m
u
lt
ip
le
d
o
c
u
m
e
n
ts
a
r
e
re
d
u
c
e
d
to
a
sh
o
rt
se
t
o
f
w
o
rd
s
o
r
p
a
ra
g
ra
p
h
t
h
a
t
c
o
n
v
e
y
s
th
e
m
a
in
m
e
a
n
in
g
o
f
th
e
tex
t
S
u
m
m
a
ri
z
a
ti
o
n
c
a
n
b
e
c
las
si
f
ied
in
to
tw
o
a
p
p
ro
a
c
h
e
s:
e
x
tra
c
ti
o
n
a
n
d
a
b
stra
c
ti
o
n
.
T
h
is
p
a
p
e
r
f
o
c
u
se
s
o
n
e
x
tra
c
ti
o
n
a
p
p
r
o
a
c
h
.
T
h
e
g
o
a
l
o
f
te
x
t
su
m
m
a
riz
a
ti
o
n
b
a
se
d
o
n
e
x
trac
ti
o
n
a
p
p
r
o
a
c
h
is
se
n
te
n
c
e
s
se
lec
ti
o
n
.
T
h
e
f
irst
ste
p
in
su
m
m
a
r
iza
ti
o
n
b
y
e
x
trac
ti
o
n
is
th
e
id
e
n
ti
f
ica
ti
o
n
o
f
im
p
o
rtan
t
f
e
a
tu
re
s.
In
o
u
r
a
p
p
r
o
a
c
h
sh
o
rt
sto
ries
a
n
d
b
io
g
ra
p
h
ies
a
re
u
se
d
a
s
tes
t
d
o
c
u
m
e
n
ts.
Eac
h
d
o
c
u
m
e
n
t
is
p
re
p
a
re
d
b
y
p
re
-
p
ro
c
e
ss
in
g
p
ro
c
e
ss
:
se
n
ten
c
e
s
e
g
m
e
n
tatio
n
,
t
o
k
e
n
iza
ti
o
n
,
sto
p
w
o
rd
re
m
o
v
a
l,
c
a
se
f
o
ld
in
g
,
le
m
m
a
ti
z
a
ti
o
n
,
a
n
d
ste
m
m
in
g
.
Th
e
n
,
u
si
n
g
im
p
o
rtan
t
f
e
a
tu
re
s,
se
n
ten
c
e
f
i
lt
e
rin
g
,
d
a
ta
c
o
m
p
re
ss
io
n
a
n
d
f
in
a
ll
y
c
a
lcu
latin
g
sc
o
re
f
o
r
e
a
c
h
se
n
ten
c
e
is
d
o
n
e
.
In
t
h
is
p
a
p
e
r
w
e
p
ro
p
o
s
e
d
v
a
rio
u
s
f
e
a
tu
re
s
o
f
S
u
m
m
a
r
y
Ex
tr
a
c
ti
o
n
a
n
d
a
lso
a
n
a
ly
z
e
d
f
e
a
tu
re
s
th
a
t
a
re
to
b
e
a
p
p
li
e
d
d
e
p
e
n
d
in
g
u
p
o
n
th
e
siz
e
o
f
th
e
Do
c
u
m
e
n
t.
T
h
e
e
x
p
e
ri
m
e
n
tatio
n
is
p
e
rf
o
r
m
e
d
w
it
h
th
e
DU
C
2
0
0
2
d
a
tas
e
t.
T
h
e
c
o
m
p
a
ra
ti
v
e
re
su
lt
s
o
f
th
e
p
ro
p
o
se
d
a
p
p
ro
a
c
h
a
n
d
th
a
t
o
f
M
S
-
W
o
rd
a
re
a
lso
p
re
se
n
ted
h
e
re
.
T
h
e
c
o
n
c
e
p
t
b
a
se
d
f
e
a
tu
re
s
a
re
g
iv
e
n
m
o
re
w
e
i
g
h
tag
e
.
F
ro
m
th
e
se
re
su
lt
s
w
e
p
ro
p
o
se
th
a
t
u
se
o
f
th
e
c
o
n
c
e
p
t
b
a
se
d
f
e
a
tu
re
s
h
e
lp
s
in
im
p
ro
v
in
g
th
e
q
u
a
li
ty
o
f
th
e
su
m
m
a
r
y
in
c
a
se
o
f
larg
e
d
o
c
u
m
e
n
ts.
K
ey
w
o
r
d
:
A
b
s
tr
ac
tio
n
DUC 2
0
0
2
E
x
tr
ac
tio
n
F
ea
tu
r
es
Su
m
m
ar
izer
s
Co
p
y
rig
h
t
©
2
0
1
4
In
stit
u
te o
f
A
d
v
a
n
c
e
d
E
n
g
i
n
e
e
rin
g
a
n
d
S
c
ien
c
e
.
Al
l
rig
h
ts
re
se
rv
e
d
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
D
.
Y
.
Sak
h
ar
e
,
R
esear
ch
s
ch
o
lar
B
h
ar
ati
V
ee
d
y
ap
ee
t
h
Dee
m
ed
Un
i
v
er
s
it
y
,
P
u
n
e,
Ma
h
ar
as
h
tr
a,
I
n
d
ia
.
E
m
ail
:
d
ip
ti
y
s
ak
h
ar
e
@
g
m
ail.
co
m
1.
I
NT
RO
D
UCT
I
O
N
No
w
ad
a
y
s
,
e
n
o
r
m
o
u
s
a
m
o
u
n
t
o
f
d
i
g
itall
y
s
to
r
ed
in
f
o
r
m
ati
o
n
is
av
ai
lab
le
o
n
in
ter
n
et.
.
I
n
o
r
d
er
to
p
r
ev
en
t
s
i
n
k
i
n
g
i
n
it,
f
ilter
i
n
g
an
d
ex
tr
ac
tio
n
o
f
in
f
o
r
m
a
tio
n
ar
e
n
ec
es
s
ar
y
.
A
s
ig
n
i
f
ica
n
t
an
d
o
p
p
o
r
tu
n
e
to
o
l
th
at
as
s
is
t
s
an
d
in
ter
p
r
ets
h
u
g
e
q
u
an
titi
es
o
f
tex
t
p
r
ese
n
te
d
in
d
o
cu
m
e
n
ts
is
a
u
to
m
atic
tex
t
s
u
m
m
ar
iza
tio
n
(
A
T
S).
T
h
e
o
b
j
ec
tiv
e
o
f
A
T
S
is
to
m
ak
e
a
b
r
ief
v
er
s
io
n
o
f
t
h
e
o
r
ig
in
al
te
x
t
w
it
h
th
e
m
o
s
t
s
ig
n
if
ican
t
in
f
o
r
m
atio
n
at
t
h
e
s
a
m
e
ti
m
e
r
etain
i
n
g
its
m
ai
n
co
n
ten
t
an
d
to
en
ab
le
th
e
u
s
er
to
q
u
ick
l
y
co
m
p
r
eh
e
n
d
h
u
g
e
q
u
an
tit
ies
o
f
in
f
o
r
m
at
io
n
[
1
]
.
T
h
e
s
u
m
m
ar
y
s
h
o
u
ld
m
ee
t
t
h
e
m
aj
o
r
co
n
ce
p
t
s
o
f
th
e
o
r
i
g
in
a
l
d
o
cu
m
e
n
t
s
et,
s
h
o
u
ld
b
e
r
ed
u
n
d
an
t
-
less
a
n
d
o
r
d
er
ed
.
T
h
ese
attr
ib
u
tes
ar
e
th
e
b
asis
o
f
th
e
g
en
er
at
io
n
p
r
o
ce
s
s
o
f
th
e
s
u
m
m
ar
y
.
T
h
e
q
u
alit
y
o
f
s
u
m
m
ar
y
i
s
s
e
n
s
i
tiv
e
f
o
r
th
o
s
e
a
ttrib
u
tes
r
elati
n
g
to
h
o
w
t
h
e
s
e
n
t
en
ce
s
ar
e
s
co
r
ed
o
n
th
e
b
asi
s
o
f
th
e
e
m
p
lo
y
ed
f
ea
t
u
r
es.
C
o
n
s
eq
u
en
t
l
y
,
t
h
e
e
s
ti
m
atio
n
o
f
th
e
e
f
f
icac
y
o
f
ea
ch
a
ttrib
u
te
co
u
ld
r
es
u
l
t
th
e
m
ec
h
a
n
is
m
to
d
is
ti
n
g
u
i
s
h
t
h
e
attr
ib
u
te
s
p
o
s
s
es
s
in
g
h
ig
h
p
r
io
r
ity
a
n
d
lo
w
p
r
io
r
it
y
[
1
]
.
Sin
g
le
d
o
cu
m
en
t
s
u
m
m
ar
izati
o
n
is
th
e
p
r
o
ce
s
s
o
f
cr
ea
tin
g
a
s
u
m
m
ar
y
f
r
o
m
a
s
in
g
le
te
x
t
d
o
cu
m
en
t.
Mu
lti
-
d
o
cu
m
en
t
s
u
m
m
ar
izati
o
n
s
h
o
r
ten
s
a
co
llectio
n
o
f
r
elate
d
d
o
cu
m
e
n
ts
;
i
n
to
s
in
g
le
s
u
m
m
ar
y
.
U
s
er
-
f
o
cu
s
ed
s
u
m
m
ar
ie
s
co
n
tai
n
in
f
o
r
m
atio
n
m
o
s
t
r
elev
a
n
t
to
th
e
in
itial
s
ea
r
ch
q
u
er
y
;
w
h
er
ea
s
g
en
er
ic
s
u
m
m
ar
ies
Evaluation Warning : The document was created with Spire.PDF for Python.
IJ
-
AI
I
SS
N:
2252
-
8938
E
ffect
o
f F
ea
tu
r
e
S
elec
tio
n
o
n
S
ma
ll a
n
d
La
r
g
e
Do
cu
men
t
S
u
mma
r
iz
a
tio
n
…
(
D
.
Y
.
S
a
kh
a
r
e
)
113
co
n
tain
in
f
o
r
m
at
io
n
ab
o
u
t
th
e
o
v
er
all
p
er
ce
p
tio
n
o
f
t
h
e
d
o
c
u
m
e
n
t
’
s
co
n
te
n
t.
A
b
s
tr
ac
ti
v
e
s
u
m
m
ar
y
m
et
h
o
d
s
g
en
er
ate
ab
s
tr
ac
ts
b
y
ex
a
m
i
n
in
g
a
n
d
i
n
ter
p
r
etin
g
th
e
tex
t
u
til
izin
g
li
n
g
u
is
tic
m
et
h
o
d
s
.
E
x
tr
ac
ti
v
e
s
u
m
m
ar
izatio
n
m
et
h
o
d
s
s
elec
t
th
e
b
est
-
s
co
r
in
g
s
e
n
ten
ce
s
f
r
o
m
t
h
e
o
r
ig
in
a
l
d
o
cu
m
e
n
t
b
ased
o
n
a
s
et
o
f
ex
tr
ac
tio
n
cr
iter
ia
an
d
p
r
esen
t th
e
m
in
t
h
e
s
u
m
m
ar
y
[
2
]
.
Au
to
m
a
tic
te
x
t
s
u
m
m
ar
izatio
n
i
s
u
tili
ze
d
i
n
a
v
ar
iet
y
o
f
a
p
p
licatio
n
s
,
i
n
cl
u
d
in
g
s
ea
r
c
h
en
g
i
n
e
h
i
t
s
u
m
m
ar
izatio
n
(
s
u
m
m
ar
izin
g
th
e
in
f
o
r
m
a
tio
n
i
n
a
h
it
l
is
t
f
e
tch
ed
b
y
ce
r
tain
s
ea
r
c
h
en
g
i
n
e)
;
p
h
y
s
icia
n
s
’
aid
s
(
to
s
u
m
m
ar
ize
a
n
d
co
m
p
ar
e
t
h
e
p
r
escr
ib
ed
tr
ea
t
m
e
n
t
s
f
o
r
a
p
atien
t)
;
cr
ea
ti
n
g
t
h
e
b
r
ie
f
o
f
a
b
o
o
k
an
d
s
o
o
n
[
3
]
.
B
est
p
er
f
o
r
m
a
n
ce
o
f
a
u
t
o
m
a
tic
tex
t
s
u
m
m
ar
izatio
n
is
ac
h
iev
ed
i
f
th
e
d
o
cu
m
e
n
t
is
w
ell
-
s
tr
u
ct
u
r
ed
,
f
o
r
ex
a
m
p
le
n
e
w
s
,
r
ep
o
r
ts
,
ar
ticles
an
d
s
cie
n
ti
f
ic
p
ap
er
s
[
4
]
.
No
r
m
a
ll
y
,
a
u
to
m
atic
d
o
cu
m
en
t
s
u
m
m
ar
izatio
n
ac
ce
p
ts
o
n
e
o
r
m
o
r
e
s
o
u
r
ce
d
o
cu
m
e
n
ts
as
in
p
u
t
a
n
d
p
r
o
v
id
es
an
ele
g
an
t
s
u
m
m
ar
y
a
s
o
u
t
p
u
t
to
t
h
e
u
s
er
b
y
ex
tr
ac
ti
n
g
th
e
g
is
t o
f
t
h
e
s
o
u
r
c
e(
s
)
.
T
h
e
p
r
o
ce
s
s
co
n
s
i
s
ts
o
f
t
h
r
ee
p
h
a
s
es,
n
a
m
el
y
,
an
al
y
s
is
,
tr
an
s
f
o
r
m
at
io
n
a
n
d
s
y
n
t
h
esi
s
.
I
n
t
h
e
a
n
al
y
s
i
s
p
h
a
s
e,
a
s
m
all
n
u
m
b
er
o
f
s
i
g
n
if
i
ca
n
t
f
ea
t
u
r
es
ar
e
ch
o
s
en
b
y
a
n
al
y
z
in
g
t
h
e
i
n
p
u
t
d
o
cu
m
en
t.
I
n
th
e
tr
an
s
f
o
r
m
a
tio
n
p
h
ase
a
s
u
m
m
ar
y
co
r
r
esp
o
n
d
in
g
to
th
e
u
s
er
’
s
n
ee
d
is
g
en
er
ate
d
b
y
tr
an
s
f
o
r
m
i
n
g
t
h
e
o
u
tp
u
t
o
f
th
e
an
al
y
s
is
p
h
ase.
Feat
u
r
es
s
e
lecte
d
ar
e
s
ig
n
i
f
ica
n
t
f
ac
to
r
s
th
at
in
f
l
u
e
n
ce
th
e
o
v
er
all
q
u
alit
y
o
f
th
e
s
u
m
m
a
r
y
.
I
n
t
h
i
s
p
r
o
p
o
s
ed
w
o
r
k
th
e
ef
f
ec
t
o
f
f
ea
t
u
r
e
s
elec
tio
n
o
n
s
u
m
m
ar
izatio
n
is
ev
alu
a
ted
.
T
h
e
r
est
o
f
th
e
p
ap
er
is
o
r
g
an
ized
as
f
o
llo
w
s
:
Sectio
n
2
d
escr
ib
es
th
e
r
ev
ie
w
o
f
r
ec
en
t
w
o
r
k
s
p
r
esen
ted
in
t
h
e
liter
atu
r
e.
Se
ctio
n
3
d
escr
ib
es
th
e
p
r
e
-
p
r
o
ce
s
s
i
n
g
s
tep
.
Sectio
n
4
p
r
esen
t
s
th
e
m
at
h
e
m
a
tical
m
o
d
ell
in
g
f
o
r
f
ea
tu
r
e
s
e
lectio
n
.
Sectio
n
5
p
r
esen
t
s
th
e
r
es
u
lt
s
an
d
d
is
cu
s
s
io
n
.
Sectio
n
6
co
n
clu
d
e
s
th
e
p
ap
er
.
2.
L
I
T
E
R
AT
U
RE
SU
RVE
Y
Au
to
m
a
ted
tex
t
s
u
m
m
ar
izatio
n
is
a
n
o
ld
e
m
in
e
n
t
r
esear
c
h
ar
ea
an
d
d
ates
b
ac
k
to
t
h
e
1
9
5
0
s
.
A
s
a
r
esu
lt
o
f
t
h
e
i
n
f
o
r
m
atio
n
o
v
er
l
o
ad
in
g
o
n
th
e
w
eb
th
er
e
is
lar
g
e
-
s
ca
le
i
n
ter
est
i
n
au
to
m
atic
tex
t
s
u
m
m
ar
izatio
n
d
u
r
in
g
th
e
s
e
d
a
y
s
.
T
h
e
ea
r
ly
w
o
r
k
o
n
s
i
n
g
le
-
d
o
cu
m
e
n
t
s
u
m
m
ar
izatio
n
w
a
s
d
o
n
e
b
y
L
u
h
n
[
3
]
.
He
p
r
esen
ted
a
m
et
h
o
d
o
f
au
to
m
at
ic
ab
s
tr
ac
tin
g
i
n
th
e
y
ea
r
1
9
5
8
.
T
h
is
alg
o
r
ith
m
s
ca
n
s
th
e
o
r
ig
i
n
al
te
x
t
d
o
cu
m
en
t
f
o
r
th
e
m
o
s
t
i
m
p
o
r
tan
t
i
n
f
o
r
m
atio
n
.
T
h
e
f
ea
tu
r
es
u
s
ed
h
er
e
ar
e
w
o
r
d
f
r
eq
u
en
c
y
a
n
d
s
e
n
ten
ce
s
co
r
in
g
.
Dep
en
d
i
n
g
o
n
a
th
r
es
h
o
ld
v
al
u
e
f
o
r
i
m
p
o
r
tan
t
f
ac
to
r
s
t
h
e
f
ea
t
u
r
ed
s
en
te
n
ce
s
ar
e
ex
tr
ac
ted
.
T
h
e
W
ea
k
n
es
s
o
f
th
is
s
y
s
te
m
i
s
t
h
e
s
u
m
m
ar
y
p
r
o
d
u
ce
d
lack
s
in
q
u
alit
y
.
T
h
e
s
y
s
te
m
w
as
r
e
s
tr
icted
to
o
f
e
w
s
p
ec
i
f
ic
d
o
m
ai
n
s
o
f
liter
at
u
r
e.
B
ax
en
d
ale
[
4
]
u
s
ed
ed
s
en
te
n
ce
p
o
s
itio
n
as
a
f
ea
t
u
r
e
to
ex
tr
ac
t
i
m
p
o
r
tan
t
p
ar
ts
o
f
d
o
cu
m
e
n
ts
.
E
d
m
u
n
d
s
o
n
[
5
]
p
r
o
p
o
s
ed
th
e
co
n
ce
p
t
o
f
cu
e
w
o
r
d
s
.
T
h
e
s
tr
en
g
th
o
f
E
d
m
u
n
d
s
o
n
’s
ap
p
r
o
ac
h
w
as
t
h
e
in
tr
o
d
u
ctio
n
to
f
ea
t
u
r
es
lik
e
s
e
n
te
n
ce
p
o
s
itio
n
i
n
tex
t,
cu
e
w
o
r
d
s
a
n
d
titl
e
an
d
h
ea
d
i
n
g
w
o
r
d
s
[
5
]
.
P
o
llo
ck
[
6
]
Used
s
en
ten
ce
r
ejec
tio
n
al
g
o
r
ith
m
.
T
h
e
ai
m
o
f
t
h
e
p
ap
er
w
as
to
d
ev
elo
p
a
s
y
s
te
m
w
h
ic
h
o
u
tp
u
ts
a
s
u
m
m
ar
y
co
n
f
o
r
m
in
g
to
th
e
s
ta
n
d
ar
d
s
o
f
th
e
C
h
e
m
ical
A
b
s
tr
ac
ts
Ser
v
ice
(
C
A
S).
T
h
e
ab
s
tr
ac
tiv
e
s
u
m
m
ar
y
g
e
n
er
atio
n
w
a
s
p
io
n
ee
r
ed
b
y
A
D
A
M
Su
m
m
ar
izer
[
7
]
.
Ma
ch
in
e
L
ea
r
n
i
n
g
f
r
a
m
e
w
o
r
k
i
s
u
s
ed
to
g
e
n
er
ate
s
u
m
m
ar
ies
u
s
i
n
g
s
e
n
te
n
ce
r
an
k
i
n
g
.
T
h
e
s
tr
en
g
t
h
o
f
t
h
is
ap
p
r
o
ac
h
w
a
s
it
’
s
p
o
ten
tial
to
h
a
n
d
le
n
e
w
d
o
m
ain
s
i
n
ad
d
itio
n
to
r
ed
u
n
d
an
c
y
eli
m
i
n
at
io
n
.
K.
R
.
Mc
Keo
w
n
i
n
h
i
s
t
h
esi
s
[
7
]
g
en
er
ated
t
h
e
s
u
m
m
ar
y
s
y
s
te
m
u
s
i
n
g
Nat
u
r
al
L
an
g
u
ag
e
P
r
o
ce
s
s
in
g
(
NL
P
)
.
T
h
e
ap
p
r
o
ac
h
w
as
b
ased
o
n
a
co
m
p
u
tatio
n
al
m
o
d
el
o
f
d
is
co
u
r
s
e
an
al
y
s
is
.
[
1
1
]
P
r
esen
ted
T
er
m
W
ei
g
h
t
in
g
a
n
d
Se
n
te
n
ce
W
eig
h
ti
n
g
as
i
m
p
o
r
tan
t
f
ea
t
u
r
es
to
r
ec
o
g
n
ize
t
h
e
f
ea
t
u
r
ed
s
en
ten
ce
s
.
I
t
h
as
al
s
o
ad
d
r
ess
ed
th
e
p
r
o
b
le
m
o
f
a
n
ap
h
o
r
a
r
eso
lu
tio
n
.
B
o
g
u
r
ae
v
&
Ken
n
ed
y
[
1
0
]
,
Me
r
ce
r
[
9
]
in
1
9
9
7
,
T
r
u
n
e
y
a
n
d
Fra
n
k
[
8
]
in
1
9
9
9
,
all
o
f
th
e
m
u
s
ed
k
e
y
p
h
r
a
s
es
ex
tr
ac
tio
n
as
a
s
u
p
er
v
is
ed
lear
n
in
g
ta
s
k
.
Fo
r
th
ese
s
y
s
t
e
m
s
a
s
ep
ar
ate
tr
ain
i
n
g
d
o
cu
m
en
t
s
et
w
it
h
alr
ea
d
y
as
s
i
g
n
ed
k
e
y
p
h
r
a
s
es
i
s
r
eq
u
ir
ed
to
f
u
n
ct
io
n
p
r
o
p
er
ly
.
T
h
is
is
ag
ai
n
an
o
p
en
c
h
alle
n
g
e
f
o
r
r
esear
ch
co
m
m
u
n
it
y
.
C
u
t
a
n
d
P
aste
[
1
2
]
is
th
e
f
ir
s
t
d
o
m
ai
n
in
d
ep
en
d
e
n
t
ab
s
tr
ac
tiv
e
s
u
m
m
ar
izatio
n
to
o
l.
T
h
is
w
a
s
d
ev
el
o
p
ed
u
s
in
g
s
en
te
n
ce
r
e
d
u
ctio
n
a
n
d
s
en
ten
ce
co
m
b
i
n
atio
n
tec
h
n
iq
u
e
s
.
Her
e
a
s
en
ten
ce
e
x
tr
ac
tio
n
alg
o
r
ith
m
w
as
i
m
p
le
m
e
n
ted
a
lo
n
g
w
it
h
o
t
h
er
f
ea
tu
r
e
s
li
k
e
lex
ical
co
h
er
en
ce
,
tf
×
id
f
s
co
r
e,
cu
e
p
h
r
ases
an
d
s
en
te
n
ce
p
o
s
itio
n
s
etc.
ME
A
D
[
1
3
]
w
as
a
m
u
lti
d
o
cu
m
e
n
t
s
u
m
m
ar
izatio
n
to
o
lk
i
t
it
h
as
u
s
ed
m
u
lt
ip
le
p
o
s
itio
n
-
b
ased
,
T
F×I
DF,
lar
g
est
co
m
m
o
n
s
u
b
s
eq
u
en
ce
,
a
n
d
k
e
y
w
o
r
d
s
f
ea
t
u
r
es.
T
h
e
m
et
h
o
d
s
f
o
r
ev
alu
at
in
g
th
e
q
u
ali
t
y
o
f
t
h
e
s
u
m
m
ar
ies
ar
e
b
o
th
i
n
tr
in
s
ic
(
s
u
c
h
as
p
er
ce
n
t
a
g
r
ee
m
e
n
t,
p
r
ec
is
io
n
/r
ec
all,
an
d
r
elati
v
e
u
t
il
it
y
)
a
n
d
ex
tr
i
n
s
ic
(
d
o
cu
m
en
t r
a
n
k
)
.
A
late
s
t v
er
s
i
o
n
o
f
ME
A
D
is
b
ased
o
n
ce
n
tr
o
id
b
ased
m
u
lti d
o
cu
m
en
t
s
u
m
m
ar
izatio
n
.
[
1
5
]
Has
p
r
o
p
o
s
ed
k
e
y
w
o
r
d
s
elec
tio
n
s
tr
ateg
y
.
T
h
is
i
s
co
m
b
in
ed
w
it
h
t
h
e
KFI
D
F
m
ea
s
u
r
e
to
s
elec
t
th
e
m
o
r
e
m
ea
n
i
n
g
f
u
l
s
en
ten
ce
s
to
b
e
in
c
lu
d
ed
i
n
t
h
e
s
u
m
m
ar
y
.
T
h
e
No
n
-
n
e
g
ati
v
e
co
n
s
tr
ain
ts
u
s
ed
he
r
e
ar
e
s
i
m
ilar
to
th
e
h
u
m
a
n
co
g
n
itio
n
p
r
o
ce
s
s
.
[
1
4
]
Pro
p
o
s
ed
a
tr
a
in
ab
le
s
u
m
m
ar
izer
b
ased
o
n
f
ea
tu
r
e
s
elec
tio
n
a
n
d
Su
p
p
o
r
t
Vec
to
r
Ma
ch
in
e
(
SV
M)
.
E
v
o
lu
tio
n
ar
y
co
n
n
ec
tio
n
i
s
t
m
o
d
el
f
o
r
A
T
S
is
d
ev
elo
p
e
d
b
y
[
1
6
]
w
h
ic
h
is
b
ased
o
n
ev
o
lu
tio
n
ar
y
,
f
u
zz
y
an
d
co
n
n
ec
tio
n
is
t
tech
n
iq
u
e
s
.
All
th
e
p
ap
er
s
d
is
c
u
s
s
ed
ab
o
v
e
u
s
e
v
ar
io
u
s
f
ea
t
u
r
es
f
o
r
s
u
m
m
ar
y
g
e
n
er
at
in
.
O
u
r
ai
m
i
n
th
is
p
ap
er
i
s
t
o
p
er
f
o
r
m
t
h
e
co
m
p
ar
ativ
e
s
t
u
d
y
o
n
t
h
e
u
s
e
o
f
v
ar
io
u
s
f
ea
t
u
r
es
u
s
ed
f
o
r
d
o
c
u
m
e
n
t
s
u
m
m
ar
izatio
n
d
ep
en
d
in
g
u
p
o
n
t
h
e
s
ize
a
n
d
t
y
p
e
o
f
th
e
d
o
cu
m
en
t.
T
h
e
f
o
llo
w
in
g
s
ec
t
io
n
d
escr
ib
es t
h
e
v
ar
io
u
s
s
tep
s
in
t
h
e
p
r
o
p
o
s
ed
s
tu
d
y
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8938
IJ
-
AI
Vo
l.
3
,
No
.
3
,
Sep
tem
b
er
201
4
:
1
1
2
–
1
2
0
114
3.
P
RE
P
RO
CE
SS
I
NG
T
h
e
p
r
o
p
o
s
ed
au
to
m
atic
te
x
t s
u
m
m
ar
izat
io
n
s
y
s
te
m
co
n
s
is
ts
o
f
th
e
f
o
llo
w
in
g
co
m
p
o
n
e
n
t
s
:
1.
P
r
ep
r
o
ce
s
s
in
g
2.
Featu
r
e
ex
tr
ac
tio
n
3.
Mo
d
el
b
u
ild
in
g
4.
Sen
te
n
ce
s
elec
t
io
n
an
d
as
s
e
m
b
ly
T
h
is
s
ec
tio
n
d
ea
ls
w
it
h
th
e
p
r
e
-
p
r
o
ce
s
s
in
g
p
h
ase.
T
h
e
in
p
u
t
d
o
cu
m
e
n
t
ca
n
b
e
o
f
an
y
d
o
cu
m
e
n
t
f
o
r
m
at
(
d
o
c,
tx
t,
p
d
f
,
h
t
m
l,
r
t
f
)
,
h
e
n
ce
th
e
s
y
s
te
m
f
ir
s
t
ap
p
lies
d
o
cu
m
e
n
t
co
n
v
er
ter
s
to
e
x
tr
ac
t
th
e
tex
t
f
r
o
m
th
e
in
p
u
t d
o
cu
m
en
t.
3.
1
.
Tex
t
Pr
ologu
i
ng
P
r
e
-
p
r
o
ce
s
s
in
g
th
e
tex
t
b
ef
o
r
e
in
ce
p
ti
n
g
to
s
u
m
m
ar
izatio
n
a
n
d
ca
teg
o
r
izatio
n
is
T
ex
t
P
r
o
lo
g
u
i
n
g
.
I
t
co
n
s
is
ts
o
f
s
ix
p
h
a
s
es
w
h
ic
h
a
r
e
lis
ted
in
th
e
f
o
llo
w
i
n
g
s
u
b
s
ec
tio
n
s
.
3.
1
.1
.
Tex
t
S
egme
nt
a
t
i
on
T
ex
t
Seg
m
e
n
tat
io
n
i
s
t
h
e
p
r
o
ce
s
s
o
f
d
ec
o
m
p
o
s
i
n
g
t
h
e
g
i
v
en
te
x
t
i
n
to
its
co
n
s
ti
tu
e
n
t
s
en
te
n
ce
s
,
ca
lcu
lati
n
g
ea
c
h
s
e
n
te
n
ce
le
n
g
th
an
d
w
o
r
d
co
u
n
t.
T
h
is
m
o
d
u
le
d
iv
id
e
s
t
h
e
d
o
cu
m
en
t
i
n
t
o
s
en
te
n
ce
s
.
At
f
ir
s
t
g
lan
ce
,
i
t
m
a
y
ap
p
ea
r
th
at
u
s
in
g
e
n
d
o
f
s
en
te
n
ce
p
u
n
c
t
u
ati
o
n
m
ar
k
s
,
s
u
c
h
as
p
er
io
d
s
,
q
u
esti
o
n
m
ar
k
s
,
a
n
d
ex
cla
m
atio
n
p
o
in
t
s
,
is
s
u
f
f
icie
n
t f
o
r
m
ar
k
i
n
g
t
h
e
s
e
n
te
n
ce
b
o
u
n
d
ar
ies.
3.
1
.2
.
N
o
r
m
a
l
iza
t
io
n
No
r
m
a
lizatio
n
is
t
h
e
p
r
o
ce
s
s
o
f
co
n
v
er
ti
n
g
w
o
r
d
s
i
n
to
n
o
r
m
al
ized
f
o
r
m
.
T
h
e
f
o
llo
w
i
n
g
ar
e
t
h
e
p
r
o
ce
s
s
es th
at
co
m
e
u
n
d
er
n
o
r
m
aliza
tio
n
tech
n
iq
u
es.
3.
1
.3
.
To
k
eni
z
a
t
i
on
I
t is th
e
p
r
o
ce
s
s
o
f
s
p
lit
tin
g
o
f
th
e
s
e
n
te
n
ce
in
to
w
o
r
d
s
3.
1
.4
.
S
t
op
wo
r
d
Rem
ov
a
l
Du
r
in
g
t
h
e
r
etr
ie
v
al
o
f
r
elev
a
n
t
i
n
f
o
r
m
atio
n
w
e
h
av
e
to
r
em
o
v
e
f
e
w
w
o
r
d
s
,
n
u
m
b
er
s
,
a
n
d
s
p
ec
ia
l
s
y
m
b
o
ls
etc.
,
w
h
ich
h
a
v
e
less
s
i
g
n
if
ica
n
ce
.
A
n
e
w
ap
p
r
o
ac
h
i
s
u
s
ed
f
o
r
s
to
p
w
o
r
d
r
e
m
o
v
al.
T
h
e
s
to
p
w
o
r
d
s
ar
e
class
i
f
ied
as
u
s
e
f
u
l a
n
d
u
s
eless
s
to
p
w
o
r
d
an
d
t
h
e
r
e
m
o
v
ed
ac
co
r
d
in
g
l
y
.
T
h
is
w
ill
h
elp
in
f
aster
o
p
er
atio
n
s
at
later
s
te
m
m
i
n
g
s
ta
g
e.
3.
1
.5
.
C
a
se
F
old
ing
C
o
n
v
er
tin
g
en
tire
w
o
r
d
s
i
n
th
e
s
en
te
n
ce
s
i
n
to
lo
w
er
ca
s
e
s
o
as
to
av
o
id
r
ep
etitio
n
o
f
s
am
e
w
o
r
d
i
n
d
if
f
er
e
n
t c
ase
s
lik
e
s
en
te
n
ce
c
ase,
ca
p
ital c
ase,
titl
e
ca
s
e,
u
p
p
er
ca
s
e
etc.
3.
1.
6.
S
t
emmi
ng
Me
ch
an
icall
y
r
e
m
o
v
in
g
o
r
ch
an
g
i
n
g
t
h
e
s
u
f
f
ix
e
s
o
f
s
o
m
e
n
o
u
n
s
o
r
v
er
b
s
.
Ste
m
m
i
n
g
i
m
p
r
o
v
es
th
e
r
etr
iev
al
p
er
f
o
r
m
a
n
ce
b
ec
au
s
e
th
e
y
r
ed
u
ce
v
ar
ian
ts
o
f
t
h
e
s
a
m
e
r
o
o
t
w
o
r
d
to
a
co
m
m
o
n
co
n
ce
p
t.
I
t
also
r
ed
u
ce
s
t
h
e
s
ize
o
f
t
h
e
i
n
d
ex
i
n
g
s
tr
u
ctu
r
e
b
ec
a
u
s
e
th
e
n
u
m
b
er
o
f
d
is
ti
n
ct
i
n
d
ex
ter
m
s
is
r
e
d
u
ce
d
.
T
h
e
d
esig
n
o
f
a
s
te
m
m
er
is
la
n
g
u
a
g
e
s
p
ec
if
ic,
a
n
d
r
eq
u
ir
es
s
o
m
e
s
i
g
n
i
f
i
ca
n
t
li
n
g
u
is
tic
ex
p
er
tis
e
i
n
t
h
e
lan
g
u
a
g
e.
Her
e
w
e
p
r
o
p
o
s
ed
an
in
te
g
r
ated
s
te
m
m
i
n
g
ap
p
r
o
ac
h
w
h
ich
in
v
o
l
v
es
b
o
th
r
u
le
b
ased
ap
p
r
o
ac
h
a
n
d
d
ictio
n
ar
y
b
ased
ap
p
r
o
ac
h
.
T
h
e
p
r
o
p
o
s
ed
in
teg
r
ated
m
o
d
el
s
h
o
w
ed
b
etter
i
m
p
ac
tin
g
r
e
s
u
lt
s
w
it
h
r
esp
ec
t
to
w
o
r
d
s
a
f
f
ec
ted
a
n
d
co
m
p
u
ti
n
g
t
i
m
e
[
1
7
]
.
4.
M
AT
H
E
M
AT
I
CAL M
O
DE
L
L
I
N
G
F
O
R
F
E
AT
URE
SE
L
E
C
T
I
O
N
Af
ter
p
r
e
-
p
r
o
ce
s
s
in
g
,
th
e
i
n
p
u
t
d
o
cu
m
en
t
i
s
s
u
b
j
ec
ted
to
f
ea
tu
r
e
ex
tr
ac
tio
n
b
y
w
h
ic
h
ea
ch
s
en
te
n
ce
in
t
h
e
tex
t
d
o
cu
m
e
n
t
o
b
tain
s
a
f
ea
tu
r
e
s
co
r
e
b
ased
o
n
its
i
m
p
o
r
tan
ce
.
T
h
e
i
m
p
o
r
tan
t
te
x
t
f
ea
tu
r
es
u
s
ed
in
t
h
e
p
r
o
p
o
s
ed
s
y
s
te
m
ar
e:
(
1
)
Fo
r
m
at
b
ased
s
co
r
e
(
2
)
Nu
m
er
ic
al
d
ata
(
3
)
T
e
r
m
w
eig
h
t
(
4
)
T
itle
f
ea
tu
r
e
(
5
)
C
o
-
r
elatio
n
a
m
o
n
g
s
e
n
te
n
ce
(
6
)
C
o
-
r
elatio
n
a
m
o
n
g
p
ar
ag
r
ap
h
,
(
7
)
C
o
n
ce
p
t
-
b
ased
f
ea
t
u
r
e
an
d
(
8
)
P
o
s
itio
n
d
ata.
T
h
e
co
n
ce
p
t b
ased
f
ea
tu
r
e
is
u
s
ed
f
o
r
th
e
f
ir
s
t ti
m
e.
4.
1
.
F
e
a
t
u
r
e
comp
u
t
a
t
io
n
On
ce
t
h
e
f
ea
t
u
r
es
ar
e
d
ec
id
ed
,
o
n
e
n
ee
d
s
to
p
r
ep
ar
e
th
e
m
a
th
e
m
atica
l
m
o
d
el
f
o
r
th
eir
co
m
p
u
tat
io
n
.
T
h
e
f
o
llo
w
i
n
g
s
u
b
s
ec
tio
n
s
d
es
cr
ib
e
th
e
m
at
h
e
m
a
tical
co
m
p
u
tatio
n
o
f
t
h
ese
f
ea
t
u
r
es.
4.
1
.1
.
F
or
m
a
t
ba
s
ed
s
co
re
:
Evaluation Warning : The document was created with Spire.PDF for Python.
IJ
-
AI
I
SS
N:
2252
-
8938
E
ffect
o
f F
ea
tu
r
e
S
elec
tio
n
o
n
S
ma
ll a
n
d
La
r
g
e
Do
cu
men
t
S
u
mma
r
iz
a
tio
n
…
(
D
.
Y
.
S
a
kh
a
r
e
)
115
T
h
e
tex
t
i
n
d
i
v
er
s
e
f
o
r
m
at
E
.
g
.
I
talics,
B
o
ld
,
u
n
d
er
li
n
ed
,
b
ig
f
o
n
t
s
ize
an
d
m
o
r
e
in
m
an
y
d
o
cu
m
e
n
ts
s
h
o
w
s
th
e
i
m
p
o
r
tan
ce
o
f
t
h
e
s
en
te
n
ce
s
.
T
h
is
f
ea
t
u
r
e
n
ev
er
d
ep
en
d
s
o
n
th
e
w
h
o
le
d
o
cu
m
en
t
i
n
s
tead
to
s
o
m
e
ex
ac
t
s
in
g
le
s
e
n
te
n
ce
.
Sco
r
e
c
an
a
s
s
i
g
n
ed
to
t
h
e
s
e
n
te
n
ce
co
n
s
id
er
in
g
th
e
f
o
r
m
at
o
f
t
h
e
w
o
r
d
s
in
th
e
te
x
t.
T
h
e
r
atio
o
f
th
e
n
u
m
b
er
o
f
w
o
r
d
s
a
v
ailab
le
i
n
t
h
e
s
en
te
n
ce
w
it
h
s
p
ec
ial
f
o
r
m
at
to
t
h
e
to
tal
n
u
m
b
er
o
f
w
o
r
d
s
i
n
th
e
s
en
te
n
ce
o
f
f
er
s
o
n
e
to
f
o
r
m
th
e
f
o
r
m
at
w
h
ich
i
s
d
ep
en
d
en
t r
elativ
e
o
n
t
h
e
s
co
r
e
o
f
t
h
e
s
en
t
en
ce
.
4.
1
.2
.
N
u
me
r
ic
a
l
da
t
a
T
h
e
i
m
p
o
r
tan
ce
s
tats
co
n
ce
r
n
in
g
th
e
v
ital
p
u
r
p
o
s
e
o
f
th
e
d
o
cu
m
e
n
t
ar
e
u
s
u
a
ll
y
s
h
o
w
n
b
y
t
h
e
n
u
m
er
ical
d
ata
w
it
h
i
n
t
h
e
s
e
n
ten
ce
a
n
d
th
is
h
as
its
o
w
n
co
n
tr
ib
u
tio
n
s
o
n
t
h
e
b
asic
th
o
u
g
h
t
o
f
t
h
e
d
o
cu
m
en
t
th
at
u
s
u
a
ll
y
m
a
k
e
w
a
y
to
s
u
m
m
ar
y
s
elec
t
io
n
.
T
h
e
r
atio
o
f
th
e
n
u
m
b
er
o
f
n
u
m
er
ical
d
ata
th
at
h
ap
p
en
s
in
s
en
te
n
ce
o
v
er
th
e
s
en
ten
ce
le
n
g
th
i
s
t
h
u
s
u
s
ed
to
ca
lcu
late
t
h
e
s
co
r
e
f
o
r
th
is
f
e
atu
r
e.
4.
1
.3
.
Ter
m
w
eig
ht
T
er
m
w
eig
h
t
i
s
a
f
ea
t
u
r
e
v
alu
e
w
h
ich
i
s
e
m
p
lo
y
ed
to
lo
o
k
in
to
th
e
p
r
o
m
i
n
e
n
t
s
e
n
ten
ce
s
f
o
r
s
u
m
m
ar
izin
g
t
h
e
tex
t
d
o
cu
m
e
n
ts
.
T
h
e
ter
m
w
ei
g
h
t
o
f
a
s
e
n
ten
ce
i
s
ca
lcu
lated
as
th
e
r
a
tio
o
f
th
e
s
e
n
ten
ce
w
ei
g
h
t
to
t
h
e
m
a
x
i
m
u
m
s
en
te
n
ce
w
ei
g
h
t
i
n
t
h
e
g
iv
e
n
tex
t
d
o
cu
m
e
n
t.
T
h
e
s
e
n
te
n
ce
w
ei
g
h
t
is
t
h
e
s
u
m
m
atio
n
o
f
th
e
w
eig
h
t
f
ac
to
r
o
f
al
l t
h
e
w
o
r
d
s
i
n
a
s
e
n
te
n
ce
.
T
h
e
w
e
ig
h
t
f
ac
to
r
is
t
h
e
p
r
o
d
u
ct
o
f
w
o
r
d
f
r
eq
u
en
c
y
a
n
d
th
e
in
v
er
s
e
o
f
t
h
e
s
en
ten
ce
f
r
eq
u
e
n
c
y
.
)
(
i
S
M
a
x
S
TW
w
D
i
w
n
j
j
w
W
S
1
)
(
T
)
/
(
)
(
N
N
og
l
t
I
S
F
W
h
er
e,
w
S
Sen
te
n
ce
w
ei
g
h
t
j
W
W
eig
h
t
f
ac
to
r
o
f
th
e
w
o
r
d
in
a
s
en
te
n
ce
n
Nu
m
b
er
o
f
w
o
r
d
s
i
n
a
s
en
te
n
ce
TF
T
h
e
n
u
m
b
er
o
f
o
cc
u
r
r
en
ce
s
o
f
th
e
ter
m
o
r
w
o
r
d
in
a
te
x
t d
o
cu
m
e
n
t
I
S
F
I
n
v
er
s
e
Se
n
te
n
ce
Fre
q
u
e
n
c
y
N
T
o
tal
n
u
m
b
er
o
f
s
e
n
te
n
ce
s
in
a
d
o
cu
m
en
t
(
T
)
N
T
o
tal
n
u
m
b
er
o
f
s
e
n
te
n
ce
s
th
at
co
n
tai
n
t
h
e
ter
m
(
T
)
4.
1
.
4.
T
it
le
f
ea
t
ures
A
s
e
n
te
n
ce
is
g
i
v
en
a
g
o
o
d
s
co
r
e
o
n
ly
w
h
e
n
t
h
e
g
i
v
e
n
s
e
n
te
n
ce
h
a
s
th
e
ti
tle
w
o
r
d
s
.
T
h
e
in
ten
tio
n
o
f
th
e
d
o
cu
m
e
n
t
i
s
s
h
o
w
n
v
ia
t
h
e
w
o
r
d
b
elo
n
g
i
n
g
to
t
h
e
t
itl
e
if
a
v
ailab
le
in
th
a
t
s
e
n
te
n
c
e.
T
h
e
r
atio
o
f
th
e
n
u
m
b
er
o
f
w
o
r
d
s
in
t
h
e
s
e
n
te
n
ce
th
at
o
cc
u
r
in
tit
le
to
th
e
to
tal
n
u
m
b
er
o
f
w
o
r
d
s
in
t
h
e
titl
e
h
elp
s
to
ca
lcu
lat
e
th
e
s
co
r
e
o
f
a
s
en
ten
ce
f
o
r
th
i
s
f
ea
tu
r
e.
4.
1
.
5.
Co
-
re
la
t
io
n a
m
o
ng
s
e
nte
nce
A
t
f
ir
s
t,
th
e
co
r
r
elatio
n
m
atr
i
x
C
is
g
en
er
ated
in
a
s
ize
o
f
N
x
M
,
in
w
h
ic
h
N
is
th
e
n
u
m
b
er
o
f
s
en
te
n
ce
a
n
d
t
h
e
M
i
s
t
h
e
n
u
m
b
er
o
f
u
n
iq
u
e
k
e
y
w
o
r
d
s
in
t
h
e
d
o
cu
m
e
n
t.
E
v
er
y
ele
m
e
n
t
o
f
t
h
e
m
atr
i
x
is
f
illed
w
i
th
ze
r
o
o
r
o
n
e,
b
as
ed
o
n
w
h
et
h
er
t
h
e
co
r
r
esp
o
n
d
in
g
k
e
y
w
o
r
d
is
p
r
esen
ted
o
r
n
o
t.
T
h
en
,
th
e
co
r
r
elatio
n
o
f
ev
er
y
v
ec
to
r
w
it
h
o
th
er
v
ec
to
r
(
s
e
n
ten
ce
w
it
h
o
th
er
s
e
n
ten
ce
)
i
s
co
m
p
u
ted
f
o
r
all
co
m
b
in
atio
n
s
s
o
th
at
th
e
m
atr
i
x
o
f
N
x
N
is
g
e
n
er
at
ed
w
h
er
e
ev
er
y
ele
m
en
t
is
th
e
co
r
r
elatio
n
o
f
t
w
o
v
ec
to
r
(
t
w
o
s
en
te
n
ce
s
)
.
T
h
en
,
ev
er
y
ele
m
en
t o
f
t
h
e
r
o
w
v
ec
to
r
is
ad
d
ed
to
g
et
th
e
s
e
n
ten
ce
s
co
r
e.
4.
1
.6
.
Co
-
re
la
t
io
n a
m
o
ng
pa
ra
g
ra
ph
I
S
F
TF
W
i
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8938
IJ
-
AI
Vo
l.
3
,
No
.
3
,
Sep
tem
b
er
201
4
:
1
1
2
–
1
2
0
116
Her
e,
th
e
co
r
r
elatio
n
is
co
m
p
u
ted
f
o
r
e
v
er
y
p
ar
ag
r
ap
h
i
n
s
te
ad
o
f
s
en
ten
ce
s
.
f
o
r
t
h
at,
t
h
e
co
r
r
elatio
n
m
atr
i
x
C
is
g
en
er
ated
in
a
s
ize
o
f
P
x
M
,
in
w
h
ic
h
P
is
th
e
n
u
m
b
er
o
f
p
ar
ag
r
ap
h
an
d
th
e
M
is
th
e
n
u
m
b
er
o
f
u
n
iq
u
e
k
e
y
w
o
r
d
s
i
n
t
h
e
d
o
cu
m
e
n
t.
E
v
er
y
ele
m
e
n
t o
f
t
h
e
m
atr
i
x
is
f
illed
w
it
h
ze
r
o
o
r
o
n
e,
b
ased
o
n
w
h
et
h
er
th
e
co
r
r
esp
o
n
d
in
g
k
e
y
w
o
r
d
is
p
r
esen
ted
o
r
n
o
t
i
n
t
h
e
p
ar
ag
r
ap
h
.
T
h
en
,
t
h
e
co
r
r
elatio
n
o
f
ev
er
y
v
ec
to
r
w
it
h
o
th
er
v
ec
to
r
(
p
ar
ag
r
ap
h
w
it
h
o
th
er
p
ar
ag
r
ap
h
)
is
co
m
p
u
ted
f
o
r
all
co
m
b
in
atio
n
s
s
o
t
h
at
th
e
m
atr
ix
o
f
P
x
P
is
g
en
er
ated
w
h
er
e
ev
er
y
ele
m
e
n
t
is
th
e
co
r
r
elatio
n
o
f
t
w
o
v
e
cto
r
(
tw
o
p
ar
ag
r
ap
h
)
.
T
h
en
,
ev
er
y
ele
m
en
t
o
f
t
h
e
r
o
w
v
ec
to
r
is
ad
d
ed
to
g
et
th
e
s
c
o
r
e
o
f
ev
er
y
p
ar
ag
r
ap
h
s
a
n
d
th
e
s
co
r
e
o
f
ev
er
y
w
i
ll
o
b
tain
th
e
s
a
m
e
s
co
r
e
o
f
w
h
at
it
s
r
elev
a
n
t p
ar
ag
r
ap
h
o
b
tain
ed
.
4.
1
.7
.
C
on
cep
t
-
ba
s
ed
f
ea
t
ure
I
n
itiall
y
,
t
h
e
co
n
ce
p
t
is
ex
tr
ac
ted
f
r
o
m
th
e
i
n
p
u
t
d
o
cu
m
en
t
u
s
i
n
g
t
h
e
m
u
t
u
al
i
n
f
o
r
m
atio
n
a
n
d
w
i
n
d
o
w
in
g
p
r
o
ce
s
s
.
A
w
i
n
d
o
w
i
n
g
p
r
o
ce
s
s
is
ca
r
r
ied
o
u
t
th
r
o
u
g
h
th
e
d
o
cu
m
e
n
t,
in
w
h
ic
h
a
v
ir
tu
al
w
i
n
d
o
w
o
f
s
ize
'
k
'
i
s
m
o
v
ed
f
r
o
m
le
f
t
to
r
ig
h
t
u
n
til
t
h
e
en
d
o
f
th
e
d
o
cu
m
en
t.
T
h
en
,
th
e
f
o
llo
w
i
n
g
f
o
r
m
u
lae
ar
e
u
s
ed
to
f
i
n
d
th
e
w
o
r
d
s
th
at
co
-
o
cc
u
r
r
ed
to
g
eth
er
w
it
h
in
ea
c
h
w
i
n
d
o
w
.
)
(
*
)
(
)
,
(
2
l
o
g
)
,
(
j
i
j
i
j
i
w
P
w
P
w
w
P
w
w
MI
W
h
er
e,
)
,
(
j
i
w
w
P
T
h
e
j
o
in
t p
r
o
b
a
b
ilit
y
t
h
at
b
o
th
k
e
y
w
o
r
d
ap
p
ea
r
ed
t
o
g
eth
er
i
n
a
tex
t
w
in
d
o
w
)
(
i
w
P
T
h
e
p
r
o
b
ab
ilit
y
t
h
at
a
k
e
y
w
o
r
d
i
w
ap
p
ea
r
s
in
a
tex
t
w
i
n
d
o
w
T
h
e
p
r
o
b
a
b
ilit
y
)
(
i
w
P
is
co
m
p
u
te
d
b
ased
o
n
sw
sw
t
,
w
h
er
e
t
sw
is
t
h
e
n
u
m
b
er
o
f
s
l
id
in
g
w
i
n
d
o
ws
co
n
tain
i
n
g
t
h
e
k
e
y
w
o
r
d
i
w
an
d
sw
is
th
e
to
tal
n
u
m
b
er
o
f
w
i
n
d
o
w
s
co
n
s
tr
u
cted
f
r
o
m
a
te
x
t
d
o
cu
m
en
t.
Si
m
i
lar
l
y
,
)
,
(
j
i
w
w
P
is
th
e
f
r
ac
tio
n
o
f
t
h
e
n
u
m
b
er
o
f
w
i
n
d
o
w
s
co
n
ta
in
i
n
g
b
o
th
k
e
y
w
o
r
d
s
o
u
t
o
f
t
h
e
to
tal
n
u
m
b
er
o
f
w
i
n
d
o
w
s
.
T
h
en
,
f
o
r
ev
er
y
co
n
ce
p
t
ex
tr
ac
ted
,
th
e
co
n
ce
p
t
w
e
ig
h
t
is
co
m
p
u
ted
b
ased
o
n
th
e
ter
m
w
ei
g
h
t
p
r
o
ce
d
u
r
e
a
n
d
t
h
e
s
e
n
ten
ce
s
co
r
e
is
also
co
m
p
u
ted
as
p
er
t
h
e
p
r
o
ce
d
u
r
e
d
escr
ib
ed
in
ter
m
w
ei
g
h
-
b
ased
f
ea
tu
r
e
co
m
p
u
tatio
n
.
4.
1
.8
.
Pos
i
t
i
on
da
t
a
P
o
s
itio
n
-
b
ased
f
ea
t
u
r
e
is
co
m
p
u
ted
w
it
h
r
elev
an
t
to
th
e
s
en
te
n
ce
lo
ca
ted
in
th
e
d
o
cu
m
en
t.
W
ith
p
er
s
p
ec
tiv
e
o
f
d
o
m
ai
n
e
x
p
er
ts
,
i
n
itial
s
e
n
ten
ce
a
n
d
th
e
las
t
s
e
n
ten
ce
o
f
th
e
d
o
cu
m
en
t
is
i
m
p
o
r
tan
t
th
a
n
th
e
ot
h
er
s
en
ten
ce
.
So
,
th
e
m
a
x
i
m
u
m
s
co
r
e
is
g
i
v
en
f
o
r
t
h
o
s
e
s
en
te
n
ce
s
an
d
t
h
e
m
ed
iu
m
v
alu
e
i
s
g
i
v
e
n
to
t
h
e
s
en
te
n
ce
lo
ca
ted
in
t
h
e
s
tar
ti
n
g
an
d
en
d
i
n
g
o
f
ev
er
y
p
ar
a
g
r
ap
h
.
5.
F
E
AT
U
RE
M
AT
RIX F
O
R
T
RAI
NIN
G
O
F
F
E
A
T
UR
E
-
B
ASE
D
N
E
URA
L
NE
T
WO
RK
T
h
is
s
ec
tio
n
d
escr
ib
es
th
e
f
e
atu
r
e
m
atr
i
x
u
s
ed
f
o
r
tr
ain
in
g
th
e
f
ea
tu
r
e
-
b
ased
n
e
u
r
al
n
e
t
w
o
r
k
.
T
h
e
f
ea
t
u
r
e
m
atr
i
x
is
r
ep
r
esen
ted
w
it
h
t
h
e
s
ize
of
Nx
F
,
w
h
er
e
N
is
th
e
n
u
m
b
er
o
f
s
e
n
te
n
ce
an
d
F
is
th
e
n
u
m
b
er
f
ea
t
u
r
e
u
s
ed
in
t
h
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
.
(
Her
e
8
F
)
.
E
v
er
y
ele
m
e
n
t
o
f
th
e
m
atr
ix
is
th
e
f
ea
t
u
r
e
s
co
r
e
o
b
tain
ed
f
o
r
th
e
co
r
r
esp
o
n
d
in
g
s
en
ten
ce
w
i
th
t
h
e
f
ea
tu
r
e.
5
.
1
.
T
ra
ini
ng
ph
a
s
e
Her
e
,
m
u
lti
-
la
y
er
p
er
ce
p
tr
o
n
s
f
ee
d
f
o
r
w
ar
d
n
eu
r
al
n
et
w
o
r
k
is
u
tili
ze
d
f
o
r
lear
n
in
g
m
ec
h
an
is
m
,
in
w
h
ic
h
t
h
e
b
ac
k
-
p
r
o
p
ag
atio
n
alg
o
r
ith
m
i
s
e
f
f
ec
ti
v
el
y
u
tili
z
ed
to
tr
ain
n
eu
r
al
n
et
w
o
r
k
s
.
T
o
tr
ain
th
e
n
eu
r
a
l
n
et
w
o
r
k
e
f
f
ec
ti
v
el
y
,
t
h
e
i
n
p
u
t
la
y
er
is
a
n
i
n
d
iv
id
u
al
(
f
ea
t
u
r
e
v
ec
to
r
)
o
b
tain
ed
f
r
o
m
t
h
e
f
ea
tu
r
e
co
m
p
u
tatio
n
s
tep
s
an
d
th
e
tar
g
et
o
u
tp
u
t
is
ze
r
o
o
r
o
n
e
th
at
s
ig
n
i
f
ies
wh
eth
er
it
s
i
m
p
o
r
tan
ce
o
r
n
o
t.
T
esti
ng
p
ha
s
e:
I
n
test
i
n
g
p
h
a
s
e,
th
e
in
p
u
t
tex
t
d
o
cu
m
en
t
i
s
p
r
e
p
r
o
ce
s
s
ed
an
d
th
e
f
ea
t
u
r
e
s
co
r
e
o
f
ev
e
r
y
s
en
te
n
ce
in
t
h
e
d
o
cu
m
en
t
is
co
m
p
u
t
ed
.
T
h
e
co
m
p
u
ted
f
ea
t
u
r
e
s
co
r
e
is
ap
p
li
ed
to
th
e
tr
ain
ed
n
et
w
o
r
k
t
h
at
r
etu
r
n
s
t
h
e
s
e
n
ten
c
e
s
co
r
e
o
f
ev
er
y
s
e
n
te
n
ce
p
r
esen
ted
in
th
e
i
n
p
u
t te
x
t d
o
cu
m
en
t
Evaluation Warning : The document was created with Spire.PDF for Python.
IJ
-
AI
I
SS
N:
2252
-
8938
E
ffect
o
f F
ea
tu
r
e
S
elec
tio
n
o
n
S
ma
ll a
n
d
La
r
g
e
Do
cu
men
t
S
u
mma
r
iz
a
tio
n
…
(
D
.
Y
.
S
a
kh
a
r
e
)
117
5
.
2
.
Ra
nk
ing
o
f
s
ent
ence
Her
e,
th
e
r
an
k
in
g
o
f
s
en
ten
ce
i
s
ca
r
r
ied
o
u
t u
s
i
n
g
th
e
s
en
ten
c
e
s
co
r
e
o
b
tain
ed
f
r
o
m
th
e
p
r
ev
io
u
s
s
tep
.
I
n
itiall
y
,
s
en
ten
ce
s
p
r
esen
ted
in
th
e
i
n
p
u
t
te
x
t
d
o
cu
m
e
n
t
ar
e
s
o
r
ted
in
d
escen
d
in
g
o
r
d
er
ac
co
r
d
in
g
to
th
e
f
i
n
al
s
en
te
n
ce
s
co
r
e.
T
h
en
,
t
h
e
to
p
-
N
s
en
te
n
ce
s
ar
e
s
elec
ted
f
o
r
th
e
s
u
m
m
ar
y
b
ased
o
n
th
e
co
m
p
r
e
s
s
io
n
r
ate
g
iv
e
n
b
y
t
h
e
in
p
u
t
u
s
er
.
Fi
n
all
y
,
th
e
s
elec
ted
to
p
-
N
s
en
ten
ce
s
ar
e
o
r
d
er
e
d
in
a
s
eq
u
e
n
tial
w
a
y
b
ased
o
n
t
h
e
o
r
d
er
o
f
th
e
r
ef
er
en
ce
n
u
m
b
er
o
r
u
n
iq
u
e
I
D
to
o
b
tain
th
e
f
i
n
a
l su
m
m
ar
y
.
1
0
0
S
N
C
N
W
h
er
e,
S
N
T
o
tal
n
u
m
b
er
o
f
s
e
n
ten
ce
s
i
n
th
e
d
o
cu
m
e
n
t
C
C
o
m
p
r
es
s
io
n
r
ate
6.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
T
h
is
s
ec
tio
n
d
escr
ib
es
t
h
e
d
etailed
th
e
e
x
p
er
i
m
e
n
tal
r
e
s
u
l
t
s
an
d
it
an
d
a
n
al
y
s
i
s
o
f
t
h
e
d
o
cu
m
en
t
s
u
m
m
ar
izatio
n
.
T
h
e
p
r
o
p
o
s
ed
s
y
n
tact
ic
an
d
s
en
te
n
ce
f
ea
tu
r
e
-
b
ased
h
y
b
r
id
ap
p
r
o
ac
h
is
i
m
p
le
m
e
n
ted
in
MA
T
L
A
B
(
Ma
tlab
7
.
1
1
)
an
d
t
h
e
ex
p
er
i
m
en
ta
tio
n
is
ca
r
r
ied
o
u
t
w
ith
i5
p
r
o
ce
s
s
o
r
h
a
v
in
g
3
GM
R
A
M.
6
.
1
.
DUC
2
0
0
2
da
t
a
s
et
Fo
r
ex
p
er
i
m
e
n
tatio
n
,
w
e
h
a
v
e
u
s
ed
DU
C
2
0
0
2
d
ataset
[
18
]
th
at
co
n
tai
n
s
d
o
cu
m
e
n
ts
o
n
d
if
f
er
en
t
ca
teg
o
r
ies an
d
ex
tr
ac
tiv
e
s
u
m
m
ar
y
p
er
d
o
cu
m
e
n
t.
6
.
2
.
E
x
peri
m
ent
a
l
Res
ults
T
ab
le
1
.
Featu
r
e
s
co
r
e
f
o
r
th
e
tex
t d
o
cu
m
en
t (
C
lu
s
ter
No
.
d
0
7
1
f
an
d
Do
cu
m
e
n
t N
o
.
A
P
8
8
0
3
1
0
-
0
0
6
2
)
S
e
n
t
e
n
c
e
ID
F
e
a
t
u
r
e
sco
r
e
F
1
F
2
F
3
F
4
F
5
F
6
F
7
F
8
1
0
0
0
.
2
5
0
0
0
.
4
0
0
2
0
.
0
6
9
5
0
.
1
8
5
0
0
.
2
3
0
7
0
.
2
5
0
0
2
0
0
0
0
.
5
6
9
5
-
0
.
0
0
4
4
0
.
1
1
8
0
0
.
3
2
8
3
0
.
2
5
0
0
3
0
.
4
5
5
0
0
1
.
0
0
0
0
-
0
.
3
5
6
8
-
0
.
1
6
4
0
0
.
5
7
6
4
0
.
2
5
0
0
4
0
0
0
0
.
3
3
8
5
0
.
0
1
4
1
-
0
.
0
7
9
0
0
.
1
9
5
1
0
5
0
0
0
0
.
2
7
3
3
0
.
2
8
3
8
-
0
.
0
7
9
0
0
.
1
5
7
5
0
.
2
5
0
0
6
0
0
0
0
.
2
4
7
0
0
.
6
6
6
1
0
.
1
3
8
6
0
.
1
4
2
4
0
7
0
.
1
0
0
0
0
.
1
0
0
0
0
0
.
4
4
2
6
0
.
0
3
7
0
0
.
1
3
8
6
0
.
2
5
5
1
0
.
2
5
0
0
8
0
0
0
0
.
5
3
1
1
0
.
3
7
9
2
0
.
4
3
6
4
0
.
3
0
6
2
0
.
2
5
0
0
T
ab
le
2
.
Neu
r
al
n
et
w
o
r
k
s
co
r
e
S
e
n
t
e
n
c
e
ID
N
e
u
r
a
l
n
e
t
w
o
r
k
sco
r
e
1.
0
.
1
5
1
8
2.
0
.
1
3
9
1
3.
0
.
1
6
4
8
4.
0
.
0
9
9
1
5.
0
.
0
7
5
2
6.
0
.
0
7
4
7
7.
0
.
1
1
6
4
8.
0
.
1
0
4
5
A
t
f
ir
s
t,
t
h
e
in
p
u
t
d
o
cu
m
e
n
t
i
s
g
iv
e
n
to
th
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
f
o
r
d
o
cu
m
e
n
t
s
u
m
m
ar
izat
io
n
.
T
h
en
,
th
e
f
ea
tu
r
e
s
co
r
e
is
co
m
p
u
te
d
f
o
r
ev
er
y
s
e
n
ten
ce
b
ased
o
n
th
e
f
ea
tu
r
e
s
u
ti
lized
in
t
h
e
p
r
o
p
o
s
ed
h
y
b
r
id
ap
p
r
o
ac
h
.
T
h
e
s
am
p
le
r
es
u
lt
s
o
b
tain
ed
f
o
r
th
e
f
ea
t
u
r
e
m
a
tr
i
x
is
g
i
v
e
n
in
tab
le
1
.
Su
b
s
eq
u
en
t
l
y
,
t
h
e
s
y
n
tact
ic
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8938
IJ
-
AI
Vo
l.
3
,
No
.
3
,
Sep
tem
b
er
201
4
:
1
1
2
–
1
2
0
118
f
ea
t
u
r
e
is
co
m
p
u
ted
f
o
r
th
e
in
p
u
t te
x
t d
o
cu
m
en
t t
h
o
s
e
s
a
m
p
l
e
r
esu
lt
is
g
i
v
en
i
n
tab
le
2
.
T
h
is
m
a
tr
ix
is
g
i
v
en
to
th
e
n
eu
r
al
n
et
w
o
r
k
to
o
b
tain
t
h
e
s
e
n
te
n
ce
s
co
r
e.
T
h
e
f
in
al
s
en
ten
ce
s
co
r
e
o
b
tain
ed
f
r
o
m
t
w
o
n
eu
r
al
n
et
w
o
r
k
s
ar
e
g
iv
en
i
n
tab
le
3
.
Her
e,
th
e
n
eu
r
al
n
e
t
w
o
r
k
is
tr
ain
ed
w
it
h
th
e
s
en
te
n
ce
s
a
v
ailab
le
in
th
e
DUC
2
0
0
2
an
d
th
e
co
r
r
esp
o
n
d
in
g
tar
g
et
lab
el
i
s
i
d
en
tifie
d
w
i
th
t
h
e
s
u
m
m
ar
y
g
i
v
en
i
n
DU
C
2
0
0
2
d
ataset.
6
.
3
.
P
er
f
o
r
m
a
nce
E
v
a
lua
t
io
n
M
ea
s
ure
Fo
r
p
er
f
o
r
m
an
ce
ev
al
u
atio
n
,
w
e
h
a
v
e
u
s
ed
t
h
e
p
er
f
o
r
m
a
n
ce
m
ea
s
u
r
e
n
a
m
el
y
,
p
r
ec
is
io
n
,
r
e
ca
ll a
n
d
F
-
m
ea
s
u
r
e.
P
r
ec
is
io
n
m
ea
s
u
r
es
th
e
r
atio
o
f
co
r
r
ec
tn
ess
f
o
r
th
e
s
e
n
ten
ce
s
in
t
h
e
s
u
m
m
ar
y
w
h
er
eb
y
r
ec
all
is
u
tili
ze
d
to
co
u
n
t
t
h
e
r
atio
o
f
r
elev
an
t
s
e
n
te
n
ce
s
i
n
cl
u
d
ed
in
s
u
m
m
ar
y
.
Fo
r
p
r
ec
is
io
n
,
th
e
h
i
g
h
er
t
h
e
v
a
lu
e
s
,
th
e
b
etter
th
e
s
y
s
te
m
i
s
i
n
e
x
cl
u
d
i
n
g
ir
r
elev
a
n
t
s
e
n
ten
ce
s
.
O
n
t
h
e
o
th
er
h
an
d
,
t
h
e
h
i
g
h
er
t
h
e
r
e
ca
ll v
a
lu
e
s
t
h
e
m
o
r
e
ef
f
ec
tiv
e
th
e
s
y
s
te
m
w
o
u
ld
b
e
in
r
etr
iev
in
g
t
h
e
r
elev
a
n
t
s
en
t
en
ce
s
.
T
h
e
w
ei
g
h
ted
h
ar
m
o
n
i
c
m
ea
n
o
f
p
r
ec
is
io
n
an
d
r
ec
all
is
ca
lled
as F
-
m
ea
s
u
r
e.
|
s
e
n
t
e
n
c
e
s
}
{
R
e
t
r
i
e
v
e
d
|
|
s
e
n
t
e
n
c
e
s
}
{
R
e
l
e
v
a
n
t
}
s
e
n
t
e
n
c
e
s
{
R
e
t
r
i
e
v
e
d
|
P
r
e
c
i
s
i
o
n
|
s
e
n
t
e
n
c
e
s
}
{
R
e
l
e
v
a
n
t
|
|
s
e
n
t
e
n
c
e
s
}
{
R
e
l
e
v
a
n
t
s
e
n
t
e
n
c
e
s
}
{
R
e
t
r
i
v
e
d
|
R
e
c
a
l
l
W
h
er
e,
s
e
n
t
e
n
c
e
s
R
e
l
e
v
a
n
t
Sen
te
n
ce
s
t
h
at
ar
e
i
d
en
tifie
d
in
t
h
e
h
u
m
an
g
en
er
at
ed
s
u
m
m
ar
y
s
e
n
t
e
n
c
e
s
R
e
t
r
i
e
v
e
d
Sen
ten
ce
s
th
a
t a
r
e
r
etr
iev
ed
b
y
t
h
e
s
y
s
te
m
R
e
c
a
l
l
P
r
e
c
i
s
i
o
n
R
e
c
a
l
l
P
r
e
c
i
s
i
o
n
2
m
e
a
s
u
r
e
-
F
6
.
4
.
P
er
f
o
r
m
a
nce
a
na
ly
s
is
As
p
er
th
e
ap
p
licatio
n
o
f
ab
o
v
e
f
ea
t
u
r
es,
th
e
an
al
y
s
is
s
h
o
w
s
t
h
at
d
if
f
er
en
t
t
y
p
es
o
f
d
o
cu
m
en
ts
r
eq
u
ir
e
d
if
f
er
en
t
co
m
b
in
a
tio
n
s
o
f
f
ea
t
u
r
es
to
g
et
p
r
ec
is
e
Su
m
m
ar
y
.
T
h
e
s
u
m
m
ar
y
e
v
a
lu
atio
n
is
d
o
n
e
o
n
d
if
f
er
e
n
t
d
o
cu
m
e
n
ts
o
f
Sta
n
d
ar
d
DUC
Fo
u
n
d
atio
n
.
Do
c
u
m
e
n
t
s
ar
e
ca
te
g
o
r
ized
as
t
y
p
e
1
an
d
t
y
p
e
2
d
o
cu
m
en
ts
.
a.
T
y
p
e
1
d
o
cu
m
e
n
ts
Do
cu
m
e
n
t
s
ab
o
u
t
a
s
in
g
le
s
h
o
r
t sto
r
y
n
o
t
m
o
r
e
th
a
n
1
5
s
en
te
n
ce
s
.
b
.
T
y
p
e
2
d
o
cu
m
en
ts
Do
cu
m
e
n
t
s
a
b
o
u
t
a
b
io
g
r
ap
h
y
o
f
a
p
er
s
o
n
m
o
r
e
t
h
an
1
5
s
en
ten
ce
s
a
n
d
les
s
t
h
an
5
0
s
en
te
n
ce
s
.
Sen
te
n
ce
s
.
W
e
h
av
e
co
m
p
ar
ed
MS
W
o
r
d
Su
m
m
ar
y
a
n
d
o
u
r
p
r
o
p
o
s
e
d
ap
p
r
o
ac
h
u
s
i
n
g
all
e
ig
h
t
f
e
atu
r
es.
T
h
e
p
r
ec
is
io
n
(
Fi
g
u
r
e
1
)
,
r
ec
all
(
F
i
g
u
r
e
2
)
a
n
d
f
-
m
ea
s
u
r
e
(
Fi
g
u
r
e
3
)
f
o
r
th
e
t
w
o
t
y
p
e
o
f
d
o
cu
m
en
t
s
ar
e
ev
al
u
ated
.
T
h
e
r
esu
lts
s
h
o
w
th
a
t o
u
r
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
(
u
s
i
n
g
all
ei
g
h
t
f
ea
t
u
r
es)
o
u
tp
er
f
o
r
m
s
th
e
M
S w
o
r
d
s
u
m
m
ar
ies.
F
ig
u
re
1
.
Ef
fe
c
t
o
n
P
re
c
isio
n
F
ig
u
re
2.
Ef
fe
c
t
o
n
re
c
a
ll
Fig
u
r
e
[
1
-
3
]
s
h
o
w
t
h
at
t
h
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
o
u
tp
er
f
o
r
m
s
th
e
MS
w
o
r
d
s
u
m
m
ar
ies.
Fig
u
r
e
4
s
h
o
ws
th
e
ef
f
ec
t o
f
in
cl
u
s
io
n
o
f
co
n
c
ep
t b
ased
f
ea
tu
r
es o
n
s
h
o
r
t a
n
d
lar
g
e
d
o
cu
m
en
ts
.
Evaluation Warning : The document was created with Spire.PDF for Python.
IJ
-
AI
I
SS
N:
2252
-
8938
E
ffect
o
f F
ea
tu
r
e
S
elec
tio
n
o
n
S
ma
ll a
n
d
La
r
g
e
Do
cu
men
t
S
u
mma
r
iz
a
tio
n
…
(
D
.
Y
.
S
a
kh
a
r
e
)
119
F
ig
u
re
3
.
Ef
fe
c
t
o
n
F
m
e
a
su
re
F
ig
u
re
4
.
C
o
m
p
a
riso
n
b
a
se
d
o
n
c
o
n
c
e
p
t
b
a
se
d
f
e
a
tu
re
7.
CO
NCLU
SI
O
N
T
h
e
re
su
lt
s
sh
o
w
th
a
t,
su
m
m
a
r
y
g
e
n
e
ra
ted
u
sin
g
p
ro
p
o
se
d
m
o
d
u
l
e
o
u
t
p
e
rf
o
rm
s
to
th
a
t
o
f
M
S
-
W
ORD
m
o
d
u
le
f
o
r
a
ll
th
e
p
e
r
f
o
r
m
a
n
c
e
p
a
ra
m
e
te
rs.
T
h
e
e
x
p
e
ri
m
e
n
t
is
a
lso
c
a
rrie
d
o
u
t
w
it
h
c
o
n
sid
e
ra
ti
o
n
o
f
c
o
n
c
e
p
t
b
a
se
d
f
e
a
tu
re
s
a
n
d
w
it
h
o
u
t
c
o
n
c
e
p
t
b
a
se
d
f
e
a
tu
re
.
T
h
e
re
su
lt
s
,
in
f
ig
u
re
4
,
s
h
o
w
th
a
t
f
o
r
larg
e
d
o
c
u
m
e
n
t
su
m
m
a
ries
th
e
c
o
n
c
e
p
t
b
a
se
d
f
e
a
tu
re
in
c
re
a
se
s
th
e
q
u
a
li
ty
c
o
n
sid
e
ra
b
ly
.
S
o
w
e
c
a
n
c
o
n
c
lu
d
e
t
h
a
t
a
d
d
it
i
o
n
o
f
th
e
c
o
n
c
e
p
t
b
a
se
d
f
e
a
tu
re
s
h
e
lp
s
in
im
p
ro
v
in
g
th
e
q
u
a
li
ty
o
f
th
e
su
m
m
a
r
y
.
T
h
e
se
re
s
u
lt
s ac
h
iev
e
d
a
re
a
p
ro
m
isi
n
g
sta
rt
t
o
w
a
rd
f
u
rth
e
r
stu
d
ie
s
.
RE
F
E
R
E
NC
E
S
[1
]
A
u
to
m
a
ti
c
tex
t
su
m
m
a
riz
a
ti
o
n
u
s
in
g
se
n
ten
c
e
F
e
a
tu
re
s:
a
r
e
v
ie
w
,
I
n
ter
n
a
ti
o
n
a
l
J
.
o
f
E
n
g
g
.
Res
e
a
rc
h
&
In
d
u
.
Ap
p
ls.
(
IJ
ER
IA)
.
IS
S
N 0
9
7
4
-
1
5
1
8
,
V
o
l.
4
,
No
.
I
V
,
No
v
e
m
b
e
r
2
0
1
1
,
p
p
.
3
1
-
42
[2
]
Oi
M
e
a
n
F
o
o
n
g
,
A
lan
Ox
le
y
a
n
d
S
u
z
iah
S
u
laim
a
n
,
‘Ch
a
ll
e
n
g
e
s
a
n
d
tren
d
s
in
a
u
to
m
a
ti
c
tex
t
su
m
m
a
riza
ti
o
n
,
IJ
IT
T
,
V
o
l
.
1
,
Iss
u
e
1
,
2
0
1
0
p
p
3
4
-
3
9
’
[3
]
L
u
h
n
H.
P
,
‘T
h
e
A
u
to
m
a
ti
c
Cre
a
ti
o
n
o
f
L
it
e
ra
tu
re
A
b
stra
c
ts’,
IBM
J
o
u
rn
a
l
A
p
ril
1
9
5
8
p
p
.
1
5
9
–
1
6
5
[4
]
Ba
x
e
n
d
a
le,
P
.
(
1
9
5
8
)
,
‘M
a
c
h
i
n
e
-
m
a
d
e
In
d
e
x
f
o
r
T
e
c
h
n
ica
l
L
it
e
ra
tu
re
’A
n
Ex
p
e
rime
n
t‘,
IBM
J
o
u
r
n
a
l
o
f
Res
e
a
rc
h
De
v
e
lo
p
me
n
t
,
V
o
l
.
2
,
No
.
4
,
p
p
.
3
5
4
-
3
6
1
[5
]
Ed
m
u
n
d
so
n
H.
P
,
‘Ne
w
M
e
th
o
d
s
in
A
u
to
m
a
ti
c
Ex
trac
ti
n
g
’,
J
o
u
rn
a
l
o
f
th
e
Asso
c
i
a
ti
o
n
f
o
r
Co
m
p
u
t
in
g
M
a
c
h
in
e
ry
,
V
o
l
1
6
,
N
o
2
,
A
p
ril
1
9
6
9
,
P
P
.
2
6
4
-
28
5
[6
]
J.J.P
o
ll
o
c
k
a
n
d
A
.
Za
m
o
ra
,
‘
A
u
t
o
m
a
ti
c
A
b
stra
c
ti
n
g
Re
se
a
rc
h
a
t
Ch
e
m
ic
a
l
A
b
stra
c
ts
S
e
r
v
ice
’
,
J
o
u
rn
a
l
o
f
C
h
e
mic
a
l
In
fo
rm
a
t
io
n
a
n
d
Co
mp
u
ter
S
c
ien
c
e
s,
1
5
(
4
),
2
2
6
-
2
3
2
(
1
9
7
5
)
[7
]
Ka
th
lee
n
R.
M
c
Ke
o
w
n
,
‘Disc
o
u
rse
S
trate
g
ie
s
f
o
r
G
e
n
e
ra
ti
n
g
N
a
tu
ra
l
L
a
n
g
u
a
g
e
T
e
x
t’,
De
p
a
rt
m
e
n
t
o
f
Co
m
p
u
ter
S
c
ien
c
e
,
Co
lu
m
b
ia Un
iv
e
rsity
,
N
e
w
Yo
rk
,
1
9
8
2
[8
]
T
u
rn
e
y
,
’
L
e
a
rn
in
g
to
e
x
trac
t
k
e
y
p
h
ra
se
s
f
ro
m
te
x
t’,
tec
h
n
ica
l
re
p
o
rt
ERB
-
1
0
5
7
.
(NRC#
4
1
6
2
2
),
N
a
ti
o
n
a
l
Re
se
a
rc
h
Co
u
n
c
il
,
I
n
stit
u
te f
o
r
I
n
f
o
rm
a
ti
o
n
T
e
c
h
n
o
lo
g
y
,
1
9
9
9
[9
]
M
a
rc
u
,
D.
‘
T
h
e
a
u
to
m
a
ti
c
c
o
n
str
u
c
ti
o
n
o
f
l
a
rg
e
-
sc
a
le
c
o
rp
o
r
a
fo
r
su
mm
a
riza
ti
o
n
re
se
a
rc
h
’
.
In
P
r
o
c
e
e
d
in
g
s
o
f
th
e
2
2
n
d
I
n
tern
a
ti
o
n
a
l
Co
n
f
e
re
n
c
e
o
n
Re
se
a
rc
h
a
n
d
De
v
e
lo
p
m
e
n
t
in
In
f
o
rm
a
ti
o
n
Re
tri
e
v
a
l,
Un
iv
e
rsit
y
o
f
Ca
li
f
o
rn
ia,
Be
rk
e
le
y
,
A
u
g
u
st
1999
[1
0
]
B
K
Bo
g
u
ra
e
v
,
C
Ke
n
n
e
d
y
,
R
Be
ll
a
m
y
,
‘
D
y
n
a
mic
p
re
se
n
ta
ti
o
n
o
f
p
h
ra
s
a
l
-
b
a
se
d
d
o
c
u
me
n
t
a
b
s
tra
c
ti
o
n
s’
,
3
2
n
d
In
tern
a
ti
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
S
y
ste
m
S
c
ien
c
e
s,
1
9
9
9
[1
1
]
Bra
n
d
o
w
,
R.
,
M
it
z
e
,
K.,
Ra
u
,
’
A
u
to
m
a
ti
c
c
o
n
d
e
n
sa
ti
o
n
o
f
e
lec
tro
n
ic
p
u
b
li
c
a
ti
o
n
s
b
y
se
n
ten
c
e
se
lec
ti
o
n
’.
In
fo
rm
a
t
io
n
Pro
c
e
ss
in
g
a
n
a
g
e
me
n
t,
3
1
(5
)
:6
7
5
-
6
8
5
,
1
9
9
5
[1
2
]
Ra
d
e
v
,
R.
,
Blair
-
g
o
ld
e
n
so
h
n
,
S
,
Zh
a
n
g
,
Z.
,
‘Exp
e
rime
n
ts
in
S
in
g
le
a
n
d
M
u
lt
i
-
Do
c
u
e
mt
n
S
u
m
ma
riza
ti
o
n
u
sin
g
M
EA
D
’.
In
F
irst
Do
c
u
m
e
n
t
Un
d
e
rsta
n
d
in
g
C
o
n
f
e
re
n
c
e
,
Ne
w
Orle
a
n
s,
L
A
,
2
0
0
1
.
[1
3
]
Jin
g
,
Ho
n
g
y
a
n
a
n
d
Ka
th
lee
n
M
c
Ke
o
w
n
.
,
‘
Cu
t
a
n
d
p
a
ste
b
a
se
d
te
x
t
su
mm
a
riza
ti
o
n
’
.
In
1
st
Co
n
f
e
r
e
n
c
e
o
f
th
e
No
rth
Am
e
rica
n
Ch
a
p
ter o
f
th
e
A
ss
o
c
iat
io
n
f
o
r
C
o
m
p
u
tatio
n
a
l
L
in
g
u
isti
c
s ,
2
0
0
0
[1
4
]
Na
d
ira
Be
g
u
m
,
M
o
h
a
m
e
d
A
b
d
e
l
F
a
tt
a
h
,
F
u
j
i
Re
n
,
‘A
u
to
m
a
ti
c
te
x
t
su
m
m
a
riz
a
ti
o
n
u
si
n
g
su
p
p
o
rt
v
e
c
to
r
m
a
c
h
in
e
’,
In
ter
n
a
t
io
n
a
l
J
o
u
rn
a
l
o
f
I
n
n
o
v
a
ti
v
e
Co
mp
u
ti
n
g
,
V
o
l
u
m
e
5
,
p
p
:
1
9
8
7
-
1
9
9
6
,
2
0
0
9
.
[1
5
]
Ra
fe
e
q
A
l
-
Ha
sh
e
m
i,
‘T
e
x
t
S
u
m
m
a
riza
ti
o
n
Ex
trac
ti
o
n
S
y
ste
m
(
T
S
ES
)Us
in
g
Ex
trac
ted
Ke
y
w
o
rd
s’,
In
ter
n
a
ti
o
n
a
l
Ara
b
J
o
u
rn
a
l
o
f
e
-
T
e
c
h
n
o
l
o
g
y
,
Vo
l.
1
,
No
.
4
,
p
p
:
1
6
4
-
1
6
8
,
2
0
1
0
[1
6
]
Ra
jes
h
S
h
a
rd
a
n
a
n
d
P
ra
sa
d
,
Ud
a
y
Ku
lk
a
rn
i,
Im
p
le
m
e
n
tatio
n
a
n
d
Ev
a
lu
a
ti
o
n
o
f
Ev
o
lu
ti
o
n
a
ry
Co
n
n
e
c
ti
o
n
ist
A
p
p
ro
a
c
h
e
s
to
A
u
to
m
a
ted
T
e
x
t
S
u
m
m
a
riza
ti
o
n
’
J
o
u
r
n
a
l
o
f
Co
mp
u
ter
S
c
ie
n
c
e
6
(1
1
):,
2
0
1
0
I
S
S
N
1
5
4
9
-
3
6
3
6
,
p
p
1
3
6
6
-
1
3
7
6
.
[1
7
]
D.Y.S
a
k
h
a
re
,
Dr.Rak
ju
m
a
r
‘S
y
n
t
a
c
ti
c
a
l
Kn
o
w
led
g
e
b
a
se
d
S
tem
m
e
r
f
o
r
A
u
to
m
a
ti
c
Do
c
u
m
e
n
t
S
u
m
m
a
riza
ti
o
n
’,
CII
T
in
ter
n
a
t
io
n
a
l
j
o
u
r
n
a
l
o
f
d
a
ta
min
in
g
k
n
o
wled
g
e
e
n
g
in
e
e
rin
g
p
ri
n
t:
IS
S
N
0
9
7
4
–
9
6
8
3
&
o
n
l
in
e
:
iss
n
0
9
7
4
–
9
5
7
8
Iss
u
e
:
m
a
r
c
h
2
0
1
2
d
o
i:
d
m
k
e
0
3
2
0
1
2
0
0
2
.
[1
8
]
d
u
c
.
n
ist.
g
o
v
/d
a
ta.h
tm
l
B
I
B
L
I
O
G
R
AP
H
Y
O
F
AUT
H
O
RS
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8938
IJ
-
AI
Vo
l.
3
,
No
.
3
,
Sep
tem
b
er
201
4
:
1
1
2
–
1
2
0
120
D
.
Y.S
a
k
h
a
re
is
re
se
a
rc
h
sc
h
o
lar
a
t
Bh
a
ra
t
V
e
e
d
y
a
p
e
e
th
,
De
e
m
e
d
Un
iv
e
rsit
y
,
P
u
n
e
,
M
a
h
a
ra
sh
tra,In
d
ia.
S
h
e
is
c
u
rre
n
t
ly
w
o
r
k
in
g
a
s
A
s
sista
n
t
P
r
o
f
e
ss
o
r
in
De
p
a
rtm
e
n
t
o
f
e
le
c
tro
n
ics
En
g
in
e
e
rin
g
a
t
M
A
E,
A
lan
d
i,
P
u
n
e
.
S
h
e
h
a
s
to
tal
Ei
g
h
t
y
e
a
r
s
tea
c
h
in
g
e
x
p
e
rien
c
e
.
He
r
tea
c
h
in
g
a
re
a
s are
:
Dig
it
a
l
s
y
ste
m
s,In
f
o
r
m
a
ti
o
n
Re
tri
v
a
l,
V
L
S
I
De
sig
n
Ra
j
Ku
m
a
r
wa
s
b
o
rn
o
n
1
4
th
M
a
y
1
9
6
3
in
M
u
z
a
ff
a
rn
a
g
a
r
U.P
.
,
I
n
d
ia.
He
h
a
s
c
o
m
p
lete
d
h
is
M
.
S
c
.
(El
e
c
tro
n
ics
)
De
g
re
e
in
1
9
8
7
f
ro
m
Un
iv
e
rsit
y
o
f
M
e
e
ru
t,
M
e
e
ru
t.
He
h
a
s
b
e
e
n
a
w
a
rd
e
d
M
.
T
e
c
h
.
a
n
d
P
h
.
D
d
e
g
re
e
in
1
9
9
2
a
n
d
1
9
9
7
re
sp
e
c
ti
v
e
l
y
f
ro
m
Un
iv
e
rsit
y
o
f
De
lh
i,
Ne
w
De
lh
i.
He
w
o
rk
e
d
a
t
CEE
RI
P
il
a
n
i
f
ro
m
1
9
9
3
t
o
1
9
9
4
a
s
a
re
se
a
rc
h
a
ss
o
c
iate
.
F
ro
m
M
a
y
1
9
9
7
t
o
Ju
n
e
1
9
9
8
,
h
e
w
o
rk
e
d
a
s
As
si
sta
n
t
P
r
o
f
e
ss
o
r
in
De
p
a
rt
m
e
n
t
El
e
c
tro
n
ics
a
n
d
Co
m
m
u
n
ica
ti
o
n
s
En
g
g
,
V
e
ll
o
re
C
o
ll
e
g
e
o
f
En
g
g
.
(No
w
V
IT
),
V
e
ll
o
re
.
H
e
w
o
rk
e
d
in
DL
RL
(DRD
O),
H
y
d
e
ra
b
a
d
a
s
S
c
ien
ti
st
f
ro
m
Ju
n
e
1
9
9
8
t
o
A
u
g
u
st
2
0
0
2
a
n
d
late
r
o
n
c
a
m
e
i
n
DIA
T
(D
U)
in
S
e
p
t
2
0
0
2
.
A
t
p
re
se
n
t,
h
e
is
S
c
ien
ti
st
‘E’
in
De
p
a
rtme
n
t
o
f
El
e
c
tro
n
ics
En
g
g
.
,
DI
A
T
(De
e
m
e
d
Un
iv
e
r
sit
y
),
P
u
n
e
.
He
e
sta
b
li
sh
e
d
a
M
icro
w
a
v
e
a
n
d
M
il
li
m
e
ter
Wav
e
A
n
ten
n
a
Lab
o
ra
to
ry
in
DI
AT
(DU
),
P
u
n
e
a
n
d
f
o
rm
u
late
d
th
e
M
.
T
e
c
h
.
P
r
o
g
ra
m
m
e
in
th
e
De
p
a
rtm
e
n
t
o
f
El
e
c
tro
n
ics
En
g
g
.
in
2
0
1
0
.
He
h
a
s w
rit
ten
se
v
e
r
a
l
tec
h
n
ica
l
p
a
p
e
r
in
re
p
u
ted
I
n
tern
a
ti
o
n
a
l
J
o
u
r
n
a
l
a
n
d
c
o
n
f
e
re
n
c
e
s.
Evaluation Warning : The document was created with Spire.PDF for Python.