I
nte
rna
t
io
na
l J
o
urna
l o
f
I
nfo
rm
a
t
ics a
nd
Co
m
m
un
ica
t
io
n T
ec
hn
o
lo
g
y
(
I
J
-
I
CT
)
Vo
l.
15
,
No
.
2
,
J
u
n
e
20
26
,
p
p
.
769
~
7
7
7
I
SS
N:
2252
-
8
7
7
6
,
DOI
:
1
0
.
1
1
5
9
1
/iji
ct
.
v15
i
2
.
pp
769
-
7
7
7
769
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ict.
ia
esco
r
e.
co
m
H
y
brid d
eep n
eur
a
l net
wo
rk
mo
del
f
o
r as
pect
and o
p
inio
n
ex
trac
tion wi
th
m
ulti
-
hea
d
atten
tio
n
-
driv
en se
ntime
nt
a
na
ly
sis
Abhi
na
n
da
n Shi
ra
ha
t
t
i
1
,
Ra
m
esh
M
eda
r
2
,
Vij
a
y
Ra
j
pu
ro
hit
3
,
Sa
nje
ev
K
a
ulg
ud
4
,
M
rut
y
un
j
a
y
a
M
a
t
ha
d
Sh
iv
a
m
urt
ha
ia
h
5
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
a
n
d
E
n
g
i
n
e
e
r
i
n
g
,
K
o
l
h
a
p
u
r
I
n
st
i
t
u
t
e
o
f
Te
c
h
n
o
l
o
g
y
’
s,
C
o
l
l
e
g
e
o
f
E
n
g
i
n
e
e
r
i
n
g
,
K
o
l
h
a
p
u
r
,
I
n
d
i
a
2
D
e
p
a
r
t
me
n
t
o
f
C
S
E
(
A
I
& M
L)
,
K
a
s
e
g
a
o
n
E
d
u
c
a
t
i
o
n
S
o
c
i
e
t
y
’
s
R
a
j
a
r
a
mb
a
p
u
I
n
st
i
t
u
t
e
o
f
Te
c
h
n
o
l
o
g
y
,
a
f
f
i
l
i
a
t
e
d
t
o
S
h
i
v
a
j
i
U
n
i
v
e
r
si
t
y
,
I
sl
a
m
p
u
r
,
I
n
d
i
a
3
D
e
p
a
r
t
me
n
t
o
f
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
a
n
d
E
n
g
i
n
e
e
r
i
n
g
,
K
LS G
o
g
t
e
I
n
st
i
t
u
t
e
o
f
Te
c
h
n
o
l
o
g
y
,
B
e
l
a
g
a
v
i
,
I
n
d
i
a
4
D
e
p
a
r
t
m
e
n
t
o
f
A
r
t
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
a
n
d
M
a
c
h
i
n
e
L
e
a
r
n
i
n
g
,
N
e
w
H
o
r
i
z
o
n
C
o
l
l
e
g
e
o
f
E
n
g
i
n
e
e
r
i
n
g
,
B
e
n
g
a
l
u
r
u
,
I
n
d
i
a
5
S
c
h
o
o
l
o
f
E
n
g
i
n
e
e
r
i
n
g
Te
c
h
n
o
l
o
g
y
,
S
v
y
a
sa
D
e
e
m
e
d
t
o
b
e
U
n
i
v
e
r
si
t
y
S
a
t
v
a
G
l
o
b
a
l
C
i
t
y
C
a
m
p
u
s,
B
e
n
g
a
l
u
r
u
,
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
Dec
1
9
,
2
0
2
4
R
ev
is
ed
Dec
9
,
2
0
2
5
Acc
ep
ted
Ma
r
3
0
,
2
0
2
6
F
in
d
in
g
a
n
d
e
x
trac
ti
n
g
sig
n
ifi
c
a
n
t
fe
a
tu
re
s
fro
m
re
v
iew
se
n
ten
c
e
s
is
k
n
o
wn
a
s
a
sp
e
c
t
tri
p
let
e
x
trac
ti
o
n
,
a
n
d
it
p
r
o
v
i
d
e
s
su
c
c
in
c
t
in
f
o
rm
a
ti
o
n
o
n
th
e
e
lem
e
n
ts
th
a
t
u
se
rs
h
a
v
e
a
d
d
re
ss
e
d
.
Th
is
m
e
th
o
d
m
a
k
e
s
se
n
ti
m
e
n
t
a
n
a
ly
sis
a
n
d
o
p
i
n
i
o
n
m
in
i
n
g
e
a
sie
r,
wh
ich
h
e
lp
s
to
p
r
o
v
id
e
a
n
a
d
e
q
u
a
te
u
n
d
e
rsta
n
d
i
n
g
o
f
u
se
r
o
p
in
i
o
n
s
i
n
re
v
iew
s.
T
h
is
re
se
a
rc
h
p
re
se
n
t
s
a
n
o
v
e
l
a
p
p
ro
a
c
h
to
a
c
h
iev
e
a
sp
e
c
t
-
se
n
ti
m
e
n
t
tri
p
let
-
e
x
trac
ti
o
n
(AS
TE
)
u
s
in
g
a
d
e
e
p
n
e
u
ra
l
n
e
two
rk
a
n
d
tran
sfo
rm
e
r
-
b
a
se
d
m
u
lt
i
-
h
e
a
d
a
tt
e
n
ti
o
n
m
o
d
e
l.
Th
e
pr
o
p
o
se
d
h
y
b
ri
d
m
o
d
e
l
a
d
o
p
ts
a
p
ip
e
li
n
e
m
e
th
o
d
o
lo
g
y
,
c
o
n
c
u
rre
n
tl
y
e
x
trac
ti
n
g
o
p
i
n
i
o
n
s
a
n
d
a
sp
e
c
ts
wh
il
e
p
e
rfo
rm
i
n
g
se
n
t
ime
n
t
c
las
sifica
ti
o
n
.
Th
e
stu
d
y
a
d
d
re
ss
e
s th
e
in
tri
c
a
te ch
a
ll
e
n
g
e
o
f
id
e
n
ti
fy
in
g
tri
p
lets t
h
a
t
c
a
p
tu
re
n
u
a
n
c
e
d
re
lati
o
n
s
h
ip
s
b
e
twe
e
n
term
s
a
n
d
se
n
ten
c
e
s,
e
m
p
lo
y
in
g
a
d
e
e
p
n
e
u
ra
l
n
e
two
rk
f
o
r
jo
in
t
e
x
trac
ti
o
n
o
f
a
sp
e
c
ts
a
n
d
o
p
i
n
io
n
s
u
si
n
g
a
se
q
u
e
n
ti
a
l
tag
g
i
n
g
m
e
th
o
d
.
S
e
n
ti
m
e
n
t
c
las
sifica
ti
o
n
is
se
a
m
les
sly
in
teg
ra
ted
in
t
o
th
e
p
ip
e
li
n
e
,
trea
ti
n
g
se
n
t
ime
n
t
re
c
o
g
n
it
i
o
n
a
s
a
c
las
sifica
ti
o
n
tas
k
,
a
n
d
a
sp
e
c
t
a
n
d
o
p
i
n
io
n
e
x
trac
ti
o
n
a
s
tex
t
-
e
x
trac
ti
o
n
c
h
a
ll
e
n
g
e
s.
Ev
a
l
u
a
ti
o
n
s
wa
s
o
u
t
e
x
p
e
rime
n
tally
o
n
t
h
e
S
e
m
Ev
a
l
2
0
1
6
re
sta
u
ra
n
t
d
a
tas
e
t
d
e
m
o
n
stra
te
th
e
e
ffe
c
ti
v
e
n
e
ss
o
f
th
e
m
o
d
e
l,
d
e
sp
it
e
issu
e
s with
u
n
e
q
u
a
l
d
istr
ib
u
ti
o
n
o
f
d
a
ta.
k
ey
w
o
r
d
s
:
Asp
ec
t
Mu
lti
-
h
ea
d
-
atten
tio
n
Pip
elin
e
ap
p
r
o
ac
h
T
r
an
s
f
o
r
m
e
r
T
r
ip
let
ex
tr
ac
tio
n
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
:
Ab
h
in
an
d
a
n
Sh
ir
ah
atti
Dep
ar
tm
en
t o
f
C
o
m
p
u
ter
Scie
n
ce
an
d
E
n
g
in
ee
r
in
g
Ko
lh
ap
u
r
I
n
s
titu
te
o
f
T
ec
h
n
o
l
o
g
y
'
s
C
o
lleg
e
o
f
E
n
g
in
ee
r
in
g
Ko
lh
ap
u
r
,
I
n
d
ia
E
m
ail: a
b
h
is
h
ir
ah
atti0
7
0
7
@
g
m
ail.
co
m
1.
I
NT
RO
D
UCT
I
O
N
Asp
ec
t
-
s
en
tim
en
t
tr
ip
let
-
ex
tr
a
ctio
n
(
ASTE
)
in
v
o
l
v
es
th
e
s
e
p
ar
atio
n
o
f
m
en
tio
n
s
o
f
a
tar
g
et
en
tity
,
ass
o
ciate
d
s
en
tim
en
t,
an
d
a
n
o
p
in
io
n
r
an
g
e
p
r
o
v
id
in
g
co
n
te
x
t
f
o
r
th
e
s
en
tim
en
t.
C
u
r
r
en
t
r
esear
ch
co
m
m
o
n
ly
em
p
lo
y
s
p
ip
elin
e
m
eth
o
d
s
o
r
a
co
m
b
in
ed
a
p
p
r
o
ac
h
u
tili
zin
g
s
eq
u
en
tial
tag
g
in
g
to
a
d
d
r
ess
th
is
is
s
u
e.
I
d
en
tify
in
g
tr
ip
lets
th
at
co
n
v
e
y
r
ich
in
ter
co
n
n
ec
tio
n
s
b
etwe
en
ter
m
s
an
d
s
en
ten
ce
s
is
a
c
h
allen
g
in
g
r
esear
ch
p
r
o
b
lem
.
I
n
th
e
r
ea
lm
o
f
asp
ec
t
-
b
ased
s
en
tim
en
t
an
aly
s
is
(
AB
SA)
,
r
ec
en
t
r
esear
ch
ef
f
o
r
ts
h
av
e
ex
p
lo
r
e
d
v
ar
io
u
s
tech
n
iq
u
es
to
en
h
an
c
e
t
h
e
ex
tr
ac
tio
n
o
f
asp
ec
t
-
s
e
n
tim
en
t
tr
ip
lets
,
aim
in
g
f
o
r
m
o
r
e
ac
cu
r
ate
an
d
co
m
p
r
eh
e
n
s
iv
e
s
en
tim
en
t
u
n
d
er
s
tan
d
in
g
.
No
tab
ly
,
a
s
tu
d
y
b
y
[
1
]
in
tr
o
d
u
ce
d
a
m
u
lti
-
tu
r
n
m
ac
h
i
n
e
-
r
ea
d
in
g
-
co
m
p
r
eh
e
n
s
i
o
n
(
MT
MRC
)
task
an
d
a
b
i
-
d
ir
ec
tio
n
al
MRC
(
B
MR
C
)
ar
ch
itectu
r
e.
T
h
i
s
wo
r
k
f
o
c
u
s
ed
o
n
co
n
s
tr
u
ctin
g
lin
k
s
b
etwe
en
co
m
p
o
n
e
n
ts
u
s
in
g
n
o
n
-
r
estrictiv
e,
r
estrictiv
e,
an
d
s
en
tim
en
t
-
class
if
y
in
g
q
u
er
ies.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
7
7
6
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
,
Vo
l.
15
,
No
.
2
,
J
u
n
e
20
26
:
7
6
9
-
7
7
7
770
T
h
is
r
esear
ch
s
ee
k
s
to
im
p
r
o
v
e
th
e
p
r
ec
is
io
n
an
d
ef
f
ec
tiv
e
n
ess
o
f
ASTE
wi
th
in
s
en
tim
en
t
an
aly
s
is
b
y
d
esig
n
in
g
a
tr
a
n
s
f
o
r
m
e
r
-
b
ased
m
u
l
ti
-
h
ea
d
atten
tio
n
n
et
wo
r
k
.
T
h
e
p
r
o
p
o
s
ed
m
eth
o
d
in
teg
r
ates
a
h
y
b
r
i
d
m
o
d
el
with
in
a
p
ip
elin
e
s
tr
u
ctu
r
e
th
at
co
n
c
u
r
r
en
tly
ex
tr
ac
ts
o
p
in
io
n
s
,
asp
ec
ts
,
an
d
p
er
f
o
r
m
s
s
en
tim
en
t
class
if
icatio
n
.
T
o
ac
h
iev
e
th
is
,
we
u
tili
ze
a
d
ee
p
n
eu
r
al
n
et
wo
r
k
tailo
r
ed
f
o
r
th
e
s
im
u
lta
n
eo
u
s
ex
tr
ac
tio
n
o
f
asp
ec
ts
an
d
o
p
i
n
io
n
s
th
r
o
u
g
h
a
s
eq
u
en
tial
tag
g
i
n
g
tech
n
iq
u
e.
Sen
tim
en
t
class
if
icatio
n
is
tr
ea
ted
as
a
d
is
tin
ct
task
with
in
th
e
p
ip
elin
e,
wh
il
e
asp
ec
t
an
d
o
p
in
io
n
e
x
tr
ac
tio
n
ar
e
m
an
a
g
ed
as
tex
t
-
e
x
tr
ac
tio
n
ch
allen
g
es.
T
h
e
k
ey
co
n
tr
ib
u
tio
n
o
f
th
is
wo
r
k
is
its
ab
ilit
y
to
ca
p
tu
r
e
an
d
c
lass
if
y
in
tr
icate
r
elatio
n
s
h
ip
s
b
etwe
en
ter
m
s
an
d
s
en
ten
ce
s
,
ad
d
r
ess
in
g
im
b
alan
ce
d
d
ata
ch
allen
g
es,
as
ev
id
en
ce
d
b
y
ex
p
er
im
en
ts
o
n
t
h
e
Sem
E
v
al
2
0
1
6
r
estau
r
an
t
d
ataset.
T
h
is
m
eth
o
d
s
ig
n
if
ican
tly
ad
v
a
n
ce
s
ASTE
an
d
o
f
f
er
s
a
r
eliab
le
s
o
lu
tio
n
f
o
r
m
o
r
e
ac
cu
r
ate
s
en
tim
en
t a
n
aly
s
is
in
co
m
p
lex
r
ev
iew
d
atasets
.
Dai
an
d
So
n
g
[
2
]
,
p
r
o
p
o
s
ed
a
m
eth
o
d
f
o
r
m
i
n
in
g
e
x
tr
ac
tio
n
p
atter
n
s
f
r
o
m
p
r
e
-
e
x
is
tin
g
tr
ain
in
g
in
s
tan
ce
s
u
s
in
g
th
e
o
u
t
p
u
t
o
f
a
d
ep
en
d
en
cy
p
ar
s
er
.
T
h
is
ap
p
r
o
ac
h
,
in
co
r
p
o
r
atin
g
b
o
th
a
n
n
o
tatio
n
a
n
d
r
u
le
-
lab
eled
s
u
p
p
lem
en
tar
y
in
f
o
r
m
atio
n
,
d
em
o
n
s
tr
ated
im
p
r
o
v
ed
ac
cu
r
ac
y
co
m
p
ar
ed
to
ex
i
s
tin
g
s
tate
-
of
-
th
e
-
ar
t
m
eth
o
d
s
.
T
h
e
i
n
tr
o
d
u
ctio
n
o
f
b
id
ir
ec
tio
n
al
-
en
c
o
d
er
-
r
ep
r
esen
tatio
n
s
f
r
o
m
tr
an
s
f
o
r
m
er
s
(
B
E
R
T
)
in
[
3
]
m
ar
k
e
d
a
s
ig
n
if
ican
t
lea
p
in
lin
g
u
is
tic
r
ep
r
esen
tatio
n
.
Pre
-
tr
ain
in
g
d
ee
p
b
id
i
r
ec
tio
n
al
r
ep
r
esen
t
atio
n
s
u
s
in
g
tex
tu
al
d
ata
r
esu
lted
in
s
tate
-
of
-
t
h
e
-
a
r
t
p
er
f
o
r
m
an
ce
o
n
v
ar
io
u
s
n
at
u
r
al
lan
g
u
a
g
e
p
r
o
ce
s
s
in
g
(
NL
P)
task
s
,
in
clu
d
in
g
s
en
tim
en
t
-
r
elate
d
b
en
c
h
m
ar
k
s
.
Ad
d
r
ess
in
g
th
e
ch
allen
g
e
o
f
im
p
r
o
v
in
g
c
o
r
r
elatio
n
s
am
o
n
g
tar
g
ets
an
d
o
p
in
io
n
s
,
[
4
]
p
r
o
p
o
s
ed
a
two
-
s
tag
e
ap
p
r
o
ac
h
in
v
o
lv
in
g
s
eq
u
en
ce
tag
g
in
g
an
d
th
e
in
s
er
tio
n
o
f
ar
tific
ial
tag
s
ca
lled
p
er
ce
iv
ab
le
p
air
s
.
T
h
eir
m
o
d
el
o
u
tp
er
f
o
r
m
ed
s
tate
-
of
-
th
e
-
a
r
t a
p
p
r
o
ac
h
es a
cr
o
s
s
d
if
f
e
r
en
t d
atasets
.
A
u
n
if
ied
m
o
d
el
f
o
r
e
n
d
-
to
-
e
n
d
tar
g
et
-
b
ased
s
en
tim
en
t
an
aly
s
is
was
in
tr
o
d
u
ce
d
in
[
5
]
,
u
tili
zin
g
a
u
n
if
ied
tag
g
in
g
m
et
h
o
d
.
E
x
p
er
im
en
tal
ev
alu
ati
o
n
s
o
n
b
en
ch
m
ar
k
d
atasets
d
em
o
n
s
tr
ated
th
e
s
u
p
er
io
r
ity
o
f
th
eir
s
y
s
tem
o
v
er
c
o
m
p
etito
r
s
.
T
o
en
h
an
ce
en
tity
-
r
elatio
n
ex
tr
ac
tio
n
,
[
6
]
p
r
esen
ted
a
n
o
v
el
p
ar
ad
ig
m
,
r
ec
asti
n
g
th
e
p
r
o
b
lem
as
a
s
et
o
f
r
elate
d
q
u
esti
o
n
s
.
Pro
v
id
ed
a
two
-
s
tep
p
r
o
ce
s
s
,
u
s
in
g
a
u
n
if
ied
m
o
d
el
to
f
o
r
ec
ast
th
e
wh
at,
h
o
w,
an
d
wh
y
,
an
d
p
air
in
g
th
e
r
esu
lts
to
p
r
o
d
u
ce
tr
ip
lets
[
7
]
.
Po
r
ia
e
t
a
l.
[
8
]
,
ch
allen
g
ed
ex
is
tin
g
v
iews,
h
ig
h
lig
h
tin
g
g
ap
s
an
d
u
n
ex
p
l
o
r
ed
ter
r
ito
r
y
ess
en
tial
f
o
r
co
m
p
lete
s
en
tim
en
t
co
m
p
r
eh
e
n
s
io
n
.
T
h
ey
p
r
o
p
o
s
ed
a
tr
ajec
to
r
y
f
o
r
f
u
r
th
e
r
s
tu
d
y
,
ad
d
r
ess
in
g
ig
n
o
r
ed
a
n
d
u
n
s
o
lv
e
d
to
p
ics.
T
h
e
wo
r
k
in
[
9
]
p
r
o
p
o
s
ed
a
co
m
p
r
eh
e
n
s
iv
e
m
o
d
el,
o
f
f
er
in
g
s
tate
-
of
-
th
e
-
ar
t
o
u
tco
m
es
with
o
u
t
t
h
e
n
ee
d
f
o
r
p
ar
s
er
s
o
r
ad
d
itio
n
al
lan
g
u
ag
e
r
eso
u
r
ce
s
.
C
o
m
p
ar
ativ
e
a
n
aly
s
es
with
v
ar
io
u
s
b
aselin
es
an
d
Sem
E
v
al
C
h
allen
g
e
r
esu
lts
co
n
f
ir
m
e
d
th
e
m
o
d
el'
s
ef
f
icac
y
.
I
n
tr
o
d
u
cin
g
th
e
g
r
id
ta
g
g
in
g
s
ch
em
e
(
GT
S)
in
[
1
0
]
,
th
e
au
th
o
r
s
tack
led
th
e
asp
ec
t
-
f
o
cu
s
ed
o
p
i
n
io
n
ex
tr
ac
tio
n
(
AFOE
)
task
with
a
u
n
if
ied
g
r
id
tag
g
in
g
o
p
e
r
atio
n
.
GT
S
m
o
d
els,
b
ased
o
n
C
NN,
B
iLST
M,
an
d
B
E
R
T
,
o
u
tp
er
f
o
r
m
ed
s
tr
o
n
g
b
aselin
e
s
in
asp
ec
t
-
o
r
ien
ted
o
p
in
io
n
p
air
ex
tr
ac
t
io
n
an
d
o
p
in
io
n
tr
ip
let
ex
tr
ac
tio
n
d
ata
s
ets.
E
n
d
-
to
-
e
n
d
t
r
ip
let
ex
tr
ac
tio
n
was
ex
p
lo
r
ed
b
y
[
1
1
]
,
u
tili
zin
g
a
p
o
s
itio
n
-
awa
r
e
tag
g
in
g
tech
n
iq
u
e.
E
x
p
er
im
en
tal
f
in
d
in
g
s
d
em
o
n
s
tr
ated
s
u
p
er
io
r
p
er
f
o
r
m
a
n
ce
co
m
p
ar
ed
to
s
tate
-
of
-
th
e
-
ar
t
m
eth
o
d
s
.
R
ef
r
am
in
g
AB
SA
a
s
a
n
o
p
in
io
n
tr
i
p
let
ex
tr
ac
tio
n
task
,
p
r
o
p
o
s
ed
a
m
u
lti
-
task
lear
n
in
g
f
r
am
ewo
r
k
f
o
r
s
im
u
ltan
eo
u
s
a
s
p
ec
t
an
d
o
p
in
io
n
ter
m
ex
tr
ac
t
io
n
,
ac
h
iev
in
g
s
u
p
e
r
io
r
p
er
f
o
r
m
an
ce
o
n
Sem
E
v
al
b
en
ch
m
ar
k
s
[
1
2
]
.
Li
et
a
l.
[
1
3
]
,
th
e
a
u
th
o
r
s
in
tr
o
d
u
ce
d
a
n
in
n
o
v
ativ
e
m
eth
o
d
f
o
r
asp
e
ct
-
ter
m
ex
tr
ac
tio
n
(
AT
E
)
,
lev
er
ag
in
g
s
u
m
m
ar
ies
o
f
o
p
in
io
n
s
an
d
asp
ec
t
d
etec
tio
n
h
is
to
r
y
to
o
u
tp
er
f
o
r
m
ex
is
tin
g
s
tate
-
of
-
th
e
-
ar
t
m
eth
o
d
s
.
A
u
n
iq
u
e
d
ee
p
m
u
l
ti
-
task
lear
n
in
g
ar
ch
itectu
r
e
b
ased
o
n
lo
n
g
s
h
o
r
t
-
ter
m
m
e
m
o
r
y
(
L
STM
)
was
p
r
esen
ted
in
[
1
4
]
,
s
h
o
wca
s
in
g
ef
f
icien
cy
in
asp
ec
t
p
h
r
ase
ex
t
r
ac
tio
n
ac
r
o
s
s
two
in
d
u
s
tr
y
-
s
tan
d
ar
d
d
atasets
.
T
h
e
d
u
al
cr
o
s
s
-
s
h
ar
ed
R
NN
f
r
am
ewo
r
k
(
DOE
R
)
was
in
tr
o
d
u
ce
d
in
[
1
5
]
,
d
e
m
o
n
s
tr
atin
g
s
u
p
er
io
r
p
er
f
o
r
m
an
ce
in
g
en
e
r
atin
g
as
p
ec
t
ter
m
-
p
o
lar
ity
p
ai
r
in
g
s
ac
r
o
s
s
th
r
ee
b
en
c
h
m
ar
k
d
atasets
.
Pro
p
o
s
ed
a
n
ew
tech
n
iq
u
e
f
o
r
asp
ec
t
ter
m
ex
tr
ac
tio
n
u
s
in
g
u
n
s
u
p
er
v
is
ed
lea
r
n
in
g
o
f
d
is
tr
ib
u
te
d
r
e
p
r
esen
ta
tio
n
s
o
f
wo
r
d
s
an
d
d
ep
en
d
e
n
cy
p
at
h
s
[
1
6
]
.
A
p
ar
ad
ig
m
s
h
if
t
was
p
r
esen
ted
in
[
1
7
]
with
ASTE
-
R
L
,
tr
ea
tin
g
asp
ec
t
an
d
o
p
in
i
o
n
wo
r
d
s
as
r
ea
s
o
n
s
f
o
r
c
o
m
m
u
n
icate
d
em
o
tio
n
with
in
a
h
ier
ar
ch
ical
R
L
f
r
am
ew
o
r
k
.
E
x
p
er
im
en
tal
r
esu
lts
d
em
o
n
s
tr
ated
s
u
p
er
io
r
ity
o
v
e
r
cu
r
r
e
n
t
b
est
p
r
ac
tices
in
c
o
m
p
u
ter
an
d
r
estau
r
an
t
in
d
u
s
tr
y
d
atasets
.
T
h
ese
s
tu
d
ies
co
llectiv
ely
co
n
tr
ib
u
t
e
to
th
e
ev
o
lv
in
g
lan
d
s
ca
p
e
o
f
AB
SA,
s
h
o
wca
s
in
g
ad
v
an
ce
m
en
ts
in
tr
ip
let
ex
tr
ac
tio
n
tech
n
i
q
u
es a
n
d
p
r
o
v
id
in
g
v
alu
ab
le
in
s
ig
h
ts
f
o
r
f
u
tu
r
e
r
esear
ch
d
ir
ec
tio
n
s
.
2.
RE
S
E
ARCH
M
E
T
H
O
D
P
r
o
p
o
s
e
d
h
y
b
r
i
d
m
e
t
h
o
d
d
e
s
ig
n
e
d
f
o
r
t
h
e
e
x
t
r
a
c
t
i
o
n
o
f
t
r
i
p
l
e
t
s
i
n
as
p
e
c
t
-
b
as
e
d
s
e
n
t
i
m
en
t
a
n
a
l
y
s
is
(
A
B
S
A
)
.
Fi
g
u
r
e
1
i
l
l
u
s
t
r
a
t
es
t
h
e
a
r
c
h
i
t
e
c
t
u
r
e
o
f
t
h
e
p
r
o
p
o
s
e
d
h
y
b
r
i
d
m
e
t
h
o
d
,
w
h
i
c
h
c
o
n
s
i
s
t
s
o
f
t
w
o
c
r
u
c
i
a
l
s
t
e
p
s
.
2
.
1
.
Asp
ec
t
a
nd
o
pin
io
n
ex
t
r
a
ct
io
n
T
h
e
f
ir
s
t
s
tep
in
v
o
lv
es
th
e
ex
t
r
ac
tio
n
o
f
o
p
in
i
o
n
s
an
d
asp
ec
t
s
f
r
o
m
a
g
i
v
en
r
e
v
iew
s
en
ten
c
e.
A
d
ee
p
n
eu
r
al
n
etwo
r
k
tex
t
e
x
tr
ac
tio
n
m
eth
o
d
is
em
p
l
o
y
ed
to
ac
co
m
p
lis
h
th
is
task
.
T
h
e
in
p
u
t
to
th
is
s
tep
is
a
r
ev
iew
s
en
ten
ce
,
an
d
th
e
o
u
t
p
u
t
c
o
m
p
r
is
es
th
e
ex
tr
ac
ted
asp
ec
ts
a
n
d
o
p
in
io
n
s
.
T
h
is
s
u
b
s
ec
tio
n
d
is
cu
s
s
es
th
e
jo
in
t
ex
tr
ac
tio
n
o
f
o
p
in
io
n
s
an
d
asp
ec
ts
f
r
o
m
r
ev
iew
s
en
ten
ce
s
u
s
in
g
th
e
p
r
e
-
t
r
ain
ed
an
d
f
in
e
-
t
u
n
ed
B
E
R
T
m
eth
o
d
v
ia
Hu
g
g
in
g
-
Face
-
T
r
a
n
s
f
o
r
m
er
s
.
T
h
e
f
in
e
-
tu
n
in
g
p
r
o
ce
s
s
en
co
m
p
ass
es
s
ev
er
al
tas
k
s
,
s
tar
tin
g
with
th
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
I
SS
N:
2252
-
8
7
7
6
Hyb
r
id
d
ee
p
n
e
u
r
a
l n
etw
o
r
k
mo
d
el
fo
r
a
s
p
ec
t a
n
d
o
p
in
io
n
ex
tr
a
ctio
n
w
ith
… (
A
b
h
in
a
n
d
a
n
S
h
ir
a
h
a
tti
)
771
r
ep
r
esen
tatio
n
o
f
a
r
ev
iew
s
en
ten
ce
as,
X
=
{
1
,
2
,
3
,
.
.
.
,
}.
h
a
v
in
g
to
k
en
s
.
Usi
n
g
th
is
r
e
v
iew
s
en
ten
ce
,
th
e
Asp
ec
t
-
Sen
tim
en
t
T
er
m
E
x
tr
ac
tio
n
m
eth
o
d
f
ir
s
t
tr
ies
to
id
en
tify
all
th
e
tr
ip
lets
.
T
=
{(
,
,
)
}
|
T
|
,
wh
er
e
=
1
.
T
h
e
is
u
s
ed
to
d
e
f
in
e
th
e
asp
ec
t,
is
u
s
ed
to
d
e
f
in
e
t
h
e
o
p
in
i
o
n
,
is
u
s
ed
f
o
r
d
e
f
in
in
g
th
e
s
en
tim
en
t
an
d
|
T
|
,
is
u
s
ed
to
d
ef
in
e
th
e
m
ag
n
it
u
d
e
o
f
t
h
e
tr
i
p
lets
.
Fu
r
th
er
,
th
is
h
y
b
r
id
m
et
h
o
d
u
s
es
a
p
o
s
itio
n
-
tag
g
in
g
m
eth
o
d
.
T
h
e
p
o
s
itio
n
-
tag
g
in
g
m
eth
o
d
is
u
s
ed
f
o
r
d
e
f
in
in
g
t
h
e
asp
ec
t
at
th
e
s
tar
tin
g
p
o
i
n
t
an
d
is
u
s
ed
f
o
r
d
e
f
in
in
g
th
e
asp
ec
t
at
th
e
en
d
in
g
p
o
in
t.
Als
o
,
is
u
s
ed
f
o
r
d
e
f
in
in
g
t
h
e
o
p
in
i
o
n
at
th
e
s
tar
tin
g
p
o
in
t
an
d
is
u
s
ed
f
o
r
d
ef
in
in
g
th
e
o
p
in
io
n
at
th
e
en
d
in
g
p
o
i
n
t.
Mo
r
eo
v
er
,
as
th
e
r
ev
iew
t
ex
ts
ar
e
alr
ea
d
y
em
b
ed
d
e
d
,
th
e
p
o
s
itio
n
en
c
o
d
i
n
g
m
et
h
o
d
h
as
b
ee
n
u
s
ed
f
o
r
e
m
b
ed
d
in
g
th
e
p
o
s
itio
n
o
f
th
e
o
p
in
io
n
an
d
asp
ec
t
to
k
en
s
.
T
h
e
to
k
en
s
wh
ic
h
h
av
e
b
ee
n
co
n
tex
tu
alize
d
,
=
{
1
,
1
,
…….
,
},
ar
e
th
en
r
ep
r
esen
ted
u
s
in
g
th
e
B
E
R
T
m
eth
o
d
.
Fo
r
th
e
class
if
icatio
n
o
f
th
e
to
k
en
s
,
t
h
e
So
f
t
m
ax
m
eth
o
d
h
as
b
ee
n
u
s
ed
w
h
ich
is
r
ep
r
esen
ted
u
s
in
g
=
(
1
1
+
1
)
.
W
h
en
th
e
class
if
icatio
n
to
k
e
n
s
ar
e
class
if
ied
,
th
en
th
e
r
esu
lts
ar
e
s
ep
ar
ated
u
s
in
g
th
e
tar
g
et
wh
ich
is
s
et
u
s
in
g
th
e
o
p
in
io
n
s
et
=
{
1
,
2
,
…
,
}
an
d
asp
ec
t
s
et
=
{
1
,
2
,
…
.
,
}
.
Fig
u
r
e
2
s
h
o
ws
a
m
o
d
el
t
h
at
t
ak
es
m
u
ltip
le
in
p
u
ts
f
r
o
m
a
r
e
v
iew
s
en
ten
ce
an
d
p
r
o
ce
s
s
es
th
em
u
s
in
g
a
B
E
R
T
-
b
ased
lan
g
u
ag
e
m
o
d
el.
T
h
e
m
o
d
el
th
e
n
s
p
lits
in
to
two
p
ar
allel
p
ath
s
to
p
r
ed
i
ct
th
e
s
tar
t
an
d
en
d
p
o
s
itio
n
s
o
f
asp
ec
ts
an
d
o
p
in
io
n
s
in
th
e
tex
t.
E
ac
h
p
ath
u
s
es d
en
s
e
an
d
f
latten
lay
er
s
f
o
llo
wed
b
y
ac
tiv
atio
n
to
g
en
er
ate
f
in
al
o
u
tp
u
ts
.
I
t
h
el
p
s
id
en
tify
an
d
ex
tr
ac
t
im
p
o
r
tan
t
o
p
in
io
n
wo
r
d
s
an
d
asp
ec
t
ter
m
s
f
r
o
m
th
e
s
en
ten
ce
ef
f
icien
tly
.
Fig
u
r
e
1
.
Hy
b
r
id
m
et
h
o
d
f
o
r
t
h
e
ex
tr
ac
tio
n
o
f
tr
ip
lets
Fig
u
r
e
2
.
A
p
r
o
p
o
s
ed
m
eth
o
d
f
o
r
o
p
in
io
n
s
an
d
asp
ec
ts
tex
t e
x
tr
ac
tio
n
f
r
o
m
u
s
in
g
th
e
r
e
v
iew
s
en
ten
ce
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
7
7
6
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
,
Vo
l.
15
,
No
.
2
,
J
u
n
e
20
26
:
7
6
9
-
7
7
7
772
2
.
2
.
P
o
la
rit
y
cla
s
s
if
ica
t
io
n
T
h
e
p
r
o
p
o
s
ed
T
FB
E
R
T
m
eth
o
d
,
p
r
e
-
tr
ain
ed
a
n
d
o
p
tim
ized
with
to
k
en
an
d
p
o
s
itio
n
-
em
b
e
d
d
ed
lay
e
r
s
alo
n
g
with
tr
an
s
f
o
r
m
e
r
b
lo
ck
s
,
f
o
r
m
s
th
e
b
ac
k
b
o
n
e
o
f
th
is
class
if
icatio
n
ap
p
r
o
ac
h
.
T
h
e
in
p
u
t to
th
is
m
eth
o
d
is
a
r
ev
iew
s
en
ten
ce
r
ep
r
esen
ted
as
S
=
{s
1
,
s
2
,
s
3
...
s
1
}
.
I
n
th
is
p
r
o
ce
s
s
,
b
o
th
th
e
en
c
o
d
ed
o
p
in
io
n
s
et
Y
O
=
{o
1
,
o
2
.
.
.
o
m
}
an
d
asp
ec
t
s
et
Y
A
=
{a
1
,
a
2
.
.
.
a
m
}
ar
e
p
ass
ed
th
r
o
u
g
h
an
en
co
d
er
.
T
h
e
r
esu
ltin
g
en
co
d
ed
s
ets,
alo
n
g
with
th
e
in
p
u
t
s
en
ten
ce
S,
c
o
llectiv
ely
co
n
tr
i
b
u
te
to
th
e
class
if
icatio
n
o
f
p
o
lar
ity
.
T
h
i
s
m
eth
o
d
e
n
s
u
r
es
a
co
m
p
r
eh
e
n
s
iv
e
ev
alu
atio
n
o
f
s
en
tim
en
t
p
o
lar
ity
b
y
co
n
s
id
er
in
g
b
o
th
th
e
o
p
in
i
o
n
s
an
d
asp
ec
ts
ex
tr
ac
ted
f
r
o
m
th
e
in
p
u
t r
e
v
iew
s
en
ten
ce
,
th
u
s
en
h
an
cin
g
t
h
e
o
v
e
r
all
r
o
b
u
s
tn
ess
o
f
o
u
r
p
r
o
p
o
s
ed
f
r
am
ewo
r
k
.
T
h
e
f
u
n
d
am
en
tal
tr
a
n
s
f
o
r
m
er
-
en
co
d
er
c
o
m
p
r
is
es
two
in
teg
r
al
co
m
p
o
n
en
ts
:
th
e
m
u
lti
-
h
ea
d
-
atten
tio
n
(
MH
A)
m
eth
o
d
an
d
a
f
u
lly
c
o
n
n
ec
ted
lay
er
.
W
ith
in
th
e
tr
an
s
f
o
r
m
er
-
e
n
co
d
e
r
f
r
am
ewo
r
k
,
th
e
MH
A
p
lay
s
a
p
iv
o
tal
r
o
le
in
ca
p
tu
r
in
g
v
a
r
io
u
s
s
tates
o
f
th
e
in
p
u
t,
em
p
lo
y
i
n
g
a
s
elf
-
atten
tio
n
ap
p
r
o
ac
h
f
o
r
co
n
s
tr
u
ctin
g
b
o
th
in
p
u
t
an
d
o
u
tp
u
t
r
ep
r
esen
tatio
n
s
.
T
h
e
C
o
n
tex
t
-
Asp
ec
t
-
W
o
r
d
-
I
n
ter
ac
tio
n
is
ac
h
iev
e
d
th
r
o
u
g
h
th
e
MH
A,
wh
er
e
th
e
in
p
u
ts
f
o
r
th
e
MH
A
f
u
n
cti
o
n
co
n
s
is
t
o
f
a
v
alu
e
-
p
air
,
a
q
u
er
y
'
q
,
'
an
d
a
s
et
o
f
k
e
y
s
'k
'
d
ef
in
ed
as
k=
{k
1
,
k
2
,
k
3
,
.
.
.
,
k
n
}.
T
h
e
ca
lc
u
latio
n
o
f
s
ca
led
d
o
t
p
r
o
d
u
ct
atte
n
tio
n
in
v
o
lv
es
d
eter
m
in
in
g
weig
h
ts
th
r
o
u
g
h
th
e
d
o
t
p
r
o
d
u
ct
o
f
k
ey
s
an
d
q
u
er
y
'q
,
'
with
th
e
d
is
tr
ib
u
ti
o
n
o
f
atten
ti
o
n
co
m
p
u
ted
u
s
in
g
(
1
)
.
(
,
)
=
(
(
,
)
)
(
1
)
2
.
2
.
1
.
At
t
ent
io
n
m
ec
ha
nis
m
a
nd
hea
d inte
g
ra
t
io
n
In
(
2
)
,
th
e
s
em
an
tic
-
g
ai
n
’
s
’
is
em
p
lo
y
ed
to
ca
lcu
late
s
em
an
tic
r
elev
an
ce
b
etwe
en
th
e
asp
ec
t
-
wo
r
d
an
d
co
n
tex
t,
co
n
s
titu
tin
g
th
e
s
co
r
in
g
m
ec
h
an
is
m
in
(
2
)
:
=
ℎ
(
[
,
]
.
)
(
2
)
T
h
is
s
co
r
in
g
m
ec
h
an
is
m
is
v
ital
in
th
e
m
u
lti
-
h
ea
d
atten
tio
n
m
ec
h
an
is
m
(
MH
A)
co
m
p
u
t
atio
n
,
wh
er
e
eig
h
t
h
ea
d
s
ar
e
u
tili
ze
d
f
o
r
lear
n
in
g
in
f
o
r
m
atio
n
ac
r
o
s
s
d
if
f
er
en
t
b
atch
es.
T
h
e
h
ea
d
s
,
r
ep
r
ese
n
ted
as
h
ea
d
i
,
ar
e
ca
lcu
lated
u
s
in
g
th
e
atten
tio
n
f
u
n
ctio
n
in
(
3
)
,
(
4
)
.
ℎ
=
(
,
)
(
3
)
=
(
ℎ
1
⨁
ℎ
2
⨁
ℎ
3
…
⨁
ℎ
ℎ
)
.
(
4
)
T
h
e
MH
A
f
u
n
ctio
n
m
itig
ates
th
e
v
an
is
h
in
g
-
g
r
ad
ien
t
is
s
u
e
b
y
o
f
f
e
r
in
g
a
m
o
r
e
s
tr
aig
h
tf
o
r
war
d
p
ath
to
t
h
e
in
p
u
ts
.
Au
g
m
en
tatio
n
o
f
h
i
d
d
en
an
d
o
u
tp
u
t
s
tates
with
a
co
n
tex
t
v
ec
to
r
C
i
,
co
m
p
u
ted
th
r
o
u
g
h
(
5
)
,
a
d
d
r
ess
es
th
is
co
n
ce
r
n
u
s
in
g
(
5
)
.
=
=
1
ℎ
(
5
)
2
.
2
.
2
.
At
t
ent
io
n
ca
lcula
t
io
n a
nd
a
lig
nm
ent
m
et
ho
d
I
n
(
6
)
,
th
e
atten
tio
n
a
ij
f
o
r
th
e
i
th
o
u
tp
u
t,
wh
ich
is
d
eter
m
i
n
ed
b
y
s
o
f
tm
a
x
(
)
.
T
h
e
e
ij
v
alu
es
ar
e
o
b
tain
ed
u
s
in
g
(
7
)
,
wh
e
r
e
'f'
r
ep
r
esen
ts
th
e
alig
n
m
en
t
m
eth
o
d
,
s
co
r
in
g
in
p
u
ts
s
u
r
r
o
u
n
d
e
d
b
y
th
e
j
th
o
u
tp
u
t
at
th
e
i
th
p
o
s
itio
n
.
T
h
e
h
i
d
d
en
s
tate
S
(i
-
1)
,
d
e
r
iv
ed
f
r
o
m
th
e
p
r
e
v
io
u
s
tim
e
s
te
p
,
in
f
lu
en
ce
s
th
e
alig
n
m
en
t
m
et
h
o
d
,
wh
ich
ca
n
b
e
ad
d
itiv
e,
m
u
ltip
l
icativ
e,
o
r
a
d
o
t p
r
o
d
u
ct,
o
f
f
e
r
in
g
f
lex
ib
ilit
y
in
co
m
p
u
tatio
n
.
=
(
)
=
(
)
=
1
(
)
(
6
)
=
(
−
1
,
ℎ
)
(
7
)
T
h
is
co
m
p
r
eh
en
s
iv
e
tr
a
n
s
f
o
r
m
er
-
en
co
d
er
m
ec
h
a
n
is
m
in
teg
r
ates
atten
tio
n
,
m
u
lti
-
h
ea
d
f
u
n
ctio
n
ality
,
a
n
d
co
n
tex
t a
u
g
m
en
tatio
n
t
o
ef
f
ec
t
iv
ely
ca
p
tu
r
e
a
n
d
p
r
o
ce
s
s
s
eq
u
en
tial in
f
o
r
m
atio
n
with
in
v
a
r
io
u
s
NL
P task
s
.
3.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
T
h
e
im
p
lem
en
tatio
n
o
f
th
e
p
r
e
s
en
ted
h
y
b
r
id
m
et
h
o
d
h
as
b
ee
n
ca
r
r
ied
o
u
t
in
Py
th
o
n
,
u
tili
zin
g
th
e
Ker
as
d
ee
p
n
eu
r
al
-
n
etwo
r
k
ap
p
r
o
ac
h
f
o
r
b
o
t
h
tr
ain
in
g
an
d
test
in
g
ac
r
o
s
s
two
p
h
ases
,
T
h
e
ev
alu
atio
n
o
f
th
e
m
eth
o
d
is
p
er
f
o
r
m
e
d
o
n
th
e
Sem
E
v
al
2
0
1
6
r
estau
r
an
t
d
at
aset,
an
d
a
c
o
n
cise
o
v
er
v
iew
o
f
th
is
d
ataset
is
p
r
o
v
id
e
d
in
th
e
s
u
b
s
eq
u
en
t sec
tio
n
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
I
SS
N:
2252
-
8
7
7
6
Hyb
r
id
d
ee
p
n
e
u
r
a
l n
etw
o
r
k
mo
d
el
fo
r
a
s
p
ec
t a
n
d
o
p
in
io
n
ex
tr
a
ctio
n
w
ith
… (
A
b
h
in
a
n
d
a
n
S
h
ir
a
h
a
tti
)
773
3
.
1
.
Da
t
a
s
et
T
h
is
s
ec
tio
n
d
elv
es
in
to
a
b
r
i
ef
ex
p
lan
atio
n
o
f
th
e
Sem
E
v
al
d
ataset,
o
r
ig
in
atin
g
f
r
o
m
f
i
v
e
d
is
tin
ct
ch
allen
g
es
[
1
8
]
-
[
2
0
]
.
Op
i
n
io
n
s
ar
e
class
if
ied
f
o
r
ea
c
h
ter
m
an
d
s
en
ten
ce
as
p
r
esen
ted
i
n
[
2
1
]
,
s
er
v
in
g
as
th
e
f
o
u
n
d
atio
n
f
o
r
tr
ain
i
n
g
a
n
d
tes
tin
g
o
f
o
p
in
io
n
ter
m
s
an
d
s
en
t
en
ce
s
[
2
2
]
.
Fig
u
r
e
3
p
r
o
v
id
es
an
o
v
er
v
iew
o
f
th
e
class
d
is
tr
ib
u
tio
n
with
in
th
e
v
a
lid
atio
n
,
test
in
g
,
an
d
tr
ain
in
g
d
atasets
.
I
n
Sem
E
v
al,
th
e
cu
s
to
m
er
r
e
v
iews
f
o
r
t
h
e
r
estau
r
a
n
t
d
ata
s
et
ar
e
ca
teg
o
r
ize
d
in
t
o
th
r
e
e
class
es:
n
eu
tr
al
(
NE
U)
,
p
o
s
itiv
e
(
POS),
an
d
n
eg
ativ
e
(
NE
G)
.
T
h
e
a
n
aly
s
is
f
r
o
m
Fig
u
r
e
3
r
e
v
ea
ls
an
im
b
alan
ce
i
n
th
e
d
ataset,
ch
ar
ac
ter
ized
b
y
a
s
u
b
s
tan
tial
n
u
m
b
er
o
f
p
o
s
itiv
e
r
ev
iews,
a
s
m
aller
q
u
an
tity
o
f
n
eu
tr
al
r
ev
iews,
an
d
a
m
o
d
er
ate
n
u
m
b
er
o
f
n
e
g
ativ
e
r
ev
iews
d
u
r
i
n
g
th
e
tr
ain
in
g
p
h
ase.
T
h
is
d
is
tr
ib
u
ti
o
n
p
atter
n
p
er
s
is
ts
co
n
s
is
ten
tly
in
b
o
th
t
h
e
v
al
id
at
io
n
an
d
test
in
g
d
atasets
[
2
3
].
Fig
u
r
e
3
.
Sen
tim
en
t c
lass
d
is
tr
ib
u
tio
n
in
tr
ai
n
in
g
,
test
in
g
,
an
d
v
alid
atio
n
d
atasets
3
.
2
.
O
pin
io
n a
nd
a
s
pect
ex
t
r
a
ct
io
n:
m
e
t
ho
do
lo
g
y
f
o
r
e
v
a
l
ua
t
io
n
T
o
ass
ess
th
e
p
er
f
o
r
m
a
n
ce
o
f
th
e
o
p
in
io
n
an
d
asp
ec
t
ex
tr
ac
ti
o
n
,
a
p
a
r
am
eter
k
n
o
wn
as
"e
x
ac
t
m
atch
"
is
in
tr
o
d
u
ce
d
.
T
h
is
p
ar
am
eter
g
au
g
es
th
e
p
r
ec
is
io
n
o
f
th
e
m
eth
o
d
b
y
alig
n
i
n
g
with
th
e
s
p
ec
if
ic
lo
ca
tio
n
o
f
o
p
in
io
n
ter
m
s
an
d
asp
ec
ts
.
T
h
e
ex
tr
ac
tio
n
p
r
o
ce
s
s
is
co
n
ce
p
tu
alize
d
as
a
tex
t
ex
tr
ac
tio
n
ch
allen
g
e,
wh
er
e
th
e
lo
g
it
o
f
th
e
in
itial
an
d
en
d
in
g
lo
ca
tio
n
s
o
f
o
p
in
io
n
s
an
d
asp
e
cts
s
er
v
es
a
s
th
e
o
u
tp
u
t.
T
h
e
m
o
d
el
is
tr
ain
ed
f
o
r
2
0
iter
atio
n
s
,
o
p
tim
izin
g
f
o
u
r
cr
o
s
s
-
ca
teg
o
r
y
en
tr
o
p
y
-
lo
s
s
f
u
n
ctio
n
s
.
E
ac
h
f
u
n
ctio
n
c
o
r
r
esp
o
n
d
s
to
o
n
e
o
p
in
io
n
an
d
o
n
e
p
er
s
p
ec
tiv
e
s
tate,
f
o
cu
s
in
g
o
n
th
e
i
n
itial
an
d
en
d
in
g
l
o
ca
tio
n
s
wi
th
in
th
e
tr
ain
in
g
s
et.
Fig
u
r
e
4
p
r
esen
ts
a
lin
e
g
r
ap
h
illu
s
tr
atin
g
t
h
e
p
r
ec
is
e
m
at
ch
s
co
r
e,
lo
s
s
,
an
d
v
alid
atio
n
lo
s
s
ac
r
o
s
s
th
e
2
0
iter
atio
n
s
d
u
r
in
g
b
o
th
th
e
tr
ain
in
g
an
d
v
alid
atio
n
p
h
ases
.
Fig
u
r
e
4.
E
x
ac
t
m
atch
s
co
r
e
,
l
o
s
s
,
an
d
v
alid
atio
n
lo
s
s
d
u
r
in
g
th
e
tr
ain
in
g
p
h
ase
A
co
m
p
r
e
h
en
s
iv
e
o
v
er
v
iew
o
f
th
e
tr
ain
i
n
g
,
v
alid
atio
n
,
an
d
t
esti
n
g
r
esu
lts
f
o
r
t
h
e
f
o
u
r
lo
s
s
f
u
n
ctio
n
s
an
d
an
ex
ac
t
m
atch
is
p
r
o
v
id
e
d
in
T
ab
le
1
.
T
h
is
ev
alu
atio
n
f
r
am
ewo
r
k
en
s
u
r
es
a
th
o
r
o
u
g
h
u
n
d
er
s
tan
d
in
g
o
f
th
e
m
o
d
el'
s
p
er
f
o
r
m
a
n
ce
in
e
x
tr
ac
tin
g
asp
ec
ts
an
d
o
p
in
io
n
s
o
v
er
th
e
c
o
u
r
s
e
o
f
tr
ain
in
g
iter
atio
n
s
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
7
7
6
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
,
Vo
l.
15
,
No
.
2
,
J
u
n
e
20
26
:
7
6
9
-
7
7
7
774
T
ab
le
1
.
Asp
ec
t a
n
d
o
p
i
n
io
n
e
x
tr
ac
tio
n
r
esu
lts
f
o
r
tr
ain
in
g
,
v
alid
atio
n
,
an
d
test
in
g
P
h
a
se
A
sp
e
c
t
s
t
a
r
t
l
o
ss
A
sp
e
c
t
e
n
d
l
o
s
s
O
p
i
n
i
o
n
st
a
r
t
l
o
ss
O
p
i
n
i
o
n
e
n
d
l
o
ss
Ex
a
c
t
ma
t
c
h
sc
o
r
e
Tr
a
i
n
i
n
g
0
.
0
1
3
4
0
.
0
0
3
3
0
.
0
1
4
2
0
.
0
0
4
1
8
1
%
V
a
l
i
d
a
t
i
o
n
0
.
0
9
1
2
0
.
1
0
1
2
0
.
0
9
0
2
0
.
0
9
7
2
7
6
%
Te
st
i
n
g
-
-
-
-
7
5
%
3
.
3
.
P
o
la
rit
y
cla
s
s
if
ica
t
io
n mo
del
Utilizin
g
an
en
h
a
n
ce
d
T
FB
E
R
T
p
r
e
-
tr
ain
ed
m
eth
o
d
,
w
h
ich
in
co
r
p
o
r
ates
lo
ca
tio
n
a
n
d
to
k
en
em
b
ed
d
in
g
s
f
o
r
ea
ch
o
p
i
n
io
n
an
d
asp
ec
t,
alo
n
g
with
cu
s
to
m
er
r
ev
iew
s
en
ten
ce
s
o
r
tex
t,
tr
an
s
f
o
r
m
atio
n
b
lo
ck
s
,
a
n
d
atten
tio
n
-
h
ea
d
s
,
a
d
ee
p
n
e
u
r
al
-
n
etwo
r
k
h
as
b
ee
n
d
esig
n
e
d
f
o
r
p
o
lar
ity
class
if
icatio
n
.
T
h
e
m
o
d
el
,
tr
ain
ed
with
a
n
Ad
am
o
p
tim
iz
er
,
u
n
d
e
r
g
o
es 3
0
iter
atio
n
s
to
class
if
y
p
o
lar
ity
d
ata.
T
h
e
o
b
j
ec
tiv
e
is
to
o
p
tim
ize
ac
cu
r
ac
y
wh
ile
m
in
i
m
izin
g
th
e
Sp
ar
s
e
-
C
ateg
o
r
ical
C
r
o
s
s
-
E
n
tr
o
p
y
L
o
s
s
-
Fu
n
ctio
n
.
Up
o
n
clo
s
er
ex
am
in
ati
o
n
o
f
in
d
iv
id
u
al
class
es
d
u
r
in
g
te
s
tin
g
p
h
ase,
th
e
p
o
s
itiv
e
r
ev
iew
class
ex
h
ib
ited
a
r
ec
all
o
f
9
5
%
,
p
r
e
cisi
o
n
o
f
7
7
%
,
an
d
an
f
-
m
ea
s
u
r
e
o
f
8
5
%
.
I
n
co
n
tr
ast,
th
e
n
eu
tr
al
an
d
n
eg
ativ
e
r
ev
iew
class
es
d
is
p
lay
ed
less
f
av
o
r
a
b
le
m
etr
ics,
with
ac
cu
r
ac
y
,
r
ec
all,
f
-
m
ea
s
u
r
e,
a
n
d
p
r
ec
is
io
n
all
f
allin
g
b
elo
w
5
0
%
.
T
h
is
h
ig
h
lig
h
t
s
a
s
ig
n
if
ican
t
im
b
alan
ce
in
th
e
d
ataset,
co
m
p
r
is
in
g
2
1
0
s
am
p
les
f
o
r
th
e
p
o
s
itiv
e
class
,
9
6
s
am
p
les
f
o
r
th
e
n
eg
ativ
e
class
,
an
d
a
m
er
e
1
6
s
a
m
p
les
f
o
r
th
e
n
e
u
tr
al
class
i
n
T
ab
le
2
.
Fig
u
r
e
5
p
r
o
v
id
es
a
v
is
u
al
r
ep
r
esen
tati
o
n
o
f
th
e
class
if
icatio
n
r
esu
lts
f
o
r
ea
ch
s
tag
e.
No
tab
ly
,
d
u
r
i
n
g
tr
ain
in
g
,
m
o
s
t
o
f
th
e
s
am
p
les
wer
e
ac
cu
r
ately
id
en
tifie
d
,
r
esu
ltin
g
in
a
9
9
%
p
r
ec
is
io
n
r
ate.
Ho
wev
er
,
v
alid
a
tin
g
an
d
test
in
g
th
e
n
eu
tr
al
class
s
am
p
les ex
h
ib
ited
a
5
0
% su
cc
ess
r
ate,
u
n
d
er
s
c
o
r
in
g
th
e
ch
allen
g
es p
o
s
ed
b
y
im
b
alan
ce
d
d
ata.
T
ab
le
3
p
r
esen
ts
a
c
o
m
p
a
r
a
tiv
e
an
aly
s
is
b
etwe
en
th
e
p
r
o
p
o
s
ed
s
y
s
tem
an
d
ex
is
tin
g
m
o
d
els,
ev
alu
atin
g
k
ey
m
etr
ics
s
u
ch
as
ac
cu
r
ac
y
,
p
r
ec
is
io
n
,
r
ec
all
,
an
d
F1
-
s
co
r
e.
Ou
d
al
o
v
et
a
l,
[2
4
]
,
th
e
m
o
d
el
ac
h
iev
ed
a
p
r
ec
is
io
n
o
f
4
6
.
9
6
,
r
ec
all
o
f
6
4
.
2
4
,
an
d
an
F1
-
s
co
r
e
o
f
5
4
.
2
1
.
R
ep
o
r
ted
a
p
r
ec
is
i
o
n
o
f
6
5
.
6
5
,
r
ec
all
o
f
5
4
.
2
8
,
an
d
a
n
F1
-
s
co
r
e
o
f
5
9
.
4
2
[
1
2
]
.
Dem
o
n
s
tr
ated
a
p
r
ec
is
io
n
o
f
7
0
.
1
7
,
r
ec
all
o
f
5
5
.
9
5
,
an
d
an
F1
-
s
co
r
e
o
f
6
2
.
2
6
[
1
0
]
.
R
ep
o
r
te
d
a
n
ac
cu
r
ac
y
o
f
7
0
.
9
1
,
p
r
ec
is
io
n
o
f
6
9
.
8
7
,
r
ec
all
o
f
6
5
.
6
8
,
an
d
a
n
F1
-
s
co
r
e
o
f
6
7
.
3
5
[
2
5
]
.
Ad
d
itio
n
ally
,
[
1
7
]
ac
h
iev
ed
an
ac
cu
r
ac
y
o
f
7
2
.
1
1
,
p
r
ec
is
io
n
o
f
6
7
.
2
1
,
r
ec
all
o
f
6
9
.
6
9
,
an
d
an
F1
-
s
co
r
e
o
f
6
8
.
4
1
.
I
n
co
m
p
ar
is
o
n
,
th
e
p
r
o
p
o
s
ed
m
o
d
el
ex
h
ib
ited
s
u
p
er
i
o
r
p
er
f
o
r
m
an
ce
with
an
ac
cu
r
ac
y
o
f
7
6
,
p
r
ec
is
io
n
o
f
7
4
,
r
ec
all
o
f
7
6
,
an
d
a
n
F1
-
s
co
r
e
o
f
7
3
.
T
ab
le
2
.
Pre
cisi
o
n
,
r
ec
all
an
d
f
1
-
s
co
r
e
m
etr
ics f
o
r
test
in
g
p
h
a
s
e
Te
st
i
n
g
p
h
a
s
e
P
r
e
c
i
s
i
o
n
R
e
c
a
l
l
F1
-
sc
o
r
e
S
u
p
p
o
r
t
P
o
si
t
i
v
e
0
.
7
7
0
.
9
5
0
.
8
5
2
1
0
N
e
g
a
t
i
v
e
0
.
7
6
0
.
4
6
0
.
5
7
96
N
e
u
t
r
a
l
0
.
2
5
0
.
0
6
0
.
1
0
16
a
c
c
u
r
a
c
y
0
.
7
6
3
2
2
mac
r
o
a
v
g
0
.
5
9
0
.
4
9
0
.
5
1
3
2
2
w
e
i
g
h
t
e
d
a
v
g
0
.
7
4
0
.
7
6
0
.
7
3
3
2
2
Fig
u
r
e
5.
C
o
n
f
u
s
io
n
m
atr
i
x
f
o
r
tr
ain
in
g
,
v
alid
atio
n
a
n
d
test
in
g
p
h
ases
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
I
SS
N:
2252
-
8
7
7
6
Hyb
r
id
d
ee
p
n
e
u
r
a
l n
etw
o
r
k
mo
d
el
fo
r
a
s
p
ec
t a
n
d
o
p
in
io
n
ex
tr
a
ctio
n
w
ith
… (
A
b
h
in
a
n
d
a
n
S
h
ir
a
h
a
tti
)
775
T
ab
le
3
.
C
o
m
p
a
r
is
o
n
o
f
th
e
ex
is
tin
g
s
y
s
tem
with
p
r
o
p
o
s
ed
s
y
s
tem
M
o
d
e
l
A
c
c
u
r
a
c
y
(
%)
P
r
e
c
i
s
i
o
n
(
%)
R
e
c
a
l
l
(
%)
F1
-
S
c
o
r
e
(
%)
W
h
a
t
H
o
w
W
h
y
[
2
3
]
-
4
6
.
9
6
6
4
.
2
4
5
4
.
2
1
O
TE
-
M
TL
[
1
2
]
-
6
5
.
6
5
5
4
.
2
8
5
9
.
4
2
G
TS (
B
i
LS
TM
)
[
1
0
]
-
7
0
.
1
7
5
5
.
9
5
6
2
.
2
6
JET
t
(
B
i
LS
TM
)
[
2
4
]
5
8
.
4
5
6
3
.
5
9
5
0
.
9
7
5
6
.
5
9
JET
o
(
B
i
LST
M
)
[
2
5
]
6
0
.
9
0
7
0
.
9
4
5
7
.
0
0
6
3
.
2
1
G
TS (
B
E
R
T)
[
1
0
]
-
6
5
.
8
9
6
6
.
2
7
6
6
.
0
8
JET
t
(
B
ER
T)
[
2
3
]
6
3
.
7
5
7
0
.
4
2
5
8
.
3
7
6
3
.
8
3
JET
o
(
B
ER
T)
[
2
3
]
-
6
3
.
5
7
7
1
.
9
8
6
7
.
5
2
TO
P
[
2
5
]
7
0
.
9
1
6
9
.
8
7
6
5
.
6
8
6
7
.
3
5
A
S
TE
-
R
L
[
1
7
]
7
2
.
1
1
6
7
.
2
1
6
9
.
6
9
6
8
.
4
1
P
r
o
p
o
se
d
M
o
d
e
l
76
74
76
73
4.
CO
NCLU
SI
O
N
T
h
is
p
ap
er
in
tr
o
d
u
ce
d
a
h
y
b
r
id
m
eth
o
d
f
o
r
jo
in
t
e
x
tr
ac
tio
n
o
f
asp
ec
t
an
d
o
p
in
io
n
,
co
u
p
led
with
s
en
tim
en
t
class
if
icat
io
n
u
s
in
g
a
p
ip
elin
e
ap
p
r
o
ac
h
.
L
ev
er
a
g
in
g
a
T
r
an
s
f
o
r
m
er
-
b
ased
Mu
lti
-
Hea
d
Atten
tio
n
Netwo
r
k
,
th
e
p
r
o
p
o
s
ed
m
o
d
e
l
d
em
o
n
s
tr
ated
p
r
o
m
is
in
g
r
esu
lts
in
ex
tr
ac
tin
g
n
u
an
ce
d
r
el
atio
n
s
h
ip
s
b
etwe
en
ter
m
s
an
d
s
en
ten
ce
s
.
T
h
e
p
ip
elin
e
ap
p
r
o
ac
h
allo
wed
f
o
r
th
e
s
im
u
ltan
eo
u
s
ex
tr
ac
tio
n
o
f
asp
ec
t
an
d
o
p
in
io
n
,
as
well
as
s
en
tim
en
t
class
if
icati
o
n
,
ad
d
r
ess
in
g
th
e
ch
allen
g
e
s
p
o
s
ed
b
y
Asp
ec
t
-
Sen
tim
en
t
tr
ip
let
ex
tr
ac
tio
n
(
ASTE
)
.
T
h
e
ex
p
er
im
en
tal
ev
alu
atio
n
s
o
n
th
e
Sem
E
v
al
2
0
1
6
r
estau
r
an
t
d
ataset
s
h
o
wca
s
ed
th
e
ef
f
ec
tiv
e
n
ess
o
f
th
e
p
r
o
p
o
s
ed
m
et
h
o
d
,
ac
h
i
ev
in
g
h
ig
h
ac
cu
r
ac
y
d
u
r
i
n
g
tr
ain
in
g
.
Ho
wev
er
,
ch
allen
g
es
em
er
g
ed
d
u
r
in
g
th
e
v
alid
atio
n
an
d
test
in
g
p
h
ases
,
p
ar
ticu
lar
ly
in
h
an
d
lin
g
im
b
al
an
ce
d
d
ata
d
is
tr
ib
u
tio
n
ac
r
o
s
s
s
en
tim
en
t
class
es
.
T
h
e
in
s
ig
h
ts
g
ain
ed
f
r
o
m
t
h
is
s
tu
d
y
n
o
t
o
n
l
y
co
n
tr
i
b
u
te
to
th
e
ad
v
an
ce
m
en
t
o
f
s
en
tim
en
t
an
aly
s
is
m
eth
o
d
o
l
o
g
ies
b
u
t
also
u
n
d
er
s
co
r
e
th
e
im
p
o
r
tan
ce
o
f
ad
d
r
ess
in
g
d
ata
im
b
alan
ce
s
in
r
e
al
-
wo
r
ld
s
ce
n
ar
io
s
.
Fu
r
th
er
r
esear
ch
c
o
u
ld
d
elv
e
in
to
th
e
r
ef
i
n
em
en
t
o
f
th
e
p
r
o
p
o
s
ed
h
y
b
r
id
m
o
d
el,
ex
p
l
o
r
in
g
v
ar
iatio
n
s
in
n
etwo
r
k
ar
c
h
itectu
r
es,
h
y
p
er
p
ar
am
eter
tu
n
in
g
,
a
n
d
alter
n
ati
v
e
d
ee
p
lear
n
in
g
f
r
am
ew
o
r
k
s
.
I
n
v
esti
g
a
tin
g
th
e
ap
p
licatio
n
o
f
tr
an
s
f
er
lear
n
i
n
g
tec
h
n
iq
u
es,
s
u
ch
as
lev
er
ag
in
g
p
r
e
-
tr
ai
n
ed
lan
g
u
a
g
e
m
o
d
els,
co
u
ld
also
co
n
tr
ib
u
te
to
im
p
r
o
v
ed
p
er
f
o
r
m
an
ce
o
n
s
en
tim
en
t a
n
aly
s
is
task
s
.
ACK
NO
WL
E
DG
E
M
E
NT
S
W
e
ex
p
r
ess
o
u
r
s
in
ce
r
e
ap
p
r
ec
iatio
n
to
th
e
Dep
ar
tm
en
t
o
f
C
o
m
p
u
ter
Scien
ce
an
d
E
n
g
in
ee
r
in
g
,
Ko
lh
ap
u
r
I
n
s
titu
te
o
f
T
ec
h
n
o
lo
g
y
’
s
C
o
lleg
e
o
f
E
n
g
in
ee
r
in
g
(
E
m
p
o
wer
e
d
Au
to
n
o
m
o
u
s
)
,
Ko
lh
a
p
u
r
,
Ma
h
ar
ash
tr
a,
I
n
d
ia,
f
o
r
en
c
o
u
r
ag
in
g
f
ac
u
lty
m
em
b
e
r
s
to
s
u
b
m
it
an
d
p
r
esen
t
r
esear
ch
wo
r
k
th
r
o
u
g
h
t
h
e
I
n
-
Ho
u
s
e
R
ev
iew,
as we
ll a
s
f
o
r
p
r
o
v
id
i
n
g
f
i
n
an
cial
s
u
p
p
o
r
t.
F
UNDING
I
NF
O
R
M
A
T
I
O
N
Au
th
o
r
s
s
tate
n
o
f
u
n
d
in
g
in
v
o
lv
ed
.
AUTHO
R
CO
NT
RI
B
UT
I
O
NS ST
A
T
E
M
E
N
T
T
h
is
jo
u
r
n
al
u
s
es
th
e
C
o
n
tr
ib
u
to
r
R
o
les
T
ax
o
n
o
m
y
(
C
R
ed
iT
)
to
r
ec
o
g
n
ize
in
d
iv
id
u
al
au
th
o
r
co
n
tr
ib
u
tio
n
s
,
r
ed
u
ce
au
th
o
r
s
h
ip
d
is
p
u
tes,
an
d
f
ac
ilit
ate
co
llab
o
r
atio
n
.
Na
m
e
o
f
Aut
ho
r
C
M
So
Va
Fo
I
R
D
O
E
Vi
Su
P
Fu
Ab
h
in
an
d
a
n
Sh
ir
ah
atti
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
R
am
esh
Me
d
ar
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
Vijay
R
ajp
u
r
o
h
it
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
San
jeev
Kau
lg
u
d
✓
✓
✓
✓
✓
✓
✓
✓
Mr
u
ty
u
n
jay
a
Ma
th
ad
Sh
iv
am
u
r
th
aiah
✓
✓
✓
✓
✓
✓
✓
✓
C
:
C
o
n
c
e
p
t
u
a
l
i
z
a
t
i
o
n
M
:
M
e
t
h
o
d
o
l
o
g
y
So
:
So
f
t
w
a
r
e
Va
:
Va
l
i
d
a
t
i
o
n
Fo
:
Fo
r
mal
a
n
a
l
y
s
i
s
I
:
I
n
v
e
s
t
i
g
a
t
i
o
n
R
:
R
e
so
u
r
c
e
s
D
:
D
a
t
a
C
u
r
a
t
i
o
n
O
:
W
r
i
t
i
n
g
-
O
r
i
g
i
n
a
l
D
r
a
f
t
E
:
W
r
i
t
i
n
g
-
R
e
v
i
e
w
&
E
d
i
t
i
n
g
Vi
:
Vi
su
a
l
i
z
a
t
i
o
n
Su
:
Su
p
e
r
v
i
s
i
o
n
P
:
P
r
o
j
e
c
t
a
d
mi
n
i
st
r
a
t
i
o
n
Fu
:
Fu
n
d
i
n
g
a
c
q
u
i
si
t
i
o
n
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
7
7
6
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
,
Vo
l.
15
,
No
.
2
,
J
u
n
e
20
26
:
7
6
9
-
7
7
7
776
CO
NF
L
I
C
T
O
F
I
N
T
E
R
E
S
T
ST
A
T
E
M
E
NT
Au
th
o
r
s
s
tate
n
o
co
n
f
lict o
f
in
t
er
est
DATA AV
AI
L
AB
I
L
I
T
Y
Data
av
ailab
ilit
y
is
n
o
t
ap
p
li
ca
b
le
to
th
is
p
ap
er
as
n
o
n
e
w
d
ata
wer
e
cr
ea
ted
o
r
an
aly
ze
d
in
th
is
s
tu
d
y
.
RE
F
E
R
E
NC
E
S
[
1
]
S
.
C
h
e
n
,
Y
.
W
a
n
g
,
J.
Li
u
,
a
n
d
Y
.
W
a
n
g
,
“
B
i
d
i
r
e
c
t
i
o
n
a
l
m
a
c
h
i
n
e
r
e
a
d
i
n
g
c
o
m
p
r
e
h
e
n
si
o
n
f
o
r
a
s
p
e
c
t
s
e
n
t
i
m
e
n
t
t
r
i
p
l
e
t
e
x
t
r
a
c
t
i
o
n
,
”
Pro
c
e
e
d
i
n
g
s
o
f
t
h
e
A
AAI
C
o
n
f
e
r
e
n
c
e
o
n
Art
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
,
v
o
l
.
3
5
,
n
o
.
1
4
,
p
p
.
1
2
6
6
6
–
1
2
6
7
4
,
M
a
y
2
0
2
1
,
d
o
i
:
1
0
.
1
6
0
9
/
a
a
a
i
.
v
3
5
i
1
4
.
1
7
5
0
0
.
[
2
]
H
.
D
a
i
a
n
d
Y
.
S
o
n
g
,
“
N
e
u
r
a
l
a
sp
e
c
t
a
n
d
o
p
i
n
i
o
n
t
e
r
m
e
x
t
r
a
c
t
i
o
n
w
i
t
h
m
i
n
e
d
r
u
l
e
s
a
s
w
e
a
k
s
u
p
e
r
v
i
si
o
n
,
”
i
n
P
ro
c
e
e
d
i
n
g
s
o
f
t
h
e
5
7
t
h
A
n
n
u
a
l
Me
e
t
i
n
g
o
f
t
h
e
Ass
o
c
i
a
t
i
o
n
f
o
r
C
o
m
p
u
t
a
t
i
o
n
a
l
L
i
n
g
u
i
s
t
i
c
s
,
2
0
1
9
,
p
p
.
5
2
6
8
–
5
2
7
7
,
d
o
i
:
1
0
.
1
8
6
5
3
/
v
1
/
p
1
9
-
1
5
2
0
.
[
3
]
U
.
K
a
m
a
t
h
,
K
.
L.
G
r
a
h
a
m,
a
n
d
W
.
E
mara,
“
B
i
d
i
r
e
c
t
i
o
n
a
l
e
n
c
o
d
e
r
r
e
p
r
e
s
e
n
t
a
t
i
o
n
s
f
r
o
m
t
r
a
n
sf
o
r
m
e
r
s
(
B
ER
T)
,
”
i
n
T
r
a
n
sf
o
rm
e
rs
f
o
r
M
a
c
h
i
n
e
L
e
a
rn
i
n
g
,
C
h
a
p
m
a
n
a
n
d
H
a
l
l
/
C
R
C
,
2
0
2
2
,
p
p
.
4
3
–
7
0
.
[
4
]
L.
H
u
a
n
g
e
t
a
l
.
,
“
F
i
r
st
t
a
r
g
e
t
a
n
d
o
p
i
n
i
o
n
t
h
e
n
p
o
l
a
r
i
t
y
:
E
n
h
a
n
c
i
n
g
t
a
r
g
e
t
-
o
p
i
n
i
o
n
c
o
r
r
e
l
a
t
i
o
n
f
o
r
a
s
p
e
c
t
sen
t
i
m
e
n
t
t
r
i
p
l
e
t
e
x
t
r
a
c
t
i
o
n
,
”
a
r
Xi
v
p
r
e
p
ri
n
t
a
r
Xi
v
:
2
1
0
2
.
0
8
5
4
9
,
2
0
2
1
.
[
5
]
X
.
Li
,
L.
B
i
n
g
,
P
.
Li
,
a
n
d
W
.
La
m,
“
A
u
n
i
f
i
e
d
mo
d
e
l
f
o
r
o
p
i
n
i
o
n
t
a
r
g
e
t
e
x
t
r
a
c
t
i
o
n
a
n
d
t
a
r
g
e
t
s
e
n
t
i
m
e
n
t
p
r
e
d
i
c
t
i
o
n
,
”
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
AA
AI
C
o
n
f
e
re
n
c
e
o
n
Art
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
,
v
o
l
.
3
3
,
n
o
.
0
1
,
p
p
.
6
7
1
4
–
6
7
2
1
,
Ju
l
.
2
0
1
9
,
d
o
i
:
1
0
.
1
6
0
9
/
a
a
a
i
.
v
3
3
i
0
1
.
3
3
0
1
6
7
1
4
.
[
6
]
X
.
Li
e
t
a
l
.
,
“
En
t
i
t
y
-
r
e
l
a
t
i
o
n
e
x
t
r
a
c
t
i
o
n
a
s
mu
l
t
i
-
t
u
r
n
q
u
e
st
i
o
n
a
n
sw
e
r
i
n
g
,
”
i
n
Pro
c
e
e
d
i
n
g
s
o
f
t
h
e
5
7
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
,
2
0
1
9
,
p
p
.
1
3
4
0
–
1
3
5
0
,
d
o
i
:
1
0
.
1
8
6
5
3
/
v
1
/
p
1
9
-
1
1
2
9
.
[
7
]
H
.
P
e
n
g
,
L.
X
u
,
L
.
B
i
n
g
,
F
.
H
u
a
n
g
,
W
.
L
u
,
a
n
d
L.
S
i
,
“
K
n
o
w
i
n
g
w
h
a
t
,
h
o
w
a
n
d
w
h
y
:
a
n
e
a
r
c
o
m
p
l
e
t
e
s
o
l
u
t
i
o
n
f
o
r
a
s
p
e
c
t
-
b
a
se
d
sen
t
i
m
e
n
t
a
n
a
l
y
si
s
,
”
Pro
c
e
e
d
i
n
g
s
o
f
t
h
e
AAAI
C
o
n
f
e
r
e
n
c
e
o
n
Ar
t
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
,
A
p
r
.
2
0
2
0
,
d
o
i
:
1
0
.
1
6
0
9
/
a
a
a
i
.
v
3
4
i
0
5
.
6
3
8
3
.
[
8
]
S
.
P
o
r
i
a
,
D
.
H
a
z
a
r
i
k
a
,
N
.
M
a
j
u
m
d
e
r
,
a
n
d
R
.
M
i
h
a
l
c
e
a
,
“
B
e
n
e
a
t
h
t
h
e
t
i
p
o
f
t
h
e
i
c
e
b
e
r
g
:
c
u
r
r
e
n
t
c
h
a
l
l
e
n
g
e
s
a
n
d
n
e
w
d
i
r
e
c
t
i
o
n
s
i
n
sen
t
i
m
e
n
t
a
n
a
l
y
si
s
r
e
se
a
r
c
h
,
”
I
EE
E
T
ra
n
s
a
c
t
i
o
n
s
o
n
Af
f
e
c
t
i
v
e
C
o
m
p
u
t
i
n
g
,
v
o
l
.
1
4
,
n
o
.
1
,
p
p
.
1
0
8
–
1
3
2
,
Ja
n
.
2
0
2
3
,
d
o
i
:
1
0
.
1
1
0
9
/
t
a
f
f
c
.
2
0
2
0
.
3
0
3
8
1
6
7
.
[
9
]
W
.
W
a
n
g
,
S
.
J
.
P
a
n
,
D
.
D
a
h
l
m
e
i
e
r
,
a
n
d
X
.
X
i
a
o
,
“
C
o
u
p
l
e
d
m
u
l
t
i
-
l
a
y
e
r
a
t
t
e
n
t
i
o
n
s
f
o
r
c
o
-
e
x
t
r
a
c
t
i
o
n
o
f
a
s
p
e
c
t
a
n
d
o
p
i
n
i
o
n
t
e
r
ms,
”
Pro
c
e
e
d
i
n
g
s
o
f
t
h
e
AAAI
C
o
n
f
e
r
e
n
c
e
o
n
Art
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
,
v
o
l
.
3
1
,
n
o
.
1
,
F
e
b
.
2
0
1
7
,
d
o
i
:
1
0
.
1
6
0
9
/
a
a
a
i
.
v
3
1
i
1
.
1
0
9
7
4
.
[
1
0
]
Z.
W
u
,
C
.
Y
i
n
g
,
F
.
Z
h
a
o
,
Z.
F
a
n
,
X
.
D
a
i
,
a
n
d
R
.
X
i
a
,
“
G
r
i
d
t
a
g
g
i
n
g
sc
h
e
me
f
o
r
a
sp
e
c
t
-
o
r
i
e
n
t
e
d
f
i
n
e
-
g
r
a
i
n
e
d
o
p
i
n
i
o
n
e
x
t
r
a
c
t
i
o
n
,
”
i
n
Fi
n
d
i
n
g
s
o
f
t
h
e
A
sso
c
i
a
t
i
o
n
f
o
r
C
o
m
p
u
t
a
t
i
o
n
a
l
L
i
n
g
u
i
s
t
i
c
s:
E
MN
L
P
2
0
2
0
,
2
0
2
0
,
p
p
.
2
5
7
6
–
2
5
8
5
,
d
o
i
:
1
0
.
1
8
6
5
3
/
v
1
/
2
0
2
0
.
f
i
n
d
i
n
g
s
-
e
mn
l
p
.
2
3
4
.
[
1
1
]
L.
X
u
,
H
.
Li
,
W
.
L
u
,
a
n
d
L.
B
i
n
g
,
“
P
o
si
t
i
o
n
-
a
w
a
r
e
t
a
g
g
i
n
g
f
o
r
a
sp
e
c
t
se
n
t
i
me
n
t
t
r
i
p
l
e
t
e
x
t
r
a
c
t
i
o
n
,
”
i
n
P
ro
c
e
e
d
i
n
g
s
o
f
t
h
e
2
0
2
0
C
o
n
f
e
re
n
c
e
o
n
Em
p
i
r
i
c
a
l
M
e
t
h
o
d
s
i
n
N
a
t
u
ra
l
L
a
n
g
u
a
g
e
Pro
c
e
ssi
n
g
(
E
MN
L
P)
,
2
0
2
0
,
p
p
.
2
3
3
9
–
2
3
4
9
,
d
o
i
:
1
0
.
1
8
6
5
3
/
v
1
/
2
0
2
0
.
e
m
n
l
p
-
m
a
i
n
.
1
8
3.
[
1
2
]
C
.
Zh
a
n
g
,
Q
.
Li
,
D
.
S
o
n
g
,
a
n
d
B
.
W
a
n
g
,
“
A
mu
l
t
i
-
t
a
s
k
l
e
a
r
n
i
n
g
f
r
a
me
w
o
r
k
f
o
r
o
p
i
n
i
o
n
t
r
i
p
l
e
t
e
x
t
r
a
c
t
i
o
n
,
”
i
n
Fi
n
d
i
n
g
s
o
f
t
h
e
Asso
c
i
a
t
i
o
n
f
o
r
C
o
m
p
u
t
a
t
i
o
n
a
l
L
i
n
g
u
i
st
i
c
s:
EM
N
L
P
2
0
2
0
,
2
0
2
0
,
p
p
.
8
1
9
–
8
2
8
,
d
o
i
:
1
0
.
1
8
6
5
3
/
v
1
/
2
0
2
0
.
f
i
n
d
i
n
g
s
-
e
mn
l
p
.
7
2
.
[
1
3
]
X
.
L
i
,
L.
B
i
n
g
,
P
.
L
i
,
W
.
L
a
m,
a
n
d
Z.
Y
a
n
g
,
“
A
sp
e
c
t
t
e
r
m
e
x
t
r
a
c
t
i
o
n
w
i
t
h
h
i
st
o
r
y
a
t
t
e
n
t
i
o
n
a
n
d
s
e
l
e
c
t
i
v
e
t
r
a
n
sf
o
r
ma
t
i
o
n
,
”
i
n
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
T
w
e
n
t
y
-
S
e
v
e
n
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
r
e
n
c
e
o
n
A
rt
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
,
J
u
l
.
2
0
1
8
,
p
p
.
4
1
9
4
–
4
2
0
0
,
d
o
i
:
1
0
.
2
4
9
6
3
/
i
j
c
a
i
.
2
0
1
8
/
5
8
3
.
[
1
4
]
X
.
L
i
a
n
d
W
.
L
a
m
,
“
D
e
e
p
m
u
l
t
i
-
t
a
s
k
l
e
a
r
n
i
n
g
f
o
r
a
s
p
e
c
t
t
e
r
m
e
x
t
r
a
c
t
i
o
n
w
i
t
h
m
e
m
o
r
y
i
n
t
e
r
a
c
t
i
o
n
,
”
2
0
1
7
,
d
o
i
:
1
0
.
1
8
6
5
3
/
v
1
/
d
1
7
-
1
3
1
0
.
[
1
5
]
H
.
Lu
o
,
T.
Li
,
B
.
L
i
u
,
a
n
d
J.
Zh
a
n
g
,
“
D
O
ER
:
d
u
a
l
c
r
o
ss
-
sh
a
r
e
d
R
N
N
f
o
r
a
s
p
e
c
t
t
e
r
m
-
p
o
l
a
r
i
t
y
c
o
-
e
x
t
r
a
c
t
i
o
n
,
”
i
n
P
ro
c
e
e
d
i
n
g
s
o
f
t
h
e
5
7
t
h
A
n
n
u
a
l
Me
e
t
i
n
g
o
f
t
h
e
Ass
o
c
i
a
t
i
o
n
f
o
r
C
o
m
p
u
t
a
t
i
o
n
a
l
L
i
n
g
u
i
s
t
i
c
s
,
2
0
1
9
,
p
p
.
5
9
1
–
6
0
1
,
d
o
i
:
1
0
.
1
8
6
5
3
/
v
1
/
p
1
9
-
1
0
5
6
.
[
1
6
]
Y
.
Y
i
n
,
F
.
W
e
i
,
L.
D
o
n
g
,
K
.
X
u
,
M
.
Zh
a
n
g
,
a
n
d
M
.
Z
h
o
u
.
,
“
U
n
s
u
p
e
r
v
i
se
d
w
o
r
d
a
n
d
d
e
p
e
n
d
e
n
c
y
p
a
t
h
e
mb
e
d
d
i
n
g
s
f
o
r
a
sp
e
c
t
t
e
r
m
e
x
t
r
a
c
t
i
o
n
,
”
a
r
Xi
v
p
r
e
p
ri
n
t
a
r
Xi
v
:
1
6
0
5
.
0
7
8
4
3
,
2
0
1
6
.
[
1
7
]
S
.
Y
u
B
a
i
Ji
a
n
,
T
.
N
a
y
a
k
,
N
.
M
a
j
u
md
e
r
,
a
n
d
S
.
P
o
r
i
a
,
“
A
s
p
e
c
t
se
n
t
i
men
t
t
r
i
p
l
e
t
e
x
t
r
a
c
t
i
o
n
u
s
i
n
g
r
e
i
n
f
o
r
c
e
me
n
t
l
e
a
r
n
i
n
g
,
”
i
n
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
3
0
t
h
A
C
M
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
I
n
f
o
r
m
a
t
i
o
n
&
a
m
p
;
K
n
o
w
l
e
d
g
e
M
a
n
a
g
e
m
e
n
t
,
O
c
t
.
2
0
2
1
,
p
p
.
3
6
0
3
–
3
6
0
7
,
d
o
i
:
1
0
.
1
1
4
5
/
3
4
5
9
6
3
7
.
3
4
8
2
0
5
8
.
[
1
8
]
M
.
P
o
n
t
i
k
i
,
D
.
G
a
l
a
n
i
s
,
J.
P
a
v
l
o
p
o
u
l
o
s,
H
.
P
a
p
a
g
e
o
r
g
i
o
u
,
I
.
A
n
d
r
o
u
t
s
o
p
o
u
l
o
s,
a
n
d
S
.
M
a
n
a
n
d
h
a
r
,
“
S
e
mE
v
a
l
-
2
0
1
4
t
a
s
k
4
:
a
sp
e
c
t
b
a
s
e
d
s
e
n
t
i
m
e
n
t
a
n
a
l
y
si
s,
”
i
n
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
Wo
r
k
s
h
o
p
o
n
S
e
m
a
n
t
i
c
E
v
a
l
u
a
t
i
o
n
(
S
e
m
E
v
a
l
2
0
1
4
)
,
2
0
1
4
,
p
p
.
2
7
–
3
5
,
d
o
i
:
1
0
.
3
1
1
5
/
v
1
/
s
1
4
-
2
0
0
4
.
[
1
9
]
M
.
P
o
n
t
i
k
i
e
t
a
l
.
,
“
S
e
m
e
v
a
l
-
2
0
1
6
t
a
sk
5
:
A
sp
e
c
t
b
a
se
d
s
e
n
t
i
me
n
t
a
n
a
l
y
s
i
s,”
I
n
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
1
0
t
h
i
n
t
e
r
n
a
t
i
o
n
a
l
w
o
r
k
sh
o
p
o
n
sema
n
t
i
c
e
v
a
l
u
a
t
i
o
n
(
S
e
m
E
v
a
l
-
2
0
1
6
)
,
p
p
.
1
9
–
3
0
,
2
0
1
6
.
[
2
0
]
T.
H
e
r
c
i
g
,
“
U
W
B
a
t
S
e
mE
v
a
l
-
2
0
1
8
Ta
sk
3
:
i
r
o
n
y
d
e
t
e
c
t
i
o
n
i
n
e
n
g
l
i
s
h
t
w
e
e
t
s
,
”
i
n
Pr
o
c
e
e
d
i
n
g
s
o
f
T
h
e
1
2
t
h
I
n
t
e
r
n
a
t
i
o
n
a
l
Wo
r
k
sh
o
p
o
n
S
e
m
a
n
t
i
c
Ev
a
l
u
a
t
i
o
n
,
2
0
1
8
,
p
p
.
5
2
0
–
5
2
4
,
d
o
i
:
1
0
.
1
8
6
5
3
/
v
1
/
s
1
8
-
10
8
4
.
[
2
1
]
S
.
C
h
a
i
t
u
s
a
n
e
y
a
n
d
A
.
Y
o
k
o
y
a
m
a
,
“
A
n
a
p
p
r
o
p
r
i
a
t
e
d
i
s
t
r
i
b
u
t
e
d
g
e
n
e
r
a
t
i
o
n
si
z
i
n
g
c
o
n
s
i
d
e
r
i
n
g
r
e
c
l
o
s
e
r
-
f
u
se
c
o
o
r
d
i
n
a
t
i
o
n
,
”
i
n
2
0
0
5
I
EEE/
P
ES
T
r
a
n
sm
i
ssi
o
n
&
a
m
p
;
a
m
p
;
D
i
s
t
r
i
b
u
t
i
o
n
C
o
n
f
e
re
n
c
e
&
a
m
p
;
a
m
p
;
E
x
p
o
s
i
t
i
o
n
:
Asi
a
a
n
d
P
a
c
i
f
i
c
,
p
p
.
1
–
6
,
d
o
i
:
1
0
.
1
1
0
9
/
t
d
c
.
2
0
0
5
.
1
5
4
6
8
3
8
.
[
2
2
]
Z.
M
.
Z
.
M
a
d
h
o
u
sh
i
,
A
.
R
.
H
a
m
d
a
n
,
S
.
Za
i
n
u
d
i
n
,
“
A
sp
e
c
t
-
b
a
se
d
se
n
t
i
men
t
a
n
a
l
y
si
s
m
e
t
h
o
d
s
i
n
r
e
c
e
n
t
y
e
a
r
s
,
”
As
i
a
-
Pa
c
i
f
i
c
J
o
u
rn
a
l
o
f
I
n
f
o
rm
a
t
i
o
n
T
e
c
h
n
o
l
o
g
y
&
a
m
p
;
Mu
l
t
i
m
e
d
i
a
,
v
o
l
.
8
,
n
o
.
1
,
7
9
–
96
,
2
0
1
9
,
d
o
i
:
1
0
.
1
7
5
7
6
/
a
p
j
i
t
m
-
2
0
1
9
-
0
8
0
1
-
0
7
.
[
2
3
]
K
.
S
c
a
r
i
a
,
H
.
G
u
p
t
a
,
S
.
G
o
y
a
l
,
S
.
S
a
w
a
n
t
,
S
.
M
i
s
h
r
a
,
a
n
d
C
.
B
a
r
a
l
,
“
I
n
st
r
u
c
t
A
B
S
A
:
i
n
st
r
u
c
t
i
o
n
l
e
a
r
n
i
n
g
f
o
r
a
s
p
e
c
t
b
a
se
d
se
n
t
i
m
e
n
t
a
n
a
l
y
si
s
,
”
i
n
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
2
0
2
4
C
o
n
f
e
re
n
c
e
o
f
t
h
e
N
o
rt
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
(
Vo
l
u
m
e
2
:
S
h
o
r
t
P
a
p
e
rs)
,
2
0
2
4
,
p
p
.
7
2
0
–
7
3
6
,
d
o
i
:
1
0
.
1
8
6
5
3
/
v
1
/
2
0
2
4
.
n
a
a
c
l
-
s
h
o
r
t
.
6
3
.
[
2
4
]
J.
Z.
M
a
i
t
a
ma
,
N
.
I
d
r
i
s,
A
.
A
b
d
i
,
L.
S
h
u
i
b
,
a
n
d
R.
F
a
u
z
i
,
“
A
sy
s
t
e
m
a
t
i
c
r
e
v
i
e
w
o
n
i
m
p
l
i
c
i
t
a
n
d
e
x
p
l
i
c
i
t
a
s
p
e
c
t
e
x
t
r
a
c
t
i
o
n
i
n
sen
t
i
m
e
n
t
a
n
a
l
y
si
s
,
”
I
EEE
Ac
c
e
ss
,
v
o
l
.
8
,
p
p
.
1
9
4
1
6
6
–
1
9
4
1
9
1
,
2
0
2
0
,
d
oi
:
1
0
.
1
1
0
9
/
a
c
c
e
ss.
2
0
2
0
.
3
0
3
1
2
1
7
.
[2
5
]
A
.
O
u
d
a
l
o
v
e
t
a
l
.
,
“
N
o
v
e
l
p
r
o
t
e
c
t
i
o
n
s
y
st
e
ms f
o
r
mi
c
r
o
g
r
i
d
s,
”
2
0
0
9
.
h
t
t
p
:
/
/
w
w
w
.
mi
c
r
o
g
r
i
d
s
.
e
u
/
d
o
c
u
me
n
t
s
/
6
8
8
.
p
d
s
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
I
SS
N:
2252
-
8
7
7
6
Hyb
r
id
d
ee
p
n
e
u
r
a
l n
etw
o
r
k
mo
d
el
fo
r
a
s
p
ec
t a
n
d
o
p
in
io
n
ex
tr
a
ctio
n
w
ith
… (
A
b
h
in
a
n
d
a
n
S
h
ir
a
h
a
tti
)
777
B
I
O
G
RAP
H
I
E
S O
F
AUTH
O
RS
Dr
.
Abh
in
a
n
d
a
n
S
h
ira
h
a
tti
h
o
ld
s
a
Do
c
t
o
ra
l
d
e
g
re
e
fr
o
m
VTU
U
n
iv
e
rsit
y
,
Ka
rn
a
tak
a
In
d
ia
i
n
2
0
2
3
.
R
e
c
e
iv
e
d
B.
E.
a
n
d
M
.
Tec
h
i
n
Co
m
p
u
ter
S
c
ien
c
e
a
n
d
En
g
i
n
e
e
rig
fr
o
m
VTU
Un
i
v
e
rsit
y
,
Ka
rn
a
tak
a
I
n
d
ia
i
n
2
0
0
6
a
n
d
2
0
1
0
,
re
sp
e
c
ti
v
e
ly
.
He
is
c
u
rre
n
tl
y
a
n
a
ss
o
c
iate
p
ro
fe
ss
o
r
a
t
Co
m
p
u
ter
S
c
ien
c
e
De
p
a
rtme
n
t
in
Ko
l
h
a
p
u
r
In
stit
u
te
o
f
Tec
h
n
o
l
o
g
y
’s
C
o
ll
e
g
e
o
f
E
n
g
i
n
e
e
rin
g
,
Ko
l
h
a
p
u
r,
a
n
d
M
a
h
a
r
a
sh
tra.
Rre
se
a
rc
h
in
c
lu
d
e
s
m
a
c
h
in
e
lea
rn
in
g
d
e
e
p
lea
rn
in
g
a
n
d
n
a
t
u
ra
l
lan
g
u
a
g
e
p
ro
c
e
ss
in
g
.
He
h
a
s
p
u
b
l
ish
e
d
o
v
e
r
1
0
p
a
p
e
rs
in
in
tern
a
ti
o
n
a
l
j
o
u
r
n
a
l
s
a
n
d
c
o
n
fe
re
n
c
e
s.
F
ro
m
Ju
ly
2
0
1
2
t
o
Ju
n
e
2
0
2
4
,
a
n
d
a
lso
se
rv
e
m
a
n
y
IEE
E
c
o
n
fe
re
n
c
e
se
ss
io
n
c
h
a
irs
.
He
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
sh
irah
a
tt
i.
a
b
h
in
a
n
d
a
n
@
k
it
c
o
e
k
.
in
o
r
a
b
h
ish
ira
h
a
tt
i0
7
0
7
@g
m
a
il
.
c
o
m
.
Dr
.
Ra
m
e
sh
Meda
r
h
o
ld
s
a
Ba
c
h
e
lo
r
o
f
En
g
i
n
e
e
rin
g
(B
.
E
n
g
.
)
i
n
Co
m
p
u
ter
S
c
ien
c
e
a
n
d
En
g
in
e
e
rig
fr
o
m
S
DMCE
T
Dh
a
rwa
d
,
M
.
Tec
h
i
n
Co
m
p
u
ter
S
c
ien
c
e
a
n
d
En
g
i
n
e
e
rig
,
fro
m
G
IT
Be
lag
a
v
i,
Ka
rn
a
tak
a
,
In
d
ia.
P
h
.
D
.
in
Co
m
p
u
ter
S
c
ien
c
e
a
n
d
E
n
g
in
e
e
rig
,
fro
m
VTU
Be
lag
a
v
i
,
Ka
rn
a
tak
a
.
He
is
c
u
rre
n
tl
y
a
n
a
ss
o
c
iate
p
ro
f
e
ss
o
r
a
t
Co
m
p
u
ter
S
c
ien
c
e
(AI
a
n
d
M
L)
De
p
a
rtme
n
t
K
ES
’S
Ra
jara
m
b
a
p
u
In
sti
tu
te
o
f
Tec
h
n
o
lo
g
y
,
Isla
m
p
u
r,
M
a
h
a
ra
sh
tra.
Rre
se
a
rc
h
in
c
l
u
d
e
s
m
a
c
h
in
e
lea
rn
i
n
g
c
lo
u
d
c
o
m
p
u
t
in
g
,
a
n
d
d
a
ta
a
n
a
l
y
ti
c
s
.
He
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
m
e
d
a
r.
ra
m
e
sh
@k
it
c
o
e
k
.
in
o
r
ra
m
e
sh
m
e
d
a
r2
1
8
4
@
g
m
a
il
.
c
o
m
.
Dr
.
Vija
y
Ra
j
p
u
r
o
h
it
re
c
e
iv
e
d
th
e
P
h
.
D
.
fr
o
m
M
a
n
i
p
a
l
Un
i
v
e
rsity
,
Ka
rn
a
tak
a
In
d
ia.
He
is
a
p
ro
fe
ss
o
r
in
,
De
p
a
rtme
n
t
o
f
CS
E
a
t
G
IT,
Be
lg
a
u
m
,
Ka
rn
a
tak
a
,
In
d
ia.
Re
c
e
iv
e
d
B.
E.
i
n
Co
m
p
u
ter
S
c
ien
c
e
fro
m
KU
D
Dh
a
rwa
d
,
M
.
Tec
h
fro
m
N
.
I
.
T.
K
S
u
ra
th
k
a
l
.
His
re
se
a
rc
h
in
tere
st
in
c
l
u
d
e
ima
g
e
p
r
o
c
e
ss
in
g
,
c
lo
u
d
c
o
m
p
u
ti
n
g
,
a
n
d
d
a
ta
a
n
a
ly
ti
c
s
.
He
h
a
s
p
u
b
li
s
h
e
d
o
v
e
r
1
0
0
p
a
p
e
rs
i
n
i
n
tern
a
ti
o
n
a
l
jo
u
rn
a
ls
a
n
d
c
o
n
fe
re
n
c
e
s.
He
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
i
l
v
ij
a
y
sr2
k
@y
a
h
o
o
.
c
o
m
o
r
v
sra
jp
u
ro
h
it
@
g
it
.
e
d
u
.
Dr
.
S
a
n
je
e
v
K
a
u
l
g
u
d
is an
a
c
c
o
m
p
li
sh
e
d
a
c
a
d
e
m
ic p
ro
fe
ss
io
n
a
l
c
u
rre
n
tl
y
se
rv
in
g
a
s
a
n
As
so
c
iate
P
ro
fe
ss
o
r
a
t
D
e
p
t.
o
f
AI
&
M
L,
Ne
w
Ho
rizo
n
Co
ll
e
g
e
o
f
En
g
in
e
e
rin
g
,
Be
n
g
a
lu
r
u
.
Wi
t
h
e
x
ten
si
v
e
e
x
p
e
ri
e
n
c
e
sp
a
n
n
in
g
o
v
e
r
a
d
e
c
a
d
e
in
b
o
th
a
c
a
d
e
m
ia
a
n
d
in
d
u
str
y
.
He
h
a
s
a
stro
n
g
b
a
c
k
g
ro
u
n
d
in
so
ftwa
re
e
n
g
in
e
e
rin
g
,
b
io
i
n
fo
rm
a
ti
c
s,
a
n
d
m
a
c
h
in
e
lea
rn
in
g
.
He
h
o
l
d
s
a
P
h
.
D
.
in
Bio
i
n
fo
rm
a
ti
c
s
fro
m
REVA
Un
iv
e
rsit
y
,
wit
h
a
d
isse
rtatio
n
f
o
c
u
se
d
o
n
d
e
v
e
lo
p
in
g
e
fficie
n
t
a
l
g
o
ri
th
m
s
fo
r
p
ro
tei
n
se
q
u
e
n
c
e
a
n
a
ly
sis
a
n
d
p
ro
sta
te
c
a
n
c
e
r
d
e
tec
ti
o
n
u
sin
g
m
a
c
h
i
n
e
lea
rn
in
g
tec
h
n
i
q
u
e
s.
He
h
a
s
m
a
n
y
p
u
b
l
ica
ti
o
n
s
i
n
re
p
u
ted
I
n
tern
a
ti
o
n
a
l
Jo
u
rn
a
ls
a
n
d
Co
n
fe
re
n
c
e
s.
He
h
a
s
m
u
lt
ip
l
e
p
a
ten
ts
fo
r
h
is
in
n
o
v
a
t
iv
e
wo
rk
s.
He
c
a
n
b
e
c
o
n
tac
ted
a
t
th
e
e
m
a
il
:
sa
n
jee
v
.
k
a
u
l
g
u
d
@g
m
a
il
.
c
o
m
.
Dr
.
Mruty
u
n
j
a
y
a
M
a
t
ha
d
Sh
iv
a
m
urt
ha
ia
h
is
a
n
As
so
c
iate
P
ro
fe
ss
o
r
a
n
d
He
a
d
o
f
CS
E
(Da
ta
S
c
ien
c
e
)
a
t
R
L
Ja
lap
p
a
In
stit
u
te
o
f
Tec
h
n
o
lo
g
y
,
wit
h
1
4
y
e
a
rs
o
f
e
x
p
e
rien
c
e
in
tea
c
h
in
g
,
i
n
d
u
stry
,
a
n
d
re
se
a
rc
h
.
He
h
o
l
d
s
a
P
h
.
D.
in
C
S
E,
s
p
e
c
ializin
g
i
n
a
d
v
a
n
c
e
d
ima
g
e
p
ro
c
e
ss
in
g
a
n
d
m
a
c
h
in
e
lea
rn
in
g
f
o
r
n
o
n
-
d
e
str
u
c
ti
v
e
q
u
a
li
t
y
a
n
a
l
y
sis.
His
e
x
p
e
rti
se
in
c
lu
d
e
s
AI,
M
L
,
c
y
b
e
rse
c
u
rit
y
,
a
n
d
d
ig
it
a
l
ima
g
e
p
ro
c
e
ss
in
g
.
An
a
wa
rd
-
win
n
i
n
g
re
se
a
rc
h
e
r
a
n
d
p
a
ten
t
h
o
ld
e
r,
h
e
h
a
s
led
k
e
y
a
c
a
d
e
m
ic
in
it
iati
v
e
s,
in
c
lu
d
in
g
th
e
AICTE
ATAL
F
DP
2
0
2
4
a
n
d
2
0
2
5
.
P
a
ss
io
n
a
te
a
b
o
u
t
m
e
n
to
ri
n
g
a
n
d
in
n
o
v
a
ti
o
n
,
h
e
f
o
ste
rs
c
o
ll
a
b
o
ra
ti
o
n
b
e
twe
e
n
a
c
a
d
e
m
ia
a
n
d
in
d
u
str
y
t
o
d
ri
v
e
tec
h
n
o
l
o
g
ica
l
a
d
v
a
n
c
e
m
e
n
t
a
n
d
re
se
a
rc
h
e
x
c
e
ll
e
n
c
e
.
He
c
a
n
b
e
c
o
n
tac
ted
at
:
m
ru
ty
u
n
j
a
y
a
m
s@
g
m
a
il
.
c
o
m
.
Evaluation Warning : The document was created with Spire.PDF for Python.