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.
1
5
,
No
.
2
,
J
u
n
e
20
2
6
,
p
p
.
674
~
6
8
2
I
SS
N:
2252
-
8
7
7
6
,
DOI
:
1
0
.
1
1
5
9
1
/iji
ct
.
v
1
5
i
2
.
pp
674
-
6
8
2
674
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ict.
ia
esco
r
e.
co
m
H
a
rness
ing
NL
P
a
nd AI t
o
d
ecod
e
po
litical disco
urs
e:
s
pee
ch
pa
tt
erns,
sentim
e
nt
a
na
ly
sis
,
a
nd p
ublic per
ception
M
a
la
y
a
j
K
um
a
r,
Anuj
K
um
a
r
Sin
g
h,
So
um
it
ra
Da
s
D
e
p
a
r
t
me
n
t
o
f
C
o
mp
u
t
e
r
S
c
i
e
n
c
e
a
n
d
En
g
i
n
e
e
r
i
n
g
,
S
h
r
i
JJ
T
U
n
i
v
e
r
si
t
y
,
J
h
u
n
j
h
u
n
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
J
an
2
5
,
2
0
2
5
R
ev
is
ed
No
v
2
7
,
2
0
2
5
Acc
ep
ted
Dec
1
4
,
2
0
2
5
Us
in
g
n
a
tu
ra
l
lan
g
u
a
g
e
p
r
o
c
e
ss
in
g
(NLP
)
a
n
d
a
rti
f
icia
l
in
tel
li
g
e
n
c
e
(AI),
th
is
stu
d
y
a
n
a
ly
z
e
s
th
e
fre
q
u
e
n
c
ies
o
f
wo
r
d
s
a
n
d
p
h
ra
se
s
in
p
o
li
ti
c
a
l
lea
d
e
rs’
sp
e
e
c
h
e
s
to
trac
k
p
a
tt
e
rn
s
in
p
o
li
ti
c
a
l
d
isc
o
u
rse
.
Th
e
o
b
jec
ti
v
e
is
t
o
id
e
n
t
if
y
lan
g
u
a
g
e
p
a
tt
e
rn
s,
se
n
ti
m
e
n
ts,
a
n
d
to
p
ics
o
f
p
o
l
it
ica
l
a
d
d
re
ss
e
s
u
s
in
g
sta
te
-
of
-
th
e
-
a
rt
m
e
th
o
d
s
li
k
e
a
u
to
m
a
t
ic
tran
sc
rip
ti
o
n
(W
h
is
p
e
r),
Bi
d
ir
e
c
ti
o
n
a
l
g
a
ted
re
c
u
rre
n
t
u
n
it
(
G
RU
)
fo
r
s
e
n
ti
m
e
n
t
a
n
a
ly
sis
,
a
n
d
BERT
o
p
ic
.
Th
r
o
u
g
h
th
e
u
se
o
f
Wh
is
p
e
r’s
sta
te
-
of
-
th
e
-
a
rt
tran
sc
rip
ti
o
n
se
rv
ice
,
we
we
re
a
b
le
to
tran
sc
rib
e
t
h
e
p
o
li
t
ica
l
sp
e
e
c
h
e
s
in
to
m
a
c
h
in
e
-
re
a
d
a
b
le
tex
t,
wh
ic
h
i
n
tu
r
n
p
ro
v
id
e
s
f
o
r
o
t
h
e
r
t
y
p
e
s
o
f
a
n
a
ly
sis.
Bid
irec
ti
o
n
a
l
G
RU
c
las
sifies
se
n
ti
m
e
n
t
a
s
p
o
siti
v
e
,
n
e
g
a
ti
v
e
,
o
r
n
e
u
tral
with
t
h
e
a
im
t
o
st
u
d
y
h
o
w
p
o
li
ti
c
ian
s
u
se
se
n
ti
m
e
n
t
t
o
m
a
n
i
p
u
late
th
e
ir
li
st
e
n
e
rs.
F
u
r
th
e
rm
o
re
,
we
u
se
BER
To
p
ic
fo
r
trac
k
in
g
th
e
e
v
o
l
u
ti
o
n
o
f
rh
e
t
o
ric,
k
e
y
tre
n
d
s
u
m
m
a
risa
ti
o
n
,
a
n
d
t
o
p
ic m
in
in
g
a
n
d
a
n
a
l
y
sis.
It
il
l
u
stra
tes
h
o
w
p
o
li
ti
c
ian
s
e
m
p
l
o
y
d
isc
u
rsiv
e
stra
t
e
g
ies
a
n
d
e
p
il
in
g
u
isti
c
e
lem
e
n
ts
to
m
a
n
a
g
e
th
e
p
u
b
li
c
m
in
d
a
n
d
re
a
li
ty
.
Ac
h
i
e
v
e
m
e
n
ts
a
n
d
o
b
jec
ti
v
e
s
a
re
fra
m
e
d
with
p
o
siti
v
e
a
n
d
d
e
fe
n
si
v
e
e
m
o
ti
o
n
s
a
ime
d
a
t
th
re
a
ts
o
r
c
rit
icism
s.
Th
e
e
m
o
ti
o
n
a
l
g
ra
b
o
f
i
t
a
ll
is
stil
l
imp
o
rta
n
t
.
It
lo
c
a
tes
in
th
e
se
th
e
th
e
m
a
ti
c
c
o
h
e
re
n
c
e
a
n
d
sh
ift
in
g
se
n
ti
m
e
n
t
th
a
t
li
e
a
t
th
e
h
e
a
rt
o
f
p
o
li
t
ica
l
sto
r
y
telli
n
g
.
It
sh
o
ws
h
o
w
p
o
li
t
ica
l
c
o
m
m
u
n
ica
ti
o
n
is
e
v
o
l
v
in
g
t
o
sta
y
re
lev
a
n
t
in
th
e
d
i
g
it
a
l
m
e
d
ia
a
g
e
a
n
d
d
e
li
v
e
rs
lan
g
u
a
g
e
–
e
v
e
n
re
a
l
-
ti
m
e
lan
g
u
a
g
e
p
a
tt
e
rn
trac
k
in
g
–
v
ia
t
h
e
u
se
o
f
AI
a
n
d
b
ig
d
a
ta.
F
u
rth
e
r
stu
d
y
is
n
e
e
d
e
d
o
f
m
u
lt
imo
d
a
l
a
n
d
flex
ib
le
tec
h
n
iq
u
e
s
fo
r
a
n
a
ly
si
n
g
p
o
li
t
ica
l
d
isc
o
u
rs
e
a
c
ro
ss
lan
g
u
a
g
e
s
a
n
d
ti
m
e
p
e
rio
d
s.
K
ey
w
o
r
d
s
:
AI
B
er
to
p
ic
B
id
ir
ec
tio
n
al
GR
U
NL
P
Po
liti
ca
l d
is
co
u
r
s
e
Pu
b
lic
p
er
ce
p
tio
n
Sen
tim
en
t a
n
aly
s
is
W
h
is
p
er
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
:
Ma
lay
aj
Ku
m
ar
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
,
Sh
r
i JJ
T
Un
iv
er
s
ity
J
h
u
n
jh
u
n
u
,
R
ajasth
an
,
I
n
d
ia
E
m
ail:
k
u
m
ar
m
alay
aj@
g
m
ail.
co
m
1.
I
NT
RO
D
UCT
I
O
N
T
h
e
p
r
e
d
ictab
le
,
as
h
is
to
r
y
te
ac
h
es
u
s
,
is
th
at
p
o
liti
ca
l
s
p
ee
ch
es
ar
e
r
o
u
tin
ely
u
s
ed
as
p
ar
t
o
f
t
h
is
b
r
o
ad
e
r
p
a
n
o
p
l
y
o
f
c
o
m
m
u
n
i
ca
tio
n
s
tr
ateg
ies
th
at
m
ay
r
an
g
e
f
r
o
m
ar
ticu
latin
g
p
o
licy
p
o
s
itio
n
s
to
r
ally
i
n
g
p
u
b
lic
o
p
in
io
n
a
n
d
ch
an
g
in
g
s
o
cieta
l
attitu
d
es
[
1
]
,
[
2
]
.
A
l
o
t o
f
th
e
u
s
e
o
f
r
h
eto
r
ic
an
d
d
if
f
e
r
en
ce
s
in
lan
g
u
a
g
e
(
s
u
ch
as
to
n
e)
te
n
d
s
to
r
ev
o
lv
e
ar
o
u
n
d
ev
o
k
in
g
a
s
e
n
tim
en
t
o
r
b
u
ild
in
g
tr
u
s
t
[
2
]
,
[
3
]
.
I
n
t
h
is
in
f
o
r
m
atio
n
a
g
e
s
u
ch
as
tim
e
o
n
th
e
web
an
d
s
p
ee
ch
es
air
in
g
to
th
e
wh
o
le
s
o
ciety
,
Sp
ee
ch
Ar
ch
iv
es
’
h
is
to
r
y
a
n
d
also
em
o
tio
n
al
a
n
aly
s
is
o
f
s
p
ea
k
er
s
an
d
s
p
ee
ch
es
f
o
r
s
p
ee
ch
b
y
t
o
p
ic
ar
e
b
ec
o
m
in
g
m
o
r
e
im
p
o
r
tan
t
th
an
b
ef
o
r
e
s
o
n
ee
d
ed
[
4
]
,
[
5
]
.
R
eg
ar
d
in
g
p
u
b
lic
s
p
ee
ch
;
I
th
i
n
k
th
e
r
esear
ch
m
ay
h
as
also
in
ter
est
with
r
esp
ec
t
to
d
is
cu
s
s
iv
e
m
o
v
em
en
ts
,
esp
ec
ially
in
th
e
p
o
lar
is
ed
p
o
liti
ca
l
wo
r
ld
o
f
to
d
ay
,
r
ath
e
r
th
an
j
u
s
t
ab
o
u
t
b
r
o
ad
m
o
v
em
en
t
an
d
r
atio
n
ale
o
f
a
g
en
ts
[
6
]
–
[
8
]
.
T
h
e
i
m
p
r
o
v
e
d
a
c
c
e
s
s
t
o
c
o
m
p
u
t
a
t
i
o
n
a
l
t
o
o
l
s
h
a
s
t
r
a
n
s
f
o
r
m
e
d
t
h
e
s
t
u
d
y
o
f
p
o
l
i
t
i
c
a
l
s
p
e
e
c
h
e
s
[
9
]
–
[
1
1
]
.
Sp
ee
ch
es
n
atu
r
ally
s
h
ar
e
c
o
m
m
o
n
alities
an
d
d
is
tin
ctiv
e
f
ea
t
u
r
es,
an
d
n
atu
r
al
lan
g
u
ag
e
p
r
o
ce
s
s
in
g
(
NL
P
)
an
d
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
Ha
r
n
ess
in
g
N
LP
a
n
d
A
I
to
d
e
co
d
e
p
o
liti
ca
l d
is
co
u
r
s
e:
s
p
ee
ch
p
a
tter
n
s
,
s
en
timen
t
…
(
Ma
la
ya
j Ku
ma
r
)
675
AI
tech
n
o
lo
g
ies
h
av
e
e
v
o
lv
e
d
in
to
ess
en
tial
to
o
ls
f
o
r
d
is
s
ec
tin
g
th
e
lin
g
u
is
tic
an
d
em
o
tio
n
al
d
im
e
n
s
io
n
s
co
n
tain
ed
with
in
s
p
ee
ch
es
[
1
2
]
–
[
1
4
]
.
T
r
ad
itio
n
al
f
o
r
m
s
o
f
q
u
alitativ
e
an
aly
s
is
m
ay
m
is
s
th
e
o
p
p
o
r
tu
n
ities
av
ailab
le
th
r
o
u
g
h
th
ese
tech
n
o
lo
g
ies
wh
ich
ca
n
r
ev
ea
l
p
atter
n
s
,
s
en
tim
en
t
tr
ajec
to
r
ies
an
d
th
em
atic
s
tr
u
ctu
r
es
[
1
5
]
–
[
1
7
]
.
Fo
r
ex
a
m
p
le,
s
en
ti
m
en
t
an
aly
s
is
p
r
o
v
id
es
a
m
ea
n
s
f
o
r
g
au
g
in
g
th
e
e
m
o
tio
n
al
t
o
n
e
o
f
s
p
ee
c
h
es,
an
d
to
p
ic
m
o
d
elin
g
p
r
o
v
id
es in
s
ig
h
t in
to
k
e
y
th
em
es a
n
d
p
r
io
r
ities
o
f
p
o
liti
ca
l le
ad
er
s
[
1
8
]
–
[
2
0
]
.
Prio
r
wo
r
k
o
n
s
en
tim
en
t
an
aly
s
is
an
d
to
p
ic
m
o
d
ellin
g
o
f
p
o
li
tical
co
m
m
u
n
icatio
n
u
s
u
ally
f
o
cu
s
ed
o
n
in
d
iv
id
u
al
a
s
p
ec
ts
an
d
was
m
o
s
tly
f
o
c
u
s
s
ed
o
n
wr
itten
tex
t,
f
o
r
e
x
am
p
le,
m
an
if
esto
s
o
r
s
o
cial
m
ed
ia
m
ess
ag
es.
T
o
th
e
b
est
o
f
o
u
r
k
n
o
wled
g
e
,
n
o
wo
r
k
h
as
co
m
b
in
e
d
a
u
to
m
atic
s
p
ee
ch
tr
an
s
cr
ip
tio
n
,
d
ee
p
lear
n
in
g
-
b
ased
s
en
tim
en
t
class
if
icatio
n
,
an
d
tr
a
n
s
f
o
r
m
e
r
-
b
a
s
ed
to
p
ic
m
o
d
elin
g
i
n
to
a
c
o
h
er
en
t
p
i
p
elin
e
f
o
r
s
p
o
k
en
p
o
liti
ca
l
d
is
co
u
r
s
e
a
n
aly
s
is
.
T
h
is
r
esear
ch
ad
d
r
ess
es
th
at
g
ap
b
y
u
s
in
g
a
r
an
g
e
o
f
ad
v
an
ce
d
tr
an
s
cr
ip
tio
n
ca
p
ab
ilit
ies
o
f
w
h
is
p
er
,
th
e
B
id
ir
ec
tio
n
al
g
a
ted
r
ec
u
r
r
en
t
u
n
it
(
GR
U
)
n
et
wo
r
k
s
f
o
r
s
en
tim
en
t
d
etec
tio
n
,
an
d
B
E
R
T
o
p
ic
f
o
r
t
h
em
atic
ex
p
lo
r
atio
n
to
p
r
o
v
i
d
e
a
f
u
ll
f
r
am
ewo
r
k
th
at
ca
n
b
e
u
s
ed
to
d
em
y
s
tify
h
o
w
p
o
liti
ca
l le
ad
er
s
ar
e
i
n
f
lu
en
cin
g
lan
g
u
ag
e
d
ep
lo
y
m
en
t o
v
er
tim
e
an
d
s
p
ac
e.
T
h
is
s
tu
d
y
f
o
cu
s
es
o
n
two
m
ajo
r
g
o
als,
an
aly
s
es
s
p
ee
ch
p
atter
n
s
o
f
p
o
liti
ca
l
lead
er
s
’
s
p
ee
ch
es
th
r
o
u
g
h
NL
P a
n
d
AI
in
o
r
d
e
r
to
r
ec
o
g
n
ize
tr
e
n
d
s
an
d
an
aly
z
e
th
e
ef
f
ec
t o
f
th
eir
u
s
ag
e
o
n
p
o
liti
ca
l sp
ee
ch
,
an
d
id
en
tifie
s
lin
g
u
is
tic
m
ar
k
er
s
an
d
p
er
f
o
r
m
s
s
en
tim
en
t
an
aly
s
i
s
o
f
p
o
liti
ca
l
lead
er
’
s
s
p
ee
ch
e
s
f
o
r
u
n
d
e
r
s
tan
d
in
g
im
p
licatio
n
o
f
th
ei
r
s
p
ee
ch
o
n
p
u
b
lic
p
er
ce
p
tio
n
.
Op
e
n
AI
'
s
W
h
is
p
er
is
u
s
ed
to
p
er
f
o
r
m
s
p
ee
ch
-
to
-
tex
t
p
r
o
ce
s
s
in
g
,
an
d
a
B
id
ir
ec
tio
n
al
GR
U
m
o
d
el
f
o
r
s
en
tim
e
n
t
an
aly
s
is
an
d
B
E
R
T
o
p
ic
f
o
r
to
p
ic
m
o
d
elin
g
.
W
h
is
p
er
s
en
s
u
r
in
g
tr
an
s
cr
ib
i
n
g
p
o
liti
ca
l
s
p
ee
ch
es
r
e
g
ar
d
less
o
f
its
b
ac
k
g
r
o
u
n
d
or
th
e
n
am
e
o
f
th
e
lan
g
u
ag
e
t
o
a
co
r
r
ec
t
tr
an
s
cr
ip
tio
n
an
d
B
id
ir
ec
tio
n
al
GR
U
p
r
o
claim
s
th
e
s
en
tim
en
ts
les
s
o
n
s
th
at
r
ef
lect
s
p
ee
ch
es
em
o
tio
n
al
p
e
r
s
p
ec
tiv
es
B
E
R
T
o
p
ic
f
u
r
th
er
a
d
d
s
n
ew
d
im
e
n
s
io
n
s
o
f
an
al
y
s
is
,
s
h
o
win
g
t
h
e
th
em
atic
tr
en
d
s
ac
r
o
s
s
all
o
f
th
ese
v
ar
io
u
s
s
p
ee
ch
es,
g
iv
in
g
i
n
s
ig
h
t in
to
th
e
s
tr
ateg
ies o
f
p
o
liti
ca
l le
ad
er
s
.
T
h
e
r
esu
lts
r
ev
ea
l
th
e
n
atu
r
e
o
f
p
o
liti
ca
l
co
m
m
u
n
icatio
n
in
th
e
d
ig
ital
m
e
d
ia
er
a.
T
h
ey
r
ev
ea
l
h
o
w
p
o
liti
cian
s
cr
af
t
m
ess
ag
es
th
at
r
ef
lect
th
e
s
p
ir
it
o
f
th
eir
ti
m
es,
b
y
p
lay
in
g
o
n
ce
r
tain
e
m
o
tio
n
s
an
d
s
h
ap
in
g
p
u
b
lic
o
p
in
io
n
.
T
h
en
we
ca
n
l
o
o
k
at
s
o
m
e
m
o
r
e
eso
ter
ic
to
p
ics
th
at
o
n
ly
b
e
g
in
to
s
cr
atch
th
e
s
u
r
f
ac
e
o
f
NL
P
an
d
AI
.
T
h
is
wo
r
k
is
m
o
r
e
g
en
er
ally
o
f
a
p
iece
with
r
ec
e
n
t
s
ch
o
lar
ly
wo
r
k
u
s
in
g
AI
to
en
g
a
g
e
in
p
o
liti
ca
l
d
is
co
u
r
s
e
an
a
ly
s
is
,
in
clu
d
in
g
s
en
tim
en
t
an
aly
s
is
an
d
to
p
ic
m
o
d
elin
g
f
o
r
p
o
liti
ca
l
co
m
m
u
n
icatio
n
[
2
1
]
,
[
2
2
]
.
T
h
e
p
ap
e
r
ad
v
a
n
ce
s
th
e
f
ield
o
f
co
m
p
u
tatio
n
al
s
o
cial
s
cien
ce
th
r
o
u
g
h
th
e
d
ev
elo
p
m
en
t
o
f
m
eth
o
d
o
lo
g
y
f
o
r
s
tu
d
y
i
n
g
p
o
liti
ca
l
lan
g
u
ag
e
an
d
b
y
r
ev
ea
lin
g
d
ee
p
co
n
n
ec
tio
n
s
b
etwe
en
lan
g
u
ag
e,
s
en
t
im
en
t,
an
d
p
u
b
lic
o
p
in
io
n
.
2.
M
E
T
H
O
D
Usi
n
g
a
s
y
s
tem
atic
ap
p
r
o
ac
h
an
d
d
r
awin
g
o
n
ad
v
an
ce
d
N
L
P
an
d
AI
to
o
ls
,
th
is
s
tu
d
y
d
is
s
ec
t
s
th
e
s
p
ee
ch
p
atter
n
s
o
f
p
o
liti
ca
l
le
ad
er
s
.
T
h
e
p
i
p
eli
n
e
its
elf
co
m
p
r
is
es
th
r
ee
m
ajo
r
s
tep
s
:
s
p
ee
ch
to
tex
t,
s
en
tim
en
t
ex
tr
ac
tio
n
an
d
to
p
ic
m
o
d
eli
n
g
.
T
h
is
m
u
lti
-
tier
ed
ap
p
r
o
a
ch
was
d
esig
n
ed
to
b
e
b
o
t
h
r
ep
r
o
d
u
cib
le
an
d
in
ter
p
r
etab
le,
b
u
t
also
t
o
m
ai
n
tain
a
h
i
g
h
lev
el
o
f
ac
cu
r
ac
y
o
f
wo
r
k
e
d
p
o
liti
ca
l
r
h
eto
r
ic
.
E
v
er
y
p
ar
t
o
f
th
e
p
ip
elin
e
-
ASR
,
SA
an
d
T
M
-
was
b
u
ilt
u
s
in
g
th
e
latest
av
ailab
le
m
o
d
els
a
n
d
cr
o
s
s
-
v
alid
ated
d
atasets
to
g
en
er
ate
ac
tio
n
a
b
le
in
f
o
r
m
atio
n
ab
o
u
t
d
is
co
u
r
s
e
p
atter
n
s
.
Fo
r
a
b
etter
o
v
er
v
iew
o
f
th
e
e
n
tire
s
tu
d
y
ar
c
h
itectu
r
e,
th
e
en
v
is
ag
ed
NL
P
-
AI
wo
r
k
f
lo
w
is
illu
s
tr
ated
in
f
ig
u
r
e
.
T
h
e
p
ip
elin
e
is
in
it
iated
b
y
W
h
is
p
er
to
a
u
to
m
ati
ca
lly
tr
an
s
cr
ip
t
p
o
liti
ca
l
s
p
ee
ch
es,
B
id
ir
ec
tio
n
al
GR
U
to
class
if
y
th
e
s
en
tim
en
t
o
f
th
e
s
p
ee
ch
es,
an
d
B
E
R
T
o
p
ic
f
o
r
th
em
atizin
g
s
p
ee
ch
es.
T
h
is
s
tr
u
ctu
r
ed
p
ip
elin
e
h
el
p
s
to
p
er
f
o
r
m
r
ep
r
o
d
u
cib
le
an
d
m
o
d
u
lar
p
r
o
ce
s
s
in
g
o
f
s
p
ee
ch
d
ata,
an
d
co
m
b
i
n
es
s
tate
-
of
-
th
e
-
ar
t
d
e
e
p
l
e
a
r
n
i
n
g
a
n
d
t
o
p
i
c
e
x
t
r
a
cti
o
n
m
e
t
h
o
d
s
t
o
w
a
r
d
s
t
h
e
g
o
a
l
o
f
a
n
a
l
y
s
i
n
g
p
o
l
i
ti
c
a
l
d
i
s
c
o
u
r
s
e
s
i
n
a
h
o
l
i
s
t
i
c
wa
y
.
T
h
is
Fi
g
u
r
e
1
illu
s
tr
ates
th
e
in
ten
d
ed
t
h
r
ee
s
tag
e
p
ip
eli
n
e:
W
h
is
p
er
f
o
r
tr
an
s
cr
ip
t,
B
i
-
GR
U
f
o
r
s
en
tim
en
t
an
d
B
E
R
T
o
p
ic
f
o
r
to
p
ic
m
o
d
elin
g
.
E
ac
h
lay
er
co
n
v
er
ts
u
n
s
tr
u
ct
u
r
ed
s
p
ee
c
h
d
ata
to
s
tr
u
ctu
r
e
d
k
n
o
wled
g
e
s
p
ec
i
f
ic
to
p
o
liti
ca
l c
o
m
m
u
n
icatio
n
.
Fig
u
r
e
1
.
W
o
r
k
f
lo
w
f
o
r
AI
-
d
r
i
v
en
p
o
liti
ca
l sp
ee
ch
a
n
aly
s
is
p
ip
elin
e
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.
1
5
,
No
.
2
,
J
u
n
e
20
2
6
:
6
7
4
-
6
8
2
676
2
.
1
.
Sp
ee
ch
-
to
-
t
ex
t
c
o
nv
er
s
io
n wit
h whis
per
T
h
e
f
ir
s
t
s
tep
is
tr
an
s
cr
ib
in
g
au
d
io
clip
s
o
f
p
o
liti
ca
l
s
p
ee
ch
es
in
to
tex
t,
wh
ich
will
b
e
d
o
n
e
u
s
in
g
W
h
is
p
er
a
s
ta
te
-
of
-
th
e
-
ar
t
s
p
e
ec
h
r
ec
o
g
n
itio
n
m
o
d
el.
W
h
is
p
er
:
A
s
tate
-
of
-
th
e
-
ar
t
tr
an
s
cr
ip
tio
n
s
er
v
ice
k
n
o
wn
f
o
r
its
h
ig
h
ac
c
u
r
ac
y
a
n
d
r
esil
ien
ce
to
d
if
f
e
r
en
t
ac
ce
n
ts
,
lan
g
u
ag
es,
s
o
u
n
d
s
,
an
d
c
o
n
d
itio
n
s
(
am
b
ien
t
s
o
u
n
d
s
)
.
As
a
r
esu
lt,
th
e
tr
an
s
cr
ip
ts
ar
e
ac
cu
r
ate,
wh
ich
is
a
s
o
lid
f
o
u
n
d
atio
n
f
o
r
s
u
b
s
eq
u
e
n
t
tex
t
an
aly
s
is
.
W
h
is
p
er
’
s
ca
p
ab
ilit
ies
allo
w
d
etailed
l
in
g
u
is
tic
an
d
th
em
atic
in
v
esti
g
ati
o
n
s
b
y
tr
an
s
f
o
r
m
i
n
g
s
p
ee
ch
es in
to
s
tr
u
ctu
r
ed
tex
t
d
ata.
I
n
th
e
p
r
esen
t
wo
r
k
,
th
e
au
d
io
f
iles
wer
e
o
b
tain
e
d
f
r
o
m
p
u
b
licly
ac
ce
s
s
ib
le
p
latf
o
r
m
s
(
i.e
.
,
Yo
u
T
u
b
e)
v
ia
y
t
-
d
lp
an
d
tr
an
s
f
o
r
m
ed
to
.
wav
f
o
r
m
at.
T
h
e
f
iles
wer
e
p
r
e
-
p
r
o
ce
s
s
ed
with
lib
r
o
s
a
to
b
e
co
m
p
atib
le
with
th
e
lar
g
e
-
v
2
m
o
d
el
o
f
W
h
is
p
e
r
,
wh
ich
n
ee
d
s
to
wo
r
k
with
t
h
e
u
n
if
o
r
m
s
am
p
lin
g
r
ate
an
d
a
clea
n
wa
v
ef
o
r
m
f
o
r
m
.
T
r
an
s
cr
ip
tio
n
was
ca
r
r
ie
d
o
u
t
b
y
W
h
is
p
e
r
a
n
d
th
e
r
esu
lts
wer
e
h
ig
h
-
q
u
ality
an
d
s
en
t
en
ce
-
s
eg
m
en
ted
in
UT
F
-
8
.
T
h
ese
tim
estam
p
ed
s
tr
u
ctu
r
ed
tex
ts
wer
e
u
s
ed
to
m
ain
tain
co
n
s
is
ten
t
an
d
r
e
p
r
o
d
u
cib
le
d
o
wn
s
tr
ea
m
n
atu
r
al
lan
g
u
ag
e
p
r
o
ce
s
s
in
g
ta
s
k
s
s
u
ch
as sen
tim
en
t c
lass
if
ic
atio
n
an
d
t
o
p
ic
m
o
d
elin
g
.
2
.
2
.
Sentim
ent
a
na
ly
s
is
wit
h bid
irec
t
io
na
l
G
RU
Fo
r
s
en
tim
en
t
an
aly
s
is
in
p
o
liti
ca
l
s
p
ee
ch
es,
a
B
id
ir
ec
tio
n
al
GR
U
n
etwo
r
k
was
u
s
ed
to
a
n
aly
s
e
th
e
em
o
tio
n
al
to
n
e
em
b
e
d
d
ed
in
d
ialo
g
u
es.
T
r
ai
n
ed
o
n
a
lab
el
ed
d
atab
ase
o
f
p
o
liti
ca
l
s
p
ee
ch
tex
ts
,
th
is
d
ee
p
lear
n
in
g
m
o
d
el
was
ab
le
to
class
if
y
s
en
tim
en
ts
as
p
o
s
itiv
e,
n
eg
ativ
e,
o
r
n
eu
tr
al.
T
h
e
m
o
d
el
s
aw
th
e
co
n
tex
t
o
f
th
e
wo
r
d
s
b
ef
o
r
e
an
d
a
f
ter
th
r
o
u
g
h
its
b
id
ir
ec
tio
n
al
ar
ch
itect
u
r
e,
th
u
s
m
ak
i
n
g
it
b
etter
at
d
e
tectin
g
s
en
tim
en
ts
.
T
h
e
s
en
tim
en
t
s
co
r
es
g
en
er
ated
b
y
th
is
an
aly
s
is
r
ev
ea
led
in
f
o
r
m
atio
n
o
n
th
e
em
o
tio
n
al
tactics
p
o
liti
ca
l
lead
er
s
u
s
e
to
m
o
tiv
ate
th
eir
b
ase
an
d
in
f
lu
e
n
ce
p
u
b
lic
o
p
in
i
o
n
.
As
illu
s
tr
ated
in
Fig
u
r
e
2
,
th
e
two
-
d
ir
ec
tio
n
g
ated
cy
clic
u
n
i
t
m
o
d
el
co
n
s
is
ts
o
f
a
f
o
r
wa
r
d
cy
clic
u
n
it
an
d
a
r
ev
e
r
s
e
cy
clic
u
n
it.
S
u
ch
u
n
its
co
m
p
r
eh
en
d
r
elatio
n
s
h
ip
s
am
o
n
g
p
r
ev
io
u
s
an
d
f
u
tu
r
e
in
f
o
r
m
atio
n
,
f
ac
ilit
atin
g
a
m
o
r
e
e
x
h
au
s
tiv
e
co
m
p
r
eh
e
n
s
io
n
o
f
th
e
in
p
u
t
b
y
k
ey
in
g
in
th
e
d
ata
to
s
aid
s
e
q
u
en
ce
.
B
o
th
t
y
p
es
o
f
u
n
its
p
r
o
ce
s
s
th
e
cu
r
r
en
t
in
p
u
t
an
d
th
e
h
id
d
en
s
tate
f
r
o
m
th
e
p
r
ev
io
u
s
tim
estep
to
p
r
o
d
u
ce
th
e
cu
r
r
en
t
h
id
d
en
s
tate.
T
h
is
m
o
d
el
u
s
es
g
atin
g
m
ec
h
a
n
is
m
s
(
f
o
r
g
et,
in
p
u
t,
an
d
o
u
tp
u
t
g
ates)
to
co
n
tr
o
l
th
e
in
f
o
r
m
atio
n
f
lo
w.
T
h
ese
g
ates
em
p
lo
y
s
ig
m
o
id
f
u
n
ctio
n
s
to
d
is
ca
r
d
u
n
n
ec
ess
ar
y
in
f
o
r
m
atio
n
an
d
k
ee
p
v
alu
ab
le
o
n
es
[
2
3
]
,
[
2
4
]
.
T
h
e
f
o
r
g
et
g
ate
d
eter
m
in
es
h
o
w
m
u
ch
o
f
th
e
p
r
e
v
io
u
s
in
f
o
r
m
atio
n
s
h
o
u
ld
b
e
let
g
o
,
th
e
in
p
u
t
g
ate
tells
wh
at
n
ew
in
f
o
r
m
atio
n
s
h
o
u
ld
b
e
ad
d
e
d
to
t
h
e
ce
ll
an
d
th
e
o
u
tp
u
t
g
ate
c
o
n
tr
o
ls
wh
at
h
id
d
e
n
s
tate
in
f
o
r
m
atio
n
s
h
o
u
ld
b
e
p
ass
ed
to
th
e
n
e
x
t
l
ay
er
.
I
n
ter
esti
n
g
ly
,
th
e
b
id
ir
ec
tio
n
al
g
ated
cy
clic
ce
ll
h
as
g
o
o
d
m
em
o
r
y
,
as
it
is
ab
le
to
u
p
d
ate
h
id
d
e
n
s
tates
co
n
tin
u
o
u
s
ly
an
d
r
etain
s
in
f
o
r
m
atio
n
h
is
to
r
y
f
o
r
im
p
r
o
v
in
g
p
r
e
d
ictio
n
o
r
class
if
icatio
n
task
s
.
Al
s
o
,
its
g
atin
g
m
ec
h
an
is
m
h
elp
s
to
m
itig
ate
r
eg
is
tr
atio
n
p
r
o
b
lem
s
s
u
ch
as
g
r
ad
ien
t
v
an
is
h
in
g
o
r
e
x
p
lo
s
io
n
,
im
p
r
o
v
in
g
g
en
er
al
s
tab
ilit
y
a
n
d
tr
ain
in
g
e
f
f
icien
cy
esp
ec
ially
f
o
r
lo
n
g
s
eq
u
en
ce
d
ata.
Op
tim
al
is
an
o
th
e
r
r
ec
en
t
t
r
a
n
s
f
o
r
m
er
-
b
ased
wo
r
k
wh
ich
a
ls
o
r
ep
o
r
ts
g
o
o
d
p
er
f
o
r
m
a
n
ce
s
p
ec
if
ically
in
th
e
task
o
f
p
o
liti
ca
l
s
en
tim
en
t
class
if
icatio
n
,
an
d
is
s
h
o
wn
to
wo
r
k
p
a
r
ticu
lar
ly
well
o
n
lo
w
-
r
eso
u
r
ce
an
d
c
o
d
e
-
m
ix
ed
lan
g
u
ag
es
[
2
5
]
,
[
2
6
]
.
T
h
e
s
en
tim
en
t
class
if
ier
was
b
u
ilt
with
a
B
id
ir
ec
tio
n
al
GR
U
m
o
d
el.
T
h
e
m
o
d
el
was
in
co
r
p
o
r
ate
d
with
in
a
p
ip
elin
e
in
wh
ich
p
o
liti
ca
l
s
p
ee
c
h
es
f
ir
s
t
s
p
o
k
en
wer
e
tr
an
s
cr
i
b
ed
with
W
h
is
p
er
,
an
d
th
en
in
p
u
t
to
th
e
GR
U
f
o
r
em
o
tio
n
p
r
e
d
ictio
n
.
T
h
o
u
g
h
th
e
ex
ac
t
tr
ai
n
in
g
s
ettin
g
s
,
in
clu
d
in
g
th
e
ty
p
e
o
f
o
p
tim
izer
an
d
th
e
in
itialis
atio
n
o
f
th
e
em
b
e
d
d
in
g
a
n
d
th
e
li
k
e
ar
e
n
o
t
r
ec
o
r
d
ed
in
th
e
o
u
t
p
u
t,
th
e
f
in
al
m
o
d
el
was
ab
le
to
ac
cu
r
ately
p
r
ed
ict
th
e
s
en
tim
en
ts
a
s
b
ein
g
p
o
s
itiv
e,
n
eg
ativ
e
o
r
n
eu
tr
al.
As
a
r
esu
lt
o
f
th
e
s
en
tim
en
t
o
u
tp
u
t
as
well
as
with
th
e
to
p
ic
ch
ar
ac
ter
izatio
n
u
s
in
g
B
E
R
T
o
p
ic,
a
f
u
ll
a
n
d
i
n
ter
p
r
eted
esti
m
atio
n
o
f
th
e
tr
e
n
d
s
in
p
o
liti
ca
l sp
ee
ch
es wa
s
p
o
s
s
ib
le.
Fig
u
r
e
2
.
Ar
c
h
itectu
r
e
o
f
th
e
b
id
ir
ec
tio
n
al
g
ated
r
ec
u
r
r
e
n
t u
n
it m
o
d
el
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
Ha
r
n
ess
in
g
N
LP
a
n
d
A
I
to
d
e
co
d
e
p
o
liti
ca
l d
is
co
u
r
s
e:
s
p
ee
ch
p
a
tter
n
s
,
s
en
timen
t
…
(
Ma
la
ya
j Ku
ma
r
)
677
2
.
3
.
T
o
pic
mo
delin
g
wit
h
B
E
RT
o
p
ic
An
ad
v
an
ce
d
to
p
ic
m
o
d
elin
g
alg
o
r
ith
m
(
B
E
R
T
o
p
ic)
was
u
tili
ze
d
to
d
is
co
v
er
th
em
atic
tr
en
d
s
ac
r
o
s
s
th
e
s
p
ee
ch
es.
B
E
R
T
o
p
ic
is
a
to
p
ic
m
o
d
elin
g
f
r
am
ewo
r
k
th
at
u
tili
ze
s
m
ac
h
in
e
lear
n
in
g
a
n
d
n
atu
r
al
lan
g
u
ag
e
p
r
o
ce
s
s
in
g
tech
n
i
q
u
es
to
e
x
tr
a
ct
co
h
er
en
t
to
p
ics
alo
n
g
with
t
h
eir
d
is
tr
ib
u
tio
n
s
in
th
e
d
o
cu
m
en
ts
.
No
t
o
n
l
y
d
id
th
e
s
u
p
er
v
is
ed
ap
p
r
o
ac
h
id
en
t
if
y
th
o
s
e
k
ey
th
em
es,
b
u
t
in
d
o
in
g
s
o
it
tr
ac
k
ed
th
e
ev
o
l
u
tio
n
o
f
th
o
s
e
th
em
es
o
v
er
tim
e,
p
r
o
v
id
i
n
g
i
n
s
ig
h
t
i
n
to
r
ec
u
r
r
in
g
p
r
io
r
ities
an
d
s
tr
ateg
ies
in
p
o
liti
ca
l
r
h
eto
r
ic.
B
y
m
ak
i
n
g
t
h
em
atic
ass
o
ciatio
n
s
ex
p
licit,
B
E
R
T
o
p
ic
h
elp
ed
r
ev
ea
l
d
ee
p
er
n
ar
r
ativ
es
with
in
th
e
lead
er
s
’
u
s
e
o
f
lan
g
u
ag
e.
Mo
d
elin
g
was
p
e
r
f
o
r
m
ed
wit
h
s
en
ten
ce
-
lev
el
e
m
b
ed
d
in
g
s
,
wh
ich
wer
e
co
m
p
u
ted
with
s
en
ten
ce
-
B
E
R
T
(
all
-
Min
iLM
-
L6
-
v
2
)
.
Dim
en
s
io
n
r
ed
u
ctio
n
was
p
er
f
o
r
m
ed
o
n
t
h
ese
h
ig
h
-
d
im
e
n
s
io
n
al
em
b
ed
d
in
g
s
with
u
n
if
o
r
m
m
an
if
o
ld
ap
p
r
o
x
im
atio
n
an
d
p
r
o
jectio
n
(
UM
AP
)
,
an
d
th
e
t
o
p
ic
g
r
o
u
p
s
we
r
e
clu
s
ter
ed
with
HDBS
C
AN.
I
n
o
r
d
er
to
m
ak
e
th
e
m
o
d
els
m
o
r
e
in
ter
p
r
etab
le,
p
r
e
p
r
o
ce
s
s
in
g
s
tep
s
,
wh
ic
h
in
clu
d
ed
le
m
m
atiza
tio
n
,
to
k
en
s
tan
d
ar
d
izatio
n
,
a
n
d
s
to
p
wo
r
d
r
em
o
v
al,
wer
e
tak
e
n
.
B
E
R
T
o
p
ic’
s
d
y
n
am
ic
to
p
ic
tr
ac
k
i
n
g
also
en
ab
le
d
th
e
v
is
u
alis
atio
n
o
f
th
em
e
ch
an
g
e
s
th
r
o
u
g
h
tim
e
an
d
co
r
r
elate
d
m
ajo
r
to
p
ic
s
p
ik
es
to
ar
tef
ac
ts
o
f
p
o
liti
ca
l
ev
en
ts
,
s
u
ch
as
elec
tio
n
s
o
r
p
o
licy
an
n
o
u
n
ce
m
en
ts
.
Allo
win
g
an
in
-
d
ep
th
a
n
aly
s
is
o
f
p
o
liti
ca
l
s
p
ee
ch
es,
th
is
m
u
lti
-
s
tep
p
r
o
ce
s
s
in
co
r
p
o
r
ates
tr
an
s
cr
ip
tio
n
,
s
en
tim
en
t
a
n
aly
s
is
,
an
d
to
p
ic
m
o
d
ellin
g
to
y
ield
v
a
lu
ab
le
in
s
ig
h
ts
in
to
p
o
liti
ca
l
d
is
co
u
r
s
e.
All
s
tag
es
o
f
th
e
an
aly
s
is
m
eth
o
d
s
wer
e
em
p
ir
ically
test
ed
an
d
lo
g
ic
ally
s
eq
u
en
ce
d
an
d
ar
e
p
r
o
v
id
e
d
to
lay
th
e
g
r
o
u
n
d
wo
r
k
f
o
r
r
ep
r
o
d
u
cib
ilit
y
an
d
allo
w
o
th
er
r
esear
ch
er
s
to
r
ep
r
o
d
u
ce
th
e
p
r
o
ce
s
s
o
r
b
u
ild
u
p
o
n
it.
3.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
Ou
r
B
id
ir
ec
tio
n
al
GR
U
m
o
d
el
p
er
f
o
r
m
ed
with
h
ig
h
co
r
r
e
latio
n
b
etwe
en
co
n
tex
t
-
awa
r
e
s
eq
u
en
ce
m
o
d
elin
g
a
n
d
s
en
tim
en
t
class
if
icatio
n
with
an
o
v
er
all
test
ac
cu
r
ac
y
o
f
9
5
%.
T
h
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
o
f
th
is
s
tu
d
y
was
in
clin
ed
to
p
r
ed
ict
d
is
p
r
o
p
o
r
tio
n
ately
th
e
n
u
m
b
er
o
f
co
r
r
ec
t
class
if
icatio
n
s
o
f
n
eu
tr
al
an
d
p
o
s
itiv
e
s
en
tim
en
ts
as
ev
id
en
ce
d
b
y
th
e
co
n
f
u
s
io
n
m
atr
ix
wh
e
r
eb
y
C
las
s
2
s
co
r
ed
1
0
0
%
class
if
ic
atio
n
ac
cu
r
ac
y
an
d
C
las
s
0
well
o
v
er
9
8
%.
His
to
r
ically
,
m
is
class
if
icatio
n
s
b
as
ed
o
n
C
lass
1
(
n
eg
ativ
e)
s
en
tim
en
t
wer
e
th
e
m
o
s
t
co
m
m
o
n
,
p
o
s
s
ib
ly
b
ec
au
s
e
th
is
ty
p
e
o
f
n
eg
ativ
e
cr
iticis
m
o
r
ag
g
r
ess
iv
e
lan
g
u
ag
e
o
v
e
r
lap
p
ed
in
m
an
y
ca
s
es
with
o
n
e
an
o
th
er
.
T
h
e
m
o
d
el
’
s
p
r
ec
is
io
n
an
d
r
ec
all
ar
e
b
o
t
h
9
2
%
an
d
9
8
%,
r
esp
ec
tiv
ely
,
an
d
alo
n
g
with
its
F1
-
s
co
r
e
(
9
5
%),
s
h
o
w
s
tr
o
n
g
p
er
f
o
r
m
an
ce
o
n
em
o
tio
n
c
ateg
o
r
ies.
T
h
e
th
em
e
an
aly
s
i
s
u
s
in
g
B
E
R
T
o
p
ic
id
en
tifie
d
c
o
m
m
o
n
alities
in
t
h
e
r
h
eto
r
ic
ce
n
tr
ed
o
n
ec
o
n
o
m
ic
r
ec
o
v
er
y
,
n
atio
n
al
s
ec
u
r
ity
an
d
co
llectiv
e
id
en
tity
,
co
n
tr
ib
u
tin
g
o
r
ig
in
a
l
u
n
d
er
s
tan
d
in
g
o
f
h
o
w
p
o
li
tical
lead
er
s
tailo
r
th
eir
lan
g
u
ag
e
t
o
au
d
ien
ce
ex
p
ec
tatio
n
s
an
d
ch
an
g
in
g
s
o
c
io
p
o
liti
ca
l
lan
d
s
ca
p
es.
T
ab
le
1
p
r
o
v
id
es a
co
n
cise o
v
er
v
iew
o
f
th
e
k
ey
lay
er
s
in
th
e
m
o
d
el,
i
n
clu
d
in
g
th
eir
o
u
t
p
u
t sh
ap
es a
n
d
t
h
e
n
u
m
b
er
o
f
t
r
ain
ab
le
p
ar
a
m
eter
s
.
T
ab
le
1
.
Par
tial d
ata
La
y
e
r
(
T
y
p
e
)
O
u
t
p
u
t
s
h
a
p
e
P
a
r
a
m
#
Emb
e
d
d
i
n
g
_
2
(
Em
b
e
d
d
i
n
g
)
(
N
o
n
e
,
3
7
9
,
5
1
2
)
2
,
2
3
1
,
8
0
8
B
i
d
i
r
e
c
t
i
o
n
a
l
_
2
(
B
i
d
i
r
e
c
t
i
o
n
a
l
)
(
N
o
n
e
,
1
0
2
4
)
3
,
1
5
1
,
8
7
2
D
e
n
se
_
2
(
D
e
n
se)
(
N
o
n
e
,
3
)
3
0
7
5
Fig
u
r
e
3
d
escr
ib
es
th
e
tr
ain
in
g
ev
o
lu
tio
n
o
f
th
e
m
o
d
el
f
o
r
b
o
th
ac
cu
r
ac
y
an
d
lo
s
s
o
v
er
e
p
o
ch
s
with
ea
s
ily
d
is
tin
g
u
is
h
ab
le
m
ar
k
e
r
s
an
d
lin
e
s
ty
les.
T
h
e
ac
cu
r
ac
y
cu
r
v
e
(
s
o
lid
lin
e
with
ci
r
cu
la
r
m
ar
k
e
r
s
)
ap
p
ea
r
s
to
r
is
e
r
ap
id
ly
o
v
er
th
e
f
ir
s
t
f
e
w
ep
o
ch
s
with
s
atu
r
a
tio
n
h
ap
p
en
in
g
q
u
ick
l
y
.
T
h
is
im
p
lies
th
at
m
o
d
el
lear
n
s
th
e
u
n
d
er
ly
i
n
g
f
ea
tu
r
es
p
r
esen
t
in
th
e
tr
ain
in
g
s
et
v
er
y
f
ast.
Af
te
r
th
at,
t
h
e
cu
r
v
e
s
tay
s
f
latten
,
wh
ich
m
ea
n
s
it
h
as
co
n
v
er
g
ed
d
u
r
in
g
tr
ain
in
g
.
On
th
e
c
o
n
tr
ar
y
,
th
e
u
n
ass
u
m
i
n
g
lo
s
s
cu
r
v
e
(
i
n
d
icate
d
b
y
a
d
ash
ed
lin
e
with
s
q
u
ar
e
m
ar
k
er
s
)
q
u
ick
l
y
d
ec
r
ea
s
es
in
th
e
b
eg
i
n
n
in
g
o
f
t
h
e
tr
ain
in
g
p
h
ase
an
d
lev
els
o
f
f
o
n
ze
r
o
,
in
d
icatin
g
th
at
t
h
e
p
r
ed
ictio
n
er
r
o
r
h
as
alm
o
s
t
b
ee
n
m
in
im
ized
.
T
h
is
co
r
r
esp
o
n
d
i
n
g
d
an
ce
b
etwe
en
in
cr
ea
s
in
g
ac
cu
r
ac
y
an
d
d
e
cr
ea
s
in
g
lo
s
s
i
s
th
e
m
ar
k
o
f
a
well
-
tu
n
e
d
m
o
d
el.
T
h
e
clea
r
b
r
ea
k
-
p
o
in
ts
wh
er
e
ea
ch
cu
r
v
e
s
tar
ts
to
f
latten
m
ar
k
s
th
e
b
o
u
n
d
ar
y
b
etwe
en
u
n
d
er
-
an
d
o
v
er
-
f
itti
n
g
.
Ab
o
v
e
th
is
,
ad
d
itio
n
al
tr
ain
in
g
p
r
o
v
id
es
m
ar
g
in
al
r
etu
r
n
s
an
d
m
ay
o
v
er
f
it
if
n
o
t
co
n
tr
o
lled
.
T
h
e
B
i
-
GR
U
m
o
d
el
is
ca
p
ab
le
o
f
o
b
tain
in
g
an
d
s
u
s
tain
in
g
h
ig
h
lev
els
o
f
ac
cu
r
ac
y
with
lo
w
lo
s
s
th
r
o
u
g
h
o
u
t
th
e
g
iv
en
e
p
o
ch
s
,
in
d
icatin
g
its
ca
p
a
b
ilit
y
to
ca
p
tu
r
e
i
n
tr
icate
tem
p
o
r
al
ch
ar
ac
ter
is
tics
in
s
eq
u
en
tial
d
ata.
I
n
th
is
s
en
s
e,
its
s
tab
ilit
y
o
f
p
er
f
o
r
m
an
c
e
c
o
n
f
ir
m
s
th
e
r
o
b
u
s
tn
ess
an
d
th
e
ad
eq
u
ac
y
o
f
th
is
ap
p
r
o
ac
h
with
th
e
class
if
icatio
n
task
at
h
an
d
.
T
h
e
Bi
-
GR
U
m
o
d
el
h
as
g
r
ea
t
ab
ilit
y
to
ca
p
tu
r
e
b
id
ir
ec
tio
n
al
co
n
tex
t,
wh
ich
is
p
ar
ticu
lar
ly
ef
f
ec
tiv
e
in
s
p
ee
ch
s
en
tim
en
t
d
etec
tio
n
.
L
o
s
s
cu
r
v
e
C
o
n
v
er
g
en
ce
o
f
th
e
lo
s
s
f
u
n
ctio
n
tak
es
p
lace
an
d
th
e
c
o
n
v
in
ce
d
th
at
it
is
a
g
o
o
d
g
en
e
r
aliza
tio
n
.
T
h
e
c
o
n
f
u
s
io
n
m
atr
i
x
s
h
o
ws
v
er
y
g
o
o
d
p
r
ec
is
io
n
in
i
d
e
n
tify
in
g
n
eu
tr
ally
o
r
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.
1
5
,
No
.
2
,
J
u
n
e
20
2
6
:
6
7
4
-
6
8
2
678
p
o
s
itiv
ely
s
en
tim
en
ted
o
n
es
with
lo
w
m
is
clas
s
if
ied
in
s
tan
ce
s
,
wh
ich
ar
e
r
elativ
ely
h
ig
h
er
f
o
r
t
h
e
n
eg
ativ
e
c
lass
,
s
u
g
g
esti
n
g
s
o
m
e
lex
ical
o
v
er
lap
p
i
n
g
b
etwe
en
s
p
ee
ch
a
cts o
f
d
if
f
er
e
n
t sen
tim
en
t f
o
r
p
o
liti
ca
l sp
ee
ch
.
Fig
u
r
e
4
s
h
o
ws
m
o
d
el
co
n
f
u
s
io
n
m
atr
ix
to
v
is
u
alize
th
e
class
if
icatio
n
o
f
th
e
th
r
ee
class
es.
T
h
e
m
atr
ix
is
ar
r
an
g
e
d
ac
co
r
d
in
g
to
th
e
s
tan
d
ar
d
p
atter
n
its
d
iag
o
n
al
ce
lls
co
r
r
esp
o
n
d
to
tr
u
e
class
if
icatio
n
s
,
o
f
f
-
d
iag
o
n
al
ce
lls
to
m
is
class
if
icat
io
n
s
with
th
e
co
r
r
esp
o
n
d
in
g
ce
lls
em
p
h
asized
u
s
in
g
cr
o
s
s
h
at
ch
ed
p
atter
n
s
.
−
C
lass
0
was
we
ll
p
r
e
d
i
cte
d
,
w
i
th
4
9
t
r
u
e
-
p
o
s
it
iv
es
a
n
d
1
f
a
ls
e
-
n
e
g
at
iv
e
(
w
h
i
ch
w
as
class
if
i
e
d
as C
l
ass
2
)
.
−
C
las
s
1
was
m
o
s
t
c
o
n
f
u
s
ed
wi
th
3
9
b
ein
g
p
r
ed
icted
co
r
r
ec
tly
wh
ile
1
was
m
is
-
class
if
ied
a
s
C
lass
0
an
d
4
as C
lass
2
.
−
C
las
s
2
was
d
is
tin
g
u
is
h
e
d
p
e
r
f
ec
tly
,
with
p
r
ec
is
io
n
an
d
r
ec
a
ll
b
o
th
eq
u
al
to
1
(
ze
r
o
f
alse
p
o
s
itiv
es,
f
alse
n
eg
ativ
e
s
)
.
Fig
u
r
e
3
.
Acc
u
r
ac
y
a
n
d
lo
s
s
o
f
b
id
ir
ec
tio
n
l
GR
U
m
o
d
el
Fig
u
r
e
4
.
C
o
n
f
u
s
io
n
m
atr
i
x
o
f
s
en
tim
en
t c
lass
if
icatio
n
m
o
d
el
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
Ha
r
n
ess
in
g
N
LP
a
n
d
A
I
to
d
e
co
d
e
p
o
liti
ca
l d
is
co
u
r
s
e:
s
p
ee
ch
p
a
tter
n
s
,
s
en
timen
t
…
(
Ma
la
ya
j Ku
ma
r
)
679
T
h
e
d
a
r
k
est
ce
lls
o
n
th
e
d
iag
o
n
al
r
ep
r
esen
t
th
e
m
o
d
el’
s
h
ig
h
ef
f
ec
tiv
e
p
r
e
d
ictiv
ity
,
an
d
th
e
p
atter
n
ed
o
f
f
-
d
iag
o
n
al
ce
l
ls
id
en
tify
r
eg
io
n
s
o
f
u
n
ce
r
tai
n
ty
with
o
u
t
u
s
in
g
co
lo
r
-
ad
h
e
r
in
g
to
th
e
n
ee
d
f
o
r
th
e
r
ea
d
er
'
s
u
s
e
o
f
b
lack
an
d
wh
ite
im
ag
er
y
.
T
h
e
ab
o
v
e
r
esu
lts
ar
e
a
te
s
tim
o
n
y
to
th
e
ab
ilit
y
o
f
th
e
m
o
d
el,
to
d
is
cr
im
in
ate
well
b
etwe
en
class
es
ac
co
r
d
in
g
to
lear
n
e
d
f
ea
t
u
r
es.
T
h
e
v
is
u
al
r
esu
lts
ar
e
co
m
p
lem
en
ted
with
q
u
an
titativ
e
m
etr
ics:
th
e
m
o
d
el
o
b
tain
ed
9
5
%
ac
c
u
r
ac
y
,
9
2
%
p
r
ec
is
io
n
,
9
8
%
r
ec
all,
a
n
d
9
5
%
F1
-
s
co
r
e
i
n
a
test
s
et.
T
h
ese
r
e
s
u
lts
ev
id
en
ce
th
e
m
o
d
el
p
er
f
o
r
m
s
well
n
o
t
o
n
ly
o
n
f
itti
n
g
th
e
tr
ain
in
g
,
b
u
t
also
g
en
er
alize
s
we
ll
o
n
th
e
u
n
s
ee
n
d
ata.
T
h
e
s
lig
h
tly
lo
wer
er
r
o
r
r
ate
i
n
C
lass
1
in
d
icate
s
th
at
th
er
e
is
s
till
r
o
o
m
f
o
r
im
p
r
o
v
em
e
n
t,
h
o
wev
er
,
in
g
e
n
er
al,
th
e
m
o
d
el
s
h
o
ws
h
ig
h
d
ep
en
d
a
b
ilit
y
,
s
tr
en
g
th
,
an
d
a
p
p
licab
ilit
y
in
tex
t
-
b
ased
em
o
tio
n
class
if
icatio
n
.
Ou
r
f
in
d
in
g
s
im
p
l
y
th
at
th
e
s
u
p
er
io
r
s
en
tim
en
t
class
if
icatio
n
p
er
f
o
r
m
an
ce
with
B
id
ir
ec
tio
n
al
GR
U
d
o
es
n
o
t
c
o
m
e
at
t
h
e
co
s
t
o
f
i
m
p
air
ed
g
en
er
aliza
tio
n
o
v
er
e
m
o
tio
n
ty
p
es.
T
h
e
p
r
o
p
o
s
ed
a
p
p
r
o
ac
h
lev
er
a
g
es
th
e
ab
ilit
y
o
f
b
id
ir
ec
tio
n
al
a
r
ch
itectu
r
e
to
m
o
d
el
c
o
n
tex
t
u
al
d
ep
en
d
en
c
ies
with
o
u
t
m
ak
in
g
a
p
en
alty
i
n
tr
ain
in
g
e
f
f
icien
cy
o
r
o
v
e
r
f
itti
n
g
.
C
o
m
p
ar
ed
with
wo
r
k
o
f
Ho
s
s
ain
et
a
l.
[
2
1
]
,
wh
er
e
r
e
v
er
s
e
L
STM
m
o
d
el
was
u
s
ed
an
d
ac
h
iev
ed
f
air
class
if
icatio
n
p
er
f
o
r
m
an
ce
,
we
ar
e
ab
le
to
g
et
b
etter
F1
-
s
co
r
es
an
d
less
n
u
m
b
er
o
f
n
eu
tr
al
an
d
p
o
s
itiv
e
s
am
p
les
b
ein
g
m
is
class
if
ied
.
L
ik
ewise,
in
co
n
tr
ast
to
Al
-
Qa
b
lan
et
a
l.
[
2
2
]
,
in
wh
ich
th
ey
u
s
ed
tr
ad
itio
n
al
to
p
ic
m
o
d
elin
g
s
u
ch
as
L
DA
with
th
e
ap
p
licatio
n
o
f
B
E
R
T
o
p
ic
in
th
is
s
tu
d
y
,
h
o
wev
er
,
a
m
o
r
e
co
h
e
r
en
t
an
d
d
y
n
a
m
ic
ex
tr
ac
tio
n
o
f
to
p
ics
as
well
a
s
th
e
tr
ac
k
in
g
o
f
r
h
eto
r
ical
s
h
if
ts
was
p
o
s
s
ib
le.
T
h
ese
r
esu
lts
d
em
o
n
s
tr
ate
th
e
im
p
o
r
tan
ce
o
f
u
s
in
g
s
p
ee
ch
tr
an
s
cr
ip
tio
n
,
d
ee
p
lear
n
in
g
a
n
d
r
ec
en
t
NL
P
-
b
ased
to
p
ic
m
o
d
elin
g
in
a
n
in
teg
r
ate
d
p
ip
elin
e
f
o
r
p
o
liti
ca
l d
is
co
u
r
s
e
an
aly
s
is
.
3
.
1
.
Sp
ee
ch
pa
t
t
er
ns
a
nd
t
hem
a
t
ic
t
re
nd
s
A
s
tu
d
y
o
f
p
o
liti
ca
l
s
p
ee
ch
es,
lo
o
k
in
g
at
tr
en
d
s
in
s
p
ee
ch
w
r
itin
g
,
r
h
eto
r
ic
an
d
h
is
to
r
y
.
T
o
p
ics
s
u
ch
as
ec
o
n
o
m
ic
ad
v
an
ce
m
e
n
t,
s
o
cial
ju
s
tice
an
d
n
atio
n
al
s
ec
u
r
ity
wer
e
r
ec
u
r
r
in
g
.
T
h
is
an
aly
s
is
wa
s
co
n
d
u
cted
th
r
o
u
g
h
B
E
R
T
o
p
ic,
t
h
at
allo
wed
to
u
n
d
er
s
tan
d
n
o
t
o
n
ly
wh
en
a
n
d
th
e
co
n
tex
t
wh
e
r
e
th
o
s
e
to
p
ics
ar
e
ad
d
r
ess
ed
,
b
u
t
also
h
o
w
lea
d
e
r
s
s
elec
ted
m
ess
ag
es
ac
co
r
d
in
g
to
e
ac
h
au
d
ie
n
ce
’
s
s
p
ec
if
ic
n
ee
d
s
.
Fo
r
e
x
am
p
le,
in
ec
o
n
o
m
ic
s
tr
aits
,
th
ey
d
eli
v
er
ed
s
p
ee
ch
es
o
n
r
ec
o
v
er
y
p
lan
s
an
d
th
e
s
tate
o
f
p
u
b
lic
f
i
n
an
ce
s
.
C
o
n
v
er
s
ely
,
u
n
ity
was
to
b
e
s
p
o
k
en
o
f
m
o
r
e
in
p
er
io
d
s
o
f
r
elativ
ely
h
ig
h
p
er
ce
iv
ed
n
atio
n
al
s
ec
u
r
ity
th
r
ea
ts
;
s
o
to
o
was
d
ef
en
ce
.
T
h
e
r
h
eto
r
ical
p
lian
c
y
is
a
r
ef
lectio
n
o
f
h
o
w
m
o
b
ile
p
o
liti
ca
l
co
m
m
u
n
icatio
n
ca
n
b
e
a
n
d
h
o
w
it
tak
es
th
e
s
h
ap
es r
eq
u
ir
e
d
b
y
tu
r
n
s
in
p
u
b
lic
m
o
o
d
a
n
d
e
v
en
ts
o
n
th
e
wo
r
ld
s
tag
e.
T
h
ese
f
in
d
in
g
s
h
a
v
e
im
p
o
r
ta
n
t
im
p
licatio
n
s
f
o
r
co
m
p
u
ter
i
ze
d
p
o
lit
ical
an
aly
s
is
.
T
h
e
t
h
em
em
atic
m
o
d
el
r
e
v
ea
l
th
e
r
h
eto
r
ical
s
im
ilar
ity
o
f
is
s
u
e
f
r
am
in
g
,
s
en
tim
en
ts
m
ap
p
in
g
s
h
o
w
h
o
w
h
ap
p
in
ess
alig
n
m
en
t
co
r
r
esp
o
n
d
s
to
o
n
g
e
n
cy
.
An
o
t
h
er
f
ea
tu
r
e
p
r
esen
t
in
NL
P
-
o
b
s
er
v
ab
le
p
atter
n
s
is
h
o
w
p
o
l
itical
lan
g
u
ag
e
as
a
wh
o
le
ca
n
b
e
al
ter
ed
(
u
s
in
g
p
o
s
itiv
e
lan
g
u
ag
e
wh
e
n
th
er
e
is
g
r
o
wth
o
r
ag
g
r
ess
iv
e
lex
is
d
u
r
in
g
cr
is
is
)
to
illu
s
tr
ate
th
e
way
s
in
wh
ich
lead
er
s
attem
p
t to
em
o
tio
n
ally
r
ea
ch
th
eir
p
u
b
lics
.
3
.
2
.
Sentim
ent
dy
na
m
ics
Sen
tim
en
t
an
aly
s
is
s
h
o
wed
th
at
th
e
s
p
o
k
en
s
p
ee
ch
es
wer
e
v
er
y
d
if
f
er
e
n
t
in
em
o
tio
n
al
to
n
e.
Po
s
itiv
e
ex
p
r
ess
io
n
s
wer
e
p
o
s
itiv
ely
co
r
r
elate
d
with
p
h
r
ases
o
f
ac
h
iev
em
en
t,
lo
o
k
in
g
f
o
r
war
d
an
d
th
e
will
f
o
r
co
o
p
er
atio
n
with
o
th
e
r
s
,
p
r
es
en
tin
g
an
o
p
tim
is
tic
v
is
io
n
th
at
m
o
tiv
ates
PTT
to
tak
e
ac
ti
o
n
.
Neg
ativ
e
l
o
n
g
-
ter
m
atti
tu
d
es
r
elate
d
to
t
h
e
o
p
p
o
s
itio
n
an
d
is
s
u
es
o
f
f
ailu
r
e
s
,
an
d
f
laws
in
th
e
in
co
m
p
lete
n
atu
r
e
o
f
s
y
s
tem
.
Par
ty
lead
er
s
h
av
e
to
ca
r
r
y
b
o
t
h
em
o
tio
n
s
,
p
r
o
jecte
d
as
a
p
at
h
f
o
r
war
d
an
d
t
h
en
h
o
w
we
n
e
ed
to
m
o
v
e
o
n
t
h
at
th
er
e’
s
wo
r
k
th
at
n
ee
d
s
d
o
i
n
g
.
T
h
is
v
ar
iatio
n
b
etwe
en
s
en
ti
m
en
t
d
y
n
am
ic
also
s
u
g
g
ested
to
u
s
th
o
s
e
p
o
liti
ca
l
s
p
ee
ch
es
m
u
s
t
in
d
ee
d
b
e
d
es
ig
n
ed
to
m
o
v
e
th
e
au
d
ie
n
ce
em
o
tio
n
ally
in
ad
d
itio
n
to
estab
lis
h
in
g
s
p
ea
k
er
cr
ed
ib
ilit
y
an
d
l
o
o
k
in
g
af
ter
a
n
y
q
u
esti
o
n
s
f
r
o
m
th
e
r
esp
o
n
s
i
v
en
ess
o
f
th
e
au
d
ien
ce
.
3
.
3
.
L
ing
uis
t
ic
m
a
rk
er
s
a
nd
pu
bli
c
perc
ept
io
n
T
h
e
s
tu
d
y
id
en
tifie
d
s
p
ec
if
ic
lin
g
u
is
tic
m
ar
k
er
s
th
at
h
av
e
a
m
ea
n
in
g
f
u
l
im
p
ac
t
o
n
th
e
way
a
n
au
d
ien
ce
u
n
d
er
s
tan
d
s
in
f
o
r
m
a
tio
n
.
T
h
e
u
s
e
o
f
co
llectiv
e
n
o
u
n
s
s
u
ch
as
"we
,
"
"o
u
r
,
"
an
d
"to
g
eth
er
"
m
ad
e
it
f
ee
l
lik
e
we
ar
e
all
in
th
is
to
g
eth
er
,
a
c
o
m
m
u
n
ity
o
f
co
-
r
esp
o
n
s
ib
ilit
y
b
etwe
en
th
e
s
p
ea
k
e
r
an
d
h
is
au
d
ie
n
ce
.
I
n
co
n
tr
ast,
co
m
b
ativ
e
lan
g
u
a
g
e
aim
ed
at
o
p
p
o
s
itio
n
p
ar
ties
ten
d
ed
to
p
air
with
n
e
g
ativ
e
s
en
tim
en
ts
,
len
d
in
g
s
tr
en
g
th
to
th
e
s
p
ea
k
er
’
s
b
ase
v
is
-
a
-
v
is
th
e
o
p
p
o
s
itio
n
.
I
t
h
i
g
h
lig
h
ts
h
o
w
p
o
liti
cian
s
m
an
ip
u
late
lan
g
u
a
g
e
to
cr
af
t n
ar
r
ativ
es
an
d
m
u
s
ter
s
u
p
p
o
r
t.
T
h
is
r
esear
ch
f
o
cu
s
es
m
ain
ly
o
n
v
er
b
al
tex
t,
b
u
t
ex
clu
d
es
m
u
ltimo
d
al
in
f
o
r
m
atio
n
lik
e
au
d
io
to
n
e,
f
ac
ial
ex
p
r
ess
io
n
,
o
r
g
estu
r
al
cu
es
th
at
p
lay
an
im
p
o
r
tan
t
r
o
le
in
p
o
liti
ca
l
d
eliv
er
y
.
Sen
tim
en
t
class
if
icatio
n
was b
ased
o
n
a
s
et
o
f
2
,
5
0
0
m
an
u
ally
lab
eled
s
am
p
les in
th
e
E
n
g
lis
h
lan
g
u
ag
e,
th
u
s
in
tr
o
d
u
cin
g
p
o
ten
tial b
ias
in
th
e
p
er
f
o
r
m
a
n
ce
in
u
n
d
er
r
e
p
r
esen
ted
s
en
tim
en
t
class
es
an
d
lim
itatio
n
s
in
cr
o
s
s
-
lin
g
u
al
an
d
cr
o
s
s
-
cu
ltu
r
al
ap
p
licab
ilit
y
.
I
t
s
h
o
u
ld
b
e
n
o
t
ed
th
at
th
ese
lim
itatio
n
s
h
a
v
e
b
ee
n
ad
m
itted
in
o
r
d
er
to
m
ai
n
tain
tr
an
s
p
ar
e
n
cy
,
an
d
we
r
ec
o
m
m
en
d
th
at
f
u
t
u
r
e
r
esear
c
h
ef
f
o
r
ts
m
o
v
e
to
war
d
m
u
ltimo
d
al
an
aly
s
is
an
d
c
r
o
s
s
-
lin
g
u
is
tic
m
o
d
elin
g
s
to
in
c
r
ea
s
e
g
en
er
al
i
za
tio
n
to
r
ea
l
-
wo
r
ld
s
ce
n
ar
io
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.
1
5
,
No
.
2
,
J
u
n
e
20
2
6
:
6
7
4
-
6
8
2
680
T
h
ese
r
esu
lts
ar
e
in
lin
e
with
p
r
io
r
wo
r
k
[
2
1
]
,
[
2
2
]
,
wh
er
e
R
NN
-
f
ash
io
n
ed
m
o
d
els
o
u
tp
er
f
o
r
m
e
d
b
aselin
e
class
if
ier
s
in
p
o
liti
ca
l
s
en
tim
en
t
an
aly
s
is
.
W
e
g
o
o
n
e
s
tep
f
u
r
th
er
i
n
o
u
r
im
p
lem
en
tatio
n
,
ad
d
i
n
g
b
id
ir
ec
tio
n
al
f
lo
w
in
GR
U
lay
er
s
to
im
p
r
o
v
e
co
n
tex
tu
al
in
f
o
r
m
atio
n
.
T
h
e
f
ac
t
th
at
p
r
ev
i
o
u
s
wo
r
k
s
h
av
e
u
s
ed
eith
er
L
STM
,
o
r
s
tatic
m
o
d
els,
we
b
eliev
e
th
at
o
u
r
s
o
lu
tio
n
is
m
o
r
e
ef
f
icien
t,
s
ca
lab
le
an
d
ac
cu
r
ate,
an
d
f
u
r
th
er
r
ef
lects th
e
wr
itin
g
tr
e
n
d
o
f
th
e
p
r
esen
t p
o
liti
ca
l N
L
P.
3
.
4
.
I
m
pli
ca
t
i
o
ns
f
o
r
po
litic
a
l dis
co
urs
e
W
h
en
u
s
ed
alo
n
g
with
s
en
tim
en
t
th
is
g
iv
es
a
f
u
ll
v
iew
o
f
p
o
liti
ca
l
r
h
eto
r
ic.
W
h
er
e
th
ese
p
atter
n
s
in
ter
s
ec
t,
we
ca
n
ca
tch
a
g
lim
p
s
e
o
f
h
o
w
lead
er
s
ap
p
l
y
ce
r
t
ain
r
ec
u
r
r
i
n
g
tactics
to
elicit
e
m
o
tio
n
s
f
r
o
m
th
eir
lis
ten
er
s
.
T
h
is
ar
ticle
ties
th
e
m
atic
f
o
ci
to
d
y
n
am
ics
o
f
s
en
tim
en
t,
p
r
o
v
id
i
n
g
a
lin
e
o
f
s
ig
h
t
in
to
h
o
w
p
o
liti
ca
l
co
m
m
u
n
icatio
n
h
as
ev
o
l
v
ed
.
I
t
o
b
s
er
v
es
th
at
p
o
liti
ca
l
talk
is
a
v
eh
icle
th
r
o
u
g
h
wh
ich
th
e
p
u
b
lic
is
a
n
s
wer
ed
in
its
i
s
s
u
es
an
d
th
at
au
d
ien
ce
s
ar
e
m
o
u
ld
ed
.
T
h
ese
f
in
d
in
g
s
tell
u
s
a
lo
t
ab
o
u
t
th
e
co
g
n
itiv
e
m
ec
h
an
is
m
s
th
at
d
r
iv
e
h
o
w
p
o
liti
ca
l
d
eb
ate
wo
r
k
s
,
an
d
h
av
e
im
p
licatio
n
s
f
o
r
wh
at
it
m
ea
n
s
to
r
eso
lv
e
p
o
li
tical
d
if
f
er
en
ce
s
in
an
in
cr
ea
s
in
g
ly
c
o
m
p
lex
u
n
iv
e
r
s
e.
4.
CO
NCLU
SI
O
N
Fin
d
in
g
s
h
av
e
r
ev
ea
led
th
at
m
atch
in
g
em
o
tio
n
an
d
th
em
e
is
in
v
o
lv
ed
in
p
o
liti
ca
l
s
p
ee
ch
es
to
a
g
r
ea
t
ex
ten
t.
T
h
ese
r
esu
lts
ar
e
c
o
m
p
atib
le
with
th
e
h
y
p
o
th
esis
th
at
r
h
eto
r
ical
s
witch
in
g
is
ass
o
ciate
d
with
d
ec
is
io
n
s
to
co
n
s
cio
u
s
ly
en
c
o
d
e
c
o
n
ten
t
an
d
af
f
ec
t
in
s
p
ec
if
ic
lin
g
u
is
tic
s
y
s
tem
s
,
r
ath
er
th
an
b
ei
n
g
b
ased
s
o
lely
o
n
t
h
e
(
r
elativ
e)
f
r
eq
u
en
cy
o
f
wo
r
d
s
p
r
o
d
u
ce
d
o
r
am
o
u
n
t
s
p
o
k
en
o
v
er
all.
I
t
aim
s
to
d
ec
ip
h
er
th
e
lin
g
u
is
tic
tech
n
iq
u
es
th
at
p
o
liti
ca
l
lead
er
s
im
p
lem
en
ted
to
cu
s
to
m
is
e
th
eir
lan
g
u
a
g
e
a
n
d
in
f
lu
e
n
c
e
p
u
b
lic
o
p
in
io
n
b
y
em
p
lo
y
in
g
s
tate
-
of
-
th
e
-
ar
t
m
et
h
o
d
s
s
u
ch
as
W
h
is
p
er
f
o
r
au
d
io
tr
an
s
cr
ip
tio
n
,
B
id
ir
ec
tio
n
al
GR
U
f
o
r
s
en
tim
en
t
class
if
icatio
n
an
d
B
E
R
T
o
p
ic
f
o
r
to
p
ic
m
o
d
ellin
g
.
T
h
e
f
in
d
in
g
s
u
g
g
ests
th
at
cu
es
-
s
u
ch
as
wo
r
d
s
o
f
in
clu
s
iv
ity
an
d
em
o
tio
n
-
lad
e
n
lan
g
u
ag
e
-
a
r
e
s
tr
ateg
ic
d
ev
ices
p
o
liti
cian
s
u
s
e
to
en
g
e
n
d
er
tr
u
s
t,
s
et
th
e
m
s
elv
es
ap
ar
t
f
r
o
m
th
e
p
o
liti
ca
l
s
tatu
s
q
u
o
,
an
d
d
o
m
in
ate
th
e
n
ews
cy
cle.
T
h
e
o
v
er
la
p
p
in
g
p
atter
n
s
o
f
s
en
tim
en
t
an
d
th
em
atic
co
h
er
en
ce
illu
m
in
ate
th
e
in
tr
i
ca
te
r
elatio
n
s
h
ip
s
b
etwe
en
em
o
tio
n
,
la
n
g
u
a
g
e,
an
d
s
tr
ateg
y
i
n
p
o
liti
ca
l
talk
.
T
o
co
n
clu
d
e,
t
h
is
s
tu
d
y
s
ee
k
s
to
d
em
o
n
s
tr
ate
th
e
u
s
e
ca
s
es
f
o
r
NL
P
(
an
d
p
ar
tial
ai)
to
war
d
s
u
p
g
r
a
d
in
g
o
f
tr
an
s
p
ar
en
cy
,
tr
ac
ea
b
ilit
y
an
d
an
aly
s
is
in
co
n
tem
p
o
r
ar
y
p
o
lit
ical
d
ialo
g
u
e
in
a
d
ig
ital c
o
m
m
u
n
icatio
n
a
g
e.
5.
F
UT
U
R
E
WO
RK
Ou
r
r
esu
lts
in
d
icate
th
at
B
id
ir
ec
tio
n
al
GR
U
m
o
d
els
ar
e
m
o
r
e
r
o
b
u
s
t
th
an
u
n
i
d
ir
ec
tio
n
al
o
n
es
in
ca
p
tu
r
in
g
th
e
co
n
tex
tu
al
s
en
tim
en
t
f
ea
tu
r
es
th
er
eo
f
in
p
o
liti
ca
l
u
tter
an
ce
s
.
Su
b
s
eq
u
en
t
r
esear
ch
co
u
ld
s
tu
d
y
d
ee
p
,
co
n
tex
tu
al
e
m
b
ed
d
in
g
s
u
s
in
g
tr
an
s
f
o
r
m
er
-
b
ased
m
o
d
els
s
u
ch
as
B
E
R
T
an
d
R
o
B
E
R
T
a,
to
im
p
r
o
v
e
s
en
tim
en
t
class
if
icat
io
n
in
th
e
f
u
ll
s
p
ec
tr
u
m
o
f
em
o
tio
n
.
I
t
wo
u
ld
also
b
e
v
alu
ab
le
to
in
clu
d
e
a
m
o
r
e
lin
g
u
is
tically
an
d
cu
ltu
r
ally
d
i
v
er
s
e
d
ataset
to
g
en
er
alis
e
th
e
m
o
d
el
f
o
r
p
o
liti
ca
l settin
g
s
wo
r
ld
wid
e.
Fu
tu
r
e
r
esear
ch
ca
n
b
u
ild
o
n
t
h
ese
d
is
co
v
er
ies
b
y
r
etu
r
n
in
g
to
th
e
tem
p
o
r
al
d
ev
elo
p
m
en
t
o
f
p
o
liti
ca
l
r
h
eto
r
ic
to
l
o
o
k
f
o
r
l
o
n
g
e
r
-
ter
m
tr
en
d
s
in
d
is
cu
r
s
iv
e
s
tr
ateg
y
.
Fo
cu
s
in
g
p
u
r
ely
o
n
v
er
b
al
co
m
m
u
n
icatio
n
,
o
r
tex
t
at
b
est,
an
d
ig
n
o
r
i
n
g
th
e
m
u
lt
im
o
d
al
co
n
tex
t
(
e.
g
.
g
estu
r
es,
f
ac
ial
ex
p
r
ess
io
n
s
)
o
f
f
er
s
b
u
t
a
p
ar
tial
an
aly
s
is
o
f
p
o
liti
ca
l
co
m
m
u
n
icatio
n
.
I
n
tr
o
d
u
ctio
n
in
to
th
e
an
aly
s
is
o
f
v
er
b
al
d
eliv
er
y
in
ad
d
itio
n
to
n
o
n
-
v
er
b
al
d
eliv
er
y
f
o
r
f
u
r
t
h
er
ex
p
lo
r
atio
n
o
f
h
o
w
v
er
b
al
asp
ec
ts
o
f
s
p
ee
ch
ar
e
im
p
ac
ted
b
y
co
n
tex
t
wo
u
ld
le
n
d
its
elf
well
to
u
n
d
er
s
tan
d
in
g
th
e
lev
el
o
f
in
f
lu
en
ce
o
f
n
o
n
-
v
er
b
al
cu
es d
u
r
i
n
g
p
o
liti
ca
l a
d
d
r
e
s
s
es.
F
UNDING
I
NF
O
R
M
A
T
I
O
N
T
h
is
r
esear
ch
d
id
n
o
t
r
ec
eiv
e
an
y
s
p
ec
if
ic
g
r
an
t
f
r
o
m
f
u
n
d
i
n
g
ag
en
cies
in
th
e
p
u
b
lic,
co
m
m
er
cial,
o
r
not
-
f
or
-
p
r
o
f
it secto
r
s
.
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
Ma
lay
aj
Ku
m
ar
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
An
u
j K
u
m
ar
Sin
g
h
✓
✓
So
u
m
itra
Das
✓
✓
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
Ha
r
n
ess
in
g
N
LP
a
n
d
A
I
to
d
e
co
d
e
p
o
liti
ca
l d
is
co
u
r
s
e:
s
p
ee
ch
p
a
tter
n
s
,
s
en
timen
t
…
(
Ma
la
ya
j Ku
ma
r
)
681
C
:
C
o
n
c
e
p
t
u
a
l
i
z
a
t
i
o
n
M
:
M
e
t
h
o
d
o
l
o
g
y
So
:
So
f
t
w
a
r
e
Va
:
Va
l
i
d
a
t
i
o
n
Fo
:
Fo
r
mal
a
n
a
l
y
s
i
s
I
:
I
n
v
e
s
t
i
g
a
t
i
o
n
R
:
R
e
so
u
r
c
e
s
D
:
D
a
t
a
C
u
r
a
t
i
o
n
O
:
W
r
i
t
i
n
g
-
O
r
i
g
i
n
a
l
D
r
a
f
t
E
:
W
r
i
t
i
n
g
-
R
e
v
i
e
w
&
E
d
i
t
i
n
g
Vi
:
Vi
su
a
l
i
z
a
t
i
o
n
Su
:
Su
p
e
r
v
i
s
i
o
n
P
:
P
r
o
j
e
c
t
a
d
mi
n
i
st
r
a
t
i
o
n
Fu
:
Fu
n
d
i
n
g
a
c
q
u
i
si
t
i
o
n
CO
NF
L
I
C
T
O
F
I
N
T
E
R
E
S
T
ST
A
T
E
M
E
NT
T
h
e
au
th
o
r
s
d
ec
lar
e
n
o
co
n
f
lict o
f
in
ter
est
.
E
T
H
I
CAL AP
P
RO
V
AL
N
o
h
u
m
a
n
o
r
a
n
i
m
a
l
s
u
b
j
e
ct
s
w
e
r
e
i
n
v
o
l
v
e
d
i
n
t
h
is
s
t
u
d
y
;
t
h
er
e
f
o
r
e
,
e
t
h
i
c
a
l
a
p
p
r
o
v
a
l
wa
s
n
o
t
r
e
q
u
i
r
e
d
.
DATA AV
AI
L
AB
I
L
I
T
Y
T
h
e
d
ata
u
s
ed
i
n
th
is
s
tu
d
y
ar
e
av
ailab
le
u
p
o
n
r
ea
s
o
n
ab
le
r
eq
u
est
.
RE
F
E
R
E
NC
E
S
[
1
]
A
.
D
a
v
i
s,
N
.
F
e
n
t
o
n
,
D
.
F
r
e
e
d
m
a
n
,
a
n
d
G
.
K
h
i
a
b
a
n
y
,
M
e
d
i
a
,
D
e
m
o
c
ra
c
y
a
n
d
S
o
c
i
a
l
C
h
a
n
g
e
.
S
A
G
E
P
u
b
l
i
c
a
t
i
o
n
s
Lt
d
,
2
0
2
2
.
[
2
]
S
.
J.
F
l
u
s
b
e
r
g
,
K
.
J.
H
o
l
m
e
s,
P
.
H
.
T
h
i
b
o
d
e
a
u
,
R
.
L.
N
a
b
i
,
a
n
d
T.
M
a
t
l
o
c
k
,
“
Th
e
p
s
y
c
h
o
l
o
g
y
o
f
f
r
a
m
i
n
g
:
h
o
w
e
v
e
r
y
d
a
y
l
a
n
g
u
a
g
e
sh
a
p
e
s
t
h
e
w
a
y
w
e
t
h
i
n
k
,
f
e
e
l
,
a
n
d
a
c
t
,
”
Psy
c
h
o
l
o
g
i
c
a
l
S
c
i
e
n
c
e
i
n
t
h
e
P
u
b
l
i
c
I
n
t
e
rest
,
v
o
l
.
2
5
,
n
o
.
3
,
p
p
.
1
0
5
-
1
6
1
,
D
e
c
.
2
0
2
4
,
d
o
i
:
1
0
.
1
1
7
7
/
1
5
2
9
1
0
0
6
2
4
1
2
4
6
9
6
6
.
[
3
]
D
.
R
.
L
o
se
k
e
,
“
Ex
a
mi
n
i
n
g
e
m
o
t
i
o
n
a
s
i
s
d
i
s
c
o
r
a
g
e
:
Em
o
t
i
o
n
c
o
d
e
s
a
n
d
p
r
e
si
d
e
n
t
i
a
l
sp
e
e
c
h
e
s
j
u
s
t
i
f
y
i
n
g
w
a
r
,
”
S
o
c
i
o
l
o
g
i
c
a
l
Q
u
a
r
t
e
r
l
y
,
v
o
l
.
5
0
,
n
o
.
3
,
p
p
.
4
9
7
–
5
2
4
,
A
u
g
.
2
0
0
9
,
d
o
i
:
1
0
.
1
1
1
1
/
j
.
1
5
3
3
-
8
5
2
5
.
2
0
0
9
.
0
1
1
5
0
.
x
.
[
4
]
S
.
S
t
i
e
g
l
i
t
z
a
n
d
L.
D
a
n
g
-
X
u
a
n
,
“
S
o
c
i
a
l
m
e
d
i
a
a
n
d
p
o
l
i
t
i
c
a
l
c
o
m
mu
n
i
c
a
t
i
o
n
:
a
so
c
i
a
l
m
e
d
i
a
a
n
a
l
y
t
i
c
s
f
r
a
mew
o
r
k
,
”
S
o
c
i
a
l
N
e
t
w
o
r
k
A
n
a
l
y
si
s
a
n
d
M
i
n
i
n
g
,
v
o
l
.
3
,
n
o
.
4
,
p
p
.
1
2
7
7
-
1
2
9
1
,
2
0
1
3
,
d
o
i
:
1
0
.
1
0
0
7
/
s
1
3
2
7
8
-
0
1
2
-
0
0
7
9
-
3.
[
5
]
J.
M
.
B
a
l
k
i
n
,
“
F
r
e
e
sp
e
e
c
h
i
n
t
h
e
a
l
g
o
r
i
t
h
mi
c
s
o
c
i
e
t
y
:
b
i
g
d
a
t
a
,
p
r
i
v
a
t
e
g
o
v
e
r
n
a
n
c
e
,
a
n
d
n
e
w
sc
h
o
o
l
sp
e
e
c
h
r
e
g
u
l
a
t
i
o
n
,
”
S
S
RN
El
e
c
t
ro
n
i
c
J
o
u
r
n
a
l
,
v
o
l
.
5
1
,
p
.
1
1
4
9
,
2
0
1
7
,
d
o
i
:
1
0
.
2
1
3
9
/
ssr
n
.
3
0
3
8
9
3
9
.
[
6
]
J.
M
c
C
o
y
,
T
.
R
a
h
ma
n
,
a
n
d
M
.
S
o
mer,
“
P
o
l
a
r
i
z
a
t
i
o
n
a
n
d
t
h
e
g
l
o
b
a
l
c
r
i
si
s
o
f
d
e
mo
c
r
a
c
y
:
c
o
m
mo
n
p
a
t
t
e
r
n
s,
d
y
n
a
mi
c
s,
a
n
d
p
e
r
n
i
c
i
o
u
s
c
o
n
s
e
q
u
e
n
c
e
s
f
o
r
d
e
m
o
c
r
a
t
i
c
p
o
l
i
t
i
e
s,”
Am
e
r
i
c
a
n
Be
h
a
v
i
o
r
a
l
S
c
i
e
n
t
i
s
t
,
v
o
l
.
6
2
,
n
o
.
1
,
p
p
.
1
6
-
4
2
,
Ja
n
.
2
0
1
8
,
d
o
i
:
1
0
.
1
1
7
7
/
0
0
0
2
7
6
4
2
1
8
7
5
9
5
7
6
.
[
7
]
J.
T
.
Jo
s
t
e
t
a
l
.
,
“
H
o
w
so
c
i
a
l
me
d
i
a
f
a
c
i
l
i
t
a
t
e
s
p
o
l
i
t
i
c
a
l
p
r
o
t
e
s
t
:
i
n
f
o
r
m
a
t
i
o
n
,
m
o
t
i
v
a
t
i
o
n
,
a
n
d
s
o
c
i
a
l
n
e
t
w
o
r
k
s,
”
Po
l
i
t
i
c
a
l
Psy
c
h
o
l
o
g
y
,
v
o
l
.
3
9
,
n
o
.
S
1
,
p
p
.
8
5
–
1
1
8
,
2
0
1
8
,
d
o
i
:
1
0
.
1
1
1
1
/
p
o
p
s
.
1
2
4
7
8
.
[
8
]
P
.
S
.
H
a
r
t
,
S
.
C
h
i
n
n
,
a
n
d
S
.
S
o
r
o
k
a
,
“
P
o
l
i
t
i
c
i
z
a
t
i
o
n
a
n
d
p
o
l
a
r
i
z
a
t
i
o
n
i
n
C
O
V
I
D
-
1
9
n
e
w
s
c
o
v
e
r
a
g
e
,
”
S
c
i
e
n
c
e
C
o
m
m
u
n
i
c
a
t
i
o
n
,
v
o
l
.
4
2
,
n
o
.
5
,
p
p
.
6
7
9
-
6
9
7
,
2
0
2
0
,
d
o
i
:
1
0
.
1
1
7
7
/
1
0
7
5
5
4
7
0
2
0
9
5
0
7
3
5
.
[
9
]
J.
G
r
i
mm
e
r
a
n
d
B
.
M
.
S
t
e
w
a
r
t
,
“
Te
x
t
a
s
d
a
t
a
:
Th
e
p
r
o
mi
s
e
a
n
d
p
i
t
f
a
l
l
s
o
f
a
u
t
o
ma
t
i
c
c
o
n
t
e
n
t
a
n
a
l
y
si
s
m
e
t
h
o
d
s
f
o
r
p
o
l
i
t
i
c
a
l
t
e
x
t
s,
”
Po
l
i
t
i
c
a
l
A
n
a
l
y
s
i
s
,
v
o
l
.
2
1
,
n
o
.
3
,
p
p
.
2
6
7
-
2
9
7
,
Ja
n
.
2
0
1
3
,
d
o
i
:
1
0
.
1
0
9
3
/
p
a
n
/
mp
s
0
2
8
.
[
1
0
]
E.
F
o
u
r
n
i
e
r
-
To
mb
s
a
n
d
M
.
K
.
M
a
c
K
e
n
z
i
e
,
“
B
i
g
d
a
t
a
a
n
d
d
e
m
o
c
r
a
t
i
c
s
p
e
e
c
h
:
P
r
e
d
i
c
t
i
n
g
d
e
l
i
b
e
r
a
t
i
v
e
q
u
a
l
i
t
y
u
si
n
g
mac
h
i
n
e
l
e
a
r
n
i
n
g
t
e
c
h
n
i
q
u
e
s,
”
M
e
t
h
o
d
o
l
o
g
i
c
a
l
I
n
n
o
v
a
t
i
o
n
s
,
v
o
l
.
1
4
,
n
o
.
2
,
2
0
2
1
,
d
o
i
:
1
0
.
1
1
7
7
/
2
0
5
9
7
9
9
1
2
1
1
0
1
0
4
1
6
.
[
1
1
]
T.
M
a
r
w
a
l
a
,
Art
i
f
i
c
i
a
l
i
n
t
e
l
l
i
g
e
n
c
e
,
g
a
m
e
t
h
e
o
ry
a
n
d
m
e
c
h
a
n
i
sm
d
e
s
i
g
n
i
n
p
o
l
i
t
i
c
s
.
S
i
n
g
a
p
o
r
e
:
S
p
r
i
n
g
e
r
N
a
t
u
r
e
S
i
n
g
a
p
o
r
e
,
2
0
2
3
.
[
1
2
]
V
.
D
e
l
i
ć
e
t
a
l
.
,
“
S
p
e
e
c
h
t
e
c
h
n
o
l
o
g
y
p
r
o
g
r
e
ss
b
a
se
d
o
n
n
e
w
ma
c
h
i
n
e
l
e
a
r
n
i
n
g
p
a
r
a
d
i
g
m,”
C
o
m
p
u
t
a
t
i
o
n
a
l
I
n
t
e
l
l
i
g
e
n
c
e
a
n
d
N
e
u
ro
s
c
i
e
n
c
e
,
v
o
l
.
2
0
1
9
,
p
p
.
1
-
1
9
,
2
0
1
9
,
d
o
i
:
1
0
.
1
1
5
5
/
2
0
1
9
/
4
3
6
8
0
3
6
.
[
1
3
]
T.
M
i
k
o
l
o
v
,
K
.
C
h
e
n
,
G
.
C
o
r
r
a
d
o
,
a
n
d
J
.
D
e
a
n
,
“
Ef
f
i
c
i
e
n
t
e
s
t
i
m
a
t
i
o
n
o
f
w
o
r
d
r
e
p
r
e
se
n
t
a
t
i
o
n
s
i
n
v
e
c
t
o
r
s
p
a
c
e
,
”
1
st
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
L
e
a
rn
i
n
g
Re
p
res
e
n
t
a
t
i
o
n
s
,
I
C
L
R
2
0
1
3
-
W
o
r
k
sh
o
p
T
r
a
c
k
Pr
o
c
e
e
d
i
n
g
s
,
S
e
p
.
2
0
1
3
.
[
1
4
]
T
.
R
a
t
h
i
a
n
d
M
.
Tr
i
p
a
t
h
y
,
“
A
n
a
l
y
z
i
n
g
t
h
e
i
n
f
l
u
e
n
c
e
o
f
d
i
f
f
e
r
e
n
t
s
p
e
e
c
h
d
a
t
a
c
o
r
p
o
r
a
a
n
d
s
p
e
e
c
h
f
e
a
t
u
r
e
s
o
n
s
p
e
e
c
h
e
m
o
t
i
o
n
r
e
c
o
g
n
i
t
i
o
n
:
A
r
e
v
i
e
w
,
”
S
p
e
e
c
h
C
o
m
m
u
n
i
c
a
t
i
o
n
,
v
o
l
.
1
6
2
,
p
.
1
0
3
1
0
2
,
Ju
l
.
2
0
2
4
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
sp
e
c
o
m.
2
0
2
4
.
1
0
3
1
0
2
.
[
1
5
]
G
.
B
r
y
d
a
a
n
d
A
.
P
.
C
o
s
t
a
,
“
Q
u
a
l
i
t
a
t
i
v
e
r
e
s
e
a
r
c
h
i
n
d
i
g
i
t
a
l
e
r
a
:
i
n
n
o
v
a
t
i
o
n
s,
me
t
h
o
d
o
l
o
g
i
e
s
a
n
d
c
o
l
l
a
b
o
r
a
t
i
o
n
s
,
”
S
o
c
i
a
l
S
c
i
e
n
c
e
s
,
v
o
l
.
1
2
,
n
o
.
1
0
,
p
.
5
7
0
,
O
c
t
.
2
0
2
3
,
d
o
i
:
1
0
.
3
3
9
0
/
s
o
c
s
c
i
1
2
1
0
0
5
7
0
.
[
1
6
]
J.
W
.
P
e
n
n
e
b
a
k
e
r
a
n
d
T.
C
.
La
y
,
“
La
n
g
u
a
g
e
u
se
a
n
d
p
e
r
so
n
a
l
i
t
y
d
u
r
i
n
g
c
r
i
s
e
s:
A
n
a
l
y
s
e
s
o
f
M
a
y
o
r
R
u
d
o
l
p
h
G
i
u
l
i
a
n
i
’
s
p
r
e
ss
c
o
n
f
e
r
e
n
c
e
s,”
J
o
u
rn
a
l
o
f
R
e
se
a
rc
h
i
n
Pe
rso
n
a
l
i
t
y
,
v
o
l
.
3
6
,
n
o
.
3
,
p
p
.
2
7
1
–
2
8
2
,
2
0
0
2
,
d
o
i
:
1
0
.
1
0
0
6
/
j
r
p
e
.
2
0
0
2
.
2
3
4
9
.
[
1
7
]
B
.
L
a
w
l
e
s
s
a
n
d
Y
.
W
.
C
h
e
n
,
“
D
e
v
e
l
o
p
i
n
g
a
m
e
t
h
o
d
o
f
c
r
i
t
i
c
a
l
t
h
e
ma
t
i
c
a
n
a
l
y
si
s
f
o
r
q
u
a
l
i
t
a
t
i
v
e
c
o
mm
u
n
i
c
a
t
i
o
n
i
n
q
u
i
r
y
,
”
H
o
w
a
rd
J
o
u
rn
a
l
o
f
C
o
m
m
u
n
i
c
a
t
i
o
n
s
,
v
o
l
.
3
0
,
n
o
.
1
,
p
p
.
9
2
–
1
0
6
,
Ja
n
.
2
0
1
9
,
d
o
i
:
1
0
.
1
0
8
0
/
1
0
6
4
6
1
7
5
.
2
0
1
8
.
1
4
3
9
4
2
3
.
[
1
8
]
P
.
D
i
M
a
g
g
i
o
,
M
.
N
a
g
,
a
n
d
D
.
B
l
e
i
,
“
Ex
p
l
o
i
t
i
n
g
a
f
f
i
n
i
t
i
e
s
b
e
t
w
e
e
n
t
o
p
i
c
mo
d
e
l
i
n
g
a
n
d
t
h
e
so
c
i
o
l
o
g
i
c
a
l
p
e
r
sp
e
c
t
i
v
e
o
n
c
u
l
t
u
r
e
:
A
p
p
l
i
c
a
t
i
o
n
t
o
n
e
w
s
p
a
p
e
r
c
o
v
e
r
a
g
e
o
f
U
.
S
.
g
o
v
e
r
n
m
e
n
t
a
r
t
s
f
u
n
d
i
n
g
,
”
P
o
e
t
i
c
s
,
v
o
l
.
4
1
,
n
o
.
6
,
p
p
.
5
7
0
-
6
0
6
,
D
e
c
.
2
0
1
3
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
p
o
e
t
i
c
.
2
0
1
3
.
0
8
.
0
0
4
.
[
1
9
]
M
.
H
a
s
e
l
m
a
y
e
r
,
S
.
C
.
D
i
n
g
l
e
r
,
a
n
d
M
.
J
e
n
n
y
,
“
H
o
w
w
o
m
e
n
s
h
a
p
e
n
e
g
a
t
i
v
i
t
y
i
n
p
a
r
l
i
a
m
e
n
t
a
r
y
s
p
e
e
c
h
e
s
-
a
s
e
n
t
i
m
e
n
t
a
n
a
l
y
s
i
s
o
f
d
e
b
a
t
e
s
i
n
t
h
e
A
u
s
t
r
i
a
n
p
a
r
l
i
a
m
e
n
t
,
”
P
a
r
l
i
a
m
e
n
t
a
r
y
A
f
f
a
i
r
s
,
v
o
l
.
7
5
,
n
o
.
4
,
p
p
.
8
6
7
-
8
8
6
,
N
o
v
.
2
0
2
2
,
d
o
i
:
1
0
.
1
0
9
3
/
p
a
/
g
s
a
b
0
4
5
.
[
2
0
]
C
.
L
u
o
,
A
.
C
h
e
n
,
B
.
C
u
i
,
a
n
d
W
.
Li
a
o
,
“
Ex
p
l
o
r
i
n
g
p
u
b
l
i
c
p
e
r
c
e
p
t
i
o
n
s
o
f
t
h
e
C
O
V
I
D
-
1
9
v
a
c
c
i
n
e
o
n
l
i
n
e
f
r
o
m
a
c
u
l
t
u
r
a
l
p
e
r
s
p
e
c
t
i
v
e
:
S
e
ma
n
t
i
c
n
e
t
w
o
r
k
a
n
a
l
y
si
s
o
f
t
w
o
so
c
i
a
l
me
d
i
a
p
l
a
t
f
o
r
ms
i
n
t
h
e
U
n
i
t
e
d
S
t
a
t
e
s
a
n
d
C
h
i
n
a
,
”
T
e
l
e
m
a
t
i
c
s
a
n
d
I
n
f
o
rm
a
t
i
c
s
,
v
o
l
.
6
5
,
p
.
1
0
1
7
1
2
,
D
e
c
.
2
0
2
1
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
t
e
l
e
.
2
0
2
1
.
1
0
1
7
1
2
.
[
2
1
]
M
.
S
.
H
o
ssa
i
n
,
M
.
R
.
I
sl
a
m
,
B
.
R
i
sk
h
a
n
,
M
.
M
.
H
a
sa
n
,
a
n
d
R
.
I
sl
a
m,
“
P
o
l
i
t
i
c
a
l
se
n
t
i
m
e
n
t
a
n
a
l
y
s
i
s
u
si
n
g
n
a
t
u
r
a
l
l
a
n
g
u
a
g
e
p
r
o
c
e
ss
i
n
g
o
n
s
o
c
i
a
l
me
d
i
a
,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
rn
a
l
o
f
A
p
p
l
i
e
d
Me
t
h
o
d
s
i
n
El
e
c
t
ro
n
i
c
s
a
n
d
C
o
m
p
u
t
e
rs
,
2
0
2
4
,
d
o
i
:
1
0
.
5
8
1
9
0
/
i
j
a
mec
.
2
0
2
4
.
1
0
8
.
[
2
2
]
T.
A
.
A
l
-
Q
a
b
l
a
n
,
M
.
H
.
M
o
h
d
N
o
o
r
,
M
.
A
.
A
l
-
B
e
t
a
r
,
a
n
d
A
.
T.
K
h
a
d
e
r
,
“
A
su
r
v
e
y
o
n
se
n
t
i
me
n
t
a
n
a
l
y
si
s
a
n
d
i
t
s
a
p
p
l
i
c
a
t
i
o
n
s,”
N
e
u
ra
l
C
o
m
p
u
t
i
n
g
a
n
d
Ap
p
l
i
c
a
t
i
o
n
s
,
v
o
l
.
3
5
,
n
o
.
2
9
,
p
p
.
2
1
5
6
7
–
2
1
6
0
1
,
2
0
2
3
,
d
o
i
:
1
0
.
1
0
0
7
/
s0
0
5
2
1
-
0
2
3
-
0
8
9
4
1
-
y.
[
2
3
]
U
.
R
a
w
a
t
a
n
d
C
.
S
.
R
a
i
,
“
E
n
h
a
n
c
i
n
g
C
N
N
-
a
t
t
e
n
t
i
o
n
-
LST
M
w
i
t
h
n
o
v
e
l
L
S
TM
g
a
t
e
mec
h
a
n
i
sm:
e
v
a
l
u
a
t
i
o
n
o
n
b
e
n
c
h
mar
k
a
n
d
b
r
a
i
n
t
u
mo
r
i
ma
g
i
n
g
d
a
t
a
s
e
t
s
.
”
N
o
v
.
2
1
,
2
0
2
4
,
d
o
i
:
1
0
.
2
1
2
0
3
/
r
s.
3
.
r
s
-
5
3
9
2
4
3
4
/
v
1
.
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.
1
5
,
No
.
2
,
J
u
n
e
20
2
6
:
6
7
4
-
6
8
2
682
[
2
4
]
Y
.
Y
a
ma
d
a
,
O
.
Li
n
d
e
n
b
a
u
m,
S
.
N
e
g
a
h
b
a
n
,
a
n
d
Y
.
K
l
u
g
e
r
,
“
F
e
a
t
u
r
e
se
l
e
c
t
i
o
n
u
s
i
n
g
s
t
o
c
h
a
st
i
c
g
a
t
e
s
,
”
i
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
rn
i
n
g
,
I
C
ML 2
0
2
0
,
2
0
2
0
,
v
o
l
.
P
a
r
t
F
1
6
8
1
4
7
-
1
4
,
p
p
.
1
0
5
7
9
–
1
0
5
9
0
.
[
2
5
]
M
.
K
.
N
a
z
i
r
,
C
.
N
.
F
a
i
sal
,
M
.
A
.
H
a
b
i
b
,
a
n
d
H
.
A
h
m
a
d
,
“
Le
v
e
r
a
g
i
n
g
mu
l
t
i
l
i
n
g
u
a
l
t
r
a
n
s
f
o
r
m
e
r
f
o
r
m
u
l
t
i
c
l
a
s
s
sen
t
i
men
t
a
n
a
l
y
s
i
s
i
n
c
o
d
e
-
m
i
x
e
d
d
a
t
a
o
f
l
o
w
-
r
e
so
u
r
c
e
l
a
n
g
u
a
g
e
s,
”
I
EE
E
Ac
c
e
s
s
,
v
o
l
.
1
3
,
p
p
.
7
5
3
8
-
7
5
5
4
,
2
0
2
5
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
E
S
S
.
2
0
2
5
.
3
5
2
7
7
1
0
.
[
2
6
]
M
.
K
r
a
s
i
t
s
k
i
i
,
O
.
K
o
l
e
s
n
i
k
o
v
a
,
L
.
C
h
a
n
o
n
a
H
e
r
n
a
n
d
e
z
,
G
.
S
i
d
o
r
o
v
,
a
n
d
A
.
G
e
l
b
u
k
h
,
“
A
d
v
a
n
c
i
n
g
s
e
n
t
i
m
e
n
t
a
n
a
l
y
s
i
s
i
n
t
a
m
i
l
-
e
n
g
l
i
s
h
c
o
d
e
-
mi
x
e
d
t
e
x
t
s
:
c
h
a
l
l
e
n
g
e
s
a
n
d
t
r
a
n
s
f
o
r
m
e
r
-
b
a
s
e
d
s
o
l
u
t
i
o
n
s
,
”
p
p
.
3
0
5
–
3
1
2
,
M
a
r
.
2
0
2
5
,
d
o
i
:
1
0
.
1
8
6
5
3
/
v
1
/
2
0
2
5
.
n
l
p
4
d
h
-
1
.
2
7
.
B
I
O
G
RAP
H
I
E
S O
F
AUTH
O
RS
Ma
la
y
a
j
K
u
m
a
r
b
o
rn
1
A
u
g
u
st
1
9
8
5
,
h
e
h
o
ld
s
a
B.
E.
fr
o
m
th
e
Un
i
v
e
rsity
o
f
P
u
n
e
a
n
d
a
n
M
.
Tec
h
.
(CS
E)
fro
m
JN
TU
Hy
d
e
ra
b
a
d
(2
0
1
3
)
.
He
is cu
rre
n
tl
y
a
Re
se
a
rc
h
S
c
h
o
lar
a
t
S
h
ri
Ja
g
d
is
h
p
ra
sa
d
Jh
a
b
a
rm
a
l
Ti
b
re
wa
la
Un
i
v
e
rsity
.
His
in
tere
sts
in
c
lu
d
e
NL
P
,
AI,
M
L
,
Co
m
p
u
ter Ne
two
r
k
s,
a
n
d
I
o
T.
H
e
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
k
u
m
a
rm
a
lay
a
j@g
m
a
il
.
c
o
m
.
Dr
.
Anu
j
K
u
m
a
r
S
in
g
h
h
e
h
o
ld
s
a
P
h
.
D
.
in
C
o
m
p
u
ter
S
c
ien
c
e
a
n
d
En
g
in
e
e
rin
g
a
n
d
h
a
s
o
v
e
r
1
6
y
e
a
rs
o
f
tea
c
h
i
n
g
e
x
p
e
rien
c
e
.
His
re
se
a
rc
h
in
tere
sts
in
c
lu
d
e
M
L,
AI,
a
n
d
Io
T.
He
se
rv
e
s a
s a
re
s
e
a
rc
h
g
u
id
e
a
t
S
h
ri
Ja
g
d
ish
p
ra
sa
d
J
h
a
b
a
rm
a
l
Ti
b
r
e
wa
la Un
iv
e
rsity
.
H
e
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
m
a
il
2
a
n
u
jj
i@
g
m
a
il
.
c
o
m
.
Dr
.
S
o
u
m
itra
Da
s
h
e
h
o
ld
s a
P
h
.
D
.
i
n
Co
m
p
u
ter S
c
ien
c
e
a
n
d
E
n
g
in
e
e
rin
g
a
n
d
h
a
s
o
v
e
r
1
6
y
e
a
rs
o
f
tea
c
h
in
g
e
x
p
e
r
ien
c
e
.
His
re
se
a
rc
h
in
tere
sts
in
c
lu
d
e
M
L,
AI,
a
n
d
Io
T.
He
se
rv
e
s
a
s
a
re
se
a
rc
h
g
u
id
e
a
t
S
h
r
i
Ja
g
d
ish
p
ra
sa
d
Jh
a
b
a
rm
a
l
Ti
b
re
wa
la
Un
iv
e
rsity
.
H
e
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
so
u
m
it
ra
_
d
a
s
@y
a
h
o
o
.
c
o
m
.
Evaluation Warning : The document was created with Spire.PDF for Python.