I
A
E
S
I
n
t
e
r
n
at
io
n
al
Jou
r
n
al
of
A
r
t
if
ic
ia
l
I
n
t
e
ll
ig
e
n
c
e
(
I
J
-
AI
)
V
ol
.
14
, N
o.
4
,
A
ugus
t
2025
, pp.
3404
~
3411
I
S
S
N
:
2252
-
8938
,
D
O
I
:
10.11591/
ij
a
i.
v
14
.i
4
.pp
3404
-
3411
3404
Jou
r
n
al
h
om
e
page
:
ht
tp
:
//
ij
ai
.
ia
e
s
c
or
e
.c
om
C
om
p
ar
i
n
g
b
i
d
i
r
e
c
t
i
on
al
e
n
c
od
e
r
r
e
p
r
e
se
n
t
at
i
on
s f
r
om
t
r
an
sf
or
m
e
r
s an
d
se
n
t
e
n
c
e
-
B
E
R
T
f
or
au
t
om
at
e
d
r
e
su
m
e
sc
r
e
e
n
i
n
g
A
s
m
it
a D
e
s
h
m
u
k
h
, A
n
j
al
i
R
au
t
D
ah
ak
e
D
e
p
a
r
t
m
e
n
t
o
f
C
om
pu
t
e
r
S
c
i
e
nc
e
a
n
d
E
n
g
i
n
e
e
r
i
n
g,
H
V
P
M
’
s
C
o
l
l
e
ge
o
f
E
ng
i
ne
e
r
i
n
g
a
nd
T
e
c
h
n
ol
o
gy
,
S
a
n
t
G
a
d
ge
B
a
ba
A
m
r
a
va
t
i
U
n
i
ve
r
s
i
t
y
,
A
m
r
a
va
t
i
, I
ndi
a
A
r
t
ic
le
I
n
f
o
A
B
S
T
R
A
C
T
A
r
ti
c
le
h
is
to
r
y
:
R
e
c
e
iv
e
d
A
pr
16, 2024
R
e
vi
s
e
d
J
un 11, 2025
A
c
c
e
pt
e
d
J
ul
10, 2025
In
today’s
digital
age,
organizations
face
the
daunting
challen
ge
of
efficientl
y
screening
an
overwhelmi
ng
number
of
resumes
for
job
openings
.
This
study
investigates
the
potential
of
two
state
-
of
-
the
-
art
natural
la
nguage
processing
models,
bidirectional
encoder
representations
from
transf
ormers
(
BERT
)
and
sentence
-
BERT
(S
-
BERT)
,
to
automate
and
optimi
ze
the
resume
screening
process.
The
research
addresses
the
need
for
ac
curate,
efficient,
and
unbiased
candidate
evaluatio
n
by
leveraging
the
po
wer
of
these
transformer
-
based
language
models.
A
comprehensive
com
parison
between
BERT
and
S
-
BERT
is
performed,
evaluating
their
perfor
mance
across
multiple
metrics,
including
accuracy,
screening
time,
correlatio
n
with
job
descriptions,
and
ranking
quality.
The
findings
reveal
that
S
-
BERT
outperforms
BERT,
ac
hieving
higher
accuracy
(90%
vs.
86%),
faster
screening
time
(0.061
seconds
vs.
1
second
per
resume),
and
st
ronger
correlatio
n
with
job
descripti
ons
(0.38385
5
vs.
0.1249).
S
-
BERT
thou
gh
has
a
smaller
vector
size
of
384
enables
capturing
richer
semantic
infor
mation
compared
to
BERT’s
vector
size
of
768,
contribu
ting
to
its
s
uperior
performance.
The
study
provides
insights
into
the
strengths
and
limi
tations
of
each
model,
offering
valuable
guidance
for
organizations
seek
ing
to
streamline
their
talent
acquisit
ion
processes
and
enhance
candidate
se
lection
through automa
ted systems
.
K
e
y
w
o
r
d
s
:
B
id
ir
e
c
ti
ona
l
e
nc
ode
r
r
e
pr
e
s
e
nt
a
ti
ons
f
r
om
tr
a
ns
f
or
m
e
r
s
C
os
in
e
s
im
il
a
r
it
y
R
e
s
um
e
s
c
r
e
e
ni
ng
S
e
nt
e
nc
e
bi
di
r
e
c
ti
ona
l
e
nc
ode
r
r
e
pr
e
s
e
nt
a
ti
ons
f
r
om
tr
a
ns
f
or
m
e
r
s
T
a
le
nt
a
c
qui
s
it
io
n
This is an
open
acce
ss artic
le unde
r the
CC BY
-
SA
license.
C
or
r
e
s
pon
di
n
g A
u
th
or
:
A
s
m
it
a
D
e
s
hm
ukh
D
e
pa
r
tm
e
nt
of
C
om
put
e
r
S
c
ie
nc
e
a
nd E
ngi
ne
e
r
in
g, H
V
P
M
’
s
C
ol
le
ge
of
E
ngi
ne
e
r
in
g a
nd T
e
c
hnol
ogy
S
a
nt
G
a
dge
B
a
ba
A
m
r
a
va
ti
U
ni
ve
r
s
it
y
A
m
r
a
va
ti
, M
a
ha
r
a
s
ht
r
a
, I
ndi
a
E
m
a
il
:
a
s
m
it
a
de
s
hm
ukh7@
gm
a
il
.c
om
1.
I
N
T
R
O
D
U
C
T
I
O
N
T
he
r
a
pi
d
pr
ol
if
e
r
a
ti
on
of
di
gi
ta
l
te
c
hnol
ogi
e
s
ha
s
tr
a
ns
f
or
m
e
d
th
e
la
nds
c
a
pe
of
ta
le
nt
a
c
qui
s
it
io
n
[
1]
,
pr
e
s
e
nt
in
g
bot
h
oppo
r
tu
ni
ti
e
s
a
nd
c
ha
ll
e
nge
s
f
or
or
ga
ni
z
a
ti
ons
.
A
s
jo
b
pos
ti
ngs
r
e
c
e
iv
e
a
n
in
f
lu
x
of
a
ppl
ic
a
ti
ons
[
2]
,
th
e
tr
a
di
ti
ona
l
m
a
nua
l
a
ppr
oa
c
h
to
r
e
s
um
e
s
c
r
e
e
ni
ng
be
c
om
e
s
in
c
r
e
a
s
in
gl
y
in
e
f
f
ic
ie
nt
a
nd
s
us
c
e
pt
ib
le
to
bi
a
s
e
s
[
3]
.
T
hi
s
pr
e
di
c
a
m
e
nt
ha
s
ig
ni
te
d
a
gr
ow
in
g
de
m
a
nd
f
or
a
ut
om
a
te
d
s
ol
ut
io
ns
th
a
t
c
a
n
s
tr
e
a
m
li
ne
t
he
r
e
s
um
e
s
c
r
e
e
ni
ng pr
oc
e
s
s
[
4]
, e
ns
ur
in
g f
a
ir
c
a
ndi
da
te
e
va
lu
a
ti
on
[
5]
.
T
he
n
a
tu
r
a
l
la
ngua
ge
pr
oc
e
s
s
in
g
(
N
L
P
)
[
6]
,
a
br
a
nc
h
of
a
r
ti
f
ic
ia
l
in
te
ll
ig
e
nc
e
,
h
a
s
e
m
e
r
ge
d
a
s
a
pow
e
r
f
ul
to
ol
in
th
is
dom
a
in
[
7]
,
of
f
e
r
in
g
la
ngua
ge
m
ode
ls
c
a
pa
bl
e
of
unde
r
s
ta
ndi
ng
a
nd
a
na
ly
z
in
g
t
e
xt
ua
l
da
ta
w
it
h
r
e
m
a
r
ka
bl
e
a
c
c
ur
a
c
y
[
8]
.
A
m
ong
th
e
m
os
t
p
r
om
i
s
in
g
N
L
P
m
ode
ls
a
r
e
bi
di
r
e
c
ti
ona
l
e
nc
ode
r
r
e
pr
e
s
e
nt
a
ti
ons
f
r
om
tr
a
ns
f
or
m
e
r
s
(
B
E
R
T
)
[
9]
a
nd
s
e
nt
e
n
c
e
-
B
E
R
T
(
S
-
B
E
R
T
)
[
10]
,
w
hi
c
h
ha
ve
d
e
m
ons
tr
a
te
d
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
A
r
ti
f
I
nt
e
ll
I
S
S
N
:
2252
-
8938
C
om
par
in
g bi
di
r
e
c
ti
onal
e
n
c
ode
r
r
e
pr
e
s
e
nt
at
io
ns
f
r
om
t
r
an
s
fo
r
m
e
r
s
and
s
e
nt
e
nc
e
…
(
A
s
m
it
a D
e
s
hm
u
k
h
)
3405
im
pr
e
s
s
iv
e
pe
r
f
or
m
a
nc
e
[
11]
in
va
r
io
us
te
xt
a
n
a
ly
s
is
ta
s
k
s
[
1
2]
.
T
hi
s
s
tu
dy
a
im
s
to
c
r
it
ic
a
ll
y
e
v
a
lu
a
te
th
e
pot
e
nt
ia
l
of
B
E
R
T
[
13
]
a
nd
S
-
B
E
R
T
f
or
a
ut
om
a
te
d
r
e
s
um
e
s
c
r
e
e
ni
ng
[
14]
,
a
ddr
e
s
s
in
g
th
e
w
ha
t,
w
hy,
a
nd
how
of
th
is
r
e
s
e
a
r
c
h e
nde
a
vor
[
15]
.
S
pe
c
if
ic
a
ll
y, w
e
in
ve
s
ti
ga
te
th
e
a
bi
li
ty
of
th
e
s
e
m
ode
ls
to
e
xt
r
a
c
t
r
e
le
v
a
nt
in
f
or
m
a
ti
on
f
r
om
r
e
s
um
e
s
[
16]
, a
s
s
e
s
s
t
he
ir
s
ui
ta
bi
li
ty
f
or
j
ob
d
e
s
c
r
ip
ti
ons
[
17]
, a
nd r
a
nk c
a
ndi
da
te
s
ba
s
e
d on
th
e
ir
qua
li
f
ic
a
ti
ons
[
18]
.
B
y
c
onduc
ti
ng
a
c
om
pr
e
he
n
s
iv
e
c
om
pa
r
is
on
be
twe
e
n
B
E
R
T
a
nd
S
-
B
E
R
T
[
19]
,
w
e
s
e
e
k
to
id
e
nt
if
y
th
e
m
ode
l
[
20]
th
a
t
e
xc
e
ls
in
a
c
c
ur
a
c
y
[
21]
,
e
f
f
ic
ie
nc
y
[
22]
,
a
nd
r
a
nki
ng
qua
li
ty
,
ul
ti
m
a
te
ly
pr
ovi
di
ng or
ga
ni
z
a
ti
ons
w
it
h a
r
obus
t
s
ol
ut
io
n f
or
opt
im
iz
in
g t
h
e
ir
t
a
le
nt
a
c
qui
s
it
io
n pr
oc
e
s
s
e
s
[
23]
.
T
he
m
ot
iv
a
ti
on
b
e
hi
nd
th
is
r
e
s
e
a
r
c
h
s
te
m
s
f
r
om
th
e
pr
e
s
s
in
g
ne
e
d
to
a
ll
e
vi
a
te
th
e
ti
m
e
-
c
ons
um
in
g
a
nd
e
r
r
or
-
pr
one
na
tu
r
e
of
m
a
nua
l
r
e
s
um
e
s
c
r
e
e
ni
ng,
w
hi
le
m
it
ig
a
ti
ng
pot
e
nt
ia
l
bi
a
s
e
s
th
a
t
m
a
y
a
r
is
e
f
r
o
m
hum
a
n
s
ubj
e
c
ti
vi
ty
[
24]
.
B
y
le
ve
r
a
gi
ng
th
e
pow
e
r
of
a
dv
a
nc
e
d
N
L
P
m
ode
ls
[
25]
,
or
ga
ni
z
a
ti
ons
c
a
n
s
tr
e
a
m
li
ne
th
e
ir
hi
r
in
g
pr
oc
e
s
s
e
s
,
r
e
duc
e
ti
m
e
-
to
-
hi
r
e
,
a
nd
e
n
ha
nc
e
c
a
ndi
da
te
s
e
le
c
ti
on,
th
e
r
e
by
ga
in
in
g
a
c
om
pe
ti
ti
ve
a
dva
nt
a
ge
in
to
da
y’
s
dyna
m
ic
jo
b
m
a
r
ke
t.
T
o
a
c
hi
e
ve
our
r
e
s
e
a
r
c
h
obj
e
c
ti
ve
s
,
w
e
e
m
pl
oye
d
a
r
ig
or
ous
e
xpe
r
im
e
nt
a
l
a
ppr
oa
c
h,
ut
il
iz
in
g
a
la
r
ge
da
t
a
s
e
t
of
r
e
s
um
e
s
a
nd
jo
b
de
s
c
r
ip
ti
ons
.
T
he
m
e
th
odol
ogy
in
vol
ve
d
pr
e
pr
oc
e
s
s
in
g
th
e
da
ta
,
ge
ne
r
a
ti
ng
e
m
be
ddi
ngs
u
s
in
g
B
E
R
T
a
nd
S
-
B
E
R
T
,
a
nd
c
om
put
in
g
c
os
in
e
s
im
il
a
r
it
ie
s
be
twe
e
n
r
e
s
um
e
a
nd
jo
b
de
s
c
r
ip
ti
on
e
m
be
ddi
ngs
.
E
xt
e
ns
iv
e
qua
nt
it
a
ti
ve
a
nd
qua
li
ta
ti
ve
a
na
ly
s
e
s
w
e
r
e
c
onduc
te
d
to
e
va
lu
a
te
th
e
m
ode
l’
s
pe
r
f
or
m
a
nc
e
,
c
ons
id
e
r
in
g
m
e
tr
ic
s
s
u
c
h
a
s
a
c
c
ur
a
c
y,
s
c
r
e
e
ni
ng
ti
m
e
,
c
or
r
e
la
ti
on w
it
h j
ob de
s
c
r
ip
ti
ons
, a
nd r
a
nki
ng qua
li
ty
.
T
he
s
ubs
e
que
nt
s
e
c
ti
ons
of
th
is
pa
pe
r
pr
ovi
de
a
de
ta
il
e
d
ove
r
vi
e
w
of
th
e
r
e
s
e
a
r
c
h
m
e
th
odol
ogy,
in
c
lu
di
ng
da
ta
c
ol
le
c
ti
on,
pr
e
pr
oc
e
s
s
in
g
te
c
hni
que
s
,
a
nd
e
xpe
r
im
e
nt
a
l
pr
oc
e
dur
e
s
.
T
he
r
e
s
ul
ts
a
nd
di
s
c
us
s
io
n
s
e
c
ti
on
pr
e
s
e
nt
s
a
c
om
pr
e
he
n
s
iv
e
a
n
a
ly
s
is
of
th
e
f
in
di
ngs
,
hi
ghl
ig
ht
in
g
th
e
s
tr
e
ngt
hs
a
nd
li
m
it
a
ti
ons
of
e
a
c
h
m
ode
l,
a
nd
dr
a
w
in
g
in
s
ig
ht
s
f
r
om
a
br
oa
de
r
s
c
ie
nt
if
ic
c
ont
e
xt
. A
ddi
ti
ona
ll
y,
pot
e
nt
ia
l
li
m
it
a
ti
ons
of
th
e
s
tu
dy
a
r
e
a
ddr
e
s
s
e
d,
a
nd
im
pl
ic
a
ti
ons
f
or
f
ut
ur
e
r
e
s
e
a
r
c
h
a
r
e
e
xpl
or
e
d.
F
in
a
ll
y,
th
e
c
onc
lu
s
io
n
s
um
m
a
r
iz
e
s
th
e
k
e
y
c
ont
r
ib
ut
io
ns
a
nd
unde
r
s
c
or
e
s
th
e
s
ig
ni
f
ic
a
nc
e
of
th
is
r
e
s
e
a
r
c
h
in
a
dva
nc
in
g
th
e
f
ie
ld
of
a
ut
om
a
te
d
r
e
s
um
e
s
c
r
e
e
ni
ng a
nd t
a
le
nt
a
c
qui
s
it
io
n.
2.
M
E
T
H
O
D
A
c
om
pr
e
he
ns
iv
e
c
om
pa
r
is
on
be
twe
e
n
B
E
R
T
a
nd
S
-
B
E
R
T
f
or
a
ut
om
a
te
d
r
e
s
um
e
s
c
r
e
e
ni
ng
w
a
s
c
onduc
te
d
us
in
g
a
r
obus
t
a
nd
r
e
pl
ic
a
bl
e
m
e
th
odol
ogy.
T
hi
s
s
e
c
ti
on
out
li
ne
s
th
e
e
xpe
r
im
e
nt
a
l
pr
oc
e
dur
e
,
in
c
lu
di
ng da
ta
c
ol
le
c
ti
on, pr
e
pr
oc
e
s
s
in
g s
te
ps
,
a
nd t
he
i
m
pl
e
m
e
nt
a
ti
on de
ta
il
s
of
bot
h m
ode
ls
.
2.1. Dat
a
c
ol
le
c
t
io
n
an
d
p
r
e
p
r
oc
e
s
s
in
g
T
he
s
tu
dy
ut
il
iz
e
d
a
di
ve
r
s
e
da
t
a
s
e
t
c
om
pr
is
in
g
of
223
r
e
s
u
m
e
s
a
nd
7
jo
b
de
s
c
r
ip
ti
ons
,
obt
a
in
e
d
f
r
om
va
r
io
us
onl
in
e
pl
a
tf
or
m
s
s
uc
h
a
s
L
in
ke
dI
n,
G
oog
le
F
or
m
s
,
a
nd
F
r
e
s
he
r
s
W
or
ld
.
T
he
r
e
s
um
e
s
w
e
r
e
in
it
ia
ll
y
c
onve
r
te
d
to
P
D
F
f
or
m
a
t
a
nd
s
ubs
e
que
nt
ly
to
E
xc
e
l
f
or
m
a
t
to
e
ns
ur
e
c
om
pa
ti
bi
li
ty
w
it
h
th
e
P
yt
ho
n
pr
ogr
a
m
m
in
g
la
ngua
ge
us
e
d
f
or
da
ta
pr
oc
e
s
s
in
g.
P
r
e
pr
oc
e
s
s
in
g
s
te
ps
w
e
r
e
pe
r
f
or
m
e
d
to
pr
e
pa
r
e
th
e
da
ta
f
or
a
na
ly
s
is
.
F
or
th
e
S
-
B
E
R
T
m
ode
l,
s
e
nt
e
nc
e
s
c
ont
a
in
in
g
s
e
ts
of
10
w
or
ds
w
e
r
e
e
xt
r
a
c
te
d
f
r
om
th
e
r
e
s
um
e
s
to
ge
ne
r
a
te
e
m
be
ddi
ngs
.
C
onve
r
s
e
ly
,
th
e
B
E
R
T
m
od
e
l
un
de
r
w
e
nt
le
m
m
a
ti
z
a
ti
on
a
nd
s
te
m
m
in
g
a
s
pr
e
pr
oc
e
s
s
in
g s
te
ps
,
f
ol
lo
w
e
d
by
th
e
r
e
m
ova
l
of
s
to
p w
or
ds
a
n
d
r
e
pe
ti
ti
ons
. T
he
k
e
yw
or
ds
e
xt
r
a
c
te
d
f
r
om
th
e
r
e
s
um
e
s
w
e
r
e
th
e
n
ut
il
iz
e
d
f
or
e
m
be
ddi
ng
ge
ne
r
a
ti
on
w
it
h
B
E
R
T
.
T
hi
s
pr
oc
e
s
s
w
a
s
r
e
pl
ic
a
te
d
f
or
th
e
jo
b
de
s
c
r
ip
ti
ons
, e
ns
ur
in
g c
on
s
is
te
nc
y i
n t
he
d
a
ta
r
e
pr
e
s
e
nt
a
ti
on a
c
r
os
s
bot
h m
ode
ls
.
2.2. E
xp
e
r
im
e
n
t
al
p
r
oc
e
d
u
r
e
T
he
e
xpe
r
im
e
nt
a
l
pr
oc
e
dur
e
in
vol
ve
d
s
e
ve
r
a
l
di
s
ti
nc
t
s
t
a
g
e
s
,
e
a
c
h
d
e
s
ig
ne
d
to
e
v
a
lu
a
te
th
e
pe
r
f
or
m
a
nc
e
of
B
E
R
T
a
nd S
-
B
E
R
T
i
n t
he
c
ont
e
xt
of
a
ut
om
a
te
d
r
e
s
um
e
s
c
r
e
e
ni
ng.
‒
S
ta
ge
1
(
ke
yw
or
d
e
xt
r
a
c
ti
on
a
nd
e
m
be
ddi
ng
ge
ne
r
a
ti
on
):
i
n
th
e
f
ir
s
t
s
ta
ge
,
th
e
s
ys
te
m
e
xt
r
a
c
te
d
s
e
nt
e
nc
e
s
a
nd
to
p
ke
yw
or
ds
f
r
om
th
e
r
e
s
um
e
s
.
F
or
S
-
B
E
R
T
,
s
e
nt
e
nc
e
s
c
ont
a
in
in
g s
e
ts
of
10
w
or
ds
w
e
r
e
e
xt
r
a
c
te
d,
w
hi
le
f
or
B
E
R
T
,
to
p
ke
yw
or
ds
w
e
r
e
id
e
nt
if
ie
d
a
f
te
r
r
e
m
ovi
ng
c
om
m
on
punc
tu
a
ti
ons
a
nd
s
to
p
w
or
ds
. T
he
e
xt
r
a
c
te
d d
a
ta
w
a
s
or
ga
ni
z
e
d i
nt
o a
pa
nda
D
a
ta
F
r
a
m
e
a
nd e
xpor
te
d t
o a
n E
xc
e
l
f
il
e
.
‒
S
ta
ge
2
(
B
E
R
T
a
na
ly
s
i
s
):
in
th
e
s
e
c
ond s
ta
ge
,
th
e
B
E
R
T
m
ode
l
a
nd
a
pr
e
-
tr
a
in
e
d
to
ke
ni
z
e
r
w
e
r
e
ut
il
iz
e
d
to
c
om
put
e
t
he
c
os
in
e
s
im
il
a
r
it
y be
twe
e
n a
j
ob de
s
c
r
ip
ti
on (
s
e
a
r
c
h que
r
y)
a
nd t
he
t
op ke
yw
or
ds
e
xt
r
a
c
te
d
f
r
om
th
e
r
e
s
um
e
s
.
T
hi
s
s
t
a
ge
in
vol
ve
d
in
te
gr
a
ti
ng
va
r
io
us
li
br
a
r
ie
s
,
in
c
lu
di
ng
tr
a
ns
f
or
m
e
r
s
f
or
B
E
R
T
,
ope
npyxl
f
or
E
xc
e
l
ha
ndl
in
g,
N
um
P
y
f
or
num
e
r
ic
a
l
ope
r
a
ti
ons
,
a
nd
to
r
c
h
f
or
in
te
r
a
c
ti
on
w
it
h
P
yT
or
c
h.
T
he
c
o
s
in
e
s
im
il
a
r
it
y
be
twe
e
n
th
e
que
r
y
e
m
b
e
ddi
ngs
a
nd
e
a
c
h
r
e
s
um
e
w
a
s
c
a
lc
ul
a
te
d
us
in
g
th
e
s
kl
e
a
r
n
f
unc
ti
on.
‒
S
ta
ge
3
(
S
-
B
E
R
T
a
na
ly
s
is
):
th
e
th
ir
d
s
ta
ge
e
m
pl
oye
d
t
he
M
in
iL
M
m
ode
l
f
r
om
th
e
S
e
nt
e
nc
e
T
r
a
ns
f
or
m
e
r
s
li
br
a
r
y
to
c
om
put
e
th
e
c
os
in
e
s
im
il
a
r
it
y
be
tw
e
e
n
th
e
s
e
a
r
c
h
qu
e
r
y
a
nd
s
e
nt
e
n
c
e
s
s
to
r
e
d
in
a
n
E
xc
e
l
f
il
e
.
T
hi
s
in
vol
ve
d
lo
a
di
ng
th
e
M
in
iL
M
m
ode
l,
a
c
c
e
s
s
in
g
th
e
E
xc
e
l
f
il
e
w
it
h
ope
npyxl,
a
nd
c
a
lc
ul
a
ti
ng t
he
c
os
in
e
s
im
il
a
r
it
y be
twe
e
n t
he
que
r
y e
m
be
ddi
ng
s
a
nd e
a
c
h r
e
s
um
e
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2252
-
8938
I
nt
J
A
r
ti
f
I
nt
e
ll
,
V
ol
. 14, No. 4, A
ugus
t
2025
:
3404
-
3411
3406
‒
S
ta
ge
4
(
r
a
nki
ng
a
nd
e
va
lu
a
ti
on
):
a
f
te
r
c
om
put
in
g
c
os
in
e
s
im
i
la
r
it
ie
s
f
or
bot
h
B
E
R
T
a
nd
S
-
B
E
R
T
,
th
e
r
e
s
um
e
s
w
e
r
e
r
a
nke
d
ba
s
e
d
on
th
e
ir
s
im
il
a
r
it
y
s
c
or
e
s
.
T
he
m
ode
l’
s
0
pe
r
f
or
m
a
nc
e
w
a
s
e
va
lu
a
te
d
us
in
g
m
e
tr
ic
s
s
uc
h a
s
a
c
c
ur
a
c
y,
s
c
r
e
e
ni
ng
ti
m
e
pe
r
r
e
s
um
e
,
a
nd c
or
r
e
la
ti
on
w
it
h
jo
b
de
s
c
r
ip
ti
ons
. A
ddi
ti
ona
ll
y,
th
r
e
e
i
nde
pe
nde
nt
H
R
m
a
na
g
e
r
s
c
r
os
s
-
ve
r
if
ie
d t
he
r
e
s
ul
ts
t
o e
ns
ur
e
t
he
a
c
c
ur
a
c
y of
t
he
r
a
nki
ngs
.
T
he
f
lo
w
di
a
gr
a
m
pr
e
s
e
nt
e
d
in
F
ig
ur
e
1
de
li
ne
a
te
s
th
e
s
t
e
p
-
by
-
s
te
p
pr
oc
e
s
s
of
c
ondu
c
ti
ng
a
c
om
pa
r
a
ti
ve
a
na
ly
s
i
s
be
tw
e
e
n
B
E
R
T
a
nd
S
-
B
E
R
T
f
or
a
ut
om
a
t
e
d
r
e
s
um
e
s
c
r
e
e
ni
ng.
I
ni
ti
a
ll
y,
a
pool
of
ove
r
223
r
e
s
um
e
s
w
a
s
c
ol
le
c
te
d
f
r
om
va
r
io
us
pl
a
tf
or
m
s
li
ke
L
in
ke
d
I
n,
G
oogl
e
F
or
m
s
,
a
nd
F
r
e
s
he
r
s
W
or
ld
,
of
te
n
in
di
f
f
e
r
in
g
f
or
m
a
ts
.
T
he
s
e
r
e
s
um
e
s
unde
r
go
c
onve
r
s
io
n
to
P
D
F
a
nd
s
ub
s
e
que
nt
ly
to
E
xc
e
l
f
or
m
a
t
f
or
c
om
pa
ti
bi
li
ty
w
it
h
P
yt
hon.
S
-
B
E
R
T
is
e
m
pl
oye
d
to
e
xt
r
a
c
t
s
e
nt
e
nc
e
s
c
ont
a
in
in
g
s
e
ts
of
10
w
or
ds
f
r
om
th
e
r
e
s
um
e
s
f
or
e
m
be
ddi
ng
g
e
ne
r
a
ti
on.
C
onv
e
r
s
e
ly
,
B
E
R
T
u
nde
r
goe
s
le
m
m
a
ti
z
a
ti
on
a
nd
s
t
e
m
m
in
g
a
s
pr
e
pr
oc
e
s
s
in
g s
te
ps
,
f
ol
lo
w
e
d
by
th
e
r
e
m
ova
l
of
s
to
p w
or
ds
a
n
d
r
e
pe
ti
ti
ons
. T
he
k
e
yw
or
ds
e
xt
r
a
c
te
d
f
r
om
th
e
r
e
s
um
e
s
a
r
e
th
e
n
ut
il
iz
e
d
f
or
e
m
be
ddi
ng
ge
ne
r
a
ti
on
w
i
th
B
E
R
T
.
T
hi
s
pr
oc
e
s
s
is
r
e
pl
ic
a
te
d
f
or
th
e
jo
b
de
s
c
r
ip
ti
on, a
nd e
m
be
ddi
ng ve
c
to
r
s
f
r
om
bot
h B
E
R
T
a
nd S
-
B
E
R
T
a
r
e
e
m
pl
oye
d t
o c
om
put
e
c
o
s
in
e
s
im
il
a
r
it
y
w
it
h
th
e
jo
b
de
s
c
r
ip
ti
on,
th
e
r
e
by
f
a
c
il
it
a
ti
ng
r
e
s
um
e
r
a
nki
ng.
T
o
e
ns
ur
e
th
e
a
c
c
ur
a
c
y
of
th
e
r
a
nki
ngs
,
th
r
e
e
in
de
pe
nde
nt
H
R
m
a
na
ge
r
s
c
r
os
s
-
ve
r
if
y
th
e
r
e
s
ul
ts
.
T
hi
s
c
om
pr
e
he
ns
iv
e
a
ppr
oa
c
h
unde
r
s
c
or
e
s
th
e
s
ys
te
m
’
s
r
e
li
a
bi
li
ty
a
nd e
f
f
ic
a
c
y i
n a
ut
om
a
ti
ng t
he
r
e
s
um
e
s
c
r
e
e
ni
ng pr
oc
e
s
s
.
F
ig
ur
e
1. F
lo
w
di
a
gr
a
m
t
o c
onduc
t
c
om
pa
r
a
ti
ve
s
tu
dy of
B
E
R
T
a
nd S
-
B
E
R
T
f
or
a
ut
om
a
te
d r
e
s
um
e
s
c
r
e
e
ni
ng
3.
R
E
S
U
L
T
S
A
N
D
D
I
S
C
U
S
S
I
O
N
T
he
c
om
pa
r
is
on
be
twe
e
n
B
E
R
T
a
nd
S
-
B
E
R
T
f
or
a
ut
om
a
te
d
r
e
s
um
e
s
c
r
e
e
ni
ng
r
e
ve
a
le
d
s
e
ve
r
a
l
ke
y
f
in
di
ngs
a
c
r
os
s
m
ul
ti
pl
e
e
va
lu
a
ti
on
m
e
tr
ic
s
.
I
ni
ti
a
ll
y,
bot
h
m
ode
ls
a
na
ly
z
e
d
a
n
id
e
nt
ic
a
l
d
a
ta
s
e
t
c
om
pr
is
in
g
223 r
e
s
um
e
s
a
nd 7 job des
c
r
ip
ti
ons
t
o e
ns
ur
e
a
f
a
ir
c
om
pa
r
is
on. A nota
bl
e
di
f
f
e
r
e
nc
e
l
ie
s
i
n t
he
f
e
a
tu
r
e
ve
c
to
r
s
iz
e
e
m
pl
oye
d
by
e
a
c
h
m
ode
l.
W
hi
le
B
E
R
T
pr
oduc
e
d
e
m
be
ddi
ngs
of
s
i
z
e
768,
S
-
B
E
R
T
g
e
ne
r
a
te
d
m
or
e
c
om
pa
c
t
ve
c
to
r
s
of
s
iz
e
384.
T
hi
s
di
ve
r
ge
n
c
e
s
ugg
e
s
ts
th
a
t
S
-
B
E
R
T
m
a
y
c
a
pt
ur
e
m
or
e
de
ta
il
e
d
a
nd
c
ont
e
xt
ua
l
in
f
or
m
a
ti
on
f
r
om
th
e
r
e
s
um
e
s
,
pot
e
nt
ia
ll
y
f
a
c
il
it
a
ti
ng
be
tt
e
r
m
a
tc
he
s
be
twe
e
n
c
a
ndi
da
te
s
a
nd
jo
b
r
e
qui
r
e
m
e
nt
s
. F
ur
th
e
r
m
or
e
, ou
r
s
tu
dy
f
ound tha
t
S
-
B
E
R
T
e
xhi
bi
te
d s
upe
r
io
r
e
f
f
ic
ie
nc
y, w
it
h
a
s
c
r
e
e
ni
ng t
im
e
of
0.061
s
e
c
onds
pe
r
r
e
s
um
e
,
c
om
pa
r
e
d
to
B
E
R
T
’
s
lo
nge
r
s
c
r
e
e
ni
ng
ti
m
e
of
1
s
e
c
ond
pe
r
r
e
s
um
e
.
T
hi
s
di
s
c
r
e
pa
nc
y i
n pr
oc
e
s
s
in
g s
pe
e
d unde
r
s
c
or
e
s
S
-
B
E
R
T
’
s
pot
e
nt
ia
l
to
e
nha
nc
e
t
he
ove
r
a
ll
p
r
oduc
ti
vi
ty
of
t
a
le
nt
a
c
qui
s
it
io
n pr
oc
e
s
s
e
s
.
C
r
uc
ia
ll
y,
S
-
B
E
R
T
a
c
hi
e
ve
d
a
hi
ghe
r
a
c
c
ur
a
c
y
r
a
te
of
90%
in
s
hor
tl
is
ti
ng
r
e
s
um
e
s
,
out
pe
r
f
or
m
in
g
B
E
R
T
’
s
a
c
c
ur
a
c
y
r
a
te
of
86%
.
T
hi
s
f
in
di
ng
im
pl
ie
s
th
a
t
S
-
B
E
R
T
m
a
y
pos
s
e
s
s
a
be
tt
e
r
a
bi
li
ty
to
di
s
c
e
r
n
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
A
r
ti
f
I
nt
e
ll
I
S
S
N
:
2252
-
8938
C
om
par
in
g bi
di
r
e
c
ti
onal
e
n
c
ode
r
r
e
pr
e
s
e
nt
at
io
ns
f
r
om
t
r
an
s
fo
r
m
e
r
s
and
s
e
nt
e
nc
e
…
(
A
s
m
it
a D
e
s
hm
u
k
h
)
3407
r
e
le
va
nt
in
f
or
m
a
ti
on
a
nd
a
c
c
ur
a
te
ly
m
a
tc
h
c
a
ndi
da
te
s
to
jo
b
d
e
s
c
r
ip
ti
ons
,
th
e
r
e
by
im
pr
ovi
ng
th
e
qua
li
ty
o
f
c
a
ndi
da
te
s
e
le
c
ti
on.
M
or
e
ove
r
,
our
a
na
ly
s
is
of
s
im
il
a
r
it
y
s
c
or
e
s
r
e
ve
a
le
d
a
s
ubs
ta
nt
ia
l
g
a
p
be
twe
e
n
th
e
two
m
ode
ls
.
W
hi
le
B
E
R
T
yi
e
ld
e
d
a
s
im
il
a
r
it
y
s
c
or
e
of
0.3,
S
-
B
E
R
T
e
xhi
bi
te
d
a
s
ig
ni
f
ic
a
nt
ly
hi
ghe
r
s
c
or
e
of
0.599.
T
hi
s
r
e
s
ul
t
s
ugge
s
t
s
th
a
t
S
-
B
E
R
T
e
xc
e
ls
in
c
a
pt
ur
in
g
s
e
m
a
nt
ic
r
e
la
ti
ons
hi
ps
a
nd
c
ont
e
xt
ua
l
nua
nc
e
s
,
le
a
di
ng
to
m
or
e
p
r
e
c
is
e
m
a
tc
he
s
be
twe
e
n
r
e
s
um
e
s
a
nd
jo
b
de
s
c
r
ip
ti
ons
.
H
ow
e
ve
r
,
it
is
im
por
ta
nt
to
a
c
knowle
dge
th
e
li
m
it
a
ti
ons
of
our
s
tu
dy.
W
hi
le
our
da
ta
s
e
t
c
om
pr
is
e
d
a
di
ve
r
s
e
r
a
nge
of
r
e
s
um
e
s
a
nd
jo
b
de
s
c
r
ip
ti
ons
,
f
ur
th
e
r
r
e
s
e
a
r
c
h
w
it
h
la
r
ge
r
a
nd
m
or
e
va
r
ie
d
da
ta
s
e
ts
is
ne
c
e
s
s
a
r
y
to
va
li
da
te
a
nd
ge
ne
r
a
li
z
e
our
f
in
di
ngs
a
c
r
os
s
di
f
f
e
r
e
nt
i
ndus
tr
ie
s
a
nd j
ob r
ol
e
s
.
F
ig
ur
e
2
of
f
e
r
s
a
d
e
ta
il
e
d
c
om
pa
r
is
on
be
tw
e
e
n
B
E
R
T
a
nd
S
-
B
E
R
T
in
te
r
m
s
of
e
xe
c
ut
io
n
s
pe
e
d
in
s
e
c
onds
(
F
ig
ur
e
2(
a
)
)
a
nd
f
e
a
tu
r
e
e
xt
r
a
c
ti
on
s
pe
e
d
(
F
ig
ur
e
2(
b)
)
.
B
E
R
T
e
xhi
bi
ts
a
n
e
xe
c
ut
io
n
s
pe
e
d
of
2.362
s
e
c
ond
s
,
s
ig
ni
f
ic
a
nt
ly
s
lo
w
e
r
th
a
n
S
-
B
E
R
T
’
s
0.87
s
e
c
onds
,
de
m
on
s
tr
a
ti
ng
th
e
la
tt
e
r
’
s
s
upe
r
io
r
e
f
f
ic
ie
nc
y.
M
or
e
ove
r
,
in
f
e
a
tu
r
e
e
xt
r
a
c
ti
on
s
pe
e
d
p
e
r
r
e
s
um
e
,
B
E
R
T
r
e
qui
r
e
s
a
bout
0.078
s
e
c
onds
,
w
hi
le
S
-
B
E
R
T
s
how
c
a
s
e
s
a
n
im
pr
e
s
s
iv
e
0.029
s
e
c
ond
s
.
T
hi
s
s
ub
s
ta
nt
ia
l
di
f
f
e
r
e
nc
e
unde
r
s
c
or
e
s
S
-
B
E
R
T
’
s
e
nha
nc
e
d
c
om
put
a
ti
ona
l
c
a
pa
bi
li
ti
e
s
,
m
a
ki
ng
it
a
m
or
e
s
ui
ta
bl
e
c
hoi
c
e
f
o
r
ta
s
ks
r
e
qui
r
in
g
s
w
if
t
a
nd
a
c
c
ur
a
te
e
xe
c
ut
io
n
a
nd f
e
a
tu
r
e
e
xt
r
a
c
ti
on.
(
a
)
(
b)
F
ig
ur
e
2. C
om
pa
r
is
on of
B
E
R
T
a
nd S
-
B
E
R
T
w
it
h r
e
s
pe
c
t
to
(
a
)
e
xe
c
ut
io
n s
pe
e
d i
n
s
e
c
ond
s
a
nd
(
b)
f
e
a
tu
r
e
e
xt
r
a
c
ti
on s
pe
e
d
F
ig
ur
e
3
de
pi
c
ts
ba
r
c
ha
r
ts
c
om
pa
r
in
g
th
e
pe
r
f
or
m
a
nc
e
of
B
E
R
T
a
nd
S
-
B
E
R
T
a
c
r
os
s
two
ke
y
m
e
tr
ic
s
of
a
c
c
ur
a
c
y
(
F
ig
ur
e
3
(
a)
)
a
nd
s
im
il
a
r
it
y
s
c
or
e
(
F
ig
ur
e
3
(
b
)
)
.
T
he
a
c
c
ur
a
c
y
m
e
tr
ic
,
di
s
pl
a
ye
d
on
th
e
ve
r
ti
c
a
l
a
xi
s
la
be
le
d
“
A
c
c
ur
a
c
y
(
%
)
”
,
e
va
lu
a
te
s
th
e
e
f
f
e
c
ti
ve
n
e
s
s
of
th
e
a
lg
or
it
hm
s
in
s
hor
tl
is
ti
ng
r
e
s
um
e
s
c
om
pa
r
e
d
to
H
R
m
a
na
ge
r
r
e
s
ul
ts
.
A
hi
ghe
r
va
lu
e
s
ig
ni
f
ie
s
s
up
e
r
io
r
pe
r
f
or
m
a
nc
e
.
S
-
B
E
R
T
a
c
hi
e
ve
s
a
hi
ghe
r
a
c
c
ur
a
c
y
(
90%
)
c
om
pa
r
e
d
to
B
E
R
T
(
86%
)
,
a
s
il
lu
s
tr
a
te
d
in
th
e
c
ha
r
t.
T
he
s
im
il
a
r
it
y
s
c
or
e
,
m
e
a
s
ur
e
d
on
th
e
ve
r
ti
c
a
l
a
xi
s
la
be
le
d
“
S
im
il
a
r
it
y
S
c
or
e
”
,
a
s
s
e
s
s
e
s
th
e
c
o
s
in
e
s
im
il
a
r
it
y
be
twe
e
n
e
m
be
ddi
ngs
of
jo
b
de
s
c
r
ip
ti
ons
a
nd
r
e
s
um
e
e
m
be
ddi
ng
s
.
A
ga
in
,
a
hi
ghe
r
va
l
ue
in
di
c
a
te
s
b
e
tt
e
r
pe
r
f
or
m
a
nc
e
.
T
he
c
ha
r
t
de
m
ons
tr
a
te
s
th
a
t
S
-
B
E
R
T
a
c
hi
e
ve
s
a
hi
ghe
r
s
im
il
a
r
it
y
s
c
or
e
(
0.599)
c
om
pa
r
e
d
to
B
E
R
T
(
0.3)
.
O
ve
r
a
ll
,
th
e
f
in
di
ngs
s
ugge
s
t
th
a
t
S
-
B
E
R
T
out
pe
r
f
or
m
s
B
E
R
T
i
n t
e
r
m
s
of
b
ot
h a
c
c
ur
a
c
y a
nd s
im
il
a
r
it
y s
c
or
e
.
(
a
)
(
b)
F
ig
ur
e
3. C
om
pa
r
is
on of
B
E
R
T
a
nd S
-
B
E
R
T
w
it
h r
e
s
pe
c
t
to
(
a
)
a
c
c
ur
a
c
y a
nd (
b)
s
im
il
a
r
it
y
s
c
or
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2252
-
8938
I
nt
J
A
r
ti
f
I
nt
e
ll
,
V
ol
. 14, No. 4, A
ugus
t
2025
:
3404
-
3411
3408
F
ig
ur
e
4
c
om
pa
r
e
s
B
E
R
T
a
nd
S
-
B
E
R
T
a
c
r
os
s
th
r
e
e
ke
y
m
e
tr
ic
s
of
ve
c
to
r
s
iz
e
(
F
ig
ur
e
4
(
a
)
)
,
s
c
r
e
e
ni
ng
ti
m
e
pe
r
r
e
s
um
e
(
F
ig
ur
e
4(
b)
)
a
nd
c
or
r
e
la
ti
on
w
it
h
j
ob
de
s
c
r
ip
ti
on
(
F
ig
ur
e
4(
c
)
)
.
T
he
ve
c
to
r
s
iz
e
,
s
how
n
on
th
e
le
f
tm
os
t
ve
r
ti
c
a
l
a
xi
s
la
be
le
d
“
V
e
c
to
r
s
iz
e
”
,
in
di
c
a
te
s
th
e
c
om
pa
c
tn
e
s
s
of
th
e
r
e
pr
e
s
e
nt
a
ti
on,
w
it
h
lo
w
e
r
va
lu
e
s
s
ugge
s
ti
ng
gr
e
a
te
r
c
om
pa
c
tn
e
s
s
.
T
h
e
c
om
pa
r
is
on
r
e
ve
a
ls
th
a
t
w
hi
le
B
E
R
T
r
e
qui
r
e
s
a
ve
c
to
r
s
iz
e
of
768,
S
-
B
E
R
T
r
e
qui
r
e
s
384
byt
e
s
of
ve
c
to
r
to
ge
ne
r
a
te
e
m
be
ddi
ngs
a
t
th
e
f
in
a
l
out
put
.
S
c
r
e
e
ni
ng
ti
m
e
,
m
e
a
s
ur
e
d
on
th
e
c
e
nt
e
r
ve
r
ti
c
a
l
a
xi
s
la
be
le
d
“
S
c
r
e
e
ni
ng
ti
m
e
(
s
e
c
pe
r
r
e
s
um
e
)
”
,
r
e
pr
e
s
e
nt
s
pr
oc
e
s
s
in
g s
p
e
e
d, w
he
r
e
l
ow
e
r
va
lu
e
s
s
ig
ni
f
y f
a
s
te
r
pr
oc
e
s
s
in
g.
(
a
)
(
b)
(
c
)
F
ig
ur
e
4. C
om
pa
r
is
on of
B
E
R
T
a
nd S
-
B
E
R
T
w
it
h r
e
s
pe
c
t
to
(
a
)
v
e
c
to
r
s
iz
e
,
(
b)
s
c
r
e
e
ni
ng t
im
e
i
n
s
e
c
onds
p
e
r
r
e
s
um
e
,
a
nd (
c
)
c
or
r
e
la
ti
on
T
he
c
ha
r
t
de
m
ons
tr
a
te
s
th
a
t
S
-
B
E
R
T
(
0.061
s
e
c
ond
s
)
s
ig
ni
f
ic
a
nt
ly
out
pe
r
f
or
m
s
B
E
R
T
(
1
s
e
c
ond)
in
te
r
m
s
of
s
c
r
e
e
ni
ng
r
e
s
um
e
s
.
C
or
r
e
la
ti
on
w
it
h
jo
b
de
s
c
r
ip
ti
on,
de
pi
c
te
d
on
th
e
r
ig
ht
m
os
t
ve
r
ti
c
a
l
a
xi
s
la
be
le
d
“
C
or
r
e
la
ti
on”
,
r
e
f
le
c
ts
th
e
s
tr
e
ngt
h
of
c
or
r
e
la
ti
on,
w
it
h
hi
ghe
r
va
lu
e
s
in
di
c
a
ti
ng
a
s
tr
onge
r
c
or
r
e
la
ti
on.
T
he
a
na
ly
s
is
il
lu
s
tr
a
te
s
th
a
t
S
-
B
E
R
T
(
0.383855)
e
xhi
bi
ts
a
s
tr
onge
r
c
or
r
e
la
ti
on
w
it
h
jo
b
de
s
c
r
ip
ti
ons
c
om
pa
r
e
d
to
B
E
R
T
(
0.1249)
.
I
n
s
um
m
a
r
y,
th
e
f
ig
u
r
e
s
ugge
s
ts
th
a
t
S
-
B
E
R
T
is
m
or
e
e
f
f
ic
ie
nt
th
a
n
B
E
R
T
f
or
a
ppl
ic
a
ti
ons
li
ke
r
e
s
um
e
s
c
r
e
e
ni
ng,
w
he
r
e
pr
oc
e
s
s
in
g
s
pe
e
d
is
c
r
uc
ia
l.
H
ow
e
ve
r
,
B
E
R
T
m
a
y
be
pr
e
f
e
r
r
e
d
if
a
m
or
e
c
om
pa
c
t
ve
c
to
r
s
i
z
e
is
ne
c
e
s
s
a
r
y,
s
uc
h
a
s
f
or
e
xe
c
ut
io
n
on
a
m
obi
le
de
vi
c
e
lo
c
a
ll
y.
D
e
s
pi
te
B
E
R
T
’
s
la
r
ge
r
ve
c
to
r
s
iz
e
of
768
c
om
pa
r
e
d
to
S
-
B
E
R
T
’
s
384,
S
-
B
E
R
T
of
f
e
r
s
f
a
s
te
r
s
c
r
e
e
ni
ng
ti
m
e
s
a
nd
de
m
ons
tr
a
te
s
a
s
tr
onge
r
c
or
r
e
la
ti
on w
it
h j
ob de
s
c
r
ip
ti
ons
.
F
ig
ur
e
5
il
lu
s
tr
a
te
s
th
e
r
e
s
um
e
m
a
tc
hi
ng
pr
oc
e
s
s
us
in
g
bot
h
B
E
R
T
a
nd
S
-
B
E
R
T
a
lg
or
it
hm
s
.
I
n
F
ig
ur
e
5
(
a
)
,
a
s
c
he
m
a
ti
c
ov
e
r
vi
e
w
of
th
e
r
e
s
um
e
m
a
tc
hi
ng
pr
o
c
e
s
s
is
pr
e
s
e
nt
e
d,
w
he
r
e
bot
h
th
e
r
e
s
um
e
a
nd
jo
b
de
s
c
r
ip
ti
on
a
r
e
ut
il
iz
e
d
f
or
e
m
be
ddi
ng
c
a
lc
ul
a
ti
on.
T
he
s
e
e
m
be
ddi
ngs
a
r
e
th
e
n
c
om
pa
r
e
d
us
in
g
c
os
in
e
s
im
il
a
r
it
y,
a
nd
th
e
r
e
s
um
e
s
a
r
e
r
a
nke
d a
c
c
or
di
ngl
y
ba
s
e
d
on
th
e
ir
s
im
il
a
r
it
y
s
c
or
e
s
.
I
n
F
ig
ur
e
5
(
b)
,
a
s
na
p
s
hot
of
th
e
out
put
s
c
r
e
e
n
dur
in
g
th
e
e
xe
c
ut
io
n
of
B
E
R
T
s
how
s
th
e
e
xt
r
a
c
ti
on
a
nd
pr
oc
e
s
s
in
g
of
s
ig
ni
f
ic
a
nt
ke
yw
or
ds
.
H
ow
e
ve
r
,
th
e
pr
oc
e
s
s
in
g
s
p
e
e
d
w
it
h
th
e
B
E
R
T
a
lg
or
it
hm
a
ppe
a
r
s
to
be
s
lo
w
e
r
.
I
n
c
ont
r
a
s
t,
F
ig
ur
e
5
(
c
)
s
how
c
a
s
e
s
a
s
na
p
s
hot
of
t
he
out
put
s
c
r
e
e
n a
f
te
r
t
he
c
om
pl
e
ti
on of
S
-
B
E
R
T
e
xe
c
ut
io
n.
A
ddi
ti
ona
ll
y,
our
e
va
lu
a
ti
on
f
oc
us
e
d
pr
im
a
r
il
y
on
qua
nt
it
a
ti
ve
m
e
tr
ic
s
s
uc
h
a
s
a
c
c
ur
a
c
y,
s
c
r
e
e
ni
ng
ti
m
e
,
a
nd
s
im
il
a
r
it
y
s
c
or
e
s
.
F
ut
ur
e
s
tu
di
e
s
c
oul
d
e
xpl
or
e
qua
li
ta
ti
ve
a
s
pe
c
ts
,
s
uc
h
a
s
th
e
in
te
r
pr
e
ta
bi
li
ty
a
nd
e
xpl
a
in
a
bi
li
ty
of
th
e
m
ode
l
s
’
de
c
i
s
io
n
-
m
a
ki
ng
pr
oc
e
s
s
e
s
,
t
o
e
ns
ur
e
f
a
ir
ne
s
s
a
nd
tr
a
ns
p
a
r
e
nc
y
in
th
e
r
e
c
r
ui
tm
e
nt
pr
oc
e
s
s
.
N
on
e
th
e
le
s
s
,
our
f
in
di
ngs
hol
d
s
ig
ni
f
ic
a
nt
im
pl
ic
a
ti
ons
f
or
r
e
c
r
ui
te
r
s
a
nd
H
R
pr
of
e
s
s
io
na
ls
. B
y l
e
ve
r
a
gi
ng S
-
B
E
R
T
’
s
s
tr
e
ngt
hs
, or
ga
ni
z
a
ti
ons
c
a
n s
tr
e
a
m
li
ne
t
he
ir
hi
r
in
g pr
oc
e
s
s
e
s
, l
e
a
di
ng
to
e
nha
nc
e
d
c
a
ndi
da
te
s
e
le
c
ti
on,
r
e
duc
e
d
ti
m
e
-
to
-
hi
r
e
,
a
nd
i
m
pr
ove
d
e
m
pl
oye
e
s
a
ti
s
f
a
c
ti
on
a
nd
r
e
te
nt
io
n.
F
ur
th
e
r
m
or
e
,
th
e
pot
e
nt
ia
l
f
o
r
m
or
e
a
c
c
ur
a
te
m
a
tc
he
s
be
twe
e
n c
a
ndi
da
te
s
a
nd
jo
b
de
s
c
r
ip
ti
ons
c
a
n
c
ont
r
ib
ut
e
to
a
be
tt
e
r
a
li
gnm
e
nt
of
s
ki
ll
s
a
nd
or
ga
ni
z
a
ti
ona
l
n
e
e
ds
,
ul
ti
m
a
te
ly
dr
iv
in
g
pr
oduc
ti
vi
ty
a
nd
or
ga
ni
z
a
ti
ona
l
s
uc
c
e
s
s
. W
hi
le
our
s
tu
dy
pr
ovi
de
s
va
lu
a
bl
e
in
s
ig
ht
s
in
to
th
e
c
o
m
pa
r
a
ti
ve
pe
r
f
or
m
a
nc
e
of
B
E
R
T
a
nd
S
-
B
E
R
T
f
or
r
e
s
um
e
s
c
r
e
e
ni
ng,
it
ha
s
li
m
it
a
ti
ons
in
te
r
m
s
o
f
th
e
da
ta
s
e
t
s
iz
e
a
nd
s
c
ope
.
F
ut
ur
e
r
e
s
e
a
r
c
h
c
oul
d
e
va
lu
a
t
e
th
e
s
e
m
ode
l
s
on
la
r
ge
r
,
m
or
e
di
v
e
r
s
e
da
t
a
s
e
t
s
a
c
r
o
s
s
va
r
io
us
f
ie
ld
s
a
nd
dom
a
in
s
.
A
ddi
ti
ona
ll
y,
e
xpl
or
in
g
e
ns
e
m
bl
e
te
c
hni
que
s
c
om
bi
ni
ng
B
E
R
T
a
nd
S
-
B
E
R
T
or
in
te
g
r
a
ti
ng
ot
he
r
s
ta
te
-
of
-
th
e
-
a
r
t
la
ngua
ge
m
ode
ls
c
oul
d
f
ur
th
e
r
e
nha
nc
e
a
c
c
ur
a
c
y.
T
h
e
im
pl
ic
a
ti
ons
a
r
e
s
ig
n
if
ic
a
nt
,
a
s
im
pl
e
m
e
nt
in
g
a
ut
om
a
te
d
r
e
s
um
e
s
c
r
e
e
ni
ng
ba
s
e
d
on
S
-
B
E
R
T
c
a
n
s
tr
e
a
m
li
ne
hi
r
in
g,
r
e
duc
e
bi
a
s
,
a
nd
pr
om
ot
e
w
or
kf
or
c
e
di
ve
r
s
it
y.
H
ow
e
ve
r
,
a
ddr
e
s
s
in
g
in
te
r
pr
e
ta
bi
li
ty
,
dom
a
in
-
s
pe
c
if
ic
te
r
m
in
ol
ogy,
a
nd
e
nd
-
to
-
e
nd
s
ys
te
m
in
te
gr
a
ti
on
r
e
m
a
in
s
c
r
uc
ia
l
f
or
pr
a
c
ti
c
a
l
de
pl
oym
e
nt
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
A
r
ti
f
I
nt
e
ll
I
S
S
N
:
2252
-
8938
C
om
par
in
g bi
di
r
e
c
ti
onal
e
n
c
ode
r
r
e
pr
e
s
e
nt
at
io
ns
f
r
om
t
r
an
s
fo
r
m
e
r
s
and
s
e
nt
e
nc
e
…
(
A
s
m
it
a D
e
s
hm
u
k
h
)
3409
(
a
)
(
b)
(
c
)
F
ig
ur
e
5. F
lo
w
c
ha
r
t
w
it
h output
of
(
a
)
g
e
ne
r
a
li
z
e
d a
ppr
oa
c
h f
or
r
e
s
um
e
m
a
tc
hi
ng
,
(
b)
o
ut
put
s
c
r
e
e
n
c
a
pt
ur
e
w
hi
le
B
E
R
T
i
s
r
unni
ng
,
a
nd (
c
)
o
ut
put
s
c
r
e
e
n
c
a
pt
ur
e
a
f
te
r
S
-
B
E
R
T
e
xe
c
ut
io
n c
om
pl
e
te
d
4.
C
O
N
C
L
U
S
I
O
N
I
n
to
da
y’
s
ta
le
nt
a
c
qui
s
it
io
n
la
nd
s
c
a
p
e
,
w
he
r
e
or
ga
ni
z
a
ti
on
s
f
a
c
e
th
e
da
unt
in
g
ta
s
k
of
e
f
f
ic
ie
nt
ly
s
c
r
e
e
ni
ng
num
e
r
ous
r
e
s
um
e
s
,
our
s
tu
dy
de
m
on
s
tr
a
te
s
th
e
s
upe
r
io
r
it
y
of
th
e
S
-
B
E
R
T
m
ode
l
ove
r
B
E
R
T
f
or
a
ut
om
a
te
d r
e
s
um
e
s
c
r
e
e
ni
ng. S
-
B
E
R
T
e
xc
e
ls
i
n a
c
c
ur
a
c
y, e
f
f
ic
i
e
nc
y, a
nd c
ont
e
xt
ua
l
unde
r
s
ta
ndi
ng, a
s
s
how
n
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2252
-
8938
I
nt
J
A
r
ti
f
I
nt
e
ll
,
V
ol
. 14, No. 4, A
ugus
t
2025
:
3404
-
3411
3410
by
it
s
hi
ghe
r
s
im
il
a
r
it
y
s
c
or
e
s
.
S
-
B
E
R
T
’
s
in
s
pi
te
of
ha
vi
ng
s
m
a
ll
e
r
f
e
a
tu
r
e
ve
c
to
r
s
iz
e
a
ll
ow
s
f
or
m
o
r
e
de
ta
il
e
d
r
e
s
um
e
a
na
ly
s
is
,
a
nd
it
s
f
a
s
te
r
s
c
r
e
e
ni
ng
ti
m
e
e
nha
nc
e
s
r
e
c
r
ui
tm
e
nt
pr
oduc
ti
vi
ty
.
I
ts
hi
ghe
r
a
c
c
ur
a
c
y
a
nd
s
upe
r
io
r
s
im
il
a
r
it
y
s
c
or
e
s
e
na
bl
e
be
tt
e
r
m
a
tc
hi
ng
of
c
a
ndi
da
te
s
w
it
h
jo
b
de
s
c
r
ip
ti
ons
,
le
a
di
ng
to
m
or
e
pr
e
c
is
e
c
a
ndi
da
te
s
e
le
c
ti
on.
W
hi
le
our
f
in
di
ngs
a
r
e
pr
om
is
in
g,
f
ur
th
e
r
r
e
s
e
a
r
c
h
is
ne
e
de
d
to
va
li
da
te
a
nd
e
xpa
nd
th
e
m
a
c
r
os
s
di
ve
r
s
e
da
ta
s
e
ts
a
nd
in
du
s
tr
ie
s
.
F
ut
ur
e
s
tu
d
ie
s
c
oul
d
a
l
s
o
e
xpl
or
e
qu
a
li
ta
ti
ve
a
s
pe
c
ts
s
uc
h
a
s
m
ode
l
in
te
r
pr
e
ta
bi
li
ty
to
e
ns
ur
e
f
a
ir
ne
s
s
in
th
e
r
e
c
r
ui
tm
e
nt
pr
oc
e
s
s
.
O
ur
r
e
s
e
a
r
c
h
s
ig
ni
f
ic
a
nt
ly
a
dva
nc
e
s
ta
le
nt
a
c
qui
s
it
io
n
by
de
m
ons
tr
a
ti
ng
th
e
pot
e
nt
ia
l
o
f
a
dva
nc
e
d
m
ode
ls
li
ke
S
-
B
E
R
T
to
im
p
r
ove
hi
r
in
g
e
f
f
ic
ie
nc
y.
U
ti
li
z
in
g
S
-
B
E
R
T
c
a
n
e
nha
nc
e
c
a
ndi
da
te
s
e
le
c
ti
on,
e
m
pl
oye
e
s
a
ti
s
f
a
c
ti
on,
a
nd
r
e
te
nt
io
n,
ul
ti
m
a
te
ly
dr
iv
in
g or
ga
ni
z
a
ti
ona
l
s
uc
c
e
s
s
.
A
C
K
N
O
WL
E
D
G
M
E
N
T
S
A
ut
hor
s
w
oul
d l
ik
e
t
o t
ha
nk c
ol
le
a
gue
s
a
t
H
V
P
M
’
s
C
ol
le
g
e
of
E
ngi
ne
e
r
in
g a
nd T
e
c
hnol
ogy.
F
U
N
D
I
N
G
I
N
F
O
R
M
A
T
I
O
N
A
ut
hor
s
s
ta
te
no f
undi
ng i
nvol
ve
d.
A
U
T
H
O
R
C
O
N
T
R
I
B
U
T
I
O
N
S
S
T
A
T
E
M
E
N
T
T
hi
s
jo
ur
na
l
us
e
s
th
e
C
ont
r
ib
ut
or
R
ol
e
s
T
a
xonomy
(
C
R
e
di
T
)
to
r
e
c
ogni
z
e
in
di
vi
dua
l
a
ut
hor
c
ont
r
ib
ut
io
ns
, r
e
duc
e
a
ut
hor
s
hi
p di
s
put
e
s
,
a
nd
f
a
c
il
it
a
te
c
ol
la
bo
r
a
ti
on.
N
am
e
o
f
A
u
t
h
or
C
M
So
Va
Fo
I
R
D
O
E
Vi
Su
P
Fu
A
s
m
it
a
D
e
s
hm
ukh
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
A
nj
a
li
R
a
ut
D
a
ha
ke
✓
✓
✓
✓
✓
C
:
C
onc
e
pt
ua
l
i
z
a
t
i
on
M
:
M
e
t
hodol
ogy
So
:
So
f
t
w
a
r
e
Va
:
Va
l
i
da
t
i
on
Fo
:
Fo
r
m
a
l
a
na
l
ys
i
s
I
:
I
nve
s
t
i
ga
t
i
on
R
:
R
e
s
our
c
e
s
D
:
D
a
t
a
C
ur
a
t
i
on
O
:
W
r
i
t
i
ng
-
O
r
i
gi
na
l
D
r
a
f
t
E
:
W
r
i
t
i
ng
-
R
e
vi
e
w
&
E
di
t
i
ng
Vi
:
Vi
s
ua
l
i
z
a
t
i
on
Su
:
Su
pe
r
vi
s
i
on
P
:
P
r
oj
e
c
t
a
dm
i
ni
s
t
r
a
t
i
on
Fu
:
Fu
ndi
ng a
c
qui
s
i
t
i
on
C
O
N
F
L
I
C
T
O
F
I
N
T
E
R
E
S
T
S
T
A
T
E
M
E
N
T
A
ut
hor
s
s
ta
te
no c
onf
li
c
t
of
i
nt
e
r
e
s
t.
D
A
T
A
A
V
A
I
L
A
B
I
L
I
T
Y
T
he
a
ut
hor
s
c
onf
ir
m
th
a
t
th
e
da
ta
th
a
t
s
uppor
t
th
e
f
in
di
ngs
of
th
is
s
tu
dy
a
r
e
a
va
il
a
bl
e
f
r
om
th
e
c
or
r
e
s
ponding a
ut
hor
, [
A
D
]
, upon r
e
a
s
ona
bl
e
r
e
que
s
t.
R
E
F
E
R
E
N
C
E
S
[
1]
S
.
Z
u
a
nd
X
.
W
a
ng,
“
R
e
s
um
e
i
nf
or
m
a
t
i
on
e
xt
r
a
c
t
i
on
w
i
t
h
a
nove
l
t
e
xt
bl
o
c
k
s
e
gm
e
nt
a
t
i
on
a
l
gor
i
t
hm
,”
I
nt
e
r
nat
i
onal
J
ou
r
nal
on
N
at
ur
al
L
anguage
C
om
put
i
ng
, vol
. 8, no. 5, pp. 29
–
48, O
c
t
. 2019, doi
:
10.5121
/
i
j
nl
c
.2019.8503.
[
2]
W
.
L
a
ks
ha
n,
K
.
P
r
a
buddhi
,
E
.
B
a
nda
r
a
,
a
nd
R
.
R
.
P
.
D
e
Z
oys
a
,
“
R
e
vol
ut
i
oni
z
i
ng
t
he
hi
r
i
ng
pr
oc
e
s
s
w
i
t
h
a
ut
om
a
t
e
d
e
va
l
ua
t
i
on
a
nd
be
ha
vi
or
a
l
a
na
l
ys
i
s
-
i
nt
e
l
l
i
hi
r
e
,”
I
nt
e
r
nat
i
onal
R
e
s
e
ar
c
h
J
our
nal
of
I
nn
ov
at
i
ons
i
n
E
ngi
ne
e
r
i
ng
and
T
e
c
hnol
ogy
,
vol
.
7,
no. 11, pp. 307
–
314, 2023, doi
:
10.47001/
I
R
J
I
E
T
/
2023.711042
.
[
3]
B
.
W
a
ng
a
nd
C
.
-
C
.
J
.
K
uo,
“
S
B
E
R
T
-
W
K
:
a
s
e
nt
e
nc
e
e
m
be
ddi
ng
m
e
t
hod
by
di
s
s
e
c
t
i
ng
B
E
R
T
-
ba
s
e
d
w
or
d
m
ode
l
s
,”
I
E
E
E
/
A
C
M
T
r
ans
ac
t
i
ons
on A
udi
o, Spe
e
c
h, and L
anguage
P
r
oc
e
s
s
i
ng
, vol
. 28, pp. 2146
–
2157, 2020, doi
:
10.1109/
T
A
S
L
P
.2020.3008390.
[
4]
N
.
T
h
a
k
u
r
,
N
.
R
e
i
m
e
r
s
,
J
.
D
a
xe
n
be
r
g
e
r
,
a
nd
I
.
G
u
r
e
v
yc
h,
“
A
u
g
m
e
nt
e
d
S
B
E
R
T
:
d
a
t
a
a
ug
m
e
n
t
a
t
i
o
n
m
e
t
ho
d
f
o
r
i
m
p
r
ov
i
ng
b
i
-
e
nc
o
de
r
s
f
o
r
p
a
i
r
w
i
s
e
s
e
nt
e
n
c
e
s
c
or
i
ng
t
a
s
ks
,
”
i
n
P
r
o
c
e
e
d
i
n
gs
o
f
t
h
e
20
21
C
on
f
e
r
e
n
c
e
o
f
t
h
e
N
or
t
h
A
m
e
r
i
c
a
n
C
h
ap
t
e
r
o
f
t
he
A
s
s
o
c
i
a
t
i
o
n
f
or
C
om
p
ut
a
t
i
on
a
l
L
i
n
g
ui
s
t
i
c
s
:
H
u
m
a
n
L
an
g
ua
ge
T
e
c
h
no
l
og
i
e
s
,
2
02
1
,
pp
.
2
96
–
31
0
,
do
i
:
1
0.
1
86
53
/
v1
/
20
21
.
na
a
c
l
-
m
a
i
n.
2
8.
[
5]
V
.
Y
a
da
v
a
nd
S
.
B
e
t
ha
r
d,
“
A
s
ur
ve
y
on
r
e
c
e
nt
a
dva
n
c
e
s
i
n
na
m
e
d
e
nt
i
t
y
r
e
c
o
gni
t
i
on
f
r
om
de
e
p
l
e
a
r
ni
ng
m
ode
l
s
,
”
C
o
m
put
at
i
on
and L
anguage
, vol
. 7, no. 1, pp. 41525
–
41550, 2019.
[
6]
J
.
S
e
o,
S
.
L
e
e
,
L
.
L
i
u,
a
nd
W
.
C
hoi
,
“
T
A
-
S
B
E
R
T
:
t
oke
n
a
t
t
e
nt
i
on
s
e
nt
e
nc
e
-
B
E
R
T
f
o
r
i
m
pr
ov
i
ng
s
e
nt
e
nc
e
r
e
pr
e
s
e
nt
a
t
i
on,”
I
E
E
E
A
c
c
e
s
s
, vol
. 10, pp. 39119
–
39128, 2022, doi
:
10.1109/
A
C
C
E
S
S
.2022.3164769.
[
7]
Y
.
S
a
nt
a
nde
r
-
C
r
uz
,
S
.
S
a
l
a
z
a
r
-
C
ol
or
e
s
,
W
.
J
.
P
a
r
e
de
s
-
G
a
r
c
í
a
,
H
.
G
ue
ndul
a
i
n
-
A
r
e
na
s
,
a
nd
S
.
T
ova
r
-
A
r
r
i
a
ga
,
“
S
e
m
a
nt
i
c
f
e
a
t
ur
e
e
xt
r
a
c
t
i
on us
i
ng S
B
E
R
T
f
or
de
m
e
nt
i
a
de
t
e
c
t
i
on,”
B
r
ai
n Sc
i
e
nc
e
s
, vol
. 12, no. 2,
F
e
b. 2022, doi
:
10.3390/
br
a
i
ns
c
i
12020270.
[
8]
J
.
M
ul
l
i
s
,
C
.
C
he
n,
B
.
M
or
kos
,
a
nd
S
.
F
e
r
gus
on,
“
D
e
e
p
ne
ur
a
l
ne
t
w
or
k
s
i
n
na
t
ur
a
l
l
a
ngua
ge
pr
oc
e
s
s
i
ng
f
or
c
l
a
s
s
i
f
yi
ng
r
e
qui
r
e
m
e
nt
s
by
or
i
gi
n
a
nd
f
unc
t
i
ona
l
i
t
y:
a
n
a
ppl
i
c
a
t
i
on
of
B
E
R
T
i
n
s
ys
t
e
m
r
e
qui
r
e
m
e
nt
s
,”
J
our
nal
of
M
e
c
hani
c
al
D
e
s
i
gn
,
vol
. 146, no. 4, A
pr
. 2024, doi
:
10.1115/
1.4063764.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
A
r
ti
f
I
nt
e
ll
I
S
S
N
:
2252
-
8938
C
om
par
in
g bi
di
r
e
c
ti
onal
e
n
c
ode
r
r
e
pr
e
s
e
nt
at
io
ns
f
r
om
t
r
an
s
fo
r
m
e
r
s
and
s
e
nt
e
nc
e
…
(
A
s
m
it
a D
e
s
hm
u
k
h
)
3411
[
9]
Y
. K
i
m
e
t
al
.
, “
A
pr
e
-
t
r
a
i
ne
d B
E
R
T
f
or
K
or
e
a
n m
e
di
c
a
l
na
t
ur
a
l
l
a
ngua
ge
pr
oc
e
s
s
i
ng,”
Sc
i
e
nt
i
f
i
c
R
e
po
r
t
s
, vol
. 12, no.
1, 2022, doi
:
10.1038/
s
41598
-
022
-
17806
-
8.
[
10]
Y
.
G
u
e
t
al
.
,
“
D
om
a
i
n
-
s
pe
c
i
f
i
c
l
a
ngua
ge
m
ode
l
pr
e
t
r
a
i
ni
ng
f
or
bi
om
e
di
c
a
l
na
t
ur
a
l
l
a
ngua
ge
pr
oc
e
s
s
i
ng,”
A
C
M
T
r
an
s
ac
t
i
ons
o
n
C
om
put
i
ng f
or
H
e
al
t
hc
ar
e
, vol
. 3, no. 1, pp. 1
–
23, J
a
n. 2022, doi
:
10.1145/
3458754.
[
11]
R
.
G
u
pt
a
,
“
B
i
d
i
r
e
c
t
i
on
a
l
e
n
c
o
de
r
s
t
o
s
t
a
t
e
-
of
-
t
he
-
a
r
t
:
a
r
e
v
i
e
w
of
B
E
R
T
a
n
d
i
t
s
t
r
a
n
s
f
o
r
m
a
t
i
v
e
i
m
pa
c
t
on
na
t
u
r
a
l
l
a
ng
ua
ge
p
r
o
c
e
s
s
i
ng
,”
I
n
f
or
m
a
t
i
c
s
E
c
on
om
i
c
s
M
an
a
ge
m
e
n
t
,
v
ol
.
3
,
no
.
1
,
p
p.
3
1
1
–
32
0
, M
a
r
.
20
24
,
d
oi
:
1
0.
4
78
1
3/
2
78
2
-
5
28
0
-
20
24
-
3
-
1
-
0
31
1
-
0
3
20
.
[
12]
A
.
Ö
z
ç
i
f
t
,
K
.
A
ka
r
s
u,
F
.
Y
um
uk,
a
nd
C
.
S
öyl
e
m
e
z
,
“
A
dva
nc
i
ng
na
t
u
r
a
l
l
a
ngua
ge
pr
oc
e
s
s
i
ng
(
N
L
P
)
a
ppl
i
c
a
t
i
ons
of
m
or
phol
ogi
c
a
l
l
y
r
i
c
h l
a
ngua
ge
s
w
i
t
h
bi
di
r
e
c
t
i
ona
l
e
nc
ode
r
r
e
p
r
e
s
e
nt
a
t
i
ons
f
r
om
t
r
a
ns
f
or
m
e
r
s
(
B
E
R
T
)
:
a
n
e
m
pi
r
i
c
a
l
c
a
s
e
s
t
udy
f
or
T
ur
ki
s
h,”
A
ut
om
at
i
k
a
, vol
. 62, no. 2, pp. 226
–
238, A
pr
. 2021, doi
:
10.1080/
000
51144.2021.1922150.
[
13]
B
. M
i
n
e
t
al
.
,
“
R
e
c
e
nt
a
dva
nc
e
s
i
n na
t
ur
a
l
l
a
ngu
a
ge
pr
oc
e
s
s
i
ng vi
a
l
a
r
ge
pr
e
-
t
r
a
i
ne
d l
a
ngua
ge
m
ode
l
s
:
a
s
ur
ve
y,”
A
C
M
C
om
put
i
ng
Sur
v
e
y
s
, vol
. 56, no. 2, pp. 1
–
40, F
e
b. 2024, doi
:
10.1145/
3605943.
[
14]
A
.
G
upt
a
,
A
.
C
ha
dha
,
a
nd
V
.
T
e
w
a
r
i
,
“
A
na
t
u
r
a
l
l
a
ngua
ge
pr
oc
e
s
s
i
ng
m
o
de
l
on
B
E
R
T
a
nd
Y
A
K
E
t
e
c
hni
que
f
or
ke
yw
or
d
e
xt
r
a
c
t
i
on on s
us
t
a
i
na
bi
l
i
t
y r
e
por
t
s
,”
I
E
E
E
A
c
c
e
s
s
, vol
. 12, pp. 7942
–
7951, 2024, doi
:
10.1109/
A
C
C
E
S
S
.2024.3352742.
[
15]
R
.
V
e
da
pr
a
dha
,
R
.
H
a
r
i
ha
r
a
n,
a
nd
R
.
S
hi
va
ka
m
i
,
“
A
r
t
i
f
i
c
i
a
l
i
nt
e
l
l
i
ge
nc
e
:
a
t
e
c
hnol
ogi
c
a
l
pr
ot
ot
ype
i
n
r
e
c
r
ui
t
m
e
nt
,”
J
our
nal
of
Se
r
v
i
c
e
Sc
i
e
n
c
e
and M
anage
m
e
nt
, vol
. 12, no. 3, pp. 382
–
390, 2019, doi
:
10.42
36/
j
s
s
m
.2019.123026.
[
16]
B
. E
l
e
, A
. E
l
e
, a
nd
F
. A
ga
ba
,
“
I
nt
e
l
l
i
ge
nt
-
ba
s
e
d j
ob
a
ppl
i
c
a
nt
s
’
a
s
s
e
s
s
m
e
nt
a
nd
r
e
c
r
ui
t
m
e
nt
s
ys
t
e
m
,”
B
r
i
t
i
s
h J
ou
r
nal
of
C
om
put
e
r
,
N
e
t
w
or
k
i
ng and I
nf
or
m
at
i
on T
e
c
hnol
og
y
, vol
. 6, no. 1, pp. 25
–
46, J
a
n. 2024, doi
:
10.52589/
B
J
C
N
I
T
-
I
S
O
B
A
M
P
D
.
[
17]
A
.
U
j
l
a
ya
n,
S
.
B
ha
t
t
a
c
ha
r
ya
,
a
nd
S
ona
ks
hi
,
“
A
m
a
c
hi
ne
l
e
a
r
ni
ng
-
ba
s
e
d
A
I
f
r
a
m
e
w
or
k
t
o
opt
i
m
i
z
e
t
he
r
e
c
r
ui
t
m
e
nt
s
c
r
e
e
ni
ng
pr
oc
e
s
s
,”
I
nt
e
r
nat
i
onal
J
our
nal
of
G
l
obal
B
us
i
ne
s
s
and
C
om
pe
t
i
t
i
v
e
n
e
s
s
,
vol
.
18,
no.
S
1,
pp.
38
–
53,
J
a
n.
202
4
,
doi
:
10.1007/
s
42943
-
023
-
00086
-
y.
[
18]
B
.
K
i
nge
,
S
.
M
a
ndha
r
e
,
P
.
C
ha
v
a
n,
a
nd
S
.
C
ha
w
a
r
e
,
“
R
e
s
um
e
s
c
r
e
e
ni
ng
us
i
n
g
m
a
c
hi
ne
l
e
a
r
ni
ng
a
nd
N
L
P
:
a
pr
opo
s
e
d
s
ys
t
e
m
,
”
I
nt
e
r
nat
i
onal
J
our
nal
of
Sc
i
e
nt
i
f
i
c
R
e
s
e
ar
c
h
i
n
C
om
put
e
r
Sc
i
e
nc
e
,
E
ngi
ne
e
r
i
ng
and
I
nf
or
m
at
i
on
T
e
c
hnol
ogy
,
pp.
253
–
258,
A
pr
. 2022, doi
:
10.32628/
C
S
E
I
T
228240.
[
19]
V
.
R
a
j
a
t
h
,
R
.
T
.
F
a
r
e
e
d
,
a
nd
S
.
K
a
g
a
nu
r
m
a
t
h,
“
R
e
s
u
m
e
c
l
a
s
s
i
f
i
c
a
t
i
o
n
a
nd
r
a
nk
i
ng
us
i
ng
K
N
N
a
nd
c
os
i
ne
s
i
m
i
l
a
r
i
t
y
,”
I
nt
e
r
nat
i
o
na
l
J
our
n
al
o
f
E
n
gi
ne
e
r
i
ng
R
e
s
e
ar
c
h &
T
e
c
h
no
l
o
gy
,
vo
l
.
10
,
no.
8
, p
p.
19
2
–
195
,
202
1,
do
i
:
10
.17
57
7/
I
J
E
R
T
V
1
0I
S
0
80
057
.
[
20]
D
.
P
a
ndi
t
a
,
“
T
a
l
e
nt
a
c
qui
s
i
t
i
on:
a
n
a
l
ys
i
s
of
di
gi
t
a
l
hi
r
i
ng
i
n
or
ga
ni
z
a
t
i
ons
,”
SA
M
V
A
D
:
SI
B
M
P
une
R
e
s
e
ar
c
h
J
our
nal
,
vol
.
18
,
pp. 66
–
72, S
e
p. 2019, doi
:
10.53739/
s
a
m
va
d/
2019/
v18/
146242.
[
21]
M
. M
okoa
t
l
e
, V
.
M
a
r
i
va
t
e
, D
. M
a
pi
ye
, R
.
B
or
nm
a
n, a
nd V
. M
. H
a
ye
s
, “
A
r
e
vi
e
w
a
nd c
om
pa
r
a
t
i
ve
s
t
udy of
c
a
nc
e
r
de
t
e
c
t
i
on us
i
ng
m
a
c
hi
ne
l
e
a
r
ni
ng:
S
B
E
R
T
a
nd
S
i
m
C
S
E
a
ppl
i
c
a
t
i
on,”
B
M
C
B
i
oi
nf
or
m
at
i
c
s
,
vo
l
.
24,
no.
1,
M
a
r
.
2023,
doi
:
10.1186/
s
12859
-
023
-
05235
-
x.
[
22]
L
.
G
e
or
ge
a
nd
P
.
S
um
a
t
hy,
“
A
n
i
nt
e
gr
a
t
e
d
c
l
us
t
e
r
i
ng
a
nd
B
E
R
T
f
r
a
m
e
w
or
k
f
or
i
m
pr
ove
d
t
opi
c
m
ode
l
i
ng,”
I
nt
e
r
nat
i
onal
J
our
nal
of
I
nf
or
m
at
i
on T
e
c
hnol
ogy
, vol
. 15, no. 4, pp. 2187
–
2195, A
pr
. 2023, doi
:
10.1007/
s
41870
-
023
-
01268
-
w.
[
23]
J
.
W
a
ng
e
t
al
.
,
“
U
t
i
l
i
z
i
ng
B
E
R
T
f
or
i
nf
or
m
a
t
i
on
r
e
t
r
i
e
va
l
:
s
ur
ve
y,
a
ppl
i
c
a
t
i
ons
,
r
e
s
our
c
e
s
,
a
nd
c
ha
l
l
e
nge
s
,”
A
C
M
C
om
put
i
n
g
Sur
v
e
y
s
, vol
. 56, no. 7, pp. 1
–
33, J
ul
. 2024, doi
:
10.1145/
3648471.
[
24]
B
.
J
ua
r
t
o
a
nd
A
.
S
.
G
i
r
s
a
ng,
“
N
e
ur
a
l
c
ol
l
a
bor
a
t
i
ve
w
i
t
h
s
e
nt
e
nc
e
B
E
R
T
f
or
ne
w
s
r
e
c
om
m
e
nde
r
s
y
s
t
e
m
,”
J
O
I
V
:
I
nt
e
r
nat
i
onal
J
our
nal
on I
nf
or
m
at
i
c
s
V
i
s
ual
i
z
at
i
on
, vol
. 5, no. 4, D
e
c
. 2021, doi
:
10.30630/
j
o
i
v.5.4.678.
[
25]
N
.
R
e
i
m
e
r
s
a
nd
I
.
G
ur
e
vyc
h,
“
S
e
nt
e
nc
e
-
B
E
R
T
:
s
e
nt
e
nc
e
e
m
be
ddi
ngs
u
s
i
ng
s
i
a
m
e
s
e
B
E
R
T
-
ne
t
w
or
ks
,”
E
M
N
L
P
-
I
J
C
N
L
P
2019
-
2019
C
onf
e
r
e
nc
e
on
E
m
pi
r
i
c
al
M
e
t
hods
i
n
N
at
ur
al
L
anguage
P
r
oc
e
s
s
i
ng
and
9t
h
I
nt
e
r
nat
i
onal
J
oi
nt
C
onf
e
r
e
nc
e
on
N
at
ur
a
l
L
anguage
P
r
oc
e
s
s
i
ng, P
r
oc
e
e
di
ngs
of
t
he
C
onf
e
r
e
nc
e
, pp. 3982
–
3992, 2019, d
oi
:
10.18653/
v1/
d19
-
1410.
B
I
O
G
R
A
P
H
I
E
S
O
F
A
U
T
H
O
R
S
Asmita
Deshmukh
is
Assistant
Professor
at
K
.
J
.
Somaiy
a
Institute
of
Technology,
Mumbai,
India.
She
is
currently
pursuing
Ph.D.
in
Computer
Science
Engineering
from
Sant
Gadge
Baba
Amravati
University
.
She
o
btained
her
Master
of
Engineering
in
Computer
Enginee
ring
from
the
University
of
M
umbai
and
Bachelor
o
f
Engineering
in
Computer
Science
Enginee
ring
from
Amrava
ti
Unive
rsity
in
1996
and
2012.
Her
main
area
of
research
interests
is
natural
language
processing,
dee
p
learning,
data
science,
and
big data
. She can be contacted a
t email: asmitadeshmukh7@
gmail.com
.
Anjali
Raut
Dahake
is
Professor
and
In
-
charge
Principal
at
HVP
M’s
College
o
f
Engineering
and
Technology,
Amravati
,
India.
She
completed
her
Ph.
D.
in
Computer
Scienc
e
Engineering
from
Sant
Gadge
Baba
Amravati
University
and
o
btained
her
Master
of
Engineering
in
Computer
Science
Enginee
ring
and
Bachelo
r
of
E
ngineer
ing
in
Computer
Scienc
e
Engine
ering
from
Amrav
ati
Univer
sity
in
1994
and
2013.
H
er
main
area
of
resea
rch
interests is
data mining
. She can be contacted a
t email: anjali_dahake@rediffmail.com
.
Evaluation Warning : The document was created with Spire.PDF for Python.