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, No. 6, D
e
c
e
m
be
r
2025
, pp.
4600
~
4613
I
S
S
N
:
2252
-
8938
,
D
O
I
:
10.11591/
ij
a
i.
v
14
.i
6
.pp
4600
-
4613
4600
Jou
r
n
al
h
om
e
page
:
ht
tp
:
//
ij
ai
.
ia
e
s
c
or
e
.c
om
S
e
m
an
t
i
c
se
ar
c
h
-
e
n
h
an
c
e
d
h
e
al
t
h
c
ar
e
c
h
at
b
ot
f
or
h
osp
i
t
al
i
n
f
or
m
at
i
on
m
an
age
m
e
n
t
sys
t
e
m
u
si
n
g
ve
c
t
or
d
at
ab
ase
an
d
t
r
an
sf
or
m
e
r
m
od
e
l
s
E
r
d
a G
u
s
li
n
ar
P
e
r
d
an
a
1,
2
, A
r
ya A
d
h
i
N
u
gr
ah
a
3
1
S
of
t
w
a
r
e
E
ngi
ne
e
r
i
ng A
ppl
i
c
a
t
i
ons
S
t
udy P
r
og
r
a
m
, S
c
hool
of
A
ppl
i
e
d S
c
i
e
nc
e
, T
e
l
kom
U
ni
ve
r
s
i
t
y
, B
a
ndung, I
ndone
s
i
a
2
C
e
nt
e
r
of
E
xc
e
l
l
e
nc
e
f
or
I
ns
pi
r
i
ng D
i
gi
t
a
l
T
r
a
ns
f
or
m
a
t
i
on f
o
r
S
oc
i
a
l
I
nnova
t
i
o
n (
I
ns
P
i
R
o)
, R
e
s
e
a
r
c
h I
ns
t
i
t
ut
e
of
S
us
t
a
i
na
bl
e
S
oc
i
e
t
y,
T
e
l
kom
U
ni
ve
r
s
i
t
y, B
a
ndung, I
ndone
s
i
a
3
P
r
oduc
t
D
e
ve
l
opm
e
nt
a
nd
T
e
c
hnol
ogy D
i
vi
s
i
on, M
e
dxa
, B
a
ndung, I
ndone
s
i
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
M
a
r
5, 2025
R
e
vi
s
e
d
O
c
t
21
, 2025
A
c
c
e
pt
e
d
N
ov 8, 2025
Healthcare
chatbots
are
increasingly
used
to
assist
hospital
staff,
ye
t
most
existin
g
systems
rely
on
rule
-
based
or
generic
machine
learning
(ML)
approaches
that
lack
the
abilit
y
to
comprehend
natural
language
q
ueries,
while
proprietary
deep
lea
rning
systems
often
incur
high
licensing
costs.
This
work
addresses
this
gap
by
proposing
a
cost
-
effective
and
s
calable
semantic
vector
retrieval
solution
for
user
intent
recognition
in
a
h
ospital
information
manageme
nt
system
(
HIMS
)
helpdesk
chatbot.
The
MPNet
-
based
transforme
r
model
is
employed
to
convert
user
inquirie
s
and
predefined
intents
into
feature
vectors,
enabling
highly
accurate
natural
language
understand
ing
through
cosine
similarity
retrieva
l
wit
hin
a
dedicated vector data
base. The
proposed vector
search method was
va
lidated
via
an
ablation
study,
a
chieving
an
accuracy
of
0.70
for
intent
recog
nition,
which demonstrates a s
ignificant
performance gain
of 28.0 percentage
points
over
a
traditional
keyword
-
based
search
baseline.
Usability
testing
across
develo
per
and
doctor
groups
yielded
an
average
score
of
7.78
on
a
1
0
-
point
Likert
scale.
This
study
concludes
that
integrating
semantic
vector
re
trieval
with
a
vector
database
is
highly
effective
for
recognizing
specialized
c
linical
intents,
offering
a
more
accura
te
solution
that
significantly
reduc
es
the
manual help
desk worklo
ad and e
nhances 2
4
-
hour assistance in healthcare
.
K
e
y
w
o
r
d
s
:
H
e
a
lt
hc
a
r
e
c
ha
tb
ot
I
nt
e
nt
r
e
c
ogni
ti
on
M
ul
ti
li
ngua
l
e
m
be
ddi
ngs
S
e
m
a
nt
ic
s
e
a
r
c
h
V
e
c
to
r
da
ta
ba
s
e
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
:
E
r
da
G
us
li
na
r
P
e
r
da
na
S
of
twa
r
e
E
ngi
ne
e
r
in
g A
ppl
ic
a
ti
ons
S
tu
dy P
r
og
r
a
m
, S
c
hool
of
A
ppl
ie
d S
c
ie
nc
e
, T
e
lk
om
U
ni
ve
r
s
it
y
B
a
ndung, W
e
s
t
J
a
va
P
r
ovi
nc
e
, I
ndone
s
i
a
E
m
a
il
:
e
r
da
@
te
lk
om
uni
ve
r
s
it
y.a
c
.i
d
1.
I
N
T
R
O
D
U
C
T
I
O
N
T
he
hos
pi
ta
l
in
f
or
m
a
ti
on
m
a
na
ge
m
e
nt
s
ys
te
m
(
H
I
M
S
)
is
a
m
u
lt
if
a
c
e
te
d
in
f
or
m
a
ti
on
s
ys
te
m
due
t
o
it
s
e
nga
ge
m
e
nt
w
it
h
s
e
ve
r
a
l
us
e
r
gr
oups
f
r
om
bot
h
in
te
r
na
l
a
nd
e
xt
e
r
na
l
hos
pi
ta
l
e
nvi
r
onm
e
nt
s
.
T
he
la
r
ge
num
be
r
of
us
e
r
gr
oups
d
r
iv
e
s
th
e
ne
e
d
f
or
a
c
a
pa
bl
e
he
lp
de
s
k
s
ys
te
m
to
s
e
r
ve
th
e
ne
e
ds
of
H
I
M
S
us
e
r
s
opt
im
a
ll
y.
C
ha
tb
ot
s
m
a
y
b
e
in
te
gr
a
te
d
in
to
th
e
H
I
M
S
he
lp
d
e
s
k
s
ys
te
m
,
pr
ovi
di
ng
a
ut
om
a
ti
on
a
nd
24
-
hour
a
s
s
is
t
a
nc
e
, a
lo
ng
s
id
e
c
os
t
r
e
duc
ti
on
a
nd t
he
c
a
pa
c
it
y t
o m
a
na
ge
a
s
ubs
ta
nt
ia
l
vol
um
e
of
c
u
s
to
m
e
r
s
[
1]
–
[
3]
.
C
ha
t
bot
s
m
a
y
ty
pi
c
a
ll
y
be
c
ons
tr
uc
t
e
d
us
in
g
r
ul
e
-
b
a
s
e
d
o
r
m
a
c
hi
n
e
le
a
r
n
in
g
(
M
L
)
a
p
p
r
oa
c
he
s
[
4]
.
R
ul
e
-
ba
s
e
d
c
h
a
tb
ot
s
a
r
e
th
e
m
os
t
p
r
e
va
le
n
t
s
o
r
t
o
f
c
ha
tb
ot
s
u
ti
li
z
e
d
i
n
th
e
c
o
r
p
or
a
te
s
e
c
to
r
n
ow
a
da
ys
[
5
]
.
R
ul
e
-
ba
s
e
d
c
ha
tb
ot
s
a
r
e
c
ha
tb
o
ts
th
a
t
a
r
e
de
ve
lo
pe
d
w
it
h
a
f
ix
e
d
d
a
ta
ba
s
e
o
f
que
s
ti
on
-
a
n
d
-
a
ns
w
e
r
pa
i
r
s
,
w
hi
c
h
r
e
s
ul
ts
in
a
s
i
gn
if
ic
a
nt
l
im
it
a
ti
on
:
th
e
i
r
in
a
bi
li
ty
to
c
o
m
p
r
e
he
nd
us
e
r
i
nqu
i
r
ie
s
e
xp
r
e
s
s
e
d
in
na
tu
r
a
l
la
ng
ua
g
e
[
5
]
.
C
onv
e
r
s
e
ly
,
a
l
th
o
ug
h
M
L
-
ba
s
e
d
c
ha
tb
o
ts
l
ik
e
C
ha
t
G
P
T
o
f
f
e
r
s
i
m
i
la
r
be
ne
f
it
s
,
th
e
i
r
pot
e
n
ti
a
l
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
Se
m
ant
ic
s
e
ar
c
h
-
e
nhanc
e
d he
al
th
c
a
r
e
c
hat
bot
f
or
ho
s
pi
ta
l
in
fo
r
m
at
io
n
…
(
E
r
da G
us
li
nar
P
e
r
dana
)
4601
to
ge
n
e
r
a
te
a
r
ti
f
ic
i
a
l
ha
ll
uc
i
na
t
io
ns
m
a
ke
s
t
he
m
l
e
s
s
a
pp
r
op
r
i
a
te
f
o
r
i
nt
e
gr
a
t
io
n
i
nt
o
he
l
p
de
s
k
s
ys
t
e
m
s
[
6]
.
A
lk
a
is
s
i
a
n
d
M
c
F
a
r
l
a
ne
[
7
]
e
lu
c
id
a
te
s
th
a
t
a
r
t
if
ic
i
a
l
ha
ll
uc
i
na
t
io
ns
a
r
e
th
e
r
e
s
u
lt
o
f
m
a
c
hi
ne
s
,
s
uc
h
a
s
c
ha
t
bo
ts
,
ge
ne
r
a
t
in
g
s
e
ns
o
r
y
e
xpe
r
i
e
nc
e
s
th
a
t
a
ppe
a
r
ge
nu
i
ne
bu
t
do
n
ot
c
or
r
e
s
pon
d
t
o
a
ny
r
e
a
l
-
w
o
r
l
d
in
p
ut
.
T
h
e
c
ha
t
bo
t'
s
a
r
t
if
ic
ia
l
ha
ll
uc
i
na
ti
o
n
ou
tp
u
t
is
i
na
de
q
ua
te
f
o
r
he
l
pde
s
k
s
ys
te
m
s
th
a
t
ne
e
d
p
r
e
c
is
e
r
e
s
p
ons
e
s
.
C
om
m
e
r
c
ia
l
r
u
le
-
ba
s
e
d
c
ha
tb
o
t
f
r
a
m
e
w
o
r
ks
s
u
c
h
a
s
G
o
og
le
D
ia
lo
g
f
lo
w
,
I
B
M
W
a
ts
on
,
a
nd
R
a
s
a
p
la
t
f
or
m
ha
ve
m
a
de
s
t
r
i
de
s
in
a
d
dr
e
s
s
in
g
th
e
l
im
it
a
ti
ons
of
ba
s
ic
r
u
le
-
ba
s
e
d s
ys
te
m
s
by
o
f
f
e
r
in
g
m
o
r
e
s
oph
is
t
ic
a
t
e
d
c
a
p
a
bi
l
it
ie
s
.
H
o
w
e
ve
r
,
th
e
s
e
s
o
lu
ti
ons
t
yp
ic
a
l
l
y
o
pe
r
a
t
e
a
s
c
lo
u
d
-
ba
s
e
d
s
e
r
vi
c
e
s
a
n
d
o
f
te
n
in
c
u
r
l
ic
e
ns
in
g c
os
ts
.
T
r
a
di
ti
ona
l
ke
yw
or
d
-
ba
s
e
d
s
e
a
r
c
h
m
e
th
od
s
s
tr
uggl
e
to
in
te
r
p
r
e
t
th
e
nua
nc
e
s
of
na
tu
r
a
l
la
ngua
g
e
,
w
hi
c
h
li
m
it
s
th
e
ir
us
e
f
ul
ne
s
s
in
c
onve
r
s
a
ti
ona
l
a
r
ti
f
ic
ia
l
in
te
ll
ig
e
nc
e
(
AI
)
.
T
o
a
ddr
e
s
s
th
i
s
,
m
ode
r
n
s
ys
te
m
s
a
dopt
s
e
m
a
nt
ic
s
e
a
r
c
h
pow
e
r
e
d
by
de
ns
e
r
e
tr
ie
va
l
(
D
R
)
te
c
hni
que
s
[
8]
.
I
n
D
R
,
bot
h
us
e
r
que
r
ie
s
a
nd
in
te
nt
r
e
pr
e
s
e
nt
a
ti
ons
a
r
e
t
r
a
ns
f
or
m
e
d i
nt
o hi
gh
-
di
m
e
ns
io
na
l
ve
c
to
r
e
m
be
ddi
ngs
us
in
g de
e
p l
e
a
r
ni
ng mode
ls
.
M
os
t
D
R
s
ys
te
m
s
us
e
bi
-
e
nc
ode
r
a
r
c
hi
te
c
tu
r
e
s
[
9]
s
uc
h
a
s
t
he
M
P
N
e
t
-
ba
s
e
d
m
ode
l
us
e
d
in
th
is
s
tu
dy
w
he
r
e
t
he
que
r
y a
nd i
nt
e
nt
a
r
e
e
nc
od
e
d s
e
pa
r
a
t
e
ly
. T
hi
s
s
e
tu
p i
s
hi
ghl
y s
c
a
la
bl
e
,
a
ll
ow
in
g i
nt
e
nt
ve
c
to
r
s
to
be
pr
e
c
om
put
e
d
a
nd
qui
c
kl
y
c
om
pa
r
e
d
w
it
h
ne
w
que
r
ie
s
us
in
g
c
os
in
e
s
im
il
a
r
it
y.
F
or
ta
s
ks
th
a
t
de
m
a
nd
hi
ghe
r
a
c
c
ur
a
c
y,
a
two
-
s
ta
ge
r
e
tr
ie
va
l
pr
oc
e
s
s
is
of
te
n
e
m
pl
oye
d.
T
he
f
ir
s
t
s
ta
ge
us
e
s
a
f
a
s
t
bi
-
e
nc
ode
r
f
or
in
it
ia
l
f
il
te
r
in
g,
f
ol
lo
w
e
d
by
a
c
r
os
s
-
e
nc
ode
r
th
a
t
r
e
-
e
va
lu
a
te
s
th
e
to
p
r
e
s
ul
t
s
by
jo
in
tl
y
m
ode
li
ng
th
e
r
e
la
ti
ons
hi
p
be
twe
e
n
th
e
que
r
y
a
nd
th
e
in
te
nt
[
10]
.
A
lt
hou
gh
th
is
a
ppr
oa
c
h
e
nha
nc
e
s
pr
e
c
i
s
io
n,
it
a
l
s
o
in
c
r
e
a
s
e
s
c
om
put
a
ti
ona
l
ove
r
he
a
d, w
hi
c
h r
e
m
a
in
s
a
ke
y c
h
a
ll
e
n
ge
i
n ongoing r
e
s
e
a
r
c
h.
A
ppl
yi
ng
na
tu
r
a
l
la
ng
ua
g
e
pr
o
c
e
s
s
in
g
(
N
L
P
)
i
n
h
e
a
lt
h
c
a
r
e
in
tr
od
uc
e
s
a
ddi
t
io
na
l
c
h
a
ll
e
nge
s
,
pa
r
ti
c
ul
a
r
ly
in
m
ul
ti
li
ngu
a
l
c
o
nt
e
xt
s
w
h
e
r
e
c
li
ni
c
ia
n
s
m
a
y
s
w
it
c
h
be
t
w
e
e
n
la
ng
ua
g
e
s
[
1
1]
or
us
e
s
p
e
c
i
a
li
z
e
d
m
e
di
c
a
l
te
r
m
s
.
T
o
m
a
na
g
e
th
i
s
c
om
pl
e
xi
t
y,
m
od
e
ls
ba
s
e
d
o
n
s
e
nt
e
nc
e
-
bi
di
r
e
c
ti
ona
l
e
nc
o
de
r
r
e
pr
e
s
e
nt
a
ti
on
s
f
r
om
tr
a
ns
f
or
m
e
r
s
(
B
E
R
T
)
[
12]
,
s
uc
h
a
s
M
i
ni
L
M
[
13]
a
nd
M
P
N
e
t
[
14]
va
r
i
a
nt
s
a
r
e
tr
a
in
e
d
to
g
e
ne
r
a
te
m
ul
ti
li
ngu
a
l
s
e
nt
e
n
c
e
e
m
be
ddi
ng
s
t
ha
t
a
li
gn
s
e
m
a
nt
ic
m
e
a
ni
ng
s
a
c
r
o
s
s
la
ng
ua
g
e
s
.
T
he
s
e
e
m
b
e
ddi
n
gs
e
n
s
ur
e
th
a
t
que
r
i
e
s
in
d
if
f
e
r
e
nt
l
a
ngu
a
ge
s
c
a
n
be
und
e
r
s
to
od
a
nd
p
r
oc
e
s
s
e
d
c
on
s
i
s
te
nt
ly
.
M
ode
l
p
e
r
f
or
m
a
n
c
e
i
s
ty
pi
c
a
ll
y
e
va
lu
a
t
e
d
us
i
ng
l
a
r
ge
-
s
c
a
le
be
nc
hm
a
r
ks
li
k
e
t
he
m
a
s
s
iv
e
t
e
xt
e
m
b
e
ddi
n
g
b
e
nc
hm
a
r
k
(
M
T
E
B
)
[
15]
,
w
hi
c
h
a
s
s
e
s
s
e
s
h
ow
w
e
ll
t
h
e
s
e
e
m
be
ddi
ng
s
g
e
ne
r
a
li
z
e
a
c
r
o
s
s
la
n
gua
ge
s
a
nd
dom
a
i
ns
.
I
n
pr
oduc
t
io
n
s
y
s
te
m
s
,
t
he
s
c
a
la
b
il
it
y
a
nd
s
p
e
e
d
of
ve
c
to
r
m
a
t
c
hi
ng
pl
a
y
a
c
r
uc
i
a
l
r
ol
e
.
P
e
r
f
or
m
in
g
e
xha
u
s
ti
ve
k
-
n
e
a
r
e
s
t
ne
ig
hbor
(
K
N
N
)
s
e
a
r
c
h
e
s
on l
a
r
ge
d
a
ta
s
e
ts
c
a
n b
e
c
om
put
a
ti
on
a
ll
y
e
xp
e
n
s
iv
e
.
A
s
a
r
e
s
ul
t,
m
ode
r
n
s
y
s
te
m
s
e
m
pl
oy
v
e
c
t
or
d
a
ta
b
a
s
e
s
s
uc
h
a
s
M
il
v
us
,
w
hi
c
h
i
m
pl
e
m
e
nt
a
ppr
o
xi
m
a
t
e
n
e
a
r
e
s
t
ne
ig
hbor
(
A
N
N
)
[
16]
a
l
gor
it
hm
s
m
os
t
not
a
bl
y
hi
e
r
a
r
c
hi
c
a
l
n
a
vi
ga
bl
e
s
m
a
ll
w
or
l
ds
(
H
N
S
W
)
[
17]
.
T
he
s
e
m
e
t
hod
s
s
li
ght
l
y c
om
pr
om
i
s
e
a
c
c
ur
a
c
y but a
c
hi
e
v
e
s
ig
ni
f
ic
a
nt
pe
r
f
or
m
a
n
c
e
ga
in
s
,
m
a
ki
ng
A
N
N
-
ba
s
e
d
v
e
c
t
or
d
a
ta
ba
s
e
s
th
e
pr
e
f
e
r
r
e
d
s
ol
ut
io
n f
or
r
e
a
l
-
ti
m
e
s
e
m
a
nt
i
c
ve
c
to
r
r
e
tr
ie
v
a
l
a
pp
li
c
a
ti
o
ns
.
T
hi
s
w
or
k
pr
o
pos
e
s
a
n
a
lt
e
r
na
ti
ve
a
ppr
oa
c
h:
a
te
xt
-
b
a
s
e
d
r
ul
e
-
ba
s
e
d
c
ha
t
bot
le
v
e
r
a
gi
ng
a
v
e
c
to
r
da
ta
b
a
s
e
m
a
na
g
e
m
e
nt
s
y
s
te
m
a
nd
d
e
e
p
l
e
a
r
ni
ng
-
ba
s
e
d
N
L
P
[
18
]
.
T
hi
s
s
ol
ut
io
n
di
s
ti
n
gui
s
he
s
it
s
e
lf
by
of
f
e
r
i
ng
a
m
or
e
c
os
t
-
e
f
f
e
c
ti
ve
a
s
it
i
s
de
v
e
lo
p
e
d
us
i
ng
ope
n
-
s
our
c
e
a
nd
a
lo
w
-
c
ode
e
n
vi
r
onm
e
nt
(
O
r
a
c
l
e
A
pe
x
).
T
he
ke
y c
ont
r
ib
ut
io
ns
of
t
hi
s
w
or
k i
nc
lu
de
:
i)
D
e
m
ons
tr
a
ti
ng
th
e
e
f
f
e
c
ti
ve
ne
s
s
of
a
m
ul
ti
li
ngua
l
M
P
N
e
t
m
ode
l
in
id
e
nt
if
yi
ng
c
om
pl
e
x
bi
li
ngua
l
(
I
ndone
s
ia
n
–
E
ngl
is
h)
c
li
ni
c
a
l
in
te
nt
s
,
ii)
M
e
a
s
ur
in
g
a
28
-
pe
r
c
e
nt
a
g
e
-
poi
nt
im
pr
ove
m
e
nt
in
r
e
s
pons
e
a
c
c
ur
a
c
y
w
he
n
u
s
in
g
s
e
m
a
nt
ic
ve
c
to
r
r
e
tr
ie
va
l
c
om
pa
r
e
d t
o a
l
e
xi
c
a
l
ba
s
e
li
ne
(
J
a
c
c
a
r
d
s
im
il
a
r
it
y)
w
it
hi
n a
n H
I
M
S
he
lp
de
s
k s
e
tt
in
g,
iii)
D
e
ve
l
op
in
g
a
s
c
a
l
a
bl
e
,
f
ul
l
y
s
e
lf
-
ho
s
t
e
d
a
r
c
hi
te
c
t
ur
e
b
ui
l
t
on
O
r
a
c
le
A
p
e
x,
D
j
a
ng
o,
a
nd
M
i
lv
u
s
,
of
f
e
r
i
ng
he
a
lt
hc
a
r
e
i
n
s
ti
t
ut
io
n
s
a
n a
f
f
or
da
bl
e
a
l
te
r
n
a
ti
v
e
t
o
c
om
m
e
r
c
i
a
l
c
l
oud
-
ba
s
e
d
na
tu
r
a
l
-
l
a
n
gu
a
g
e
u
nd
e
r
s
t
a
nd
in
g
(
N
L
U
)
s
y
s
t
e
m
s
.
A
c
om
pa
r
is
on
of
di
f
f
e
r
e
nt
c
ha
tb
ot
a
r
c
hi
te
c
tu
r
e
s
,
de
ta
il
in
g
th
e
ir
s
e
a
r
c
h
m
e
th
odol
ogy,
in
de
xi
ng
f
r
a
m
e
w
or
k,
pe
r
f
or
m
a
nc
e
f
oc
us
, a
nd c
os
t/
s
c
a
la
bi
li
ty
, i
s
pr
ovi
de
d i
n T
a
bl
e
1.
T
a
bl
e
1
.
C
om
pa
r
is
on of
c
h
a
tb
ot
a
r
c
hi
te
c
tu
r
e
s
C
ha
t
bot
s
ys
t
e
m
/
a
r
c
hi
t
e
c
t
ur
e
S
e
a
r
c
h m
e
t
hodol
ogy
I
nde
xi
ng/
f
r
a
m
e
w
or
k
P
e
r
f
or
m
a
nc
e
f
oc
us
C
os
t
/
s
c
a
l
a
bi
l
i
t
y
T
r
a
di
t
i
ona
l
r
ul
e
-
ba
s
e
d
L
e
xi
c
a
l
/
e
xa
c
t
m
a
t
c
h
S
Q
L
/
R
D
B
M
S
L
ow
a
c
c
ur
a
c
y, hi
gh
s
pe
e
d
L
ow
c
os
t
, poor
s
c
a
l
a
bi
l
i
t
y
B
a
s
i
c
ML
I
nt
e
nt
c
l
a
s
s
i
f
i
c
a
t
i
on
(
N
L
U
)
-
H
i
gh
a
c
c
ur
a
c
y,
m
e
di
um
s
pe
e
d
C
os
t
va
r
i
e
s
, poor
s
c
a
l
a
bi
l
i
t
y
M
ode
r
n ve
c
t
or
s
e
a
r
c
h
S
e
m
a
nt
i
c
r
e
t
r
i
e
va
l
(
B
i
-
e
nc
ode
r
)
V
e
c
t
or
da
t
a
ba
s
e
s
(
e
.g., M
i
l
vus
, F
A
I
S
S
,
w
e
a
vi
a
t
e
, e
l
a
s
t
i
c
s
e
a
r
c
h
)
H
i
gh
a
c
c
ur
a
c
y,
s
c
a
l
a
bl
e
pe
r
f
or
m
a
nc
e
M
ode
r
a
t
e
c
os
t
(
i
nf
r
a
s
t
r
uc
t
ur
e
)
, hi
gh
l
y
s
c
a
l
a
bl
e
T
hi
s
s
t
udy
S
e
m
a
nt
i
c
r
e
t
r
i
e
va
l
(
M
P
N
e
t
)
M
i
l
vus
d
a
t
a
ba
s
e
M
i
l
vus
ve
c
t
or
da
t
a
ba
s
e
s
,
O
r
a
c
l
e
A
pe
x
O
pt
i
m
i
z
e
d
a
c
c
ur
a
c
y f
or
H
I
M
S
i
nt
e
nt
s
(
a
c
c
:
0.70)
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. 6, D
e
c
e
m
be
r
2025
:
4600
-
4613
4602
2.
M
E
T
H
O
D
2.1. Us
e
r
in
t
e
n
t
d
at
as
e
t
p
r
e
p
ar
at
io
n
W
e
pe
r
f
or
m
e
d
a
n
a
na
ly
s
is
of
th
e
H
I
M
S
us
e
r
m
a
nua
l
to
e
xt
r
a
c
t
in
f
or
m
a
ti
on
pe
r
ti
ne
nt
to
doc
to
r
s
in
s
id
e
th
e
H
I
M
S
f
r
a
m
e
w
or
k.
T
he
da
ta
w
a
s
s
ubs
e
que
nt
ly
a
gg
r
e
ga
te
d
in
to
a
c
om
pi
la
ti
on
of
11
us
e
r
in
te
nt
s
a
s
s
how
n
in
T
a
bl
e
2.
F
or
e
a
c
h
us
e
r
in
te
nt
,
13
r
e
a
li
s
ti
c
us
e
r
que
r
y
s
a
m
pl
e
s
w
e
r
e
m
a
nu
a
ll
y
f
or
m
ul
a
te
d
(
10 f
or
t
r
a
in
in
g, 3
f
or
t
e
s
ti
ng)
ba
s
e
d on c
om
m
on doc
to
r
i
nqui
r
ie
s
, r
e
s
ul
ti
ng i
n 143 tot
a
l
r
e
c
or
ds
.
T
a
bl
e
2
.
U
s
e
r
in
te
nt
da
ta
s
e
t
No
U
s
e
r
i
nt
e
nt
T
r
a
i
ni
ng s
a
m
pl
e
s
T
e
s
t
i
ng s
a
m
pl
e
s
1
S
ur
ge
r
y
s
c
he
dul
e
10
3
2
D
oc
t
or
pr
a
c
t
i
c
e
hour
s
10
3
3
O
ut
pa
t
i
e
nt
nur
s
e
s
hi
f
t
s
c
he
dul
e
10
3
4
D
oc
t
or
vi
s
i
t
s
c
he
dul
e
10
3
5
M
e
di
c
a
l
s
e
r
vi
c
e
pa
ym
e
nt
de
t
a
i
l
10
3
6
O
nl
i
ne
r
e
s
e
r
va
t
i
on pa
t
i
e
nt
l
i
s
t
10
3
7
C
ons
ul
t
e
d
pa
t
i
e
nt
i
nf
or
m
a
t
i
on
10
3
8
P
ol
yc
l
i
ni
c
pa
t
i
e
nt
l
i
s
t
10
3
9
M
e
di
c
a
l
pr
oc
e
dur
e
de
t
a
i
l
s
10
3
10
D
oc
t
or
f
e
e
dba
c
k on c
ha
t
bot
us
a
ge
10
3
11
C
ha
t
bot
f
e
a
t
ur
e
ove
r
vi
e
w
10
3
2.2. E
m
b
e
d
d
in
g
m
od
e
ls
e
val
u
at
io
n
E
m
be
ddi
ng
m
ode
ls
w
e
r
e
ut
il
iz
e
d
to
c
r
e
a
te
ve
c
to
r
da
ta
f
r
om
us
e
r
in
te
nt
a
nd
us
e
r
que
r
y
to
m
a
ke
s
e
m
a
nt
ic
ve
c
to
r
r
e
tr
ie
va
l
w
it
h
c
os
in
e
s
im
il
a
r
it
y
be
in
g
pos
s
ib
l
e
.
F
iv
e
e
m
be
ddi
ng
m
ode
ls
w
e
r
e
s
e
le
c
te
d
a
s
c
a
ndi
da
te
s
ba
s
e
d
on
th
e
ir
r
e
por
te
d
pe
r
f
or
m
a
nc
e
on
M
T
E
B
a
nd
th
e
ir
doc
um
e
nt
e
d
m
ul
ti
li
ngua
l
c
a
pa
bi
li
ti
e
s
[
15]
.
T
he
in
c
lu
s
io
n
of
m
ul
ti
li
ngua
l
m
ode
ls
is
c
r
uc
ia
l
gi
ve
n
th
e
pot
e
nt
ia
l
f
or
c
ode
-
s
w
it
c
hi
ng
or
va
r
ia
ti
ons
i
n
c
li
ni
c
a
l
te
r
m
in
ol
ogy of
te
n f
ound in hea
lt
hc
a
r
e
c
ont
e
xt
s
[
11]
.
T
o
de
te
r
m
in
e
th
e
m
os
t
s
ui
ta
bl
e
m
ode
l
f
or
us
e
r
in
te
nt
da
ta
s
e
t,
pe
r
f
or
m
a
nc
e
e
va
lu
a
ti
on
w
a
s
ne
e
d
e
d.
T
he
pe
r
f
or
m
a
nc
e
e
va
lu
a
ti
on
w
a
s
c
onduc
te
d
by
tr
a
in
in
g
a
lo
gi
s
ti
c
r
e
gr
e
s
s
io
n
c
la
s
s
if
ie
r
to
a
s
s
e
s
s
th
e
qua
li
ty
of
th
e
ge
ne
r
a
te
d
u
s
e
r
in
te
nt
e
m
be
ddi
ngs
[
19]
.
T
hi
s
c
la
s
s
if
ic
a
ti
on
ta
s
k
s
e
r
ve
s
a
s
a
n
obj
e
c
ti
ve
pr
oxy
to
de
te
r
m
in
e
th
e
e
m
be
ddi
ng
m
ode
l
b
e
s
t
c
a
pa
bl
e
of
di
s
ti
ngui
s
hi
ng
be
tw
e
e
n
th
e
11
u
s
e
r
in
te
nt
s
in
th
e
v
e
c
to
r
s
pa
c
e
.
T
he
m
ode
l
s
e
le
c
t
e
d
ba
s
e
d
on
pr
e
c
is
io
n,
r
e
c
a
ll
,
a
nd
F
1
-
s
c
or
e
w
il
l
th
e
n
be
im
pl
e
m
e
nt
e
d
to
f
a
c
il
it
a
te
th
e
f
in
a
l
s
e
m
a
nt
ic
ve
c
to
r
r
e
tr
ie
va
l.
T
he
f
iv
e
e
m
b
e
ddi
ng
m
ode
ls
’
c
a
ndi
d
a
te
s
a
r
e
:
i)
a
ll
-
M
in
iL
M
-
L6
-
v2
,
ii
)
pa
r
a
phr
a
s
e
-
m
ul
ti
li
ngua
l
-
M
in
iL
M
-
L
12
-
v2
,
ii
i)
pa
r
a
phr
a
s
e
-
m
ul
ti
li
ngua
l
-
m
pne
t
-
ba
s
e
-
v2
,
iv
)
di
s
ti
lu
s
e
-
ba
s
e
-
m
ul
ti
li
ngua
l
-
c
a
s
e
d
-
v2
, a
nd v)
s
e
nt
e
nc
e
-
tr
a
ns
f
or
m
e
r
s
/L
a
B
S
E
.
2.3. S
e
m
an
t
ic
ve
c
t
or
r
e
t
r
ie
val
v
s
k
e
yw
or
d
m
at
c
h
in
g c
o
m
p
ar
is
on
W
e
c
onduc
t
e
d
a
n
a
bl
a
ti
on
s
tu
dy
to
qua
nt
if
y
th
e
im
pr
ove
m
e
nt
in
in
te
nt
r
e
c
ogni
ti
on
ga
in
e
d
by
u
s
in
g
s
e
m
a
nt
ic
ve
c
to
r
r
e
tr
ie
va
l
w
it
h
c
os
in
e
s
im
il
a
r
it
y
ove
r
a
tr
a
di
ti
ona
l
ke
yw
or
d
-
ba
s
e
d
s
e
a
r
c
h
ba
s
e
li
ne
.
F
or
th
e
s
e
m
a
nt
ic
ve
c
to
r
r
e
tr
ie
va
l
c
om
pone
nt
,
th
e
s
e
le
c
te
d
e
m
be
ddi
ng
m
ode
l
w
a
s
us
e
d
to
ve
c
to
r
iz
e
a
ll
us
e
r
in
te
nt
s
a
m
pl
e
s
.
N
e
a
r
e
s
t
n
e
ig
hbor
c
la
s
s
if
ic
a
ti
on
w
a
s
th
e
n
a
ppl
ie
d,
ut
il
iz
in
g
c
os
in
e
s
im
il
a
r
it
y
to
f
in
d
th
e
s
in
gl
e
k=
1
ne
a
r
e
s
t
ne
ig
hbor
(
th
e
pr
e
di
c
te
d i
nt
e
nt
c
la
s
s
)
f
or
e
a
c
h t
e
s
t
s
a
m
pl
e
.
F
or
th
e
ke
yw
or
d
-
ba
s
e
d
s
e
a
r
c
h
ba
s
e
li
ne
,
th
e
J
a
c
c
a
r
d
s
im
il
a
r
it
y
a
lg
or
it
hm
w
a
s
c
hos
e
n
to
m
e
a
s
ur
e
th
e
to
ke
n
ove
r
la
p
be
tw
e
e
n
th
e
us
e
r
qu
e
r
y
a
nd
th
e
tr
a
in
in
g
in
te
n
t
s
a
m
pl
e
s
.
T
h
e
in
te
nt
c
la
s
s
w
it
h
th
e
hi
ghe
s
t
J
a
c
c
a
r
d
s
c
or
e
w
a
s
s
e
le
c
t
e
d
a
s
th
e
pr
e
di
c
ti
on.
A
c
c
ur
a
c
y
w
a
s
c
a
lc
ul
a
te
d
f
or
bot
h
m
e
th
ods
,
a
ll
ow
in
g
us
t
o
qua
nt
if
y t
he
pe
r
f
or
m
a
nc
e
ga
in
of
t
he
s
e
m
a
nt
ic
ve
c
to
r
r
e
tr
ie
va
l
a
ppr
oa
c
h f
or
i
nt
e
nt
r
e
c
ogni
ti
on.
2.4. T
h
r
e
s
h
ol
d
s
e
n
s
it
iv
it
y an
al
ys
is
T
o
de
te
r
m
in
e
th
e
opt
im
a
l
c
os
in
e
s
im
il
a
r
it
y
th
r
e
s
hol
d
f
or
in
te
nt
m
a
tc
hi
ng
a
nd
to
a
na
ly
z
e
it
s
e
f
f
e
c
t
on
c
ha
tb
ot
pe
r
f
or
m
a
nc
e
,
a
th
r
e
s
hol
d
s
e
n
s
it
iv
it
y
a
na
ly
s
i
s
w
a
s
p
e
r
f
or
m
e
d.
T
he
s
e
le
c
te
d
e
m
be
ddi
ng
m
ode
l
w
a
s
ut
il
iz
e
d t
o c
onve
r
t
in
te
nt
s
a
m
pl
e
da
ta
s
e
t
in
to
e
m
be
ddi
ng ve
c
to
r
. F
or
e
a
c
h i
nt
e
nt
s
a
m
pl
e
t
e
s
t
da
ta
s
e
t,
t
he
c
o
s
in
e
s
im
il
a
r
it
y
be
twe
e
n
th
e
te
s
t
e
m
be
ddi
ng
a
nd
it
s
to
p
r
e
tr
ie
ve
d
c
a
ndi
da
te
w
a
s
c
om
put
e
d
us
in
g
th
e
K
N
N
a
lg
or
it
hm
w
it
h
c
os
in
e
s
im
il
a
r
it
y. A
s
e
r
ie
s
of
t
hr
e
s
hol
ds
r
a
ngi
ng
f
r
om
0.40 to
0.95
(
in
i
nc
r
e
m
e
nt
s
of
0.01)
w
a
s
e
va
lu
a
te
d.
A
m
a
tc
h w
a
s
onl
y
a
c
c
e
pt
e
d
if
th
e
hi
ghe
s
t
s
im
il
a
r
it
y
s
c
or
e
a
m
ong
th
e
k
c
a
ndi
da
te
s
w
a
s
gr
e
a
te
r
th
a
n
or
e
qua
l
to
th
e
te
s
te
d
th
r
e
s
hol
d
(
τ)
.
T
he
in
te
nt
c
or
r
e
s
ponding
t
o
th
e
hi
ghe
s
t
s
c
or
e
a
bove
τ
w
a
s
de
s
ig
na
te
d
a
s
th
e
pr
e
di
c
ti
on;
ot
he
r
w
is
e
,
th
e
que
r
y
w
a
s
c
l
a
s
s
if
ie
d a
s
"
N
o
m
a
tc
h
f
ound
"
(
r
e
je
c
ti
on)
.
F
or
e
a
c
h
th
r
e
s
hol
d
va
lu
e
,
pr
e
c
is
io
n, r
e
c
a
ll
, a
nd F
1
-
s
c
or
e
w
e
r
e
c
om
put
e
d.
2.5. S
of
t
w
ar
e
d
e
ve
lo
p
m
e
n
t
T
h
e
a
g
il
e
d
e
v
e
l
opm
e
n
t
m
e
th
od
ol
o
gy w
a
s
e
m
pl
oy
e
d t
o de
ve
lo
p
t
he
c
h
a
t
bot
a
ppl
ic
a
ti
on
. T
h
is
m
e
th
od i
s
c
h
a
r
a
c
t
e
r
i
z
e
d
by
i
ts
i
nc
r
e
m
e
n
ta
l
na
tu
r
e
(
s
m
a
l
l
r
e
l
e
a
s
e
v
e
r
s
io
n
s
w
it
h
f
a
s
t
c
y
c
l
e
s
)
,
c
oop
e
r
a
ti
ve
n
a
t
ur
e
(
d
e
ve
l
op
e
r
s
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
Se
m
ant
ic
s
e
ar
c
h
-
e
nhanc
e
d he
al
th
c
a
r
e
c
hat
bot
f
or
ho
s
pi
ta
l
in
fo
r
m
at
io
n
…
(
E
r
da G
us
li
nar
P
e
r
dana
)
4603
a
nd
u
s
e
r
s
w
or
k
to
g
e
t
he
r
w
it
h
i
nt
e
n
s
e
c
o
m
m
u
ni
c
a
ti
o
n)
,
a
nd
a
da
pt
a
bi
l
it
y
[
2
0]
,
[
2
1]
.
T
h
e
s
ol
u
ti
o
n
ut
i
li
z
e
s
a
lo
w
c
od
e
t
e
c
hn
ol
o
gy
p
la
t
f
or
m
t
o
im
pr
ov
e
th
e
e
f
f
ic
ie
nc
y
of
s
of
t
w
a
r
e
d
e
v
e
l
opm
e
n
t
[
22]
,
s
pe
c
if
ic
a
ll
y
e
m
pl
oyi
ng
th
e
O
r
a
c
l
e
A
p
e
x
d
e
v
e
lo
pm
e
nt
pl
a
tf
or
m
f
or
t
h
e
f
r
on
t
-
e
nd
i
nt
e
r
f
a
c
e
a
n
d c
or
e
b
u
s
in
e
s
s
l
ogi
c
in
t
e
g
r
a
t
io
n.
2.6. Ch
at
b
ot
e
val
u
at
io
n
U
s
a
bi
li
ty
e
va
lu
a
ti
on
w
a
s
e
m
pl
oye
d
to
a
s
s
e
s
s
th
e
r
e
s
ul
ti
ng
c
ha
tb
ot
a
ppl
ic
a
ti
on.
U
s
a
bi
li
ty
is
de
f
in
e
d
by
I
S
O
9241
-
11:
1998
a
s
"
th
e
de
gr
e
e
to
w
hi
c
h
a
pr
oduc
t
c
a
n
be
ut
il
iz
e
d
by
s
pe
c
if
ic
us
e
r
s
to
a
c
hi
e
ve
s
p
e
c
if
ic
obj
e
c
ti
ve
s
w
it
h
e
f
f
ic
ie
nc
y,
e
f
f
e
c
ti
ve
ne
s
s
,
a
nd
s
a
ti
s
f
a
c
ti
on
in
a
s
pe
c
if
ic
c
ont
e
xt
of
us
e
"
[
23]
.
A
c
c
or
di
ng
to
th
e
de
f
in
it
io
n,
th
r
e
e
m
e
tr
ic
s
a
r
e
ut
il
iz
e
d
in
th
e
e
v
a
lu
a
ti
on
of
th
is
s
t
udy:
e
f
f
e
c
ti
ve
ne
s
s
, e
f
f
ic
ie
nc
y,
a
nd s
a
ti
s
f
a
c
ti
on.
E
a
c
h
m
e
a
s
ur
e
in
c
lu
de
s
th
r
e
e
pr
ope
r
ti
e
s
th
a
t
a
r
e
c
he
c
ke
d
us
in
g
a
que
s
ti
onna
ir
e
w
it
h
a
L
ik
e
r
t
s
c
a
le
r
a
ngi
ng
f
r
om
1
to
10
[
24]
.
T
he
s
e
ni
n
e
a
tt
r
ib
ut
e
s
r
e
f
e
r
s
to
a
s
tu
dy c
ondu
c
te
d
by
N
ic
ol
e
a
nd
M
or
ga
n
[
25]
in
c
onduc
ti
ng
us
a
bi
li
ty
te
s
ti
ng
on
c
ha
tb
ot
s
.
T
he
que
s
ti
onna
ir
e
r
e
s
ponde
nt
s
c
o
m
pr
is
e
d
two
gr
oups
:
s
ix
in
di
vi
dua
ls
f
r
om
th
e
H
I
M
S
a
ppl
ic
a
ti
on
de
ve
lo
pm
e
nt
te
a
m
(
f
or
te
c
hni
c
a
l
va
li
da
ti
o
n)
a
nd
f
if
te
e
n
m
e
m
be
r
s
of
th
e
doc
to
r
gr
oup
(
f
or
e
nd
-
us
e
r
va
li
da
ti
on)
.
2.7. Ac
t
iv
e
le
ar
n
in
g l
oop
T
o
e
ns
ur
e
th
e
s
ys
te
m
r
e
m
a
in
s
a
c
c
ur
a
te
a
nd
a
da
pt
iv
e
w
hi
le
ut
i
li
z
in
g
a
f
ix
e
d
e
m
be
ddi
ng
m
ode
l,
w
e
pr
opos
e
im
pl
e
m
e
nt
in
g
a
n
a
c
ti
ve
l
e
a
r
ni
ng
(
A
L
)
lo
op
to
c
ont
in
uous
ly
r
e
f
in
e
th
e
us
e
r
in
te
nt
knowle
dg
e
ba
s
e
in
th
e
ve
c
to
r
da
ta
ba
s
e
.
T
hi
s
m
e
c
h
a
ni
s
m
is
c
r
it
ic
a
l
f
or
a
ddr
e
s
s
in
g
i
ns
ta
nc
e
s
of
s
e
m
a
nt
ic
a
m
bi
gui
ty
a
nd
c
ove
r
a
ge
ga
ps
.
T
h
e
A
L
lo
op
f
oc
us
e
s
on
id
e
nt
if
yi
ng
hi
gh
-
va
lu
e
que
r
ie
s
by
m
oni
to
r
in
g
th
r
e
e
s
ig
na
l
s
:
c
onf
id
e
nc
e
th
r
e
s
hol
d
vi
ol
a
ti
on
(
que
r
y
is
to
o
di
s
ta
nt
f
r
om
a
ll
c
ur
r
e
nt
in
te
n
ts
)
;
a
m
bi
gui
ty
s
a
m
pl
in
g
(
to
p
s
im
il
a
r
it
y
s
c
or
e
s
a
r
e
to
o
c
lo
s
e
)
;
a
nd
us
e
r
e
s
c
a
la
ti
on/
f
e
e
dba
c
k
s
ig
na
l
(
th
e
que
r
y
r
e
s
ul
ts
in
im
m
e
di
a
te
ne
ga
ti
ve
us
e
r
f
e
e
dba
c
k)
.
T
he
s
e
f
la
gge
d
qu
e
r
ie
s
a
r
e
r
out
e
d
to
th
e
H
I
M
S
c
ont
e
nt
m
a
na
ge
m
e
nt
in
te
r
f
a
c
e
w
h
e
r
e
a
n
e
xpe
r
t
a
s
s
ig
ns
th
e
c
or
r
e
c
t
in
te
nt
a
nd
a
dds
th
e
que
r
y
a
s
a
ne
w
in
te
nt
s
a
m
pl
e
.
F
in
a
ll
y,
th
e
'
V
e
c
to
r
iz
e
a
ll
in
te
nt
'
f
unc
ti
on
is
e
xe
c
ut
e
d
vi
a
th
e
D
ja
ngo
A
P
I
to
ge
ne
r
a
te
th
e
ve
c
to
r
f
or
th
e
ne
w
s
a
m
pl
e
a
nd
in
s
e
r
t
it
in
to
th
e
M
il
vus
ve
c
to
r
da
ta
ba
s
e
.
T
hi
s
pr
oc
e
s
s
it
e
r
a
ti
ve
ly
e
xpa
nds
a
nd
c
la
r
if
ie
s
th
e
b
ounda
r
ie
s
of
th
e
in
te
nt
c
lu
s
te
r
s
in
th
e
v
e
c
to
r
s
pa
c
e
, e
nh
a
nc
in
g s
e
m
a
nt
ic
r
e
tr
ie
va
l
a
c
c
ur
a
c
y ov
e
r
t
im
e
w
it
hout
t
he
ne
c
e
s
s
it
y of
e
xpe
ns
iv
e
m
ode
l
r
e
tr
a
in
in
g.
2.8. S
ys
t
e
m
p
e
r
f
or
m
an
c
e
e
val
u
at
io
n
S
ys
te
m
-
le
ve
l
pe
r
f
or
m
a
nc
e
m
e
tr
ic
s
w
e
r
e
e
va
lu
a
te
d
to
e
ns
ur
e
th
a
t
th
e
s
e
m
a
nt
ic
ve
c
to
r
r
e
tr
ie
va
l
c
a
n
ope
r
a
te
e
f
f
ic
ie
nt
ly
in
r
e
a
l
-
ti
m
e
he
a
lt
hc
a
r
e
c
h
a
tb
ot
s
c
e
na
r
io
s
.
P
e
r
f
or
m
a
nc
e
e
va
lu
a
ti
on
f
oc
u
s
e
d
on
l
a
te
nc
y
a
nd
th
r
oughput
c
ons
is
te
n
c
y
unde
r
c
onc
ur
r
e
nt
que
r
y
lo
a
ds
.
L
a
te
nc
y
r
e
f
e
r
s
to
th
e
a
ve
r
a
g
e
ti
m
e
ta
ke
n
f
or
M
il
vus
to
r
e
tr
ie
ve
th
e
to
p
-
1
s
e
m
a
nt
ic
m
a
tc
h
f
r
om
a
ve
c
to
r
c
ol
le
c
ti
on
a
f
te
r
r
e
c
e
iv
in
g a
u
s
e
r
que
r
y. T
hr
oughput
r
e
pr
e
s
e
nt
s
th
e
num
b
e
r
of
s
uc
c
e
s
s
f
ul
r
e
tr
ie
va
l
ope
r
a
ti
on
s
p
e
r
s
e
c
ond.
T
he
b
e
nc
hm
a
r
ki
ng
e
nvi
r
onm
e
nt
c
ons
is
te
d
of
a
M
il
vus
2.3.21
in
s
ta
nc
e
hos
te
d
on
a
c
lo
ud
s
e
r
ve
r
e
qui
ppe
d
w
it
h
16
G
B
of
R
A
M
,
8
vC
P
U
s
,
a
nd
a
n
S
S
D
-
ba
s
e
d
s
to
r
a
ge
ba
c
ke
nd.
L
oa
d
te
s
ti
ng
w
a
s
s
im
ul
a
te
d
us
in
g
a
P
yt
hon
-
ba
s
e
d
a
s
ync
hr
onous
c
li
e
nt
bui
lt
w
it
h
lo
c
us
t
a
nd
a
io
ht
tp
,
ge
ne
r
a
ti
ng
c
onc
ur
r
e
nt
que
r
y
r
e
que
s
ts
r
a
ngi
ng
f
r
om
1
to
100
pa
r
a
ll
e
l
us
e
r
s
.
E
a
c
h
c
li
e
nt
is
s
ue
d
r
a
ndom,
pr
e
-
e
m
be
dd
e
d
que
r
ie
s
f
r
om
th
e
in
te
nt
s
a
m
pl
e
te
s
t
da
ta
s
e
t,
a
nd
th
e
a
ve
r
a
ge
e
nd
-
to
-
e
nd
la
te
nc
y (
f
r
om
r
e
que
s
t
to
r
e
tr
ie
va
l
r
e
s
ul
t)
w
a
s
m
e
a
s
ur
e
d ove
r
1,0
00 que
r
ie
s
pe
r
c
onc
ur
r
e
nc
y l
e
ve
l.
3.
R
E
S
U
L
T
S
A
N
D
D
I
S
C
U
S
S
I
O
N
3.1. E
m
b
e
d
d
in
g
m
od
e
l
e
val
u
at
io
n
W
e
tr
a
in
e
d
lo
gi
s
ti
c
r
e
gr
e
s
s
io
n
c
la
s
s
if
ie
r
f
r
om
our
us
e
r
in
te
nt
tr
a
in
in
g
s
a
m
pl
e
w
hi
c
h
ha
s
b
e
e
n
ge
ne
r
a
te
d
in
to
ve
c
to
r
da
ta
by
e
a
c
h
c
a
ndi
da
te
m
ode
l.
V
e
c
to
r
iz
e
d
us
e
r
in
te
nt
te
s
ti
ng
s
a
m
pl
e
th
e
n
w
a
s
u
s
e
d
to
e
va
lu
a
te
th
e
tr
a
in
e
d
c
la
s
s
if
ie
r
.
T
he
e
va
lu
a
ti
on
r
e
s
ul
t
a
s
s
h
ow
n
in
T
a
bl
e
3
in
di
c
a
te
s
th
a
t
pa
r
a
phr
a
s
e
-
m
ul
ti
li
ngua
l
-
m
pne
t
-
ba
s
e
-
v2 outpe
r
f
or
m
e
d a
not
he
r
c
a
ndi
da
te
m
ode
ls
i
n a
ll
e
va
lu
a
ti
on me
tr
ic
.
T
a
bl
e
3. E
m
be
ddi
ng mode
l
e
va
lu
a
ti
on r
e
s
ul
t
No
M
ode
l
A
c
c
ur
a
c
y
F1
-
s
c
or
e
P
r
e
c
i
s
i
on
R
e
c
a
l
l
1
a
l
l
-
M
i
ni
L
M
-
L6
-
v2
0.5455
0.5102
0.5795
0.5455
2
pa
r
a
phr
a
s
e
-
m
ul
t
i
l
i
ngua
l
-
M
i
ni
L
M
-
L
12
-
v
0.6364
0.5686
0.6011
0.6364
3
pa
r
a
phr
a
s
e
-
m
ul
t
i
l
i
ngua
l
-
m
pne
t
-
ba
s
e
-
v2
0.7576
0.7102
0.7212
0.7576
4
di
s
t
i
l
us
e
-
ba
s
e
-
m
ul
t
i
l
i
ngua
l
-
c
a
s
e
d
-
v2
0.6061
0.5554
0.5697
0.6061
5
s
e
nt
e
nc
e
-
t
r
a
ns
f
or
m
e
r
s
/
L
a
B
S
E
0.6364
0.5962
0.6515
0.6364
3.2.
E
r
r
or
a
n
al
ys
is
F
ur
th
e
r
a
na
ly
s
is
w
a
s
c
onduc
te
d
f
or
pa
r
a
phr
a
s
e
-
m
ul
ti
li
ngua
l
-
m
pne
t
-
ba
s
e
-
v2
to
id
e
nt
if
y
in
te
nt
s
m
is
c
la
s
s
if
ic
a
ti
on
s
.
W
e
f
ound
th
a
t
6
of
11
in
te
nt
s
w
e
r
e
c
la
s
s
if
ie
d
w
it
h
100%
a
c
c
ur
a
c
y.
T
he
ope
r
a
ti
ona
l
a
nd
pa
ti
e
nt
li
s
t
in
te
nt
s
s
ur
ge
r
y
s
c
he
dul
e
,
doc
to
r
pr
a
c
ti
c
e
h
our
s
,
out
pa
ti
e
nt
nur
s
e
s
hi
f
t
s
c
he
dul
e
,
onl
in
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. 6, D
e
c
e
m
be
r
2025
:
4600
-
4613
4604
r
e
s
e
r
va
ti
on
pa
ti
e
nt
li
s
t,
pol
yc
li
ni
c
pa
ti
e
nt
li
s
t,
a
nd
c
ons
ul
te
d
pa
ti
e
nt
in
f
or
m
a
ti
on
w
e
r
e
c
le
a
r
ly
s
e
pa
r
a
bl
e
in
th
e
e
m
be
ddi
ng
s
pa
c
e
.
T
he
s
e
in
te
nt
s
r
e
pr
e
s
e
nt
c
or
e
c
li
ni
c
a
l
a
nd
a
d
m
in
is
tr
a
ti
ve
que
r
ie
s
th
a
t
th
e
e
m
be
ddi
ng
m
ode
l
s
uc
c
e
s
s
f
ul
ly
di
s
ti
ngui
s
he
d.
H
ig
h
-
c
onf
us
io
n
in
te
nt
s
,
w
hi
c
h
e
xpe
r
ie
nc
e
d
th
e
m
a
jo
r
it
y
of
th
e
e
r
r
or
s
,
w
e
r
e
m
e
di
c
a
l
pr
oc
e
dur
e
de
ta
il
s
,
m
e
di
c
a
l
s
e
r
vi
c
e
pa
ym
e
nt
de
ta
il
,
doc
to
r
f
e
e
dba
c
k
on
c
ha
tb
ot
us
a
ge
,
a
nd
c
ha
tb
ot
f
e
a
tu
r
e
ove
r
vi
e
w
. T
he
m
is
c
la
s
s
if
ic
a
ti
on s
ugge
s
ts
a
s
tr
ong s
e
m
a
nt
ic
ove
r
la
p be
twe
e
n t
h
e
s
e
i
nt
e
nt
s
. T
he
c
la
s
s
if
ic
a
ti
on r
e
s
ul
ts
f
or
t
he
s
e
pr
obl
e
m
a
ti
c
in
te
nt
s
a
r
e
de
t
a
il
e
d
in
T
a
bl
e
4
.
F
ig
ur
e
1
s
how
s
th
e
c
om
pl
e
te
c
onf
us
io
n
m
a
tr
ix
.
T
a
bl
e
5
f
ur
th
e
r
il
lu
s
tr
a
te
s
th
e
s
pe
c
if
ic
na
tu
r
e
of
th
e
s
e
e
r
r
or
s
by
pr
ovi
di
ng
t
he
m
is
c
la
s
s
if
ie
d
te
s
t
que
r
ie
s
a
nd
th
e
m
ode
l'
s
in
c
or
r
e
c
t
pr
e
di
c
ti
on.
T
he
a
na
ly
s
is
s
how
s
th
a
t,
w
hi
le
th
e
m
ode
l
is
r
obus
t
f
or
th
e
m
a
jo
r
i
ty
of
ope
r
a
ti
ona
l
ta
s
ks
,
th
e
ove
r
la
p
be
tw
e
e
n
s
pe
c
if
ic
de
ta
il
-
or
ie
nt
e
d
que
s
ti
on
s
(
li
ke
pr
oc
e
dur
e
s
te
p
s
or
m
e
di
c
a
l
s
e
r
vi
c
e
f
e
e
)
a
nd
s
ys
te
m
-
m
e
ta
que
s
ti
ons
(
c
ha
tb
ot
f
e
a
tu
r
e
s
/f
e
e
dba
c
k)
r
e
qui
r
e
s
f
ur
th
e
r
a
tt
e
nt
io
n.
T
a
bl
e
4. M
i
s
c
la
s
s
if
ic
a
ti
on a
na
ly
s
i
s
T
r
ue
i
nt
e
nt
C
or
r
e
c
t
l
y
c
l
a
s
s
i
f
i
e
d
M
i
s
c
l
a
s
s
i
f
i
e
d
E
r
r
or
s
a
nd a
na
l
ys
i
s
D
oc
t
or
vi
s
i
t
s
c
he
dul
e
1
2
H
i
gh
e
r
r
or
r
a
t
e
(
2/
3
m
i
s
c
l
a
s
s
i
f
i
e
d)
.
1
s
a
m
pl
e
w
a
s
m
i
s
c
l
a
s
s
i
f
i
e
d
a
s
"D
oc
t
or
pr
a
c
t
i
c
e
hour
s
"
a
nd
a
not
he
r
a
s
"O
nl
i
ne
r
e
s
e
r
va
t
i
on
pa
t
i
e
nt
l
i
s
t
".
T
hi
s
s
ugge
s
t
s
s
e
m
a
nt
i
c
ove
r
l
a
p a
nd poor
l
y l
a
be
l
e
d t
e
s
t
s
a
m
pl
e
M
e
di
c
a
l
pr
oc
e
dur
e
de
t
a
i
l
2
1
1
s
a
m
pl
e
w
a
s
m
i
s
c
l
a
s
s
i
f
i
e
d
a
s
"C
ons
ul
t
e
d
pa
t
i
e
nt
i
nf
or
m
a
t
i
on
".
T
hi
s
s
ugge
s
t
s
s
e
m
a
nt
i
c
ove
r
l
a
p w
i
t
h pa
t
i
e
nt
c
on
s
ul
t
a
t
i
on.
M
e
di
c
a
l
s
e
r
vi
c
e
pa
ym
e
nt
de
t
a
i
l
2
1
1
s
a
m
pl
e
m
i
s
c
l
a
s
s
i
f
i
e
d
a
s
"S
ur
ge
r
y
s
c
he
dul
e
".
T
hi
s
s
ugge
s
t
s
s
e
m
a
nt
i
c
ove
r
l
a
p w
he
n us
e
r
que
r
y a
bout
s
ur
ge
r
y s
e
r
vi
c
e
pa
ym
e
nt
.
D
oc
t
or
f
e
e
dba
c
k on
c
ha
t
bot
us
a
ge
2
1
1
s
a
m
pl
e
m
i
s
c
l
a
s
s
i
f
i
e
d
a
s
"M
e
di
c
a
l
s
e
r
vi
c
e
pa
ym
e
nt
de
t
a
i
l
."
T
hi
s
i
ndi
c
a
t
e
s
t
ha
t
us
e
r
f
e
e
dba
c
k
m
i
ght
c
ont
a
i
ns
t
e
r
m
s
w
hi
c
h
ove
r
l
a
p
w
i
t
h
m
e
di
c
a
l
s
e
r
vi
c
e
i
nt
e
nt
.
C
ha
t
bot
f
e
a
t
ur
e
ove
r
vi
e
w
0
3
H
i
ghe
s
t
e
r
r
or
r
a
t
e
(
3
m
i
s
c
l
a
s
s
i
f
i
e
d)
.
T
hi
s
i
ndi
c
a
t
e
s
t
ha
t
que
r
i
e
s
a
bout
t
he
c
ha
t
bot
'
s
f
e
a
t
ur
e
s
m
i
ght
c
ont
a
i
n t
e
r
m
s
w
hi
c
h ove
r
l
a
p w
i
t
h a
not
he
r
i
nt
e
nt
.
F
ig
ur
e
1.
C
onf
us
io
n
m
a
tr
ix
3.3.
S
e
m
an
t
ic
ve
c
t
or
r
e
t
r
ie
val
v
s
k
e
yw
or
d
m
at
c
h
in
g c
o
m
p
ar
is
on
W
e
c
onduc
te
d
a
n
a
bl
a
ti
on
s
tu
dy
to
c
om
pa
r
e
th
e
pe
r
f
or
m
a
nc
e
of
th
e
pr
opos
e
d
s
e
m
a
nt
ic
ve
c
to
r
r
e
tr
ie
va
l
m
e
th
od
a
ga
in
s
t
a
tr
a
di
ti
ona
l
le
xi
c
a
l
s
e
a
r
c
h
ba
s
e
li
ne
f
o
r
us
e
r
in
te
nt
r
e
c
ogni
ti
on.
T
he
s
e
m
a
nt
ic
v
e
c
to
r
r
e
tr
ie
va
l
ut
il
iz
e
d
th
e
be
s
t
-
pe
r
f
or
m
in
g
e
m
be
ddi
ng
m
ode
l,
pa
r
a
phr
a
s
e
-
m
ul
ti
li
ngua
l
-
m
pne
t
-
ba
s
e
-
v2,
c
om
bi
ne
d
w
it
h
k=
1
ne
a
r
e
s
t
ne
ig
hbor
c
la
s
s
if
ic
a
ti
on
us
in
g
c
os
in
e
s
im
il
a
r
it
y.
T
he
le
xi
c
a
l
ba
s
e
li
ne
e
m
pl
oye
d
J
a
c
c
a
r
d
s
im
il
a
r
it
y
f
or
ke
yw
or
d
m
a
tc
hi
ng
be
twe
e
n
th
e
us
e
r
que
r
y
a
nd
th
e
tr
a
in
in
g
in
te
nt
s
a
m
pl
e
s
.
B
ot
h
m
e
th
ods
w
e
r
e
e
va
lu
a
te
d
us
in
g
th
e
a
c
c
ur
a
c
y
m
e
tr
ic
on
th
e
in
te
nt
te
s
t
da
ta
s
e
t.
T
he
r
e
s
ul
ts
of
th
e
a
bl
a
ti
on
s
tu
dy
a
s
s
how
n
in
T
a
bl
e
6
,
de
m
ons
tr
a
te
a
c
le
a
r
s
upe
r
io
r
it
y
o
f
th
e
s
e
m
a
nt
ic
ve
c
to
r
r
e
tr
ie
va
l
a
ppr
oa
c
h
in
te
r
m
s
o
f
r
e
c
ogni
ti
on a
c
c
ur
a
c
y.
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
Se
m
ant
ic
s
e
ar
c
h
-
e
nhanc
e
d he
al
th
c
a
r
e
c
hat
bot
f
or
ho
s
pi
ta
l
in
fo
r
m
at
io
n
…
(
E
r
da G
us
li
nar
P
e
r
dana
)
4605
T
he
s
e
m
a
nt
ic
ve
c
to
r
r
e
tr
ie
va
l
a
c
hi
e
ve
d a
n a
c
c
ur
a
c
y of
0.70. I
n
c
ont
r
a
s
t,
t
he
ke
yw
or
d s
e
a
r
c
h (
le
xi
c
a
l)
ba
s
e
li
ne
onl
y
r
e
a
c
he
d
a
n
a
c
c
ur
a
c
y
of
0.42.
T
hi
s
r
e
s
ul
t
c
onf
ir
m
s
th
a
t
s
e
m
a
nt
ic
ve
c
to
r
r
e
tr
ie
va
l
pr
ovi
de
s
a
s
ig
ni
f
ic
a
nt
pe
r
f
or
m
a
nc
e
ga
in
f
o
r
us
e
r
in
te
nt
r
e
c
ogni
ti
on
[
26]
,
im
pr
ovi
ng
th
e
a
c
c
ur
a
c
y
by
28.0
pe
r
c
e
nt
a
ge
poi
nt
s
. T
hi
s
i
m
pr
ove
m
e
nt
i
s
a
tt
r
ib
ut
e
d t
o
th
e
a
bi
li
ty
o
f
t
he
e
m
b
e
ddi
ng mode
l
to
c
a
pt
ur
e
t
he
s
e
m
a
nt
ic
m
e
a
ni
ng
of
th
e
us
e
r
'
s
que
s
ti
ons
,
w
hi
c
h
ove
r
c
om
e
s
th
e
li
m
it
a
ti
ons
of
s
im
pl
e
ke
yw
or
d
or
to
ke
n
m
a
tc
hi
ng
in
h
e
r
e
nt
in
th
e
J
a
c
c
a
r
d
s
im
il
a
r
it
y m
e
th
od.
W
hi
le
t
he
s
e
m
a
nt
ic
ve
c
to
r
r
e
tr
ie
va
l
m
e
th
od of
f
e
r
e
d s
ubs
ta
nt
ia
l
a
c
c
ur
a
c
y i
m
pr
ove
m
e
nt
s
, i
t
in
tr
oduc
e
d
a
c
le
a
r
tr
a
de
-
of
f
in
pr
oc
e
s
s
in
g
ti
m
e
[
27]
.
K
e
yw
or
d
s
e
a
r
c
h
w
a
s
s
ig
ni
f
ic
a
nt
ly
f
a
s
te
r
,
w
it
h
a
n
a
ve
r
a
ge
la
te
nc
y
of
13.622
m
s
pe
r
s
a
m
pl
e
;
it
r
a
n
ne
a
r
ly
8x
f
a
s
te
r
c
om
pa
r
e
d
to
s
e
m
a
nt
ic
ve
c
to
r
r
e
tr
ie
va
l
w
it
h
a
n
a
ve
r
a
ge
la
te
nc
y
of
109.043
m
s
pe
r
s
a
m
pl
e
.
T
hi
s
hi
gh
e
f
f
ic
ie
nc
y
is
e
xpe
c
te
d,
a
s
J
a
c
c
a
r
d
s
im
il
a
r
it
y
in
vol
ve
s
onl
y
s
im
pl
e
s
tr
in
g
to
ke
ni
z
a
ti
on
a
nd
s
e
t
c
om
pa
r
is
on,
w
h
e
r
e
a
s
s
e
m
a
nt
ic
ve
c
to
r
r
e
tr
ie
va
l
r
e
qui
r
e
s
a
c
om
put
a
ti
ona
ll
y
in
te
n
s
iv
e
m
ode
l
in
f
e
r
e
nc
e
s
te
p
[
28]
.
T
a
bl
e
5. M
i
s
c
la
s
s
if
ie
d t
e
s
t
s
a
m
pl
e
s
T
r
ue
i
nt
e
nt
P
r
e
di
c
t
e
d i
nt
e
nt
O
r
i
gi
na
l
que
r
y (
w
i
t
h E
ng
l
i
s
h t
r
a
ns
l
a
t
i
on i
n pa
r
e
nt
he
s
e
s
)
D
oc
t
or
vi
s
i
t
s
c
he
dul
e
D
oc
t
or
pr
a
c
t
i
c
e
hour
s
Si
apa
s
aj
a
pas
i
e
n
s
ay
a
y
ang
ha
r
us
di
v
i
s
i
t
e
ha
r
i
i
ni
(
W
ho
a
r
e
m
y
pa
t
i
e
nt
s
s
c
h
e
dul
e
d
f
or
r
ounds
/
vi
s
i
t
s
t
oda
y)
D
oc
t
or
vi
s
i
t
s
c
he
dul
e
O
nl
i
ne
r
e
s
e
r
va
t
i
on
pa
t
i
e
nt
l
i
s
t
B
e
r
apa
j
um
l
ah
pas
i
e
n
y
ang
har
u
s
s
a
y
a
k
unj
ungi
s
o
r
e
i
ni
(
W
ha
t
i
s
t
he
num
b
e
r
of
pa
t
i
e
nt
s
I
ne
e
d t
o vi
s
i
t
t
hi
s
a
f
t
e
r
noon)
M
e
di
c
a
l
pr
oc
e
dur
e
de
t
a
i
l
s
C
ons
ul
t
e
d pa
t
i
e
nt
i
nf
or
m
a
t
i
on
B
agai
m
ana
pr
os
e
dur
pe
r
m
i
nt
aan
k
ons
ul
s
pe
s
i
al
i
s
m
e
nu
r
ut
panduan
k
l
i
ni
s
(
W
ha
t
i
s
t
he
pr
oc
e
dur
e
f
or
r
e
que
s
t
i
ng
a
s
pe
c
i
a
l
i
s
t
c
on
s
ul
t
a
t
i
on
a
c
c
or
di
ng
t
o
c
l
i
ni
c
a
l
gui
de
l
i
ne
s
)
C
ha
t
bot
f
e
a
t
ur
e
ove
r
vi
e
w
S
ur
ge
r
y s
c
he
dul
e
A
pak
ah
m
e
m
i
l
i
k
i
f
i
t
ur
unt
uk
m
e
l
i
ha
t
j
adw
al
ope
r
as
i
(
D
oe
s
i
t
ha
ve
a
f
e
a
t
ur
e
t
o
vi
e
w
t
he
s
ur
ge
r
y s
c
he
dul
e
)
C
ha
t
bot
f
e
a
t
ur
e
ove
r
vi
e
O
ut
pa
t
i
e
nt
nur
s
e
s
hi
f
t
s
c
he
dul
e
A
pak
ah
s
i
s
t
e
m
bi
s
a
m
e
ngi
r
i
m
not
i
f
i
k
as
i
k
e
pe
r
a
w
at
(
C
a
n
t
he
s
ys
t
e
m
s
e
nd
not
i
f
i
c
a
t
i
ons
t
o t
he
nur
s
e
s
)
C
ha
t
bot
f
e
a
t
ur
e
ove
r
vi
e
w
D
oc
t
or
f
e
e
dba
c
k on
c
ha
t
bot
us
a
g
C
hat
bot
i
ni
bi
s
a
m
e
nangani
pe
r
t
any
aan
apa
s
aj
a
(
W
ha
t
ki
nd
of
que
s
t
i
ons
c
a
n
t
hi
s
c
ha
t
bot
ha
ndl
e
)
M
e
di
c
a
l
s
e
r
vi
c
e
pa
ym
e
nt
de
t
a
i
l
S
ur
ge
r
y s
c
he
dul
e
B
e
r
apa
i
ns
e
nt
i
f
y
ang
s
ay
a
t
e
r
i
m
a
da
r
i
t
i
ndak
an
ope
r
as
i
m
i
nggu
l
al
u
(
W
ha
t
i
s
t
he
i
nc
e
nt
i
ve
/
r
e
m
une
r
a
t
i
on I
r
e
c
e
i
ve
d f
or
l
a
s
t
w
e
e
k'
s
s
ur
gi
c
a
l
pr
oc
e
dur
e
s
)
D
oc
t
or
f
e
e
dba
c
k on
c
ha
t
bot
us
a
ge
M
e
di
c
a
l
s
e
r
vi
c
e
pa
ym
e
nt
de
t
a
i
l
Say
a
t
any
a
t
e
nt
ang
j
as
a
m
e
di
s
t
api
j
aw
abanny
a
t
i
dak
j
e
l
as
(
I
a
s
ke
d
a
bout
m
e
di
c
a
l
s
e
r
vi
c
e
f
e
e
s
but
t
he
a
n
s
w
e
r
w
a
s
unc
l
e
a
r
)
T
a
bl
e
6
.
C
om
pa
r
is
on of
s
e
m
a
nt
ic
vs
k
e
yw
or
d s
e
a
r
c
h pe
r
f
or
m
a
n
c
e
M
e
t
r
i
c
V
e
c
t
or
s
e
a
r
c
h (
s
e
m
a
nt
i
c
,
C
os
i
ne
)
K
e
yw
or
d
s
e
a
r
c
h (
l
e
xi
c
a
l
,
J
a
c
c
a
r
d
)
A
c
c
ur
a
c
y
0.70
0.42
A
vg
.
l
a
t
e
nc
y
(
m
s
/
s
a
m
pl
e
)
109.043
13.622
3.4.
I
m
p
ac
t
of
s
im
il
ar
it
y t
h
r
e
s
h
ol
d
on
p
e
r
f
o
r
m
an
c
e
T
he
in
f
lu
e
nc
e
of
th
e
c
os
in
e
s
im
il
a
r
it
y
th
r
e
s
hol
d
(
τ)
on
th
e
s
e
m
a
nt
ic
ve
c
to
r
r
e
tr
ie
va
l
s
ys
te
m
'
s
pe
r
f
or
m
a
nc
e
w
a
s
a
s
s
e
s
s
e
d
by
pl
ot
ti
ng
th
e
pr
e
c
is
io
n
a
nd
r
e
c
a
l
l
a
c
r
os
s
a
r
a
nge
of
th
r
e
s
hol
ds
.
T
he
r
e
s
ul
ti
ng
c
ur
ve
s
,
s
ho
w
n
in
F
ig
ur
e
2,
de
m
ons
tr
a
te
th
e
c
la
s
s
ic
pr
e
c
i
s
io
n
-
r
e
c
a
ll
tr
a
de
-
of
f
.
A
s
th
e
th
r
e
s
hol
d
in
c
r
e
a
s
e
s
,
th
e
s
ys
te
m
be
c
om
e
s
s
tr
ic
te
r
,
c
a
us
in
g
r
e
c
a
ll
to
de
c
r
e
a
s
e
s
te
a
di
ly
,
w
hi
le
p
r
e
c
is
io
n
in
it
ia
ll
y
r
is
e
s
be
f
or
e
de
c
li
ni
ng
s
ig
ni
f
ic
a
nt
ly
a
t
ve
r
y hi
gh t
hr
e
s
hol
ds
.
T
he
m
os
t
ba
la
n
c
e
d
pe
r
f
or
m
a
nc
e
,
m
e
a
s
ur
e
d
by
th
e
F
1
-
s
c
or
e
,
is
a
c
hi
e
ve
d
w
he
r
e
th
e
two
c
ur
ve
s
in
te
r
s
e
c
t
or
a
r
e
c
lo
s
e
s
t.
T
hi
s
opt
im
a
l
ope
r
a
ti
ng
r
e
gi
on
w
a
s
id
e
nt
if
ie
d
w
it
hi
n
th
e
lo
w
e
r
th
r
e
s
hol
d
r
a
nge
,
s
pe
c
if
ic
a
ll
y
a
r
ound
τ=
0.51.
A
t
th
is
poi
nt
,
th
e
s
ys
te
m
m
a
xi
m
iz
e
s
bot
h
th
e
tr
ue
in
te
nt
c
a
pt
ur
e
r
a
te
(
r
e
c
a
ll
≈0.70)
a
nd
th
e
a
c
c
ur
a
c
y
of
th
o
s
e
pr
e
di
c
ti
ons
(
p
r
e
c
is
io
n≈0.68)
.
C
r
uc
ia
l
ly
,
th
r
e
s
hol
ds
a
bove
τ
=
0.70
r
e
s
ul
te
d
in
a
r
a
pi
d
c
ol
la
ps
e
of
r
e
c
a
ll
, i
ndi
c
a
ti
ng t
ha
t
th
e
s
y
s
te
m
w
oul
d f
r
e
que
nt
ly
f
a
il
t
o c
la
s
s
if
y l
e
gi
ti
m
a
te
us
e
r
que
r
ie
s
.
3.5. S
ys
t
e
m
d
e
s
ig
n
H
I
M
S
is
a
c
om
pl
ic
a
t
e
d
s
y
s
te
m
th
a
t
in
vol
ve
s
s
e
ve
r
a
l
u
s
e
r
s
w
it
h
di
ve
r
s
e
t
a
s
ks
.
U
s
e
r
-
f
r
ie
ndl
in
e
s
s
is
a
c
r
uc
ia
l
c
om
pone
nt
f
or
th
e
s
uc
c
e
s
s
f
ul
im
pl
e
m
e
nt
a
ti
on
of
H
I
M
S
[
29]
.
T
hi
s
pa
pe
r
c
onc
e
nt
r
a
te
s
on
de
v
e
lo
pi
ng
a
c
ha
tb
ot
a
ppl
ic
a
ti
on t
ha
t
f
ul
f
il
ls
a
c
r
it
ic
a
l
he
lp
de
s
k f
unc
ti
on f
or
doc
to
r
s
.
3.5.1.
C
h
at
b
ot
ge
n
e
r
al
ar
c
h
it
e
c
t
u
r
e
T
he
ge
ne
r
a
l
a
r
c
hi
te
c
tu
r
e
of
th
e
c
ha
tb
ot
is
il
lu
s
tr
a
te
d
in
F
ig
ur
e
3.
U
pon
r
e
c
e
iv
in
g
a
us
e
r
in
qui
r
y
,
th
e
s
ys
t
e
m
e
xa
m
in
e
s
th
e
r
e
que
s
t
to
a
s
c
e
r
ta
in
th
e
in
te
nde
d
pur
p
os
e
(
us
e
r
in
t
e
nt
)
.
T
he
s
ub
s
e
que
nt
a
c
ti
vi
ti
e
s
a
nd
a
nt
ic
ip
a
te
d
r
e
s
pon
s
e
s
,
w
hi
c
h
a
r
e
c
ont
in
ge
nt
upon
th
e
id
e
nt
if
ie
d
us
e
r
in
te
nt
,
c
a
n
th
e
n
be
e
xe
c
ut
e
d
a
s
pot
e
nt
ia
l
s
ol
ut
io
ns
.
R
e
s
pons
e
s
m
a
y
be
de
r
iv
e
d
f
r
om
s
e
ve
r
a
l
s
our
c
e
s
, s
uc
h
a
s
a
n
a
n
s
w
e
r
r
e
pos
it
or
y
w
it
hi
n
th
e
r
e
la
ti
ona
l
da
ta
ba
s
e
m
a
n
a
ge
m
e
nt
s
y
s
te
m
(
R
D
B
M
S
)
,
in
te
r
a
c
ti
on
w
it
h
a
n
e
xt
e
r
na
l
A
P
I
,
or
th
e
e
xe
c
ut
io
n
of
S
Q
L
que
r
ie
s
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. 6, D
e
c
e
m
be
r
2025
:
4600
-
4613
4606
a
ga
in
s
t
th
e
H
I
M
S
da
ta
ba
s
e
.
T
he
r
e
s
pon
s
e
ge
ne
r
a
ti
on
c
om
pone
nt
tr
a
ns
m
it
s
a
r
e
pl
y
to
th
e
us
e
r
ba
s
e
d
on
th
e
s
e
obt
a
in
e
d
c
a
ndi
da
te
r
e
s
pon
s
e
s
.
W
hi
le
th
e
c
ur
r
e
nt
s
ys
t
e
m
r
e
li
e
s
on
pr
e
de
f
in
e
d
or
r
e
tr
ie
ve
d
da
ta
,
a
ge
ne
r
a
ti
ve
tr
a
ns
f
or
m
e
r
m
ode
l
m
a
y
be
e
m
pl
oye
d
f
or
f
ut
ur
e
te
xt
ge
ne
r
a
ti
o
n
to
pr
oduc
e
va
r
ie
d
phr
a
s
e
f
or
m
s
[
30]
,
[
31]
,
e
nha
nc
in
g c
onve
r
s
a
ti
ona
l
n
a
tu
r
a
ln
e
s
s
.
F
ig
ur
e
2
.
T
hr
e
s
hol
d vs
ve
c
to
r
r
e
tr
ie
va
l
pe
r
f
or
m
a
nc
e
F
ig
ur
e
3
.
C
ha
tb
ot
ge
ne
r
a
l
a
r
c
hi
te
c
tu
r
e
3.5.2.
S
e
m
an
t
ic
ve
c
t
or
r
e
t
r
ie
val
s
ys
t
e
m
ar
c
h
it
e
c
t
u
r
e
F
ig
ur
e
4
il
lu
s
tr
a
te
s
th
e
s
pe
c
if
ic
a
r
c
hi
te
c
tu
r
e
us
e
d
f
or
th
e
c
or
e
in
te
nt
r
e
c
ogni
ti
on
ta
s
k.
T
he
s
ys
te
m
r
e
li
e
s
on
a
ve
c
to
r
da
ta
ba
s
e
m
a
na
g
e
m
e
nt
s
ys
te
m
(
M
il
vus
)
a
nd
t
he
c
e
nt
r
a
l
R
D
B
M
S
.
U
s
e
r
in
te
nt
de
f
in
it
io
ns
a
r
e
tr
a
ns
f
or
m
e
d
in
to
f
e
a
tu
r
e
ve
c
to
r
s
vi
a
a
tr
a
ns
f
or
m
e
r
-
ba
s
e
d
de
e
p
le
a
r
ni
ng
m
ode
l
a
nd
m
a
in
ta
in
e
d
in
th
e
M
il
vus
d
a
ta
ba
s
e
.
U
pon
r
e
c
e
iv
in
g
a
u
s
e
r
que
r
y,
th
e
tr
a
ns
f
or
m
e
r
m
ode
l
c
onve
r
ts
th
e
in
qui
r
y
in
to
a
f
e
a
tu
r
e
ve
c
to
r
a
nd
pe
r
f
or
m
s
a
s
im
il
a
r
it
y
ve
c
to
r
r
e
t
r
ie
va
l
w
it
hi
n
M
il
vus
to
id
e
nt
i
f
y
th
e
us
e
r
in
te
nt
m
os
t
a
ki
n
to
th
e
que
s
ti
on.
T
he
id
e
nt
if
ie
d
us
e
r
in
te
nt
I
D
is
th
e
n
tr
a
ns
m
it
te
d
to
th
e
R
D
B
M
S
to
f
e
tc
h
th
e
m
os
t
s
ui
ta
bl
e
c
or
r
e
s
ponding
a
c
ti
on or
r
e
s
pons
e
.
3.6. S
ys
t
e
m
d
e
ve
lo
p
m
e
n
t
an
d
i
m
p
le
m
e
n
t
at
io
n
T
he
c
ha
tb
ot
a
ppl
ic
a
ti
on
w
a
s
de
ve
lo
pe
d
us
in
g
th
e
O
r
a
c
le
A
pe
x
lo
w
-
c
ode
pl
a
tf
or
m
f
or
th
e
us
e
r
in
te
r
f
a
c
e
a
nd
qui
c
k
a
ppl
ic
a
ti
on
d
e
pl
oym
e
nt
[
32]
.
T
hi
s
s
e
le
c
ti
o
n
f
a
c
il
it
a
te
s
d
e
ve
lo
pm
e
nt
s
pe
e
d
a
nd
a
ll
ow
s
th
e
a
ppl
ic
a
ti
on
to
be
in
s
ta
ll
e
d
on
s
m
a
r
tp
hone
s
u
s
in
g
th
e
pr
ogr
e
s
s
iv
e
w
e
b
a
pps
(
P
W
A
)
f
e
a
tu
r
e
[
33]
,
[
34]
.
C
r
it
ic
a
l
ML
c
om
pone
nt
s
a
r
e
m
a
na
ge
d by a
P
yt
hon
-
ba
s
e
d A
P
I
[
35
]
us
in
g t
he
D
ja
ngo we
b f
r
a
m
e
w
or
k, w
hi
c
h pe
r
f
or
m
s
in
f
e
r
e
nc
e
a
nd ve
c
to
r
r
e
tr
ie
va
l.
3.6.1.
S
e
m
an
t
ic
r
e
t
r
ie
val
m
od
e
l
an
d
c
on
f
ig
u
r
at
io
n
T
he
de
e
p
le
a
r
ni
ng
m
ode
l
e
m
pl
oye
d
f
or
ve
c
to
r
iz
a
ti
on
i
s
th
e
pa
r
a
phr
a
s
e
-
m
ul
ti
li
ngua
l
-
m
pne
t
-
ba
s
e
-
v2
tr
a
ns
f
or
m
e
r
m
ode
l
[
14]
,
w
hi
c
h
c
onve
r
ts
te
xt
in
to
768
-
di
m
e
ns
io
na
l
f
e
a
tu
r
e
ve
c
to
r
s
.
F
or
th
e
ve
c
to
r
s
to
r
a
ge
a
nd
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
Se
m
ant
ic
s
e
ar
c
h
-
e
nhanc
e
d he
al
th
c
a
r
e
c
hat
bot
f
or
ho
s
pi
ta
l
in
fo
r
m
at
io
n
…
(
E
r
da G
us
li
nar
P
e
r
dana
)
4607
r
e
tr
ie
va
l,
th
e
ope
n
-
s
our
c
e
M
il
vus
ve
c
to
r
da
ta
ba
s
e
is
ut
il
iz
e
d,
e
m
pl
oyi
ng
a
c
os
in
e
s
im
il
a
r
it
y
m
e
tr
ic
f
or
th
e
s
e
m
a
nt
ic
ve
c
to
r
r
e
tr
ie
va
l
[
36]
.
T
he
e
nt
ir
e
a
r
c
hi
te
c
tu
r
e
,
c
om
pr
is
in
g
th
e
O
r
a
c
le
A
pe
x
s
e
r
ve
r
,
th
e
D
ja
ngo
A
P
I
,
a
nd t
he
M
il
vus
da
ta
b
a
s
e
, w
a
s
de
pl
oye
d
w
it
hi
n a
de
di
c
a
te
d G
oogl
e
C
lo
ud P
la
tf
or
m
(
G
C
P
)
e
nvi
r
onm
e
nt
.
A
s
e
ns
it
iv
it
y
a
na
ly
s
is
w
a
s
c
onduc
t
e
d
to
de
te
r
m
in
e
th
e
opt
im
a
l
s
im
il
a
r
it
y
th
r
e
s
hol
d
f
or
de
pl
oym
e
nt
.
B
a
s
e
d
on
th
is
a
na
ly
s
i
s
,
th
e
opt
im
a
l
th
r
e
s
hol
d
w
a
s
d
e
te
r
m
in
e
d
t
o
be
0.51.
T
hi
s
va
lu
e
w
a
s
s
e
le
c
te
d
be
c
a
us
e
it
m
a
xi
m
iz
e
s
th
e
F
1
-
s
c
or
e
,
r
e
pr
e
s
e
nt
in
g
th
e
b
e
s
t
op
e
r
a
ti
ona
l
ba
l
a
nc
e
be
twe
e
n
r
e
c
a
ll
(
th
e
a
bi
li
ty
to
r
e
tr
ie
ve
a
ll
r
e
le
va
nt
in
te
nt
s
)
a
nd
pr
e
c
is
io
n
(
th
e
a
c
c
ur
a
c
y
of
th
e
r
e
tr
ie
ve
d
i
nt
e
nt
s
)
.
T
he
c
hos
e
n
th
r
e
s
hol
d
of
0.51
e
ns
ur
e
s
r
obus
t
in
te
nt
c
a
pt
ur
e
w
hi
le
m
a
in
ta
in
in
g hi
gh pr
e
di
c
ti
ve
r
e
li
a
bi
li
ty
.
F
ig
ur
e
4. S
e
m
a
nt
ic
ve
c
to
r
r
e
tr
ie
va
l
a
r
c
hi
te
c
tu
r
e
3.6.2.
C
h
at
b
ot
ap
p
li
c
at
io
n
i
n
t
e
r
f
ac
e
s
T
he
c
ha
tb
ot
s
y
s
te
m
in
c
lu
de
s
bot
h
a
m
a
na
ge
m
e
nt
in
te
r
f
a
c
e
a
n
d
a
us
e
r
in
te
r
f
a
c
e
.
T
he
m
a
na
ge
m
e
nt
in
te
r
f
a
c
e
is
us
e
d
to
ove
r
s
e
e
,
upda
te
,
a
nd
ve
c
to
r
iz
e
th
e
knowle
dge
ba
s
e
.
I
t
a
ls
o
s
e
r
ve
s
a
s
th
e
in
te
r
f
a
c
e
f
or
th
e
AL
lo
op
,
r
out
in
g
f
la
gge
d,
hi
gh
-
va
lu
e
us
e
r
que
r
ie
s
(
id
e
nt
if
ie
d
vi
a
unc
e
r
ta
in
ty
m
e
tr
ic
s
or
us
e
r
e
s
c
a
la
ti
on)
to
e
xpe
r
ts
f
or
c
or
r
e
c
ti
on
a
nd
la
be
li
ng.
T
he
us
e
r
in
te
r
f
a
c
e
is
th
e
pl
a
tf
or
m
f
or
na
tu
r
a
l
la
ngua
ge
in
te
r
a
c
ti
on.
I
t
a
ll
ow
s
t
he
us
e
r
t
o i
nqui
r
e
a
nd r
e
c
e
iv
e
r
e
s
pon
s
e
s
b
a
s
e
d on th
e
id
e
nt
if
ie
d i
nt
e
nt
.
M
a
na
ge
m
e
nt
in
te
r
f
a
c
e
:
th
e
us
e
r
in
te
nt
m
a
na
ge
m
e
nt
f
or
m
(
F
ig
ur
e
5)
ove
r
s
e
e
s
in
te
nt
d
e
f
in
it
io
ns
,
r
e
que
s
t
s
a
m
pl
e
s
,
a
nd
a
ns
w
e
r
s
.
T
he
'
V
e
c
to
r
iz
e
a
ll
in
te
nt
'
but
to
n
c
onve
r
ts
a
ll
in
te
nt
s
a
m
pl
e
d
a
ta
in
to
f
e
a
tu
r
e
ve
c
to
r
s
f
or
s
to
r
a
ge
in
M
il
vus
.
T
he
in
te
nt
r
e
s
pons
e
in
te
r
f
a
c
e
(
F
ig
ur
e
6)
gove
r
ns
th
e
r
e
s
pons
e
ty
pe
s
(
e
.g.,
te
xt
,
S
Q
L
que
r
y,
A
P
I
,
U
R
L
,
or
b
ut
to
n)
a
nd
th
e
ir
s
e
que
nc
e
(
F
ig
ur
e
7)
.
U
s
e
r
i
nt
e
r
f
a
c
e
:
th
e
pr
im
a
r
y
a
ppl
ic
a
ti
on
in
te
r
f
a
c
e
(
F
ig
ur
e
8)
a
ll
ow
s
th
e
doc
to
r
to
in
put
in
qui
r
ie
s
in
na
tu
r
a
l
la
ngua
ge
,
de
m
ons
tr
a
ti
ng
th
e
s
ys
t
e
m
'
s
a
bi
li
ty
t
o r
e
s
pond to r
e
que
s
ts
a
r
ti
c
ul
a
te
d i
n a
bl
e
nd of
I
ndone
s
ia
n a
nd E
ngl
is
h.
F
ig
ur
e
5. U
s
e
r
i
nt
e
nt
m
a
na
ge
m
e
nt
:
in
te
nt
s
a
m
pl
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. 6, D
e
c
e
m
be
r
2025
:
4600
-
4613
4608
F
ig
ur
e
6. I
nt
e
nt
r
e
s
pons
e
F
ig
ur
e
7
.
I
nt
e
nt
r
e
s
pons
e
t
ype
F
ig
ur
e
8. C
ha
tb
ot
a
ppl
ic
a
ti
on
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
Se
m
ant
ic
s
e
ar
c
h
-
e
nhanc
e
d he
al
th
c
a
r
e
c
hat
bot
f
or
ho
s
pi
ta
l
in
fo
r
m
at
io
n
…
(
E
r
da G
us
li
nar
P
e
r
dana
)
4609
3.7.
E
val
u
at
io
n
U
s
a
bi
li
ty
e
va
lu
a
ti
on
is
c
onduc
te
d
by
th
e
f
or
m
ul
a
ti
on
of
a
que
s
ti
onna
ir
e
th
a
t
e
n
c
om
pa
s
s
e
s
e
f
f
ic
ie
nc
y,
e
f
f
e
c
ti
ve
ne
s
s
, a
nd
s
a
ti
s
f
a
c
ti
on.
E
a
c
h
f
a
c
e
t
ha
s
th
r
e
e
que
s
ti
ons
,
r
e
s
ul
ti
ng
in
a
to
ta
l
of
ni
ne
que
s
ti
ons
pe
r
ta
in
in
g
to
us
a
bi
li
ty
.
A
s
ur
ve
y
c
om
pl
e
te
d
by
th
e
r
e
s
ponde
nt
a
f
te
r
th
e
y
us
e
th
e
c
ha
tb
ot
a
ppl
ic
a
ti
on.
A
L
ik
e
r
t
s
c
a
le
r
a
ngi
ng
f
r
om
1
to
10
[
24]
w
a
s
e
m
pl
oye
d
to
a
s
s
e
s
s
r
e
s
po
nde
nt
s
'
le
ve
l
of
a
gr
e
e
m
e
nt
w
it
h
s
ta
te
m
e
nt
s
a
bout
us
a
bi
li
ty
.
T
he
ni
ne
que
s
ti
onna
ir
e
que
s
ti
ons
a
r
e
p
a
r
t
of
t
he
38 us
a
bi
li
ty
t
e
s
ti
ng a
tt
r
ib
ut
e
s
i
de
nt
if
ie
d i
n a
s
tu
dy by
N
ic
ol
e
a
nd
M
or
ga
n
[
25]
.
N
in
e
of
th
e
th
ir
ty
-
e
ig
ht
que
s
ti
ons
w
e
r
e
c
hos
e
n
ba
s
e
d
on
th
e
ir
r
e
le
va
nc
e
to
th
e
de
ve
lo
pm
e
nt
of
th
e
he
lp
de
s
k
c
ha
tb
ot
,
s
in
c
e
th
e
s
tu
dy
by
N
ic
ol
e
a
nd
M
or
ga
n
[
25]
a
im
e
d
to
a
s
s
e
s
s
c
onve
r
s
a
ti
ona
l
c
ha
tb
ot
s
.
T
he
li
s
t
of
th
e
ni
ne
que
s
ti
ons
c
a
n
be
s
e
e
n
in
T
a
bl
e
7.
T
h
e
r
e
s
ponde
nt
s
w
e
r
e
c
a
te
gor
iz
e
d
in
to
two
gr
oups
:
th
e
H
I
M
S
de
ve
lo
pm
e
nt
t
e
a
m
,
c
om
pr
is
in
g
6
in
di
vi
dua
ls
,
a
nd
doc
to
r
s
,
to
ta
li
ng
15
in
di
vi
dua
ls
.
T
h
e
s
e
le
c
ti
on
of
doc
to
r
s
a
s
r
e
s
ponde
nt
s
w
a
s
d
ue
to
th
e
c
ha
tb
ot
'
s
a
w
a
r
e
ne
s
s
of
th
e
s
pe
c
if
ie
d
us
e
r
i
nt
e
nt
, w
hi
c
h a
im
e
d t
o a
ddr
e
s
s
va
r
io
us
que
r
ie
s
f
r
om
doc
to
r
s
.
T
a
bl
e
7
.
L
is
t
of
que
s
ti
onna
ir
e
s
A
s
pe
c
t
C
ode
Q
ue
s
t
i
on/
s
t
a
t
e
m
e
nt
E
f
f
i
c
i
e
nc
y
EI
-
01
Q
ui
c
k i
n gi
vi
ng r
e
s
pons
e
E
f
f
i
c
i
e
nc
y
EI
-
02
A
bl
e
t
o ha
ndl
e
une
xpe
c
t
e
d r
e
que
s
t
s
E
f
f
i
c
i
e
nc
y
EI
-
03
F
a
c
i
l
i
t
a
t
e
s
r
e
que
s
t
e
s
c
a
l
a
t
i
on i
ns
i
de
t
he
t
i
c
k
e
t
i
ng pr
oc
e
dur
e
f
or
hum
a
n i
nt
e
r
ve
nt
i
on.
E
f
f
e
c
t
i
ve
ne
s
s
ES
-
01
A
c
c
ur
a
t
e
l
y i
nt
e
r
pr
e
t
us
e
r
r
e
que
s
t
s
E
f
f
e
c
t
i
ve
ne
s
s
ES
-
02
E
a
s
y t
o us
e
E
f
f
e
c
t
i
ve
ne
s
s
ES
-
03
A
bl
e
t
o pr
ovi
de
c
onvi
nc
i
ng, s
a
t
i
s
f
yi
ng,
a
nd
na
t
ur
a
l
i
nt
e
r
a
c
t
i
ons
S
a
t
i
s
f
a
c
t
i
on
KP
-
01
C
a
n a
s
c
e
r
t
a
i
n t
he
m
e
a
ni
ng or
i
nt
e
nt
of
a
us
e
r
'
s
i
nqui
r
y
S
a
t
i
s
f
a
c
t
i
on
KP
-
02
G
i
vi
ng gr
e
e
t
i
ngs
, pr
ovi
di
ng pl
e
a
s
a
nt
i
nt
e
r
a
c
t
i
ons
S
a
t
i
s
f
a
c
t
i
on
KP
-
03
P
r
ovi
di
ng a
di
ve
r
s
e
r
e
a
c
t
i
on
T
he
us
a
bi
li
ty
te
s
t
r
e
s
pond
e
nt
s
w
e
r
e
c
a
r
e
f
ul
ly
c
hos
e
n
f
r
om
two
di
s
ti
nc
t
gr
oups
to
pr
ovi
de
a
c
om
pr
e
he
ns
iv
e
e
va
lu
a
ti
on
pe
r
s
pe
c
ti
ve
:
6
in
di
vi
dua
ls
f
r
om
t
he
H
I
M
S
a
ppl
ic
a
ti
on
de
ve
lo
pm
e
nt
te
a
m
a
nd
15
m
e
m
be
r
s
of
th
e
doc
to
r
gr
oup
.
T
he
H
I
M
S
de
ve
lo
pm
e
nt
te
a
m
,
c
om
pr
is
in
g
s
of
twa
r
e
e
ngi
ne
e
r
s
a
nd
s
ys
te
m
a
r
c
hi
te
c
ts
,
w
a
s
s
e
le
c
te
d
due
to
th
e
ir
in
-
de
pt
h
te
c
hni
c
a
l
unde
r
s
ta
ndi
ng
of
th
e
s
ys
te
m
'
s
unde
r
ly
in
g
a
r
c
hi
te
c
tu
r
e
a
nd
f
unc
ti
ona
li
ty
.
T
he
ir
f
e
e
dba
c
k
is
c
r
uc
ia
l
f
o
r
id
e
nt
i
f
yi
ng
te
c
hni
c
a
l
us
a
bi
li
ty
is
s
ue
s
a
nd
va
li
da
ti
ng
th
e
s
ys
te
m
'
s
a
dhe
r
e
n
c
e
to
de
s
ig
n
s
pe
c
if
ic
a
ti
ons
.
T
he
ir
ba
c
kgr
ound
m
ig
ht
le
a
d
th
e
m
to
f
oc
us
m
or
e
on
e
f
f
ic
ie
nc
y,
pe
r
f
or
m
a
nc
e
,
a
nd
te
c
hni
c
a
l
r
obu
s
tn
e
s
s
.
C
onve
r
s
e
ly
,
th
e
doc
t
or
gr
oup
,
c
ons
is
ti
ng
of
m
e
di
c
a
l
pr
of
e
s
s
io
na
ls
w
ho
a
r
e
pr
im
a
r
y
e
nd
-
us
e
r
s
of
th
e
H
I
M
S
a
nd
th
e
ta
r
ge
t
us
e
r
s
f
or
th
is
he
lp
de
s
k
c
ha
tb
ot
,
w
e
r
e
c
ho
s
e
n
to
e
va
lu
a
te
th
e
c
ha
tb
ot
'
s
pr
a
c
ti
c
a
l
ut
il
it
y,
e
a
s
e
of
us
e
,
a
nd
e
f
f
e
c
ti
ve
ne
s
s
in
a
ddr
e
s
s
in
g
th
e
ir
da
il
y
ope
r
a
ti
ona
l
que
r
ie
s
.
T
o
obj
e
c
ti
ve
ly
de
te
r
m
in
e
if
th
e
obs
e
r
ve
d
di
f
f
e
r
e
nc
e
s
in
m
e
a
n
s
c
or
e
s
be
twe
e
n
th
e
two
gr
oups
w
e
r
e
s
ta
ti
s
ti
c
a
ll
y
r
e
li
a
bl
e
,
a
n
in
de
p
e
nde
nt
s
a
m
pl
e
s
t
-
te
s
t
(
s
pe
c
if
ic
a
ll
y
W
e
lc
h'
s
t
-
te
s
t,
due
to
une
qu
a
l
s
a
m
pl
e
s
iz
e
s
)
w
a
s
c
onduc
te
d
f
or
e
a
c
h
of
th
e
ni
ne
us
a
bi
li
ty
a
tt
r
ib
ut
e
s
.
T
he
t
-
te
s
t
c
om
pa
r
e
d
th
e
m
e
a
n
r
a
ti
ngs
of
th
e
H
I
M
S
d
e
ve
lo
pe
r
gr
oup (
N
=
6)
a
ga
in
s
t
th
e
doc
to
r
gr
oup
(
N
=
15)
, w
it
h a
s
ig
ni
f
ic
a
nc
e
l
e
ve
l
s
e
t
a
t
p<
0.05.
T
he
r
e
s
ul
t
s
of
th
e
que
s
ti
onna
ir
e
,
a
dm
in
is
te
r
e
d
to
21
r
e
s
ponde
nt
s
a
nd
il
lu
s
tr
a
te
d
in
F
ig
ur
e
9,
s
how
e
d
a
m
e
a
n
s
c
or
e
x
̅
=
7.78,
w
hi
c
h
is
in
te
r
pr
e
te
d
a
s
'
m
os
tl
y
a
gr
e
e
'
[
37]
.
H
ow
e
ve
r
,
th
e
in
de
pe
nde
nt
s
a
m
pl
e
s
t
-
te
s
t,
w
it
h
r
e
s
ul
ts
de
ta
il
e
d
in
T
a
bl
e
8,
c
onf
ir
m
e
d
th
a
t
none
of
th
e
s
e
obs
e
r
ve
d
di
f
f
e
r
e
nc
e
s
in
m
e
a
n
s
c
or
e
s
w
e
r
e
s
ta
ti
s
ti
c
a
ll
y
s
ig
ni
f
ic
a
nt
(
p>
0.05)
f
or
a
ny
of
th
e
ni
ne
us
a
bi
li
ty
a
tt
r
ib
ut
e
s
.
T
he
ove
r
a
ll
m
e
a
n s
c
or
e
s
,
a
s
s
how
n
in
T
a
bl
e
9,
in
di
c
a
te
d
th
a
t
th
e
e
f
f
e
c
ti
ve
ne
s
s
a
s
pe
c
t
e
xhi
bi
te
d
th
e
hi
ghe
s
t
m
e
a
n
va
lu
e
a
nd
e
f
f
ic
ie
nc
y
th
e
lo
w
e
s
t
i
n
bot
h r
e
s
ponde
nt
gr
oups
.
F
ig
ur
e
9. T
e
s
t
r
e
s
ul
t
Evaluation Warning : The document was created with Spire.PDF for Python.