T
E
L
K
O
M
NIKA
T
elec
o
mm
un
ica
t
io
n Co
m
pu
t
i
ng
E
lect
ro
nics
a
nd
Co
ntr
o
l
Vo
l.
23
,
No
.
6
,
Dec
em
b
er
20
25
,
p
p
.
1
4
9
5
~1
505
I
SS
N:
1
6
9
3
-
6
9
3
0
,
DOI
: 1
0
.
1
2
9
2
8
/
T
E
L
KOM
NI
K
A
.
v
2
3
i
6
.
27400
1495
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//jo
u
r
n
a
l.u
a
d
.
a
c.
id
/in
d
ex
.
p
h
p
/TELK
OM
N
I
K
A
Retr
iev
a
l
-
a
ug
m
en
ted
g
enera
tion for
Arabic leg
a
l in
for
m
a
tion
:
t
he f
a
m
ily
co
de c
a
se study
J
a
m
a
l H
ri
m
ec
h,
M
o
ha
m
m
e
d M
g
ha
ri,
Yo
us
s
ef
Z
a
z
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
,
F
a
c
u
l
t
y
o
f
S
c
i
e
n
c
e
,
A
b
d
e
l
mal
e
k
Essa
â
d
i
U
n
i
v
e
r
si
t
y
,
T
e
t
o
u
a
n
,
M
o
r
o
c
c
o
Art
icle
I
nfo
AB
ST
RA
C
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
J
u
l
15
,
2025
R
ev
i
s
ed
Sep
19
,
2
0
2
5
A
cc
ep
ted
Oct
19
,
2
0
2
5
T
h
is
d
o
c
u
m
e
n
t
d
e
sc
rib
e
s
t
h
e
im
p
lem
e
n
tatio
n
a
n
d
e
v
a
lu
a
ti
o
n
o
f
a
re
tri
e
v
a
l
-
a
u
g
m
e
n
ted
g
e
n
e
ra
ti
o
n
(RAG
)
s
y
s
tem
to
im
p
ro
v
e
a
c
c
e
s
s
to
a
n
d
u
n
d
e
rsta
n
d
in
g
o
f
M
o
ro
c
c
a
n
law
,
p
a
rt
icu
larly
t
h
e
f
a
m
il
y
c
o
d
e
in
A
ra
b
ic.
T
h
e
re
se
a
rc
h
a
d
d
re
ss
e
s
th
e
d
ra
w
b
a
c
k
s
o
f
th
e
w
id
e
l
y
u
s
e
d
li
n
g
u
isti
c
m
o
d
e
l
a
p
p
li
e
d
to
c
o
m
p
lex
l
e
g
a
l
ter
m
in
o
lo
g
y
in
A
ra
b
ic
a
n
d
a
im
s
to
h
e
lp
c
it
ize
n
s
a
c
c
e
ss
c
ru
c
ial
leg
a
l
d
a
ta.
We
b
u
il
t
a
n
e
w
c
u
sto
m
d
a
tas
e
t
w
it
h
2
.
5
k
q
u
e
sti
on
-
a
n
sw
e
r
p
a
irs
w
h
il
e
p
re
p
ro
c
e
ss
in
g
a
n
d
u
sin
g
th
e
BG
E
-
m
3
e
m
b
e
d
d
in
g
m
o
d
e
l
in
t
h
is
e
x
p
e
ri
m
e
n
t.
P
e
rf
o
rm
a
n
c
e
m
e
tri
c
s,
su
c
h
a
s
m
e
a
n
re
c
ip
ro
c
a
l
ra
n
k
(
M
RR
)
,
Re
c
a
ll
@k
,
a
n
d
F
1
-
sc
o
re
,
in
d
ic
a
te
th
a
t
th
e
RAG
a
p
p
ro
a
c
h
is
e
ffe
c
ti
v
e
c
o
m
p
a
re
d
to
th
e
u
se
o
f
sta
n
d
a
lo
n
e
larg
e
lan
g
u
a
g
e
m
o
d
e
l
s
(
LL
M
s
)
.
M
o
re
o
v
e
r,
a
n
e
v
a
lu
a
ti
o
n
o
n
m
e
tri
c
s
su
c
h
a
s
th
e
b
lu
e
s
c
o
re
,
f
id
e
li
ty
,
re
sp
o
n
se
re
lev
a
n
c
e
,
a
n
d
c
o
n
tex
tu
a
l
re
lev
a
n
c
e
in
d
ica
ted
th
a
t
th
e
m
a
tch
in
g
o
f
m
e
a
n
in
g
s
a
n
d
c
o
n
tex
t
w
e
r
e
we
ll
c
a
p
tu
re
d
,
w
h
ich
sig
n
if
ies
a
v
e
r
y
g
o
o
d
se
m
a
n
ti
c
u
n
d
e
rst
a
n
d
i
n
g
.
T
h
e
re
se
a
rc
h
h
ig
h
li
g
h
ts
th
e
n
e
e
d
f
o
r
lan
g
u
a
g
e
-
sp
e
c
if
ic
m
o
d
e
l
sp
e
c
ializa
ti
o
n
i
n
A
ra
b
ic
a
n
d
p
re
se
n
ts
it
s
m
a
in
c
h
a
ll
e
n
g
e
s,
su
c
h
a
s
d
iale
c
tal
v
a
riatio
n
s
a
n
d
a
p
p
ro
p
riate
e
v
a
lu
a
ti
o
n
m
e
a
su
re
s.
T
h
e
re
su
lt
s
in
d
ica
te
t
h
a
t
w
e
ll
-
d
e
v
e
lo
p
e
d
RAG
s
y
ste
m
s
o
ff
e
r
a
p
ro
m
isin
g
a
p
p
ro
a
c
h
to
im
p
ro
v
in
g
a
c
c
e
s
s
to
leg
a
l
in
f
o
rm
a
ti
o
n
in
A
ra
b
ic
-
sp
e
a
k
in
g
p
ra
c
ti
c
e
c
o
m
m
u
n
it
ies
a
n
d
to
g
u
i
d
in
g
f
u
tu
re
re
se
a
rc
h
a
n
d
d
e
v
e
lo
p
m
e
n
t
in
t
h
is
f
ield
.
K
ey
w
o
r
d
s
:
A
r
ab
ic
-
n
at
u
r
al
la
n
g
u
a
g
e
p
r
o
ce
s
s
in
g
L
ar
g
e
la
n
g
u
a
g
e
m
o
d
el
L
e
g
al
ac
ce
s
s
ib
ilit
y
Mo
r
o
cc
an
la
w
R
etr
iev
al
-
a
u
g
m
e
n
ted
g
e
n
er
ati
o
n
Se
m
a
n
tic
s
ea
r
c
h
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
:
J
am
al
Hr
i
m
ec
h
Dep
ar
t
m
en
t
o
f
C
o
m
p
u
ter
Scie
n
ce
,
Facu
l
t
y
o
f
Scie
n
ce
,
A
b
d
el
m
ale
k
E
s
s
aâ
d
i U
n
i
v
er
s
it
y
P
.
O.
B
o
x
.
2
1
2
1
,
M
’
Han
n
ec
h
I
I
,
T
etu
an
,
9
3
0
3
0
,
Mo
r
o
cc
o
E
m
ail:
j
a
m
al.
h
r
i
m
ec
h
@
et
u
.
u
a
e.
ac
.
m
a
1.
I
NT
RO
D
UCT
I
O
N
R
etr
iev
al
-
a
u
g
m
e
n
ted
g
e
n
er
ati
o
n
(
R
AG)
r
ep
r
esen
ts
o
n
e
s
u
c
h
p
o
w
er
f
u
l
an
d
cr
ea
tiv
e
r
esp
o
n
s
e
to
th
e
li
m
ita
tio
n
s
lar
g
e
p
r
e
-
tr
ain
ed
lan
g
u
a
g
e
m
o
d
els
s
u
f
f
er
f
r
o
m
w
h
e
n
ap
p
lied
to
k
n
o
w
led
g
e
-
r
i
ch
n
at
u
r
al
lan
g
u
a
g
e
p
r
o
ce
s
s
in
g
(
NL
P
)
tas
k
s
.
Alt
h
o
u
g
h
t
h
ese
m
o
d
els
h
a
v
e
s
h
o
w
n
r
e
m
ar
k
ab
le
ab
ilit
y
to
m
e
m
o
r
i
ze
f
ac
ts
,
th
e
y
o
f
ten
h
av
e
d
i
f
f
icu
lt
y
p
r
ec
is
el
y
q
u
er
y
in
g
t
h
at
s
a
m
e
k
n
o
w
led
g
e
,
wh
ich
m
a
n
i
f
est
s
as
“
h
all
u
ci
n
ati
o
n
s
”
(
th
e
g
e
n
er
atio
n
o
f
f
alse,
u
n
s
u
p
p
o
r
ted
f
ac
ts
)
a
n
d
o
p
ac
ity
in
th
eir
d
ec
is
io
n
-
m
ak
in
g
.
T
h
e
R
A
G
m
o
d
els
ad
d
r
ess
th
i
s
p
r
o
b
lem
b
y
co
m
b
i
n
i
n
g
p
ar
a
m
etr
ic
an
d
n
o
n
-
p
ar
a
m
etr
ic
m
e
m
o
r
y
s
y
s
te
m
s
.
I
n
p
ar
ticu
lar
,
R
A
G
is
t
h
e
co
m
p
o
s
itio
n
o
f
a
p
r
e
-
tr
ain
ed
g
en
er
ati
v
e
m
o
d
el
(
p
ar
a
m
etr
ic
m
e
m
o
r
y
)
an
d
a
r
etr
ie
v
ab
le
k
n
o
w
led
g
e
b
ase,
w
h
ic
h
is
u
s
u
all
y
a
d
en
s
e
v
ec
to
r
in
d
ex
o
f
d
o
cu
m
e
n
t
s
(
n
o
n
-
p
ar
a
m
etr
ic
m
e
m
o
r
y
)
.
S
u
c
h
a
h
y
b
r
id
m
ec
h
a
n
is
m
e
n
ab
les
m
o
d
els
to
d
y
n
a
m
icall
y
r
etr
iev
e
an
d
u
tili
ze
k
n
o
w
l
ed
g
e
f
r
o
m
th
e
e
x
ter
n
al
s
o
u
r
ce
w
h
il
e
g
en
er
ati
n
g
te
x
t,
w
h
ich
ca
n
m
ak
e
t
h
e
g
e
n
er
ated
r
esp
o
n
s
e
m
o
r
e
f
ac
t
u
al,
r
elate
d
an
d
v
ar
ied
.
T
h
is
h
a
s
a
f
e
w
i
m
p
o
r
tan
t
ad
v
a
n
tag
e
s
:
le
s
s
h
allu
ci
n
atio
n
,
b
etter
p
er
f
o
r
m
a
n
ce
o
n
a
v
ar
iet
y
o
f
NL
P
tas
k
s
,
ea
s
ier
w
o
r
ld
k
n
o
w
led
g
e
u
p
d
at
es
b
y
r
ep
laci
n
g
th
e
n
o
n
-
p
ar
a
m
etr
ic
m
e
m
o
r
y
i
n
d
ex
,
an
d
m
o
r
e
in
ter
p
r
etab
ilit
y
b
ec
au
s
e
w
e
ar
e
p
u
ll
in
g
u
p
h
u
m
a
n
-
r
ea
d
ab
le
d
o
cu
m
en
ts
.
As
a
b
en
e
f
it,
R
A
G
h
a
s
m
u
ch
f
e
w
er
tr
ain
ab
l
e
p
ar
am
eter
s
t
h
an
t
h
e
lar
g
e
p
ar
a
m
etr
ic
-
o
n
l
y
m
o
d
els an
d
ac
h
i
ev
es
m
o
r
e
ef
f
ec
ti
v
e
p
er
f
o
r
m
a
n
ce
[
1
]
.
A
d
d
it
i
o
n
a
lly
,
R
A
G
c
an
s
ig
n
if
i
c
an
t
ly
r
e
d
u
c
e
g
r
a
p
h
i
cs
p
r
o
c
e
s
s
in
g
u
n
it
(
G
PU
)
a
n
d
r
an
d
o
m
a
cc
e
s
s
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
1
6
9
3
-
6930
T
E
L
KOM
NI
K
A
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l
C
o
n
tr
o
l
,
Vo
l.
23
,
No
.
6
,
Dec
em
b
er
20
25
:
1
4
9
5
-
1
505
1496
m
em
o
r
y
(
R
A
M
)
c
o
s
ts
w
h
e
n
o
u
ts
o
u
r
c
in
g
k
n
o
w
l
e
d
g
e
t
o
s
ca
l
ab
l
e
d
a
t
a
s
t
o
r
es
(
as
o
p
p
o
s
e
d
t
o
m
o
d
e
l
w
e
ig
h
ts
)
,
an
d
r
o
b
u
s
t
s
c
a
lin
g
la
w
an
d
em
p
i
r
i
ca
l
r
e
s
u
l
ts
s
u
g
g
es
t
th
a
t
s
m
a
ll
e
r
m
o
d
e
ls
w
it
h
la
r
g
e
r
e
t
r
i
ev
a
l
i
n
d
i
c
e
s
a
r
e
c
o
m
p
e
ti
t
iv
e
w
ith
o
r
ev
en
o
u
t
p
e
r
f
o
r
m
la
r
g
e
r
f
u
l
ly
p
a
r
am
et
r
i
c
m
o
d
e
ls
a
t
a
f
r
ac
t
i
o
n
o
f
t
h
e
G
PU
m
em
o
r
y
c
o
s
t
[
2
]
.
Fu
r
t
h
er
m
o
r
e,
th
e
R
A
G
s
y
s
te
m
ca
n
ac
ce
s
s
an
d
in
te
g
r
ate
r
elate
d
in
f
o
r
m
atio
n
in
s
tr
u
ct
u
r
ed
d
ata,
d
o
cu
m
en
ts
o
r
d
atab
ases
,
s
o
t
h
at
th
e
i
n
f
o
r
m
atio
n
is
n
o
t
o
n
l
y
s
u
itab
le
f
o
r
th
e
p
r
e
-
s
ett
in
g
co
n
tex
t
o
f
t
h
e
m
o
d
el,
an
d
th
e
d
ata
s
o
u
r
ce
ca
n
b
e
r
ep
lace
d
f
lex
ib
l
y
ac
co
r
d
in
g
to
th
e
ac
tu
al
n
ee
d
f
o
r
th
e
co
n
v
er
s
io
n
o
f
t
h
e
k
n
o
w
led
g
e
o
f
th
e
lar
g
e
lan
g
u
ag
e
m
o
d
el
(
L
L
M
)
to
a
s
p
ec
if
ic
k
n
o
w
led
g
e
d
o
m
ai
n
.
Fro
m
a
R
A
G
s
tr
u
ct
u
r
e
p
er
s
p
ec
tiv
e,
t
h
er
e
ar
e
t
w
o
f
u
n
d
a
m
en
tal
co
m
p
o
n
en
t
s
.
First,
t
h
e
r
etr
iev
er
is
th
e
p
r
o
ce
s
s
o
f
f
i
n
d
in
g
an
d
ex
tr
ac
tin
g
r
elev
a
n
t
in
f
o
r
m
atio
n
f
r
o
m
a
lar
g
e
d
atas
et
o
r
k
n
o
w
led
g
e
b
ase
b
y
co
m
p
ar
in
g
t
h
e
q
u
e
s
tio
n
v
ec
to
r
an
d
d
ata
v
ec
to
r
s
to
f
i
n
d
th
e
clo
s
est
o
n
es
i
n
ter
m
s
o
f
m
ea
n
in
g
.
I
n
o
r
d
er
to
tr
a
n
s
f
o
r
m
t
h
e
d
ata
in
to
a
v
ec
to
r
,
a
p
r
elim
i
n
ar
y
p
r
o
ce
s
s
ca
lled
in
d
ex
i
n
g
is
m
a
n
d
ato
r
y
f
o
r
s
to
r
in
g
t
h
e
d
ata
in
a
v
ec
to
r
d
atab
ase.
T
h
is
s
er
v
e
s
to
o
p
tim
ize
th
e
s
ea
r
ch
f
u
n
ctio
n
alit
y
s
o
th
at
r
etr
ie
v
al
is
a
s
f
ast
a
n
d
e
f
f
icien
t
as
p
o
s
s
ib
le.
Seco
n
d
,
th
e
g
en
er
ato
r
is
u
s
u
all
y
t
h
e
L
L
M
th
at
le
v
er
ag
e
s
t
h
e
in
f
o
r
m
atio
n
r
etr
ie
v
ed
b
y
th
e
r
e
tr
iev
er
an
d
q
u
er
y
to
g
e
n
er
ate
a
co
h
er
en
t a
n
d
f
ac
tu
al
a
n
s
w
er
[
3
]
.
Giv
e
n
th
e
ab
ilit
y
o
f
R
A
G
s
y
s
te
m
s
i
n
i
m
p
r
o
v
i
n
g
th
e
ac
cu
r
a
c
y
o
f
L
L
M
s
,
th
e
y
h
av
e
b
ee
n
ap
p
lied
in
v
ar
io
u
s
f
ield
s
,
s
u
c
h
as
th
e
b
an
k
i
n
g
s
ec
to
r
,
a
R
A
G
s
y
s
te
m
n
a
m
ed
“
U
n
i
Ask
”
i
n
ten
d
ed
f
o
r
E
u
r
o
p
ea
n
b
an
k
e
m
p
lo
y
ee
s
ac
ce
s
s
in
g
d
o
cu
m
e
n
tatio
n
r
elate
d
to
p
o
licies,
r
eg
u
la
ti
o
n
s
an
d
p
r
o
ce
s
s
e
s
[
4
]
.
I
n
th
e
h
ea
l
th
a
n
d
m
ed
icin
e
s
ec
to
r
,
i
n
p
ar
ticu
lar
f
o
r
t
h
e
p
u
r
p
o
s
e
o
f
in
ter
p
r
etin
g
r
ec
o
m
m
en
d
a
tio
n
s
,
a
s
s
i
s
ti
n
g
w
it
h
d
ia
g
n
o
s
i
s
,
s
elec
ti
n
g
elig
ib
ilit
y
f
o
r
clin
ic
al
tr
ials
,
s
ea
r
ch
i
n
g
f
o
r
clin
ic
al
in
f
o
r
m
at
io
n
a
n
d
ex
tr
ac
tin
g
in
f
o
r
m
atio
n
f
r
o
m
s
cien
t
if
ic
liter
at
u
r
e
[
5
]
.
I
n
th
e
f
i
n
an
cia
l
s
ec
to
r
,
a
R
A
G
p
ip
elin
e
h
as
b
ee
n
d
ev
elo
p
ed
f
o
r
a
c
o
m
p
ar
ati
v
e
s
t
u
d
y
o
f
r
etr
iev
al
an
d
p
r
o
m
p
ti
n
g
s
tr
at
eg
ies
i
n
f
i
n
a
n
cial
q
u
ali
t
y
as
s
u
r
an
ce
tas
k
s
,
i
n
f
o
r
m
i
n
g
p
r
o
d
u
ctio
n
d
ep
lo
y
m
en
t
r
ec
o
m
m
e
n
d
atio
n
s
,
alt
h
o
u
g
h
th
e
m
ain
f
o
cu
s
is
o
n
t
h
e
an
al
y
s
i
s
o
f
co
n
tr
o
lled
co
m
p
o
n
e
n
t
s
[
6
]
.
I
n
th
e
leg
al
f
ield
,
an
i
m
p
le
m
e
n
tatio
n
o
f
t
h
e
R
A
G
s
y
s
te
m
to
ass
i
s
t
co
n
te
n
t
cr
e
ato
r
s
in
d
is
p
u
te
s
r
elate
d
to
th
e
f
air
u
s
e
d
o
ctr
in
e
in
Am
er
ica
n
co
p
y
r
ig
h
t
la
w
,
w
h
i
ch
co
m
b
i
n
es
s
ev
er
al
e
le
m
e
n
t
s
(
s
e
m
a
n
tic
s
ea
r
ch
,
j
u
d
icial
cit
atio
n
n
e
t
w
o
r
k
s
,
a
n
d
leg
al
k
n
o
w
led
g
e
g
r
ap
h
s
)
,
th
is
w
o
r
k
i
m
p
r
o
v
es
t
h
e
leg
al
r
ele
v
an
ce
o
f
th
e
d
o
cu
m
e
n
ts
f
o
u
n
d
,
w
h
ic
h
is
cr
u
cial
f
o
r
a
s
o
lid
f
air
u
s
e
d
e
f
en
s
e
[
7
]
.
I
n
th
is
p
ap
er
w
e
w
il
l
i
m
p
le
m
e
n
t
a
R
A
G
s
y
s
te
m
i
n
t
h
e
Mo
r
o
cc
an
le
g
al
co
n
tex
t,
p
ar
tic
u
lar
l
y
th
e
f
a
m
il
y
co
d
e,
w
h
ic
h
w
ill
f
ac
ilit
ate
cit
i
ze
n
s
’
u
n
d
er
s
ta
n
d
in
g
o
f
t
h
e
la
w
.
W
e
w
ill
u
s
e
m
u
lt
ili
n
g
u
al
e
m
b
ed
d
in
g
m
o
d
els
(
o
p
en
s
o
u
r
ce
)
p
r
o
v
in
g
th
e
ir
p
er
f
o
r
m
an
ce
i
n
t
h
e
A
r
ab
ic
lan
g
u
ag
e
f
o
r
th
e
r
ec
o
v
er
y
o
f
r
elev
a
n
t d
o
cu
m
e
n
t
s
,
an
d
a
lar
g
e
lan
g
u
a
g
e
m
o
d
el
as
a
g
e
n
er
ato
r
.
T
h
e
o
b
j
ec
tiv
e
o
f
th
is
wo
r
k
is
p
ar
t
o
f
th
e
r
ig
h
t
o
f
a
cc
es
s
to
in
f
o
r
m
a
tio
n
an
d
th
e
s
i
m
p
li
f
icat
io
n
o
f
j
u
d
icial
p
r
o
ce
d
u
r
es a
n
d
p
r
o
ce
s
s
es.
De
s
p
ite
its
p
r
o
v
e
n
s
u
cc
es
s
,
R
AG
h
as
y
et
to
b
e
ap
p
lied
to
th
e
m
o
r
p
h
o
s
y
n
tactic
a
n
d
s
e
m
an
tic
in
tr
icac
ie
s
o
f
A
r
ab
ic
leg
al
lan
g
u
ag
e.
C
h
alle
n
g
es
i
n
c
lu
d
e
h
a
n
d
lin
g
l
eg
a
l
ter
m
i
n
o
lo
g
y
,
m
ai
n
tai
n
i
n
g
th
e
co
n
ten
t
v
alid
it
y
o
f
leg
a
l c
lau
s
es r
ep
r
o
d
u
ce
d
f
r
o
m
r
etr
ie
v
al
s
y
s
te
m
s
,
a
n
d
cr
ea
ti
n
g
r
eliab
le
ev
al
u
atio
n
s
ets,
as w
e
l
l
as
t
h
e
s
ca
r
cit
y
o
f
a
n
n
o
tated
A
r
ab
ic
le
g
al
co
r
p
o
r
a.
T
o
f
ill
t
h
is
lac
u
n
a,
th
i
s
p
ap
er
ai
m
s
to
an
s
w
er
t
h
e
f
o
ll
o
w
i
n
g
r
esear
ch
q
u
e
s
tio
n
s
:
w
h
at
i
s
th
e
b
est
w
a
y
to
ch
u
n
k
le
g
al
d
ata
in
a
w
a
y
t
h
at
ac
k
n
o
w
led
g
es
s
e
m
a
n
tic
r
ep
r
es
en
tatio
n
s
?
d
o
s
tan
d
ar
d
ized
m
et
r
ics
p
r
o
p
er
ly
m
ea
s
u
r
e
t
h
e
q
u
ali
t
y
o
f
t
h
e
g
e
n
er
ated
leg
ald
o
cu
m
en
ts
i
n
A
r
ab
ic?
an
d
w
h
ic
h
o
p
en
-
s
o
u
r
ce
m
u
l
tili
n
g
u
al
e
m
b
ed
d
in
g
m
o
d
el
ac
h
ie
v
e
s
t
h
e
b
est
p
er
f
o
r
m
a
n
ce
f
o
r
th
e
r
etr
iev
al
o
f
th
e
r
ele
v
an
t a
r
ticles o
f
t
h
e
f
a
m
il
y
co
d
e?
T
o
ad
d
r
ess
th
ese
q
u
es
tio
n
s
,
t
h
e
k
e
y
co
n
tr
ib
u
tio
n
s
o
f
t
h
i
s
p
ap
er
ar
e
as
f
o
llo
w
s
:
t
o
th
at
en
d
,
w
e
co
m
p
ile
a
n
e
w
co
r
p
u
s
o
f
2
,
5
0
0
ar
ab
ic
q
u
esti
o
n
-
a
n
s
w
er
p
air
s
f
o
c
u
s
ed
o
n
t
h
e
Mo
r
o
cc
an
f
a
m
il
y
c
o
d
e,
o
b
tain
ed
b
y
a
s
e
m
i
-
au
to
m
at
ic
p
ip
elin
e
a
n
d
m
an
u
a
ll
y
c
h
ec
k
ed
.
T
h
e
s
tr
u
ct
u
r
e
o
f
t
h
is
d
ata
s
et,
w
h
ic
h
i
n
clu
d
es
th
e
q
u
esti
o
n
,
s
o
u
r
ce
,
an
d
r
ef
er
en
ce
a
n
s
w
er
,
is
ill
u
s
tr
ated
in
Fig
u
r
e
1
.
F
ig
u
r
e
1
.
D
a
t
as
e
t
s
t
r
u
ct
u
r
e
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KO
MN
I
K
A
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l
C
o
n
tr
o
l
R
etri
ev
a
l
-
a
u
g
men
ted
g
en
era
ti
o
n
fo
r
A
r
a
b
ic
leg
a
l in
fo
r
ma
tio
n
:
t
h
e
fa
mily
co
d
e
ca
s
e
s
tu
d
y
(
Ja
ma
l H
r
imec
h
)
1497
W
e
p
r
o
p
o
s
e
a
clau
s
e
-
b
ased
d
o
cu
m
e
n
t
d
iv
is
io
n
ap
p
r
o
ac
h
th
at
is
m
o
r
e
s
e
m
a
n
tica
ll
y
r
elev
a
n
t
th
a
n
th
e
s
tan
d
ar
d
ch
a
r
ac
ter
o
r
to
k
en
-
b
a
s
ed
ap
p
r
o
ac
h
f
o
r
leg
al
te
x
ts
.
W
e
p
r
esen
t
a
co
m
p
ar
ativ
e
a
n
a
l
y
s
i
s
o
f
s
tate
-
of
-
t
h
e
-
ar
t
m
u
lt
ilin
g
u
al
e
m
b
ed
d
in
g
m
o
d
els
o
n
leg
al
r
etr
iev
al
f
o
r
th
e
A
r
ab
ic
lan
g
u
a
g
e
an
d
a
d
is
cu
s
s
io
n
o
f
ev
a
lu
at
io
n
m
etr
ics
w
i
th
f
o
cu
s
o
n
th
e
c
h
al
len
g
e
s
o
f
t
h
e
A
r
ab
ic
leg
al
d
o
m
ai
n
.
T
h
e
r
est
o
f
th
is
p
ap
er
f
o
llo
ws
a
p
atter
n
:
s
ec
tio
n
2
d
i
s
cu
s
s
es
r
elate
d
w
o
r
k
.
Sectio
n
3
ad
o
p
ts
th
e
m
et
h
o
d
o
lo
g
y
u
s
ed
i
n
t
h
i
s
p
a
p
er
.
Sectio
n
4
p
r
esen
t
s
t
h
e
e
x
p
er
i
m
e
n
ts
an
d
e
x
p
lai
n
s
t
h
e
r
esu
lt
s
.
Sectio
n
5
s
u
m
m
ar
izes t
h
e
co
n
cl
u
s
io
n
s
a
n
d
k
e
y
id
ea
s
,
as
w
ell
as
s
u
g
g
e
s
tio
n
s
th
at
f
u
tu
r
e
r
esear
c
h
er
s
c
o
u
ld
f
o
llo
w
.
2.
RE
L
AT
E
D
WO
RK
R
ec
en
t
y
ea
r
s
h
a
v
e
s
ee
n
t
h
e
e
m
er
g
e
n
ce
o
f
r
esear
ch
ap
p
l
y
in
g
au
g
m
e
n
ted
r
etr
iev
al
-
g
en
er
at
io
n
(
A
R
G
)
ar
ch
itect
u
r
es
s
y
s
te
m
s
t
h
at
co
m
b
in
e
n
e
u
r
al
r
etr
ie
v
al
o
n
s
e
m
an
tic
e
m
b
ed
d
in
g
’
s
w
i
th
lar
g
e
lan
g
u
a
g
e
m
o
d
el
tex
t
g
en
er
atio
n
f
o
r
A
r
ab
ic
lan
g
u
a
g
e
task
s
.
El
-
B
elta
g
y
a
n
d
A
b
d
all
ah
[
8
]
p
r
esen
t
a
co
m
p
r
eh
e
n
s
i
v
e
ca
s
e
s
tu
d
y
o
n
th
e
i
m
p
le
m
en
ta
tio
n
a
n
d
e
v
al
u
atio
n
o
f
R
AG
f
o
r
A
r
ab
ic
tex
ts
.
T
h
eir
w
o
r
k
ex
p
lo
r
es
t
h
e
p
er
f
o
r
m
an
ce
o
f
d
if
f
er
en
t
s
e
m
a
n
tic
e
m
b
ed
d
in
g
m
o
d
els (
a
m
o
n
g
th
e
m
m
u
lt
ili
n
g
u
al
m
o
d
els lik
e
A
r
aB
er
t a
n
d
C
o
h
er
e
1
)
f
o
r
r
etr
iev
al,
an
d
a
f
e
w
lo
n
g
-
ter
m
tr
an
s
latio
n
L
L
M
m
o
d
els
f
o
r
g
en
er
at
io
n
.
T
h
is
w
o
r
k
al
s
o
in
v
e
s
ti
g
ates
t
h
e
d
i
f
f
ic
u
lties
o
f
d
ialec
tal
v
ar
iatio
n
s
b
et
w
ee
n
d
o
cu
m
e
n
ts
a
n
d
q
u
er
ies,
an
d
s
h
o
w
s
th
at
d
esp
ite
th
e
ch
alle
n
g
es,
it
w
as
p
o
s
s
ib
le
to
co
m
p
o
s
e
R
A
G
p
ip
elin
e
f
o
r
A
r
ab
ic
b
y
f
u
s
i
n
g
ex
i
s
ti
n
g
s
e
m
a
n
tic
e
m
b
ed
d
in
g
s
,
a
n
d
L
L
Ms.
A
b
d
elaz
i
m
et
a
l.
[
9
]
t
h
e
tea
m
co
n
ce
n
tr
ated
o
n
t
h
e
co
r
e
g
en
er
atio
n
p
ar
t
o
f
A
r
ab
ic
R
A
G
s
y
s
te
m
s
,
a
n
d
co
m
p
ar
ed
ten
s
ta
te
-
of
-
th
e
-
ar
t
m
u
ltil
i
n
g
u
al
s
e
m
a
n
tic
e
m
b
e
d
d
in
g
m
o
d
els
to
j
u
d
g
e
a
v
ailab
le
A
r
ab
ic
s
ea
r
ch
-
g
e
n
er
atio
n
o
u
tp
u
ts
a
n
d
p
er
f
o
r
m
a
n
ce
.
On
th
e
A
r
ab
ic
r
ea
d
in
g
co
m
p
r
eh
e
n
s
io
n
d
ataset
(
AR
C
D
)
test
b
ed
,
th
e
y
r
ep
o
r
ted
th
at
th
e
Mic
r
o
s
o
f
t
E
5
s
en
ten
ce
e
m
b
ed
d
in
g
m
o
d
el
ac
h
iev
ed
t
h
e
h
i
g
h
est
r
ec
all
r
a
te
o
f
1
0
,
w
ith
ab
o
v
e
9
0
%.
T
h
is
w
o
r
k
e
m
p
h
asize
s
th
e
n
ec
e
s
s
it
y
o
f
u
s
i
n
g
A
r
ab
ic
-
ap
p
r
o
p
r
iate
m
u
ltil
in
g
u
al
w
o
r
d
e
m
b
ed
d
in
g
’
s
f
o
r
A
r
ab
ic
s
ea
r
ch
tas
k
s
.
A
l
s
h
a
m
m
ar
y
et
a
l.
[
1
0
]
t
h
ey
in
tr
o
d
u
ce
t
h
e
r
etr
iev
al
f
ac
t
-
c
h
ec
k
i
n
g
p
r
o
m
p
t
g
en
er
atio
n
(
R
FP
G
)
f
r
a
m
e
w
o
r
k
,
a
f
ac
t
-
a
w
ar
e
R
A
G
ap
p
r
o
ac
h
to
q
u
esti
o
n
an
s
w
er
i
n
g
i
n
A
r
ab
ic,
a
lo
w
-
r
e
s
o
u
r
ce
l
an
g
u
a
g
e.
T
h
e
f
ac
t
-
ch
ec
k
i
n
g
is
p
er
f
o
r
m
ed
in
an
a
s
y
m
m
etr
ic
m
u
lt
i
-
s
ta
g
e
p
ip
elin
e
r
etr
iev
al
s
et
u
p
an
d
th
e
g
en
er
atio
n
is
co
n
d
itio
n
ed
o
n
p
er
s
o
n
alize
d
p
r
o
m
p
ts
.
T
h
e
R
FP
G
m
o
d
el
d
e
m
o
n
s
tr
ated
h
ig
h
ac
c
u
r
ac
y
(
1
0
0
%)
in
an
s
w
er
in
g
1
2
3
A
r
ab
ic
q
u
esti
o
n
s
,
as
w
ell
as
a
s
o
u
r
ce
citatio
n
ac
cu
r
ac
y
o
f
9
8
%,
o
u
tp
er
f
o
r
m
in
g
t
h
e
s
tan
d
ar
d
R
AG
an
d
r
ec
en
t
L
L
Ms
s
u
c
h
as
GP
T
-
4
an
d
GPT
-
4
o
i
n
th
eir
e
x
p
er
i
m
e
n
ts
.
Ho
w
e
v
er
,
th
e
s
co
p
e
o
f
th
eir
ev
al
u
atio
n
w
as
li
m
ited
to
a
co
n
tr
o
lled
q
u
esti
o
n
s
e
t r
ath
er
t
h
an
o
p
en
o
r
leg
al
co
r
p
o
r
a.
C
u
r
r
e
n
tl
y
r
esear
ch
t
h
at
tar
g
e
ts
R
A
G
in
t
h
e
leg
al
f
ie
ld
in
v
o
lv
e
s
m
o
s
tl
y
r
etr
ie
v
al
r
ath
e
r
th
an
f
u
ll
g
en
er
atio
n
p
ip
elin
e
s
.
J
af
ar
et
a
l.
[
1
1
]
p
r
esen
t
a
n
e
w
ap
p
r
o
ac
h
to
au
to
m
ate
r
etr
ie
v
al
o
f
A
r
a
b
ic
leg
al
te
x
ts
u
s
in
g
u
n
s
u
p
er
v
is
ed
to
p
ic
m
o
d
eli
n
g
(
T
o
p
2
Vec
)
an
d
d
en
s
it
y
-
b
ased
clu
s
te
r
i
n
g
(
HDB
S
C
A
N)
.
T
h
e
au
th
o
r
s
ad
d
r
ess
t
h
e
m
o
r
p
h
o
lo
g
ical
co
m
p
lex
it
y
a
n
d
a
m
b
ig
u
it
y
o
f
A
r
ab
ic
leg
al
tex
t
s
th
r
o
u
g
h
p
r
ep
r
o
ce
s
s
i
n
g
s
tep
s
s
u
c
h
as
n
o
r
m
aliza
t
io
n
a
n
d
to
k
e
n
izatio
n
.
B
u
t
th
e
ir
s
y
s
te
m
w
h
i
le
ac
h
iev
in
g
a
d
o
cu
m
e
n
t
r
etr
iev
al
r
ate
o
f
8
7
%
a
n
d
a
co
v
er
ag
e
r
ate
f
o
r
8
0
%,
lack
s
t
h
e
g
e
n
er
ati
v
e
ca
p
ab
ilit
y
n
ee
d
e
d
f
o
r
f
u
ll
R
A
G.
S
e
v
e
r
a
l
m
o
r
e
g
e
n
e
r
a
l
l
e
g
a
l
t
e
c
h
n
o
l
o
g
y
s
t
u
d
i
e
s
h
a
v
e
i
n
v
e
s
t
i
g
a
t
e
d
o
r
p
r
o
p
o
s
e
d
R
AG
p
i
p
e
l
i
n
e
s
f
o
r
l
e
g
a
l
t
e
x
t
s
,
b
u
t
g
e
n
e
r
a
l
l
y
n
o
t
f
o
c
u
s
e
d
o
n
Ar
a
b
i
c
.
F
o
r
e
x
a
m
p
l
e
,
H
i
n
d
i
e
t
a
l.
[
1
2
]
p
r
o
v
i
d
e
a
s
y
s
t
e
m
a
t
i
c
s
u
r
v
e
y
o
f
e
x
i
s
t
i
n
g
R
AG
a
r
c
h
i
t
e
c
t
u
r
e
s
i
n
t
h
e
l
e
g
a
l
d
o
m
a
i
n
,
d
i
s
c
u
s
s
i
n
g
r
e
t
r
i
e
v
a
l
m
e
t
h
o
d
s
,
e
v
a
l
u
a
t
i
o
n
m
e
t
r
i
c
s
,
a
n
d
e
t
h
i
c
a
l
c
o
n
s
i
d
er
a
t
io
n
s
,
b
u
t d
o
n
o
t
r
ep
o
r
t d
i
r
e
c
t e
x
p
e
r
i
m
e
n
t
s
o
n
Ar
a
b
i
c
l
e
g
a
l
c
o
r
p
o
r
a
.
W
a
h
i
d
u
r
e
t
a
l
.
[
1
3
]
a
n
d
K
a
l
r
a
e
t
a
l
.
[
1
4
]
i
n
t
r
o
d
u
c
e
R
A
G
f
r
a
m
e
w
o
r
k
s
t
a
i
l
o
r
ed
to
t
h
e
le
g
a
l
d
o
m
a
i
n
,
w
i
t
h
i
n
n
o
v
a
t
i
o
n
s
s
u
c
h
a
s
r
e
c
u
r
s
i
v
e
f
e
e
d
b
a
c
k
o
r
a
d
a
p
t
i
v
e
h
y
b
r
i
d
r
e
t
r
i
e
v
a
l
,
b
u
t
t
h
e
i
r
w
o
r
k
d
o
e
s
n
o
t
f
o
c
u
s
o
n
Ar
a
b
i
c
o
r
u
s
e
p
r
i
m
a
r
y
Ar
a
b
i
c
l
e
g
a
l
d
o
c
u
m
e
n
t
s
.
Giv
e
n
t
h
e
lac
k
o
f
p
r
ev
io
u
s
s
tu
d
ies
o
n
t
h
e
ap
p
licatio
n
o
f
R
AG
s
y
s
te
m
s
in
t
h
e
f
ield
o
f
Mo
r
o
cc
an
la
w
,
a
d
ir
ec
t
co
m
p
ar
is
o
n
is
h
ar
d
l
y
p
o
s
s
ib
le.
Ho
w
e
v
er
,
w
e
ca
n
p
u
t
o
u
r
r
esu
lt
s
in
t
h
e
b
r
o
ad
er
co
n
te
x
t
o
f
leg
al
ar
ti
f
icial
in
telli
g
e
n
ce
(
AI
)
r
esear
ch
.
F
o
r
ex
a
m
p
le,
w
o
r
k
o
n
th
e
le
g
al
g
en
er
al
la
n
g
u
a
g
e
u
n
d
er
s
tan
d
in
g
e
v
al
u
atio
n
(
L
ex
GL
UE
)
b
en
ch
m
ar
k
h
as
a
lr
ea
d
y
n
o
ted
th
e
co
m
p
lex
i
t
y
o
f
E
n
g
lis
h
leg
al
la
n
g
u
a
g
e
f
o
r
s
tan
d
ar
d
lan
g
u
a
g
e
m
o
d
el
s
[
1
5
]
.
On
th
e
o
th
er
h
an
d
,
th
e
cr
ea
tio
n
o
f
s
p
ec
ialized
q
u
esti
o
n
-
a
n
s
w
er
in
g
co
r
p
o
r
a
f
r
o
m
co
m
p
le
x
E
u
r
o
p
ea
n
r
eg
u
lat
io
n
s
s
u
c
h
as
th
e
g
e
n
er
al
d
ata
p
r
o
tectio
n
r
e
g
u
la
tio
n
(
GDP
R
)
[
1
6
]
,
is
co
n
s
id
er
ed
an
im
p
o
r
tan
t
s
tep
f
o
r
th
e
d
e
v
elo
p
m
en
t
o
f
r
eliab
le
q
u
esti
o
n
a
n
d
an
s
w
er
i
n
g
(
Q
&A
)
s
y
s
te
m
s
.
O
u
r
ap
p
r
o
ac
h
is
i
n
li
n
e
w
it
h
s
y
s
te
m
s
s
u
ch
as
C
h
at
L
a
w
[
1
7
]
fo
r
C
h
in
ese
la
w
.
Un
s
u
r
p
r
is
in
g
l
y
,
C
h
at
L
a
w
also
co
n
f
ir
m
s
th
e
n
ee
d
to
in
teg
r
ate
ex
ter
n
al
k
n
o
w
led
g
e
to
r
ed
u
ce
f
ac
t
u
al
er
r
o
r
s
m
ad
e
b
y
L
L
M
s
i
n
C
h
i
n
a.
Fin
a
ll
y
,
th
e
p
er
f
o
r
m
a
n
ce
o
f
o
u
r
r
esear
c
h
m
o
d
el
s
ca
n
b
e
co
m
p
ar
ed
to
th
e
r
ep
o
r
ts
o
f
p
r
ev
io
u
s
lar
g
e
in
f
o
r
m
at
io
n
r
etr
i
ev
al
te
s
t
s
id
en
ti
f
ied
.
T
h
is
d
em
o
n
s
tr
ate
s
th
e
r
o
b
u
s
tn
e
s
s
o
f
th
e
s
e
m
o
d
els e
v
e
n
w
h
e
n
ap
p
lied
to
a
f
ield
as sp
ec
ialized
as
la
w
.
T
o
p
r
o
v
id
e
a
clea
r
co
m
p
ar
is
o
n
w
i
th
p
r
io
r
ar
t,
T
a
b
le
1
s
u
m
m
a
r
izes
th
e
m
o
s
t
r
ele
v
an
t
s
t
u
d
ies d
is
cu
s
s
ed
in
t
h
e
p
r
ev
io
u
s
s
ec
tio
n
.
T
h
is
T
ab
le
1
h
ig
h
lig
h
t
s
th
e
k
e
y
d
i
f
f
e
r
en
ce
s
i
n
d
ataset
s
ize,
la
n
g
u
a
g
e
f
o
cu
s
,
e
v
al
u
atio
n
m
etr
ics,
a
n
d
tec
h
n
ical
li
m
ita
ti
o
n
s
,
t
h
er
eb
y
p
o
s
itio
n
in
g
o
u
r
c
o
n
tr
ib
u
tio
n
w
it
h
i
n
t
h
e
e
x
is
t
in
g
lan
d
s
ca
p
e
o
f
le
g
al
A
I
r
esear
ch
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
1
6
9
3
-
6930
T
E
L
KOM
NI
K
A
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l
C
o
n
tr
o
l
,
Vo
l.
23
,
No
.
6
,
Dec
em
b
er
20
25
:
1
4
9
5
-
1
505
1498
T
ab
le
1
.
C
o
m
p
ar
ativ
e
r
es
u
lt
s
o
f
s
u
m
m
ar
izi
n
g
p
r
e
v
io
u
s
w
o
r
k
A
sp
e
c
t
L
e
x
G
L
U
E
(
C
h
a
l
k
i
d
i
s
e
t
a
l
.
[
1
5
]
)
P
r
i
v
a
t
e
I
n
t
e
r
n
a
t
i
o
n
a
l
L
a
w
(
P
I
L
)
d
a
t
a
se
t
(
S
o
v
r
a
n
o
e
t
a
l
.
[
1
6
]
)
C
h
a
t
L
a
w
(
C
u
i
e
t
a
l
.
[
1
7
]
)
D
a
t
a
se
t
s
i
z
e
7
d
a
t
a
se
t
s,
2
0
0
K
+
t
o
t
a
l
i
n
st
a
n
c
e
s
1
7
q
u
e
st
i
o
n
s (
9
n
e
w
+
8
f
r
o
m p
r
e
v
i
o
u
s w
o
r
k
)
4
M
s
a
mp
l
e
s a
c
r
o
ss
1
0
m
a
j
o
r
c
a
t
e
g
o
r
i
e
s
L
a
n
g
u
a
g
e
En
g
l
i
sh
o
n
l
y
En
g
l
i
sh
(
EU
r
e
g
u
l
a
t
i
o
n
s)
C
h
i
n
e
se
l
e
g
a
l
t
e
x
t
s
D
o
mai
n
c
o
v
e
r
a
g
e
M
u
l
t
i
p
l
e
:
Eu
r
o
p
e
a
n
C
o
n
v
e
n
t
i
o
n
o
n
H
u
ma
n
R
i
g
h
t
s (E
C
H
R
)
,
U
S
l
a
w
,
EU
l
a
w
,
c
o
n
t
r
a
c
t
s
P
I
L
o
n
l
y
B
r
o
a
d
l
e
g
a
l
d
o
ma
i
n
s
Ev
a
l
u
a
t
i
o
n
me
t
r
i
c
s
M
i
c
r
o
/
mac
r
o
F
1
,
mu
l
t
i
p
l
e
t
a
sk
-
s
p
e
c
i
f
i
c
me
t
r
i
c
s
T
o
p
5
-
r
e
c
a
l
l
,
T
o
p
5
-
p
r
e
c
i
si
o
n
,
T
o
p
5
-
F1
A
c
c
u
r
a
c
y
,
F
1
,
e
x
p
e
r
t
e
v
a
l
u
a
t
i
o
n
(
4
d
i
me
n
s
i
o
n
s:
c
o
m
p
l
e
t
e
n
e
ss,
c
o
r
r
e
c
t
n
e
ss,
g
u
i
d
a
n
c
e
,
a
n
d
a
u
t
h
o
r
i
t
y
)
B
e
n
c
h
mar
k
s
u
se
d
7
s
t
a
n
d
a
r
d
i
z
e
d
l
e
g
a
l
t
a
sk
s (E
C
H
R
,
S
u
p
r
e
me
C
o
u
r
t
o
f
t
h
e
U
n
i
t
e
d
S
t
a
t
e
s (S
C
O
T
U
S
)
,
a
n
d
Eu
r
o
p
e
a
n
U
n
i
o
n
L
a
w
A
c
c
e
ss Po
r
t
a
l
(
EU
R
-
L
E
X
)
)
C
u
s
t
o
m PI
L
q
u
e
st
i
o
n
s
w
i
t
h
e
x
p
e
r
t
v
a
l
i
d
a
t
i
o
n
L
a
w
B
e
n
c
h
,
l
e
g
a
l
p
r
o
f
e
ssi
o
n
a
l
e
x
a
m
H
u
ma
n
e
v
a
l
u
a
t
i
o
n
L
i
mi
t
e
d
h
u
ma
n
e
v
a
l
u
a
t
i
o
n
b
a
se
l
i
n
e
I
n
d
e
p
e
n
d
e
n
t
l
e
g
a
l
e
x
p
e
r
t
v
a
l
i
d
a
t
i
o
n
L
e
g
a
l
e
x
p
e
r
t
a
sse
ssm
e
n
t
o
n
r
e
a
l
c
a
se
s
P
e
r
f
o
r
man
c
e
r
e
su
l
t
s
T
a
sk
-
d
e
p
e
n
d
e
n
t
:
5
0
-
9
5
%
F
1
3
8
.
0
5
%
T
o
p
5
-
F
1
o
v
e
r
a
l
l
6
0
.
0
8
%
a
v
g
o
n
L
a
w
B
e
n
c
h
(
v
s G
P
T
-
4
:
5
2
.
3
5
%)
G
e
o
g
r
a
p
h
i
c
sco
p
e
U
S
/
EU
-
f
o
c
u
se
d
l
e
g
a
l
sy
st
e
ms
EU
-
fo
c
u
se
d
(
P
I
L
)
C
h
i
n
a
-
f
o
c
u
se
d
l
e
g
a
l
s
y
st
e
m
L
a
n
g
u
a
g
e
l
i
mi
t
a
t
i
o
n
s
En
g
l
i
sh
o
n
l
y
En
g
l
i
sh
o
n
l
y
C
h
i
n
e
se
p
r
i
maril
y
T
e
c
h
n
i
c
a
l
l
i
mi
t
a
t
i
o
n
s
M
i
ss
i
n
g
h
u
ma
n
b
a
se
l
i
n
e
s,
c
o
p
y
r
i
g
h
t
r
e
st
r
i
c
t
i
o
n
s
L
o
w
b
a
se
l
i
n
e
p
e
r
f
o
r
man
c
e
,
r
e
q
u
i
r
e
s
mu
l
t
i
-
h
o
p
r
e
a
so
n
i
n
g
P
r
i
v
a
c
y
c
o
n
c
e
r
n
s,
h
a
l
l
u
c
i
n
a
t
i
o
n
r
i
s
k
s
d
e
sp
i
t
e
mi
t
i
g
a
t
i
o
n
3.
M
E
T
H
O
D
T
h
is
w
o
r
k
ai
m
s
to
ex
p
lo
it
th
e
R
A
G
ar
ch
itect
u
r
e
in
t
h
e
Mo
r
o
cc
an
leg
al
co
n
te
x
t
an
d
to
th
i
s
en
d
w
e
w
ill
ev
alu
a
te
t
h
e
p
er
f
o
r
m
a
n
ce
o
f
m
u
ltil
i
n
g
u
al
e
m
b
ed
d
in
g
m
o
d
e
ls
(
o
p
en
w
ei
g
h
ts
)
i
n
th
e
A
r
ab
ic
leg
al
co
n
te
x
t,
i
n
p
ar
ticu
l
ar
f
a
m
il
y
co
d
e
ter
m
i
n
o
lo
g
y
as
w
ell
as
th
e
e
v
al
u
ati
o
n
o
f
t
h
e
ca
p
ac
it
y
o
f
lar
g
e
m
u
ltil
i
n
g
u
a
l
lan
g
u
ag
e
m
o
d
el
s
f
o
r
g
en
er
ati
n
g
co
m
p
let
e
an
d
p
r
ec
is
e
r
esp
o
n
s
es.
T
h
e
m
et
h
o
d
ad
o
p
ted
in
th
is
p
ap
er
is
as
f
o
llo
w
s
:
s
tar
tin
g
w
it
h
th
e
ex
p
er
i
m
en
tatio
n
o
f
m
u
ltil
i
n
g
u
a
l
e
m
b
e
d
d
in
g
m
o
d
els
in
th
e
leg
al
co
n
tex
t
in
A
r
ab
ic
lan
g
u
ag
e,
th
e
n
th
e
ex
p
er
i
m
e
n
tati
o
n
o
f
lar
g
e
lan
g
u
ag
e
m
o
d
el
s
b
ased
o
n
o
n
e
o
f
th
e
e
m
b
ed
d
in
g
m
o
d
els p
r
o
v
in
g
its
h
ig
h
q
u
alit
y
an
d
s
p
ee
d
r
atio
.
3
.
1
.
Da
t
a
s
et
I
n
th
i
s
s
u
b
s
ec
tio
n
,
w
e
ex
p
lo
r
e
th
e
d
ataset
g
en
er
at
io
n
m
e
th
o
d
u
s
ed
in
t
h
is
w
o
r
k
a
s
w
ell
a
s
it
s
p
r
ep
r
o
ce
s
s
in
g
.
T
h
e
d
ataset
u
s
ed
in
th
i
s
w
o
r
k
co
n
tai
n
s
2
.
5
k
q
u
est
io
n
s
a
n
d
an
s
w
er
s
o
n
Mo
r
o
cc
an
la
w
w
a
s
d
esig
n
ed
to
ev
al
u
ate
o
u
r
R
AG
s
y
s
te
m
o
n
t
h
e
A
r
ab
ic
le
g
al
co
n
tex
t.
I
t
w
a
s
cr
ea
ted
w
it
h
t
h
e
as
s
is
ta
n
ce
o
f
a
r
o
b
u
s
t
L
L
M
f
o
r
co
n
ten
t
g
en
er
atio
n
s
,
it
w
a
s
s
tr
u
ctu
r
ed
w
it
h
th
r
ee
co
lu
m
n
s
:
{
q
u
e
s
tio
n
s
},
{so
u
r
ce
}
an
d
{r
ef
er
en
ce
}.
{Qu
est
io
n
s
}
: q
u
esti
o
n
s
as
k
ed
b
y
t
h
e
u
s
er
;
{So
u
r
ce
}:
th
e
d
o
cu
m
e
n
t r
elati
n
g
to
t
h
e
q
u
est
io
n
as
k
ed
;
{Ref
er
e
n
ce
}:
t
h
e
co
r
r
ec
t a
n
d
co
m
p
lete
a
n
s
w
er
to
th
e
q
u
est
i
o
n
s
.
T
h
e
q
u
esti
o
n
s
co
lu
m
n
co
n
tai
n
s
all
p
o
s
s
ib
le
q
u
esti
o
n
s
f
o
r
ea
ch
ar
ti
cle
in
t
h
e
A
r
ab
ic
leg
al
d
o
cu
m
en
t.
T
h
ese
q
u
esti
o
n
s
w
er
e
g
en
er
ate
d
u
s
in
g
GE
MI
NI
2
.
5
,
u
s
in
g
a
w
ell
-
d
etailed
an
d
s
tr
u
ct
u
r
ed
p
r
o
m
p
t
A
l
g
o
r
ith
m
1
,
in
o
r
d
er
to
o
b
tain
th
e
m
ax
i
m
u
m
n
u
m
b
er
o
f
q
u
es
tio
n
m
o
d
ali
ties
th
a
t
an
i
n
d
iv
id
u
al
o
r
leg
al
ex
p
er
t
co
u
ld
ask
.
An
s
w
er
g
e
n
er
atio
n
w
as p
er
f
o
r
m
ed
b
y
an
o
t
h
er
L
L
M
(
GP
T
-
4
o
)
u
s
in
g
a
s
p
ec
i
f
ic
p
r
o
m
p
t
Alg
o
r
ith
m
2
,
to
an
s
w
er
q
u
esti
o
n
s
g
e
n
er
ated
b
y
GE
MI
NI
2
.
5
.
Div
er
s
if
y
i
n
g
L
L
Ms
in
t
h
e
q
u
est
io
n
a
n
s
w
er
g
en
er
atio
n
p
r
o
ce
d
u
r
e
o
v
er
co
m
es
th
e
p
r
o
b
lem
o
f
p
r
o
p
ag
ated
er
r
o
r
s
;
th
at
is
,
if
an
L
L
M
m
a
k
e
s
an
er
r
o
r
in
a
q
u
esti
o
n
(
a
m
b
ig
u
it
y
,
m
is
in
ter
p
r
etatio
n
)
,
it
w
ill
li
k
el
y
b
e
r
ep
ea
ted
in
th
e
an
s
w
er
.
H
o
w
e
v
er
,
w
e
co
n
ce
d
e
th
at
g
en
er
atin
g
th
e
d
ata
u
s
in
g
L
L
Ms
m
a
y
i
n
tr
o
d
u
ce
s
o
m
e
b
i
as.
T
h
es
e
b
iases
ca
n
r
e
v
ea
l
t
h
e
m
s
el
v
es
as
a
n
i
n
ter
p
r
etiv
e
b
i
as,
w
h
er
e
t
h
e
m
o
d
el
m
a
y
ex
h
ib
it
a
p
r
ef
er
e
n
ce
f
o
r
a
d
o
m
i
n
a
n
t
o
r
m
ai
n
s
tr
ea
m
i
n
ter
p
r
etatio
n
o
f
a
le
g
al
cla
u
s
e
’
i
n
its
v
as
t
t
h
o
u
g
h
u
n
co
n
tr
o
lled
tr
ain
i
n
g
d
ata
’
m
is
s
in
g
o
u
t
o
n
k
e
y
s
u
b
tle
ties
.
T
h
er
e
is
a
f
u
r
t
h
er
d
an
g
er
o
f
f
r
a
m
i
n
g
b
ias;
th
e
g
e
n
er
ated
q
u
esti
o
n
s
m
a
y
s
er
v
e
to
r
ed
u
ce
co
m
p
le
x
leg
a
l
is
s
u
es
to
v
er
y
s
i
m
p
le
f
a
u
lt
y
f
o
r
m
o
r
lead
to
w
ar
d
s
s
o
m
e
p
ar
ticu
lar
k
in
d
o
f
a
n
s
w
er
.
Of
co
u
r
s
e,
d
i
v
er
s
i
f
y
in
g
L
L
Ms
in
q
u
esti
o
n
-
a
n
s
w
er
g
e
n
e
r
atio
n
ca
n
lead
to
in
co
n
s
i
s
t
en
c
y
i
s
s
u
es
s
te
m
m
i
n
g
f
r
o
m
t
h
e
d
if
f
er
en
t
a
n
al
y
s
i
s
an
d
in
ter
p
r
etatio
n
s
t
y
l
es
o
f
ea
ch
L
L
M,
w
h
ich
ca
n
th
u
s
cr
ea
te
q
u
esti
o
n
-
an
s
w
er
m
is
al
ig
n
m
en
ts
.
T
o
ad
d
r
ess
th
i
s
p
o
ten
tial
f
o
r
b
ias
an
d
in
co
n
s
i
s
te
n
c
y
,
w
e
m
an
u
all
y
ch
ec
k
t
h
e
q
u
e
s
tio
n
s
an
d
th
eir
an
s
w
er
s
.
T
h
is
d
ataset
h
as
b
ee
n
c
h
ec
k
ed
a
n
d
co
r
r
ec
t
ed
b
y
p
r
iv
ate
la
w
la
u
r
ea
tes,
all
th
e
er
r
o
r
s
f
o
u
n
d
i
n
th
is
d
ataset
ar
e
g
en
er
all
y
r
elate
d
to
th
e
s
tr
u
ctu
r
in
g
o
f
th
e
q
u
es
t
io
n
s
,
th
e
an
s
w
er
s
an
d
th
e
s
o
u
r
ce
tex
t
(
la
w
clau
s
e)
,
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KO
MN
I
K
A
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l
C
o
n
tr
o
l
R
etri
ev
a
l
-
a
u
g
men
ted
g
en
era
ti
o
n
fo
r
A
r
a
b
ic
leg
a
l in
fo
r
ma
tio
n
:
t
h
e
fa
mily
co
d
e
ca
s
e
s
tu
d
y
(
Ja
ma
l H
r
imec
h
)
1499
in
p
ar
ticu
lar
w
e
f
in
d
s
ev
er
al
an
s
w
er
s
g
en
er
ated
in
a
w
a
y
t
h
at
is
to
o
co
n
cise,
w
h
ic
h
ca
u
s
es
a
m
b
i
g
u
i
t
y
i
n
th
e
u
n
d
er
s
ta
n
d
in
g
o
f
th
e
co
n
tex
t,
t
h
u
s
w
e
f
in
d
a
d
is
o
r
d
er
at
th
e
le
v
el
o
f
m
ap
p
in
g
b
et
w
ee
n
q
u
esti
o
n
s
an
d
it
s
an
s
w
er
an
d
its
s
o
u
r
ce
tex
t.
Algorithm 1. Prompt used to generate questions of
d
ataset
yo
u
ar
e
an
ex
pe
rt
in
ar
a
b
ic
la
w
te
xt
,
fr
om
th
is
ar
ti
cl
e
in
at
ta
ch
ed
fi
le
s,
cr
ea
te
qu
es
ti
on
s
for any article with respecting the following instructions:
Output:
A TXT file with:
article_number: The article number (e.g 65).
questions: A list of question in Arabic:
Short articles (
1
-
2 sentences): 1 questions
Lon
ge
r
ar
ti
cl
es
(m
ul
ti
pl
e
se
nt
en
ce
s,
cl
au
se
s,
or
p
oi
nt
s)
:
At
le
as
t
on
e
qu
es
ti
on
pe
r
di
st
in
ct
po
in
t
(e
.g
,
cl
au
se
,
co
nd
i
ti
on
,
ex
ce
pt
io
n,
st
ep
),
w
it
h
a
mi
ni
mu
m
of
3
qu
es
ti
on
s
an
d
no
up
pe
r
limit.
Guidelines:
1.Article Analysis:
-
Fo
r
{s
ho
rt
ar
ti
cl
es
}
(1
-
2
se
nt
e
nc
es
),
id
en
ti
fy
th
e
co
re
id
ea
an
d
ge
ne
ra
te
on
e
question.
-
Fo
r
{l
on
ge
r
ar
ti
cl
es
},
br
ea
k
th
e
te
xt
in
to
di
st
in
ct
po
in
ts
(e
.g
,
ea
ch
cl
au
se
,
condition,
exception,
or
procedural
step)
and
ge
nerate
at
least
one
ques
tion
per
point.
Examples of distinct points:
-
A separate sentence or clause.
-
Each numbered condition or requirement (e.g, 1), 2), 3)).
-
Exceptions or special cases (e.g, if unable to speak).
1.Question Generation:
-
{Short articles}: Generate exactly one question.
-
{Longer articles}:
Generate at least one question
per distinct point ident
ified. If
fe
we
r
th
an
3
po
in
ts
ex
is
t,
pa
d
wi
th
ad
di
ti
on
al
qu
es
ti
on
s
to
re
ac
h
5.
If
mo
r
e
th
an
5
po
in
ts
exist, generate questions for all points, ensuring more than 5 questions.
Use var
ied question types
(e.g, direct, yes/no, hypothetical).
Algorithm
2.
Prompt used to generate answers of
d
ataset
yo
u
ar
e
an
ex
pe
rt
in
ar
ab
ic
la
w
te
xt
,
fr
om
th
is
qu
es
ti
on
in
at
ta
ch
ed
fi
le
s
na
me
d,
ge
ne
ra
te
a
concise
answer
for
the
que
stions
from
the
law
text
e
in
fi
le
“
LAW
”
wi
th
re
sp
ec
t
in
g
th
e
fo
ll
ow
in
g
instructions:
the answers should be arabic only;
ea
ch
qu
es
ti
on
as
so
ci
at
ed
wi
th
it
s
it
em
nu
mb
er
an
d
th
e
an
sw
er
mu
st
be
ge
ne
ra
te
d
fr
om
th
e
te
xt
of the same item number in the law text file;
output format:
article
_1:
answers:
1
-
“
ةرسا ةنودم مسا نوناقلا اذه ىلع قلطي
”
article_2:
answers:
ىر
خأ
ةي
سن
جل
ا
ن
ي
لماح
اون
ا
ك
ولو
ةب
ر
اغ
م
لا
ع
ي
م
ج
ىلع
”
-
1
”
ىلع
يف ةخر
ؤم
لا
ف
ي
ن
ج
ةي
ق
افت اقبط
،
ة
ي
سن
جلا
وم
ي
دع
مهيف
ن
م
ب
ن
ي
ئ
جلا
28
ز
ويل
وي
1951
ن
ي
ئ
ج
لا ةي
ع
ضوب
ةقلعتملا
”
-
2
”
اه
ي
ف ن
وك
ي
يتلا
تاق
ع
لا
ىلع
اي
ب
ر
غ
م
ن
ي
ف
ر
ط
لا
دحأ
”
-
3
”
ةي
ب
ر
غ
م
لا
ةي
ر
ب
ع
لا
ةي
ص
خشل
ا لاوحا
د
ع
او
ق مه
ي
لع
ير
ست
ف
ةب
ر
اغ
م
لا
دوه
ي
لا
ام
أ
”
-
4
”
3
.
1
.
1
.
P
re
pro
ce
s
s
ing
o
f
leg
a
l
t
ex
t
T
y
p
icall
y
,
le
g
al
d
o
cu
m
e
n
tatio
n
is
m
a
n
ag
ed
b
y
th
e
r
ele
v
an
t
Mo
r
o
cc
an
m
in
is
tr
y
,
w
h
ic
h
u
s
e
s
a
co
m
m
o
n
f
o
r
m
at
f
o
r
all
leg
a
l
d
o
cu
m
e
n
t
s
,
in
cl
u
d
in
g
t
h
e
m
i
n
is
tr
y
’
s
w
a
ter
m
ar
k
,
h
ea
d
er
,
f
o
o
ter
,
an
d
p
r
ea
m
b
le.
All
t
h
ese
d
o
cu
m
en
ts
ar
e
p
u
b
lis
h
ed
o
n
li
n
e
in
P
DF
f
o
r
m
at.
T
o
ex
tr
ac
t
o
n
l
y
t
h
e
le
g
al
c
lau
s
es
f
r
o
m
th
e
s
e
d
o
cu
m
en
ts
,
w
e
u
s
e
th
e
P
y
t
h
o
n
lib
r
ar
y
“
P
y
M
u
P
D
F
”
f
o
r
ad
v
an
ce
d
P
DF
m
an
ip
u
latio
n
,
as
w
ell
as
its
p
o
w
er
f
u
l
h
an
d
lin
g
o
f
A
r
ab
ic
ch
ar
ac
ter
s
an
d
r
ig
h
t
-
to
-
le
f
t
(
R
T
L
)
r
ea
d
in
g
d
ir
ec
tio
n
.
T
h
is
ex
tr
ac
ti
o
n
s
h
o
u
ld
s
i
g
n
if
ican
t
l
y
i
m
p
r
o
v
e
r
etr
iev
a
l
ac
cu
r
ac
y
,
as
it
en
s
u
r
es
th
at
t
h
e
v
ec
to
r
d
atab
ase
co
n
tain
s
o
n
l
y
t
ex
t
th
at
i
s
s
e
m
an
ticall
y
r
ele
v
a
n
t
to
p
o
ten
tial
leg
al
q
u
er
ies,
th
u
s
i
m
p
r
o
v
in
g
t
h
e
s
i
g
n
al
-
to
-
n
o
is
e
r
atio
.
T
o
f
ac
ilit
ate
th
e
r
etr
iev
a
l
to
o
l
’
s
ex
p
lo
itatio
n
o
f
t
h
e
d
atase
t,
w
e
r
e
m
o
v
ed
d
iacr
itical
m
ar
k
s
(
ta
s
h
k
ee
l)
,
w
h
ic
h
ch
a
n
g
e
th
e
p
r
o
n
u
n
cia
tio
n
b
u
t
n
o
t
t
h
e
f
u
n
d
a
m
e
n
tal
m
ea
n
in
g
.
T
h
e
ex
p
ec
ted
i
m
p
ac
t
i
s
a
n
i
n
cr
ea
s
e
in
t
h
e
r
etr
iev
al
r
ate,
as
t
h
i
s
av
o
id
s
in
co
n
s
i
s
ten
c
ies
d
u
e
to
th
e
p
r
ese
n
ce
o
f
d
iacr
itical
m
ar
k
s
in
t
h
e
s
o
u
r
ce
tex
t,
b
u
t a
b
s
e
n
t f
r
o
m
th
e
u
s
er
q
u
er
y
.
W
e
th
en
s
ta
n
d
ar
d
ized
th
e
A
r
a
b
ic
ch
ar
ac
ter
s
b
y
r
ep
lacin
g
th
e
d
if
f
er
e
n
t
f
o
r
m
s
o
f
t
h
e
A
r
ab
i
c
lette
r
s
،ة
)
آ
،
أ
،
إ
)
w
it
h
a
s
ta
n
d
ar
d
s
y
llab
ar
y
(
ا
،
ه
)
.
T
h
is
s
tep
s
ta
n
d
ar
d
izes
co
m
m
o
n
s
p
ellin
g
v
ar
iatio
n
s
i
n
t
h
e
A
r
ab
ic
s
cr
ip
t
w
it
h
t
h
e
ai
m
o
f
f
u
r
t
h
er
i
m
p
r
o
v
in
g
d
ata
r
etr
iev
al,
e
n
s
u
r
in
g
t
h
a
t
r
elev
an
t
d
o
cu
m
e
n
t
s
ar
e
n
o
t m
is
s
ed
d
u
e
to
m
in
o
r
an
d
s
e
m
a
n
tica
ll
y
ir
r
ele
v
an
t c
h
ar
ac
ter
v
ar
iatio
n
s
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
1
6
9
3
-
6930
T
E
L
KOM
NI
K
A
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l
C
o
n
tr
o
l
,
Vo
l.
23
,
No
.
6
,
Dec
em
b
er
20
25
:
1
4
9
5
-
1
505
1500
3
.
2
.
E
m
bed
din
g
m
o
del
s
E
m
b
ed
d
in
g
m
o
d
els
ar
e
p
r
o
v
i
n
g
to
b
e
v
al
u
ab
le
to
o
ls
f
o
r
A
r
ab
ic
NL
P
,
allo
w
i
n
g
to
ca
p
t
u
r
e
th
e
s
e
m
a
n
tic
an
d
s
y
n
tactic
n
u
an
ce
s
o
f
w
o
r
d
s
.
T
h
er
e
ar
e
d
if
f
er
en
t
v
ar
ian
t
s
o
f
th
ese
m
o
d
els,
s
u
ch
as
A
r
aB
E
R
T
,
E
5
(
Mic
r
o
s
o
f
t)
m
u
lt
ili
n
g
u
al
a
n
d
o
th
er
s
th
a
t
o
f
f
er
v
ec
to
r
r
ep
r
esen
tatio
n
s
o
f
A
r
ab
ic
w
o
r
d
s
,
t
h
u
s
f
ac
il
itatin
g
i
n
f
o
r
m
atio
n
ex
tr
ac
tio
n
[
1
8
]
.
I
n
t
h
is
w
o
r
k
w
e
u
s
e
t
h
r
ee
d
if
f
er
en
t
t
y
p
es
o
f
m
u
lt
ili
n
g
u
al
e
m
b
ed
d
in
g
m
o
d
el
(
o
p
en
w
ei
g
h
t
s
)
h
av
i
n
g
o
p
ti
m
al
d
i
m
en
s
io
n
f
o
r
s
i
m
ilar
it
y
s
ea
r
ch
ta
s
k
i
n
A
r
ab
ic
leg
al
co
n
te
x
t.
T
h
e
B
GE
-
m
3
m
o
d
el
p
r
o
v
es
its
s
tr
en
g
t
h
s
w
er
e
its
m
u
ltil
i
n
g
u
al
ca
p
ab
ilit
ies,
its
ab
ilit
y
to
h
a
n
d
le
lo
n
g
an
d
n
o
is
y
te
x
ts
,
a
n
d
its
ex
ce
lle
n
t
p
er
f
o
r
m
a
n
ce
i
n
s
e
m
an
tic
s
i
m
ilar
it
y
s
ea
r
c
h
,
esp
ec
iall
y
s
e
m
an
tic
s
i
m
i
lar
ities
f
o
r
th
e
leg
a
l
d
o
cu
m
e
n
t
r
etr
iev
a
l
task
,
it
r
el
ati
v
el
y
a
n
e
w
m
o
d
el
an
d
p
r
o
v
id
es
m
ea
n
i
n
g
f
u
l
r
ep
r
esen
tatio
n
s
i
n
o
v
er
1
0
0
lan
g
u
ag
e
s
an
d
at
m
u
l
tip
le
d
ep
th
s
:
w
o
r
d
,
p
ar
ag
r
ap
h
,
an
d
f
u
ll
te
x
t
u
p
to
8
1
9
2
t
o
k
en
s
,
also
it
is
w
ell
s
u
ited
f
o
r
a
w
id
e
r
an
g
e
o
f
r
ec
o
v
er
y
tas
k
s
a
n
d
s
u
p
p
o
r
ts
m
u
lti
-
g
r
an
u
lar
it
y
te
x
t
in
te
g
r
atio
n
s
[
1
9
]
.
Mic
r
o
s
o
f
t
’
s
E
5
m
u
ltil
i
n
g
u
al
m
o
d
el
al
s
o
s
u
p
p
o
r
ts
A
r
ab
ic
la
n
g
u
a
g
e
a
n
d
i
s
o
p
t
i
m
ized
f
o
r
th
e
tas
k
o
f
s
e
m
a
n
tic
s
i
m
ilar
it
y
a
n
d
p
ass
ag
e
r
etr
iev
a
l,
w
h
o
s
e
e
m
b
e
d
d
in
g
d
i
m
e
n
s
io
n
is
7
6
8
(E5
-
s
m
al
l)
an
d
1
0
2
4
(
E
5
-
lar
g
e)
.
T
h
e
ch
o
ice
o
f
th
e
e
m
b
ed
d
in
g
m
o
d
el
is
an
i
m
p
o
r
tan
t
s
tep
,
es
p
ec
iall
y
i
n
t
h
e
A
r
ab
ic
leg
al
co
n
tex
t,
w
h
ic
h
h
as
a
s
p
ec
if
ic
ch
ar
ac
te
r
is
t
ic
s
u
ch
as
lo
n
g
an
d
co
m
p
lex
s
e
n
te
n
ce
s
,
th
e
in
te
n
s
iv
e
u
s
e
o
f
th
e
p
ass
iv
e,
th
e
m
ix
tu
r
e
b
et
w
ee
n
m
o
d
er
n
le
g
al
ter
m
in
o
lo
g
y
an
d
t
h
e
co
n
ce
p
t
o
f
tr
ad
itio
n
al
I
s
la
m
ic
la
w
.
T
h
is
c
h
ar
ac
ter
is
tic,
k
n
o
w
n
i
n
leg
al
te
x
ts
,
lead
s
u
s
to
th
e
e
v
al
u
atio
n
o
f
e
m
b
ed
d
in
g
m
o
d
els
b
ef
o
r
e
in
te
g
r
ati
n
g
t
h
e
m
i
n
to
th
e
R
A
G
ar
ch
itectu
r
e
.
I
n
th
is
w
o
r
k
,
w
e
u
s
e
t
h
r
ee
e
m
b
ed
d
in
g
m
o
d
els
in
le
g
al
co
n
te
x
t
r
etr
iev
al
in
o
r
d
er
to
d
is
tin
g
u
is
h
o
n
e
a
m
o
n
g
t
h
e
th
r
ee
m
o
d
els t
h
at
is
t
h
e
m
o
s
t e
f
f
icien
t in
ter
m
s
o
f
ac
cu
r
ac
y
a
n
d
lig
h
t i
n
u
s
e.
3
.
3
.
I
nd
ex
ing
a
nd
s
p
litt
ing
I
n
d
ex
i
n
g
is
a
p
r
eli
m
i
n
ar
y
i
n
R
A
G
s
y
s
te
m
s
,
it f
ac
il
itates t
h
e
s
to
r
ag
e,
r
etr
iev
al,
an
d
s
i
m
ilar
it
y
s
ea
r
ch
o
f
v
ec
to
r
r
ep
r
esen
tatio
n
s
es
s
en
ti
al
f
o
r
th
e
r
ap
id
s
ea
r
c
h
o
f
r
el
ev
an
t
d
o
cu
m
e
n
t
s
[
2
0
]
.
I
n
th
i
s
r
eg
ar
d
,
ef
f
icie
n
t
in
d
ex
i
n
g
allo
w
s
R
A
G
s
y
s
te
m
s
to
a
d
ap
t
p
er
f
ec
tl
y
to
lar
g
e
d
at
ab
ases
an
d
r
etu
r
n
th
e
m
o
s
t
r
ele
v
an
t
i
n
f
o
r
m
atio
n
to
u
s
er
q
u
er
ies,
th
u
s
en
ab
li
n
g
f
as
ter
an
d
m
o
r
e
ac
cu
r
ate
r
esp
o
n
s
es.
Sp
litt
in
g
is
an
es
s
en
tia
l
an
d
d
is
tin
ctiv
e
asp
ec
t
o
f
th
e
s
y
s
te
m
to
m
an
a
g
e
t
h
e
c
o
n
s
id
er
ab
le
s
ize
o
f
th
e
i
n
f
o
r
m
atio
n
te
x
t
an
d
,
co
n
s
eq
u
e
n
tl
y
,
t
o
m
ak
e
d
o
cu
m
e
n
t
p
r
o
ce
s
s
in
g
a
n
d
r
ap
id
id
en
tif
ic
atio
n
o
f
u
s
e
f
u
l i
n
f
o
r
m
atio
n
p
o
s
s
ib
le
[
2
1
]
.
T
h
er
e
a
r
e
tw
o
m
ai
n
d
o
cu
m
e
n
t
-
s
p
litt
in
g
tech
n
iq
u
e
s
u
s
ed
in
t
h
e
R
A
G
s
y
s
te
m
:
th
e
r
ec
u
r
s
i
v
e
ch
ar
ac
ter
s
p
litt
er
(
R
C
S)
an
d
t
h
e
to
k
e
n
-
b
ased
s
p
litt
er
(
T
T
S),
b
u
t
in
th
is
w
o
r
k
R
C
S
a
n
d
T
T
S
m
e
th
o
d
ar
e
n
o
t
s
u
itab
le
f
o
r
leg
al
d
o
cu
m
e
n
t
s
,
w
e
s
p
litt
i
n
g
b
y
c
lau
s
es
to
p
r
eser
v
es
t
h
e
i
n
t
eg
r
it
y
o
f
ea
c
h
co
m
p
lete
clau
s
e
,
r
esp
ec
ts
th
e
lo
g
ical
s
tr
u
ct
u
r
e
o
f
t
h
e
le
g
al
d
o
cu
m
en
t,
av
o
id
s
s
p
litt
i
n
g
a
cla
u
s
e
i
n
t
h
e
m
id
d
le
an
d
m
ai
n
tai
n
s
th
e
le
g
al
co
n
te
x
t o
f
ea
c
h
s
ec
tio
n
.
3
.
4
.
G
ener
a
t
o
r
I
n
a
R
A
G
s
y
s
te
m
a
r
c
h
itect
u
r
e,
th
e
g
e
n
er
ato
r
is
b
ased
o
n
an
L
L
M
t
h
at
tak
e
s
as
i
n
p
u
t
t
h
e
u
s
er
’
s
q
u
er
y
(
q
u
esti
o
n
)
a
n
d
th
e
co
n
te
x
t
s
e
l
ec
ted
b
y
th
e
r
etr
ie
v
al,
i.e
.
t
h
e
tex
t
p
o
ten
tiall
y
co
n
ta
in
i
n
g
t
h
e
ele
m
e
n
ts
o
f
t
h
e
r
esp
o
n
s
e
i
n
o
r
d
er
to
p
r
o
d
u
ce
an
i
n
f
o
r
m
ati
v
e
o
u
tp
u
t
t
h
at
r
esp
o
n
d
s
to
th
e
q
u
e
r
y
e
n
ter
ed
b
y
th
e
u
s
er
.
Fo
r
a
g
en
er
ato
r
to
g
en
er
ate
an
in
f
o
r
m
ati
v
e
an
s
w
er
to
a
q
u
esti
o
n
f
r
o
m
a
s
p
ec
if
ic
co
n
tex
t,
a
s
tr
u
ct
u
r
ed
p
r
o
m
p
t
p
lay
s
a
v
ital
r
o
le
in
h
elp
i
n
g
th
e
L
L
M
d
eter
m
i
n
e
th
e
co
n
tex
t
an
d
t
h
e
q
u
esti
o
n
,
as
w
ell
a
s
t
h
e
g
u
id
eli
n
es
to
f
o
l
lo
w
i
n
th
e
ir
an
s
w
er
s
[
2
2
]
.
I
n
th
is
w
o
r
k
an
d
g
iv
e
n
th
e
s
p
ec
i
f
icit
y
o
f
t
h
e
leg
al
f
ield
,
in
p
ar
ticu
lar
th
e
ac
cu
r
ac
y
o
f
t
h
e
in
f
o
r
m
atio
n
to
b
e
g
en
er
ated
,
w
e
u
s
e
an
ad
ap
ted
p
r
o
m
p
t
w
h
ich
d
ir
ec
ts
t
h
e
g
e
n
er
ato
r
to
f
o
llo
w
th
e
f
o
llo
w
in
g
d
ir
ec
tiv
es:
1
.
“
Yo
u
ar
e
a
sp
ec
ia
li
ze
d
as
si
st
an
t
in
Mo
ro
cc
an
fa
mi
ly
l
aw
.
Yo
u
mu
st
an
s
we
r
on
ly
ba
se
d
on
the
information contained in the excerpts of context provided below.
”
2.
“
If
yo
u
do
no
t
fi
nd
th
e
in
fo
rm
at
io
n
in
th
es
e
ex
ce
rp
ts
,
st
at
e
cl
ea
rl
y
th
a
t
th
e
in
fo
rm
at
io
n
is n
ot
available in the provided documents.
”
3.
“
Never invent legal content not explicitly mentioned in the excerpts.
”
4.
“
An
sw
er
in
Ar
ab
ic
on
l
y
an
d
us
e
a
ma
xi
mu
m
of
th
r
ee
se
nt
en
ce
s
an
d
ma
ke
t
he
an
sw
er
ac
cu
ra
te
and concise.
”
5.
“
Use the same terms
contained in
the reference text when possible.
”
Retrieval
and
Generation
for
a
Legal
Query
{I
np
ut
:}
Us
er
qu
er
y
Q,
Ve
ct
or
DB
Vc
la
us
es
,
LL
M
Mgen,
embedding
mo
de
l
Ee
m
b
{Output:}
An
sw
er
A
qe
mb
←
Ee
mb
(Q
)
Em
be
d
th
e
us
er
qu
er
y
Cr
et
ri
ev
ed
←
SimilaritySearch
(
qemb
,
Vc
la
us
es
,
k
=
5)
Re
tr
ie
ve
to
p
-
k
cl
au
se
s
Pp
ro
mp
t
←
Bu
il
d
Pr
om
pt
(
Cretrieved, Q) Use prompt A ←
Mgen (
Prompt) Generate answer with JAIS
{Return}
A
.
4.
E
XP
E
R
I
M
E
NT
A
T
I
O
N
AN
D
RE
SU
L
T
S
4
.
1
.
Ret
rie
v
er
pa
rt
I
n
th
is
s
u
b
s
ec
tio
n
,
w
e
as
s
ess
th
e
r
etr
iev
er
p
ar
t
o
n
th
e
M
o
r
o
cc
an
leg
al
d
o
cu
m
e
n
t.
T
h
e
e
m
b
ed
d
in
g
m
o
d
el
s
w
e
u
s
e
in
th
i
s
w
o
r
k
ar
e
B
GE
-
m
3
,
E
5
-
lar
g
e,
an
d
E
5
-
s
m
all
to
d
eter
m
i
n
e
th
e
b
est
m
o
d
el
am
o
n
g
th
e
th
r
e
e
w
it
h
t
h
e
ab
ilit
y
to
h
an
d
le
t
h
e
l
eg
al
co
n
te
x
t i
n
A
r
ab
ic.
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KO
MN
I
K
A
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l
C
o
n
tr
o
l
R
etri
ev
a
l
-
a
u
g
men
ted
g
en
era
ti
o
n
fo
r
A
r
a
b
ic
leg
a
l in
fo
r
ma
tio
n
:
t
h
e
fa
mily
co
d
e
ca
s
e
s
tu
d
y
(
Ja
ma
l H
r
imec
h
)
1501
T
h
e
ass
es
s
m
e
n
t
p
r
o
ce
s
s
b
eg
i
n
s
b
y
s
to
r
in
g
t
h
e
le
g
al
d
o
cu
m
en
tatio
n
in
t
h
e
v
ec
to
r
d
atab
ase
af
te
r
p
r
ep
r
o
ce
s
s
in
g
an
d
s
p
litt
i
n
g
t
h
e
d
o
cu
m
e
n
t.
T
h
en
,
w
e
u
s
e
th
e
{
q
u
esti
o
n
}
a
n
d
{so
u
r
ce
}
co
lu
m
n
s
o
f
o
u
r
d
ataset
to
ca
lcu
late
t
h
e
n
ec
e
s
s
ar
y
b
e
n
ch
m
ar
k
s
to
q
u
an
ti
f
y
t
h
e
ef
f
ec
ti
v
e
n
es
s
o
f
th
e
r
etr
ie
v
er
in
a
R
AG
s
y
s
te
m
.
T
h
e
ass
ess
m
e
n
t
m
etr
ic
s
u
s
e
d
ar
e
MRR
,
w
h
ic
h
i
s
w
id
el
y
u
s
ed
i
n
i
n
f
o
r
m
atio
n
r
et
r
iev
al
a
n
d
r
ec
o
m
m
e
n
d
atio
n
s
y
s
te
m
s
.
MR
R
ev
alu
a
tes
h
o
w
w
e
ll
a
co
r
r
ec
t
an
s
w
er
o
r
r
elev
an
t
r
eso
u
r
ce
is
class
i
f
ied
af
ter
a
q
u
er
y
is
m
ad
e.
MRR
i
s
an
a
v
e
r
ag
e
b
ased
o
n
th
e
p
o
s
itio
n
s
o
f
th
e
f
ir
s
t
co
r
r
ec
t
r
ec
ip
r
o
c
al
an
s
w
er
o
v
er
a
n
u
m
b
er
o
f
q
u
er
y
q
u
esti
o
n
s
an
d
th
er
e
f
o
r
e
p
r
o
v
id
es
a
s
in
g
le
s
co
r
e.
R
ec
all@
k
is
u
s
ed
to
ev
alu
ate
f
o
r
in
f
o
r
m
atio
n
r
etr
iev
al
s
y
s
te
m
s
.
I
t
in
v
esti
g
ate
s
th
e
f
r
a
ctio
n
o
f
r
elev
a
n
t
d
o
cu
m
en
t
s
t
h
at
ar
e
r
etr
iev
ed
in
th
e
to
p
k
d
o
cu
m
en
ts
r
etu
r
n
ed
b
y
a
s
y
s
te
m
.
T
h
er
ef
o
r
e,
it
q
u
an
ti
f
ies
h
o
w
e
f
f
ec
ti
v
e
a
s
y
s
te
m
is
at
s
u
r
f
ac
in
g
ite
m
s
t
h
at
ar
e
tr
u
l
y
o
f
in
ter
es
t
to
a
u
s
er
.
T
h
e
F1
s
co
r
e
is
a
p
er
f
o
r
m
a
n
ce
m
e
tr
ic
th
at
is
p
r
i
m
ar
il
y
u
s
ed
to
ev
al
u
ate
cla
s
s
i
f
icatio
n
m
o
d
els
,
p
ar
ticu
lar
l
y
i
n
th
e
ca
s
e
o
f
i
m
b
alan
ce
d
d
ata.
I
t
is
th
e
m
ea
n
h
ar
m
o
n
ic
o
f
p
r
ec
is
io
n
a
n
d
r
ec
all,
allo
w
i
n
g
u
s
to
o
b
tain
a
co
m
b
in
ed
s
co
r
e
th
a
t
is
b
alan
ce
d
f
o
r
f
alse
p
o
s
itiv
es
an
d
f
al
s
e
n
e
g
ati
v
es.
T
ab
le
2
p
r
esen
ts
th
e
co
m
p
ar
ati
v
e
r
es
u
lts
o
f
th
e
t
h
r
e
e
e
m
b
ed
d
in
g
m
o
d
el
s
ev
al
u
ate
d
o
n
th
ese
m
et
r
ics
.
T
ab
le
2
.
C
o
m
p
ar
ativ
e
r
es
u
lt
s
o
f
e
m
b
ed
d
in
g
m
o
d
els E
5
-
l
ar
g
e,
E
5
-
s
m
al
l,
B
GE
-
m3
M
e
t
r
i
c
s
E5
-
l
a
r
g
e
E5
-
smal
l
B
G
E
-
m3
B
G
E
-
m3
v
s E
5
-
l
a
r
g
e
B
G
E
-
m3
v
s E
5
-
smal
l
R
e
c
a
l
l
@
1
0
.
9
6
7
0
0
.
9
4
7
0
0
.
9
5
3
0
-
1
.
4
5
%
+
0
.
6
3
%
R
e
c
a
l
l
@
3
0
.
9
8
6
0
0
.
9
8
4
0
0
.
9
8
5
0
-
0
.
1
0
%
+
0
.
1
0
%
R
e
c
a
l
l
@
5
0
.
9
9
0
0
0
.
9
8
7
0
0
.
9
8
8
0
-
0
.
2
0
%
+
0
.
1
0
%
F1
0
.
9
6
7
0
0
.
9
4
7
0
0
.
9
5
3
0
-
1
.
4
5
%
+
0
.
6
3
%
M
R
R
0
.
9
7
6
3
0
.
9
6
4
5
0
.
9
6
8
8
-
0
.
7
7
%
+
0
.
4
5
%
R
e
t
r
i
e
v
e
r
l
a
t
e
n
c
y
0
.
6
5
7
s
0
.
1
4
5
s
0
.
1
7
8
s
-
0
.
4
7
9
s
+
0
.
0
3
3
s
Af
ter
an
a
l
y
zi
n
g
t
h
e
r
esu
l
t
ac
co
r
d
in
g
to
T
ab
le
2
,
w
e
s
ee
i
n
t
h
e
R
ec
all
@
1
m
etr
ic
o
f
th
e
B
GE
-
m
3
m
o
d
el
r
ea
ch
in
g
9
5
.
3
%,
w
h
ic
h
is
les
s
th
an
1
.
4
%
co
m
p
ar
ed
to
E
5
-
lar
g
e
an
d
0
.
6
%
m
o
r
e
ac
cu
r
ate
th
an
E
5
-
s
m
all.
T
h
u
s
,
in
th
e
R
ec
all
@
5
m
etr
ic
o
f
t
h
e
s
a
m
e
m
o
d
el,
a
v
al
u
e
o
f
9
8
.
8
%,
w
h
ic
h
is
p
r
ac
ticall
y
eq
u
iv
alen
t
t
o
E
5
-
l
ar
g
e
(
a
d
if
f
er
e
n
ce
o
f
0
.
2
%),
an
d
s
u
p
er
io
r
to
E
5
-
s
m
al
l.
A
t
t
h
e
M
R
R
m
etr
ic
lev
el,
th
e
B
GE
-
m
3
m
o
d
el
ac
h
iev
e
s
9
7
%
lo
w
er
b
y
0
.
8
%
t
h
a
n
E
5
-
lar
g
e
an
d
h
i
g
h
er
b
y
0
.
4
%
t
h
an
E
5
-
lar
g
e,
i
n
d
icati
n
g
a
g
o
o
d
o
v
er
all
p
o
s
itio
n
i
n
g
o
f
r
elev
an
t
r
es
u
lts
.
A
co
n
s
is
ten
c
y
b
et
w
ee
n
R
ec
all
@
1
,
F1
@
1
a
n
d
MRR
w
o
u
ld
m
ea
n
th
at
f
o
r
ea
ch
q
u
esti
o
n
,
th
e
co
r
r
ec
t la
w
ar
tic
le
i
s
al
w
a
y
s
t
h
e
f
ir
s
t r
es
u
lt.
An
e
x
ce
lle
n
t
p
r
ec
is
io
n
o
f
r
elev
an
t
ar
ticle
s
i
n
t
h
e
f
ir
s
t
p
o
s
itio
n
an
d
a
b
ala
n
ce
o
f
p
er
f
o
r
m
a
n
ce
/
r
eso
u
r
ce
s
an
d
n
ea
r
-
p
er
f
ec
t c
o
v
er
ag
e,
as
w
ell
as e
x
ce
lle
n
t
s
u
p
p
o
r
t f
o
r
t
h
e
A
r
ab
ic
la
n
g
u
a
g
e,
w
as o
b
s
er
v
ed
in
th
e
B
GE
-
m
3
e
m
b
ed
d
in
g
m
o
d
el.
T
o
p
r
o
v
id
e
a
clea
r
er
v
is
u
a
l
r
ep
r
esen
tatio
n
o
f
t
h
e
s
e
f
in
d
i
n
g
s
,
Fi
g
u
r
e
2
g
r
ap
h
icall
y
s
u
m
m
ar
izes
th
e
p
er
f
o
r
m
a
n
ce
o
f
th
e
e
m
b
ed
d
in
g
m
o
d
el
s
.
Sp
ec
if
icall
y
,
Fig
u
r
e
2
(
a)
illu
s
tr
at
es
th
e
co
m
p
ar
ativ
e
r
esu
lt
s
ac
r
o
s
s
th
e
k
e
y
p
er
f
o
r
m
an
ce
m
etr
ics
d
i
s
cu
s
s
ed
,
v
is
u
all
y
co
n
f
ir
m
i
n
g
th
e
co
m
p
eti
tiv
e
p
er
f
o
r
m
an
ce
o
f
B
GE
-
m
3
a
g
ain
s
t
t
h
e
E
5
m
o
d
els.
I
n
p
ar
allel,
Fig
u
r
e
2
(
b
)
p
r
esen
ts
t
h
e
r
etr
iev
a
l
late
n
c
y
f
o
r
ea
c
h
m
o
d
el,
h
ig
h
li
g
h
ti
n
g
t
h
e
ef
f
ic
ien
c
y
o
f
B
GE
-
m
3
,
w
h
ic
h
o
f
f
er
s
a
s
tr
o
n
g
b
alan
ce
b
et
w
ee
n
s
p
ee
d
an
d
ac
cu
r
ac
y
.
B
GE
-
m
3
is
an
o
p
ti
m
a
l c
h
o
ice
f
o
r
a
p
r
o
d
u
ctio
n
d
ep
lo
y
m
en
t o
f
ar
ab
ic
leg
al
r
etr
ie
v
al
s
y
s
te
m
b
ec
a
u
s
e
it
:
a)
Of
f
er
s
ex
ce
l
len
t
p
er
f
o
r
m
an
ce
;
b)
Of
f
er
s
a
g
o
o
d
b
alan
ce
b
et
w
ee
n
p
er
f
o
r
m
a
n
ce
an
d
r
eso
u
r
ce
c
o
n
s
u
m
p
tio
n
;
c)
B
en
ef
it
s
f
r
o
m
a
p
u
r
p
o
s
e
-
b
u
il
t r
etr
iev
al
ar
ch
itect
u
r
e
;
d)
De
m
o
n
s
tr
ates e
x
ce
lle
n
t
A
r
ab
ic
lan
g
u
a
g
e
s
u
p
p
o
r
t
.
(
a)
(
b
)
Fig
u
r
e
2
.
C
o
m
p
ar
is
o
n
o
f
e
m
b
e
d
d
in
g
m
o
d
els
;
(
a)
m
etr
ics a
n
d
(
b
)
laten
c
y
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
1
6
9
3
-
6930
T
E
L
KOM
NI
K
A
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l
C
o
n
tr
o
l
,
Vo
l.
23
,
No
.
6
,
Dec
em
b
er
20
25
:
1
4
9
5
-
1
505
1502
4
.
2
.
G
ener
a
t
o
r
pa
rt
I
n
th
i
s
s
u
b
s
ec
tio
n
,
w
e
as
s
es
s
t
h
e
R
AG
s
y
s
te
m
o
n
g
e
n
er
ati
n
g
an
s
w
er
s
to
q
u
esti
o
n
s
i
n
t
h
e
Mo
r
o
cc
an
le
g
al
co
n
te
x
t
u
s
in
g
t
h
e
B
GE
-
m
3
e
m
b
ed
d
in
g
m
o
d
el.
Ge
n
er
all
y
,
t
h
e
g
e
n
er
ato
r
r
elie
s
o
n
a
n
L
L
M,
p
r
o
m
p
t
a
n
d
r
etr
iev
er
s
u
ch
th
a
t
L
L
M
r
ec
ei
v
es
t
h
e
u
s
er
’
s
q
u
esti
o
n
a
n
d
th
e
r
etr
iev
ed
d
o
cu
m
e
n
t
an
d
t
h
e
n
u
s
es
a
g
u
id
ed
p
r
o
m
p
t
to
g
en
er
ate
a
f
ac
tu
al
a
n
d
n
o
n
-
h
all
u
cin
a
to
r
y
an
s
w
er
,
th
i
s
is
wh
y
t
h
e
u
s
e
o
f
a
r
o
b
u
s
t
L
L
M
i
s
ess
e
n
tial
i
n
a
R
AG
s
y
s
te
m
.
A
cc
o
r
d
in
g
to
Sen
g
u
p
ta
et
a
l
.
[
2
3
]
,
J
ais
A
r
ab
ic
in
s
tr
u
ctio
n
-
b
ased
s
y
s
te
m
(
J
A
I
S
)
is
a
s
tate
-
of
-
t
h
e
-
ar
t,
A
r
ab
ic
-
ce
n
tr
ic,
i
n
s
tr
u
ctio
n
-
o
p
t
i
m
ized
o
p
en
g
en
er
ati
v
e
la
n
g
u
ag
e
m
o
d
el
w
i
th
1
3
B
p
ar
a
m
ete
r
s
an
d
b
u
il
t
o
n
t
h
e
GPT
-
3
d
ec
o
d
e
r
ar
ch
itectu
r
e.
J
A
I
S b
ei
n
g
d
esi
g
n
ed
an
d
tr
ain
e
d
f
r
o
m
th
e
g
r
o
u
n
d
u
p
w
ith
a
v
er
y
lar
g
e
a
m
o
u
n
t o
f
A
r
ab
ic
d
ata
(
alo
n
g
s
id
e
E
n
g
li
s
h
)
o
f
ten
g
i
v
es
it
a
b
etter
in
tr
in
s
ic
u
n
d
er
s
ta
n
d
in
g
o
f
n
u
a
n
ce
s
,
m
o
r
p
h
o
lo
g
y
,
s
y
n
ta
x
,
an
d
p
er
h
ap
s
d
ialec
ts
(
d
esp
it
e
b
ein
g
p
r
i
m
ar
il
y
MS
A
)
.
T
h
e
J
A
I
S
m
o
d
el
is
e
v
al
u
ated
ac
r
o
s
s
a
w
id
e
r
an
g
e
o
f
A
r
ab
ic
NL
P
b
en
ch
m
ar
k
s
,
co
v
er
i
n
g
r
ea
s
o
n
i
n
g
,
k
n
o
w
led
g
e,
m
is
i
n
f
o
r
m
at
io
n
,
an
d
b
ias,
a
n
d
h
a
s
b
ee
n
f
o
u
n
d
to
h
a
v
e
s
ig
n
i
f
ica
n
tl
y
s
u
p
er
io
r
A
r
ab
ic
k
n
o
w
led
g
e
an
d
r
ea
s
o
n
in
g
ca
p
ab
ilit
ies
co
m
p
ar
e
d
to
all
ex
is
tin
g
o
p
en
A
r
ab
ic
an
d
m
u
ltil
i
n
g
u
al
m
o
d
els.
I
n
th
is
ass
es
s
m
en
t,
w
e
u
s
e
s
i
x
m
etr
i
cs
to
ass
e
s
s
t
h
e
L
L
M,
n
a
m
el
y
co
s
i
n
e
s
i
m
ilar
i
t
y
,
b
ilin
g
u
al
ev
a
lu
at
io
n
u
n
d
er
s
t
u
d
y
s
co
r
e
(
B
L
E
U
)
s
co
r
e,
F1
s
co
r
e,
f
aith
f
u
l
n
es
s
,
an
s
w
er
r
elev
an
ce
,
an
d
co
n
te
x
t
r
elev
an
ce
[
2
4
]
.
T
h
e
p
er
f
o
r
m
a
n
ce
r
esu
lts
o
f
o
u
r
R
AG
s
y
s
te
m
's
g
en
er
ato
r
,
ev
alu
ated
o
n
t
h
e
Mo
r
o
cc
an
la
w
d
ataset
u
s
i
n
g
th
e
s
e
s
i
x
m
e
tr
ics,
ar
e
p
r
esen
t
ed
i
n
T
ab
le
3
.
T
ab
le
3
.
Pe
r
f
o
r
m
a
n
ce
r
esu
l
t o
f
th
e
g
e
n
er
ato
r
o
n
th
e
Mo
r
o
cc
an
la
w
d
ataset
C
o
si
n
e
-
si
mi
l
a
r
i
t
y
B
L
EU
sco
r
e
B
ER
T
s
c
o
r
e
C
h
r
F
+
+
F
1
sco
r
e
F
a
i
t
h
f
u
l
n
e
ss
A
n
s
w
e
r
r
e
l
e
v
a
n
c
e
C
o
n
t
e
x
t
r
e
l
e
v
a
n
c
e
Jai
s
-
13B
6
9
%
5
.
7
%
73
%
4
2
,
7
%
2
1
.
9
%
6
7
%
7
9
.
0
9
%
7
3
.
9
%
Af
ter
a
n
a
l
y
zi
n
g
t
h
e
p
er
f
o
r
m
a
n
ce
r
esu
lt
o
f
t
h
e
g
en
er
ato
r
,
w
e
f
i
n
d
a
g
o
o
d
ca
p
tu
r
e
o
f
m
ea
n
in
g
a
n
d
co
n
tex
t
o
f
t
h
e
an
s
w
er
s
w
h
ic
h
in
d
icate
s
a
n
e
x
ce
lle
n
t
s
e
m
a
n
tic
u
n
d
er
s
tan
d
i
n
g
,
at
th
e
lev
e
l
o
f
m
ea
s
u
r
in
g
t
h
e
s
i
m
ilar
it
y
b
et
w
ee
n
a
g
e
n
er
ated
an
s
w
er
an
d
a
r
e
f
er
en
ce
a
n
s
w
er
.
I
n
t
h
e
f
o
llo
w
i
n
g
,
w
e
ca
n
n
o
tice
t
h
at
t
h
e
lo
w
v
alu
e
o
b
tai
n
ed
f
o
r
th
e
leg
a
l
tex
t
w
r
itte
n
in
A
r
ab
ic
is
n
o
r
m
al
,
b
ec
au
s
e,
o
n
th
e
o
n
e
h
a
n
d
,
we
f
in
d
an
e
x
te
n
s
i
v
e
s
y
n
o
n
y
m
y
-
th
a
t
is
,
s
e
v
er
al
s
y
n
o
n
y
m
o
u
s
w
o
r
d
s
in
a
s
i
n
g
le
la
n
g
u
a
g
e
ca
n
e
x
p
r
ess
t
h
e
s
a
m
e
n
o
tio
n
o
r
id
ea
,
a
n
d
,
o
n
th
e
o
th
er
h
a
n
d
,
in
th
e
f
ie
ld
o
f
la
w
,
ea
c
h
ter
m
u
s
ed
h
as
a
l
eg
al
a
n
d
p
r
ec
is
e
m
ea
n
in
g
,
t
h
er
ef
o
r
e,
B
L
E
U
s
co
r
e
d
o
es
n
o
t
d
if
f
er
en
tiate
b
et
w
ee
n
ap
p
ar
en
t
s
y
n
o
n
y
m
s
th
a
t
m
a
y
h
av
e
d
if
f
er
en
t
leg
a
l
i
m
p
licatio
n
s
.
I
n
ad
d
itio
n
,
w
e
r
ec
all
th
at
in
g
e
n
er
al,
th
e
A
r
ab
ic
lan
g
u
a
g
e
h
a
s
a
r
ath
er
co
m
p
licated
g
r
a
m
m
atica
l
s
tr
u
ct
u
r
e,
w
it
h
ac
ce
p
tab
le
w
o
r
d
o
r
d
er
v
ar
iatio
n
s
.
T
h
is
also
in
f
lu
e
n
ce
s
t
h
e
B
L
E
U
s
co
r
e,
b
ec
au
s
e,
b
ein
g
b
ased
o
n
n
-
g
r
a
m
s
,
it
ca
n
also
p
en
alize
f
o
r
m
u
lat
io
n
s
t
h
at
ar
e
co
r
r
ec
t,
b
u
t si
m
p
l
y
d
i
f
f
er
[
2
5
]
.
Fu
r
t
h
er
m
o
r
e,
i
n
t
h
e
co
n
tex
t
o
f
NL
P
d
ea
lin
g
w
it
h
t
h
e
A
r
ab
ic
c
o
r
p
u
s
o
f
le
g
al
tex
ts
,
w
h
er
e
th
e
v
ar
iab
ilit
y
o
f
th
e
tr
a
n
s
la
ted
ter
m
i
n
o
lo
g
y
an
d
th
e
co
m
p
le
x
m
o
r
p
h
o
s
y
n
ta
x
p
o
s
e
m
aj
o
r
p
r
o
b
lem
s
,
t
h
e
F
1
s
co
r
e
s
u
g
g
es
ts
t
h
at
th
e
s
y
s
te
m
is
co
m
p
ete
n
t
e
n
o
u
g
h
to
p
ar
tiall
y
ca
p
tu
r
e
t
h
e
co
n
c
ep
tio
n
at
s
tak
e
b
y
m
ea
n
s
o
f
a
d
i
v
er
s
it
y
o
f
s
u
b
s
u
m
p
tio
n
w
h
i
le
b
ein
g
tr
an
s
lated
b
y
an
ad
ap
ted
s
o
u
r
ce
l
ex
ico
n
ev
e
n
if
it
is
d
if
f
er
en
t
f
r
o
m
th
e
r
ef
er
e
n
ce
f
o
r
m
u
latio
n
s
.
w
e
also
d
ed
u
ce
a
s
tr
o
n
g
co
r
r
elatio
n
an
d
g
o
o
d
co
h
er
en
ce
b
etw
ee
n
co
s
in
e
m
etr
ic
s
<
-
>
B
lu
e,
C
o
s
i
n
e
<
-
>
F1
an
d
B
lu
e
<
-
>
F1
,
w
h
ic
h
in
d
icate
s
r
esp
ec
ti
v
el
y
th
at
t
h
e
R
AG
s
y
s
te
m
h
as
a
n
u
n
d
e
r
s
tan
d
in
g
t
h
at
aid
s
g
en
er
atio
n
an
d
an
u
n
d
er
s
tan
d
i
n
g
a
n
d
v
o
ca
b
u
lar
y
r
elate
d
to
l
ex
ical
co
h
er
e
n
ce
.
PRACTICAL CASE: WHAT DO YOUR CORRELATIONS REVEAL?
Question:
ام
طورش يه
وزلا ةحص
؟ جا
Reference
:
ُ
ق
ا
د
ص
ل
ا
و
ِ
ن
ا
د
ِ
ه
ا
ّ
ش
ل
ا
و
ي
ِ
ل
َ
و
ل
ا
ي
ه
ِ
ج
ا
و
ز
ل
ا
ِ
ة
ح
ِ
ص
ُ
ط
و
ر
ُ
ش
Genrerated :
ر
ه
َ
م
و
د
و
ه
ُ
ش
و
ّ
ي
ِ
ل
َ
و
َ
د
و
ج
ُ
و
ُ
ح
ي
ح
ص
ل
ا
ُ
ج
ا
و
ز
ل
ا
ُ
ب
ل
َ
ط
َ
ت
َ
ي
#Resulting metrics:
Cosinus: 0.85 → Excellent (same meaning, identical
concepts)
BLEU: 0.15
→
L
ow
(different Word:
(
ا
يلول
يلو
,
ِ
ن
ا
د
ِ
ه
ا
ّ
ش
د
و
ه
ُ
ش
,
ر
ه
َ
م
ق
ا
د
ص
)
F1: 0.30
→
Moderate
(some common words:
ُ
ج
ا
و
ز
ل
ا
,
َ
د
و
ج
ُ
و
)
T
h
en
,
af
ter
a
n
al
y
zin
g
t
h
e
r
e
s
u
lt
s
o
b
tain
ed
o
n
th
e
“
f
a
ith
f
u
ln
e
s
s
,
a
n
s
w
er
r
elev
a
n
ce
,
an
d
co
n
tex
t
r
elev
an
ce
m
etr
ic
s
”
.
I
n
ter
m
s
o
f
m
ea
s
u
r
i
n
g
w
h
et
h
er
th
e
g
en
er
ated
r
esp
o
n
s
e
is
f
ait
h
f
u
l
to
th
e
r
etr
iev
ed
co
n
tex
t,
a
m
o
d
er
ate
f
id
elit
y
i
n
d
icate
s
t
h
at
th
e
m
o
d
el
g
e
n
er
all
y
r
e
m
ai
n
s
co
n
s
is
te
n
t
w
it
h
A
r
ab
ic
leg
al
d
o
cu
m
en
t
s
.
Fo
r
“
a
n
s
w
er
r
ele
v
a
n
ce
”
,
w
e
o
b
s
e
r
v
e
an
ex
ce
llen
t
s
co
r
e
o
f
7
9
.
0
9
%,
s
h
o
w
i
n
g
t
h
at
t
h
e
m
o
d
el
g
e
n
er
ates
r
ele
v
a
n
t
r
esp
o
n
s
es
to
th
e
p
o
s
ed
q
u
es
ti
o
n
s
,
in
ad
d
itio
n
to
a
g
o
o
d
ab
i
lit
y
o
f
t
h
e
s
y
s
te
m
to
id
en
ti
f
y
an
d
u
s
e
ap
p
r
o
p
r
iate
co
n
tex
t
u
al
p
ass
a
g
e
s
,
s
u
g
g
esti
n
g
an
e
f
f
ec
tiv
e
f
u
n
c
tio
n
i
n
g
o
f
t
h
e
r
etr
iev
al
m
o
d
u
le.
T
h
e
d
ed
u
ce
d
r
esu
lts
th
e
n
lea
d
u
s
to
id
en
t
if
y
t
h
at
B
GE
-
m
3
is
th
e
b
est,
w
h
ich
m
ea
n
s
t
h
at
d
e
v
elo
p
m
e
n
t
co
m
p
a
n
ies
th
at
cr
ea
te
s
i
m
ilar
ap
p
licatio
n
s
i
n
t
h
e
A
r
ab
w
o
r
l
d
n
o
w
h
a
v
e
an
o
p
en
w
ei
g
h
t,
e
f
f
icien
t
an
d
p
r
o
v
e
n
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KO
MN
I
K
A
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l
C
o
n
tr
o
l
R
etri
ev
a
l
-
a
u
g
men
ted
g
en
era
ti
o
n
fo
r
A
r
a
b
ic
leg
a
l in
fo
r
ma
tio
n
:
t
h
e
fa
mily
co
d
e
ca
s
e
s
tu
d
y
(
Ja
ma
l H
r
imec
h
)
1503
m
o
d
el,
w
h
ich
ac
ce
ler
ates
t
h
e
d
ev
elo
p
m
en
t
o
f
n
e
w
le
g
al
tec
h
n
o
lo
g
ies.
T
h
u
s
,
th
e
g
o
o
d
f
u
n
ctio
n
in
g
o
f
o
u
r
R
A
G
s
y
s
te
m
o
n
Mo
r
o
cc
an
la
w
p
r
o
v
es
th
at
w
e
h
a
v
e
a
r
eliab
le
m
et
h
o
d
th
at
m
a
k
es
tex
t
s
th
at
ar
e
co
n
tex
tu
al
l
y
co
m
p
le
x
(
lik
e
th
e
la
w
)
ac
ce
s
s
ib
le
to
th
e
g
en
er
al
p
u
b
lic,
s
o
m
eth
in
g
th
at
h
as
a
s
o
cial
i
m
p
ac
t
h
elp
s
p
eo
p
le
ac
ce
s
s
leg
al
in
f
o
r
m
atio
n
.
Mo
r
eo
v
er
,
th
e
l
o
w
B
L
E
U
s
co
r
e
ca
lls
o
n
th
e
r
esear
ch
co
m
m
u
n
it
y
to
d
elv
e
in
to
th
is
n
ich
e
o
f
d
ev
elo
p
in
g
e
v
alu
a
tio
n
m
e
th
o
d
s
f
o
r
A
r
ab
ic,
esp
ec
iall
y
in
s
p
e
cialize
d
f
ield
s
lik
e
la
w
.
T
h
e
ex
p
er
im
e
n
t
s
w
er
e
r
u
n
o
n
a
m
ac
h
in
e
w
i
th
t
h
e
f
o
llo
w
i
n
g
s
p
ec
if
icat
io
n
s
:
−
GP
U:
T
4
×
2
−
GP
U
m
e
m
o
r
y
(
V
R
A
M)
: 1
6
GB
−
S
y
s
te
m
m
e
m
o
r
y
(
R
A
M)
: 3
2
G
B
−
Fra
m
e
w
o
r
k
: P
y
T
o
r
ch
T
ab
le
4
p
r
o
v
id
e
q
u
alitativ
e
ex
a
m
p
le
s
to
s
h
o
w
th
e
f
i
n
er
p
o
in
ts
o
f
t
h
e
m
o
d
el
’
s
p
er
f
o
r
m
an
ce
.
T
h
e
in
co
r
r
ec
t
g
en
er
atio
n
s
il
lu
s
tr
ate
d
ea
d
ly
f
ail
u
r
e
m
o
d
es li
k
e
r
etu
r
n
in
g
a
f
ac
tu
al
l
y
i
n
co
r
r
ec
t a
n
s
w
er
(
h
al
lu
ci
n
atio
n
)
in
E
x
a
m
p
le
1
o
r
p
r
o
v
id
in
g
t
h
e
r
ig
h
t
b
u
t
p
er
ilo
u
s
l
y
i
n
co
m
p
lete
an
s
w
er
in
E
x
a
m
p
le
2
.
C
o
r
r
ec
t
g
en
er
atio
n
s
,
h
o
w
ev
er
,
i
n
d
icate
t
h
at
th
e
s
y
s
te
m
i
s
ca
p
ab
le
o
f
g
e
n
er
ati
n
g
a
r
esp
o
n
s
e
w
h
ic
h
i
s
s
e
m
a
n
tic
all
y
eq
u
iv
a
len
t
an
d
f
ac
t
u
all
y
co
r
r
ec
t
f
o
r
th
is
q
u
es
tio
n
.
S
u
ch
a
q
u
alitati
v
e
a
n
al
y
s
is
i
s
i
m
p
o
r
ta
n
t
to
g
et
a
n
i
n
s
ig
h
t
o
n
r
ea
l
-
w
o
r
ld
r
eliab
ilit
y
o
f
t
h
e
m
o
d
el
b
esid
e
q
u
an
tita
tiv
e
p
er
f
o
r
m
an
ce
.
T
ab
le
4
.
Q
u
alitativ
e
e
x
a
m
p
le
s
o
f
co
r
r
ec
t a
n
d
in
co
r
r
ec
t g
en
er
a
tio
n
s
I
n
c
o
r
r
e
c
t
g
e
n
e
r
a
t
i
o
n
s
C
o
r
r
e
c
t
g
e
n
e
r
a
t
i
o
n
s
Ex
e
mp
l
e
1
:
Q
u
e
st
i
o
n
:
ن
م
ر
د
ص
ي
أ
ب
ج
ي
ي
ذ
لا
ف
ر
ص
ت
لا
وه ام
ر
ذ
ع
ت
ب
هم
لع د
ع
ب
خ
س
ف
ل
ا
ب
لا
ط
؟
ه
ب
ل
ط
ل
ب
ق
ُ
ي
ل
ب
ي
ع
ل
ا
ن
م
ء
ا
ف
ش
ل
ا
R
e
f
e
r
e
n
c
e
:
جي
م
لع
لا
د
ع
ب
ب
ي
ع
لا
ب
ا
ض
ر
ل
ا
ى
لع ل
د
ي
ا
م
خ
س
ف
ل
ا
ب
ل
ا
ط
ن
م
ر
د
ص
ي
ا
ب
هئ
ا
ف
ش
ر
ذ
ع
ت
ب
G
e
n
e
r
a
t
e
d
r
e
sp
o
n
se
:
ب
لا
ل
ب
ق
هد
ق
ع
ل
د
س
ا
ف
لا
ج
او
ز
لا
خ
س
ف
ي
ي
ف
ك
لذ
و
،
هد
ع
ب
و
ء
ا
ن
ي
ت
ا
ت
ا
ح
لا
:
ة
-
ي
ف
ج
اوز
لا
ن
ا
ك
ا
ذ
إ
ى
ف
ش
ي
ن
أ
إ
،
ن
ي
ج
و
ز
لا
د
ح
فوخ
م
لا
ض
ر
م
لا
؛
ج
اوز
لا
د
ع
ب
ض
ي
ر
م
لا
-
؛
ا
ث
ث
ا
ه
ق
لط
ن
م
ل
ةت
وت
ب
م
لا
لي
لح
ت
ج
او
ز
لا
ب
ج
و
ز
لا
د
ص
ق
ا
ذ
إ
-
اذ
إ
ت
ا
ح
ل
ا
ي
ف
عقاول
ا
قي
لطت
لا
وأ
ق
طلا
ب
د
ت
ع
ي
هب
وج
و
ةلا
ح
ي
ف
ي
لو
ن
ود
ب
ج
اوز
لا
ن
ا
ك
ك
ذ
م
لا
خ
س
ف
لا
ب
م
ك
ح
لا
ر
ود
ص
لب
ق
،
هع
أ
ةر
و
.
Ex
e
mp
l
e
2
:
Q
u
e
st
i
o
n
s
:
؟
ةونب
لا
و
ةوب
ا
ت
وب
ث
ى
لع ب
ت
ر
ت
ت
ي
ت
لا
ت
ا
ب
ج
او
لا
و
ق
وق
ح
لا
ي
ه ا
م
R
e
f
e
r
e
n
c
e
:
،
هن
ي
د
و
هب
س
ن
ي
ف
ها
ب
أ
عب
ت
ي
،
ب
س
ن
لا
ل
وه
ج
م
د
لو
ةونب
ت
ت
ب
ث
ى
ت
م
ن
ب
لا
و
ةوب
ا
ت
ا
ب
ج
اوو
قوق
ح
ه
ي
لع ب
ت
ر
ت
ي
و
،
ج
ا
وز
لا
عن
اوم
هن
ع
ج
ت
ن
ي
و
ن
ا
ث
ر
اوتي
و
ةو
·
g
e
n
e
r
a
t
e
d
r
e
sp
o
n
se
:
ث
ر
ا
وت
لا
قح
هب
ت
ب
ث
ي
Ex
e
mp
l
e
1
:
Q
u
e
st
i
o
n
s
:
ة
ب
طخ
ل
ا
ف
ي
ر
ع
ت
وه ام
R
e
f
e
r
e
n
c
e
:
ةب
طخ
لا
يه
ر
ي
ب
ع
ت
ي
أ
ب
ن
ي
فر
طل
ا
د
ع
اوتلا
د
ي
ف
ت
ا
ه
ي
لع ف
ر
ا
ع
ت
م
ةلي
س
و
ج
ا
وز
لا
ى
لع
G
e
n
e
r
a
t
e
d
r
e
sp
o
n
se
:
ةب
طخ
لا
ق
ع
وهو
ج
اوز
لا
ى
لع
لج
ر
لا
و
ةأ
ر
م
لا
د
ع
اوت ي
ه
د
طخ
لا
ها
ض
ت
ق
م
ب
م
ز
ت
لي
م
ا
ب
ةب
ي
طخ
ل
ا
و
أ
ب
ي
،
ا
هري
غ
و
أ
هر
ي
غ
ى
لع ةب
طخ
لا
ن
ع
ع
ا
ن
ت
Ex
e
mp
l
e
2
:
Q
u
e
st
i
o
n
s
:
لا
وح
ا
ب
قلع
ت
ي
ا
م
ي
ف
ةب
ر
ا
غ
م
لا
د
وه
ي
لا
ى
لع ي
ر
س
ت
ي
ت
لا
ةد
ع
ا
ق
لا
ي
ه ا
م
؟
ةي
ص
خ
ش
لا
R
e
f
e
r
e
n
c
e
:
ةي
ر
ب
ع
لا
ةي
ص
خ
ش
لا
لا
وح
ا
د
ع
ا
وق
م
ه
ي
لع ي
ر
س
ت
ف
ةب
ر
ا
غ
م
لا
د
وه
ي
لا
ةي
ب
ر
غ
م
لا
G
e
n
e
r
a
t
e
d
r
e
sp
o
n
se
:
وق
ي
ر
س
ت
ا
د
ع
ا
ى
لع ةي
ب
ر
غ
م
لا
ةي
ر
ب
ع
لا
ةي
ص
خ
ش
لا
لا
وح
ةب
ر
ا
غ
م
لا
د
وه
ي
لا
.
5.
CO
NCLU
SI
O
N
T
h
is
r
esear
ch
d
e
m
o
n
s
tr
ated
th
at
a
p
r
o
p
er
ly
co
n
f
i
g
u
r
ed
R
A
G
co
u
ld
s
ig
n
i
f
ica
n
tl
y
i
m
p
r
o
v
e
ac
ce
s
s
to
an
d
u
n
d
er
s
ta
n
d
in
g
o
f
Mo
r
o
cc
an
le
g
al
i
n
f
o
r
m
atio
n
,
p
ar
tic
u
lar
l
y
t
h
e
f
a
m
i
l
y
co
d
e
,
an
u
n
d
er
ex
p
lo
r
ed
ar
ea
,
s
u
r
p
ass
in
g
th
e
ca
p
ab
ilit
ies
o
f
s
tan
d
alo
n
e
L
L
Ms.
Sp
ec
i
f
ical
l
y
,
o
u
r
s
y
s
te
m
ac
h
ie
v
ed
a
r
etr
iev
al
r
ec
all
o
f
9
8
%
f
o
r
th
e
r
elev
an
t
leg
al
ar
ticles
a
n
d
a
f
i
n
al
an
s
w
er
r
elev
a
n
ce
s
co
r
e
o
f
7
9
.
0
9
%,
u
n
d
er
s
co
r
i
n
g
it
s
ef
f
ec
ti
v
e
n
es
s
w
h
er
e
b
aseli
n
e
m
o
d
el
s
o
f
te
n
f
ail.
B
ased
o
n
t
h
e
q
u
a
n
titat
iv
e
r
e
s
u
l
ts
,
t
h
e
i
n
teg
r
ated
B
GE
-
m
3
m
o
d
el
p
r
o
v
ed
to
b
e
th
e
m
o
s
t
s
u
itab
le
f
o
r
th
i
s
ap
p
licatio
n
,
o
f
f
er
in
g
a
g
o
o
d
b
alan
ce
b
et
w
ee
n
p
er
f
o
r
m
a
n
ce
,
ef
f
icie
n
c
y
,
an
d
A
r
ab
ic
lan
g
u
ag
e.
T
h
e
id
en
tif
icatio
n
o
f
th
e
B
GE
-
m
3
m
o
d
el
as
o
p
ti
m
al
p
r
o
v
id
es
a
s
o
lid
s
tar
tin
g
p
o
in
t
f
o
r
f
u
t
u
r
e
w
o
r
k
o
n
s
e
m
a
n
tic
r
etr
iev
al
task
s
in
A
r
ab
ic
leg
al
co
n
tex
t.
T
h
is
w
o
r
k
r
ep
r
esen
ts
a
s
tep
to
w
ar
d
s
th
e
r
ea
l
d
e
m
o
cr
atiza
tio
n
o
f
ac
ce
s
s
to
leg
al
in
f
o
r
m
atio
n
f
o
r
th
e
Mo
r
o
cc
an
p
eo
p
le.
A
s
th
e
s
e
s
y
s
te
m
s
w
ill
b
e
ab
le
to
p
r
o
v
id
e
p
r
ec
is
e
an
d
u
n
d
er
s
ta
n
d
ab
le
an
s
w
er
s
,
t
h
e
y
ca
n
h
elp
p
eo
p
le
b
etter
u
n
d
er
s
t
a
n
d
th
eir
r
i
g
h
t
s
a
n
d
o
b
lig
atio
n
s
a
n
d
b
r
ea
k
d
o
w
n
th
e
b
ar
r
ier
b
et
w
ee
n
citize
n
s
a
n
d
th
e
la
w
a
n
d
it
h
i
g
h
l
ig
h
ted
th
e
n
ee
d
to
ad
ap
t
a
m
a
x
i
m
u
m
len
g
th
g
e
n
er
ato
r
to
A
r
ab
ic
in
th
e
B
GE
-
m
3
m
o
d
el
a
n
d
to
o
p
tim
ize
tr
ain
i
n
g
tec
h
n
iq
u
e
s
.
F
u
r
th
er
m
o
r
e,
th
is
r
esear
ch
r
esu
lted
i
n
th
e
cr
ea
tio
n
o
f
a
s
et
o
f
q
u
e
s
tio
n
s
a
n
d
an
s
w
er
s
,
w
h
ic
h
co
u
ld
b
e
r
ef
i
n
ed
an
d
d
ev
elo
p
ed
f
u
r
t
h
er
.
Ou
r
i
m
m
ed
iate
p
r
io
r
it
y
w
i
ll
b
e
to
i
m
p
r
o
v
e
th
e
k
e
y
co
m
p
o
n
en
ts
o
f
th
e
R
A
G
s
y
s
te
m
.
T
h
is
in
cl
u
d
e
s
o
p
tim
izin
g
th
e
B
GE
-
m
3
in
te
g
r
atio
n
m
o
d
el
o
n
o
u
r
s
p
ec
if
ic
leg
al
co
r
p
u
s
to
f
u
r
th
er
s
p
ec
ialize
its
s
ea
r
ch
ca
p
ab
ilit
ies.
W
e
also
p
lan
t
o
ex
p
lo
r
e
m
o
r
e
ad
v
an
ce
d
R
A
G
ar
ch
itectu
r
e
s
,
s
u
c
h
as
t
h
e
in
te
g
r
atio
n
o
f
a
r
ec
lass
i
f
icatio
n
m
o
d
el
to
r
e
f
in
e
th
e
r
etr
ie
v
ed
d
o
cu
m
e
n
t
s
b
ef
o
r
e
th
eir
tr
an
s
m
is
s
io
n
to
t
h
e
g
en
er
ato
r
.
On
t
h
e
o
n
e
h
an
d
,
al
t
h
o
u
g
h
b
r
o
ad
en
in
g
t
h
e
s
co
p
e
is
o
n
e
o
f
t
h
e
g
u
id
eli
n
es
f
o
r
f
u
t
u
r
e
w
o
r
k
,
it
is
e
s
s
en
tial
to
v
er
i
f
y
o
u
r
m
et
h
o
d
o
lo
g
y
in
o
t
h
er
ar
ea
s
o
f
la
w
,
s
u
c
h
as
cr
i
m
in
al
la
w
,
co
m
m
er
cial
la
w
,
an
d
ad
m
i
n
i
s
tr
at
iv
e
la
w
,
as
t
h
is
w
i
ll
allo
w
test
i
n
g
t
h
e
s
y
s
te
m
’
s
ab
ilit
y
to
p
r
o
ce
s
s
m
o
r
e
d
iv
er
s
e
a
n
d
s
p
ec
ialized
le
g
al
ter
m
s
a
n
d
r
ea
s
o
n
i
n
g
m
o
d
els,
th
u
s
co
n
f
ir
m
i
n
g
th
e
g
e
n
er
aliz
ab
ilit
y
o
f
o
u
r
m
eth
o
d
o
lo
g
y
.
Fu
r
t
h
er
m
o
r
e,
an
d
i
n
o
r
d
er
to
m
iti
g
ate
t
h
e
b
ia
s
es
in
h
er
e
n
t
in
th
e
d
ata
g
e
n
er
ated
b
y
th
e
L
L
M,
w
e
p
la
n
to
d
e
v
elo
p
a
r
ef
er
en
ce
co
r
p
u
s
,
f
u
ll
y
a
n
n
o
tated
an
d
v
al
id
ated
b
y
le
g
al
ex
p
er
ts
.
T
h
is
ex
p
er
t
-
cu
r
ated
d
ataset
w
ill
n
o
t
o
n
l
y
s
er
v
e
as
a
m
o
r
e
r
eliab
le
b
en
ch
m
ar
k
b
u
t
w
ill
also
en
ab
le
a
m
o
r
e
s
y
s
te
m
atic
s
tu
d
y
o
f
g
e
n
er
atio
n
b
ias
.
Fu
r
t
h
er
m
o
r
e,
w
e
p
lan
to
ex
p
lo
r
e
th
e
d
ev
elo
p
m
en
t
o
f
n
e
w
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
1
6
9
3
-
6930
T
E
L
KOM
NI
K
A
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l
C
o
n
tr
o
l
,
Vo
l.
23
,
No
.
6
,
Dec
em
b
er
20
25
:
1
4
9
5
-
1
505
1504
d
o
m
ai
n
-
s
p
ec
i
f
ic
ev
al
u
atio
n
m
ea
s
u
r
es
t
h
at
g
o
b
e
y
o
n
d
s
e
m
a
n
tic
s
i
m
ilar
it
y
to
ass
es
s
th
e
f
ac
tu
al
ac
cu
r
ac
y
a
n
d
lo
g
ical
co
n
s
is
te
n
c
y
o
f
g
e
n
er
at
ed
leg
al
r
esp
o
n
s
es.
ACK
NO
WL
E
D
G
M
E
NT
S
T
h
e
au
th
o
r
s
w
o
u
ld
lik
e
to
th
an
k
th
e
a
n
o
n
y
m
o
u
s
r
ev
ie
wer
s
f
o
r
th
eir
v
alu
ab
le
f
ee
d
b
ac
k
,
w
h
ic
h
s
ig
n
i
f
ica
n
tl
y
i
m
p
r
o
v
ed
t
h
e
q
u
a
lit
y
o
f
t
h
i
s
m
a
n
u
s
cr
ip
t.
F
UNDIN
G
I
NF
O
RM
AT
I
O
N
Au
t
h
o
r
s
s
tate
n
o
f
u
n
d
i
n
g
i
n
v
o
l
v
ed
.
AUTHO
R
CO
NT
RIB
UT
I
O
NS ST
A
T
E
M
E
NT
T
h
is
j
o
u
r
n
al
u
s
e
s
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
i
v
id
u
al
au
th
o
r
co
n
tr
ib
u
tio
n
s
,
r
ed
u
c
e
au
t
h
o
r
s
h
ip
d
is
p
u
tes,
an
d
f
ac
ilit
ate
co
lla
b
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
J
am
al
Hr
i
m
ec
h
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
Mo
h
a
m
m
ed
Mg
h
ar
i
✓
✓
✓
✓
✓
✓
✓
Yo
u
s
s
e
f
Z
az
✓
✓
✓
✓
✓
✓
✓
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
si
s
I
:
I
n
v
e
st
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
si
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
Au
t
h
o
r
s
s
tate
n
o
co
n
f
lic
t o
f
i
n
t
er
est.
DATA AV
AI
L
AB
I
L
I
T
Y
T
h
e
d
ata
th
at
s
u
p
p
o
r
t
th
e
f
i
n
d
in
g
s
o
f
t
h
i
s
s
t
u
d
y
,
s
p
ec
i
f
icall
y
t
h
e
2
,
5
0
0
q
u
esti
o
n
-
a
n
s
w
er
p
air
s
g
e
n
er
ated
f
o
r
th
is
r
esear
c
h
,
ar
e
av
ailab
le
f
r
o
m
t
h
e
co
r
r
esp
o
n
d
in
g
a
u
t
h
o
r
,
J
.
H.
,
u
p
o
n
r
ea
s
o
n
ab
le
r
eq
u
est.
RE
F
E
R
E
NC
E
S
[
1
]
M
.
S
i
g
a
l
a
,
A
.
B
e
e
r
,
L
.
H
o
d
g
s
o
n
,
a
n
d
A
.
O
’
C
o
n
n
o
r
,
“
B
i
g
d
a
t
a
f
o
r
m
e
a
s
u
r
i
n
g
t
h
e
i
m
p
a
c
t
o
f
t
o
u
r
i
s
m
e
c
o
n
o
m
i
c
d
e
v
e
l
o
p
m
e
n
t
p
r
o
g
r
a
m
m
e
s
:
A
p
r
o
c
e
s
s
a
n
d
q
u
a
l
i
t
y
c
r
i
t
e
r
i
a
f
r
a
m
e
w
o
r
k
f
o
r
u
s
i
n
g
b
i
g
d
a
t
a
,
”
B
i
g
D
a
t
a
a
n
d
I
n
n
o
v
a
t
i
o
n
i
n
T
o
u
r
i
s
m
,
T
r
a
v
e
l
,
a
n
d
H
o
s
p
i
t
a
l
i
t
y
.
S
p
r
i
n
g
e
r
S
i
n
g
a
p
o
r
e
,
p
p
.
5
7
–
7
3
,
2
0
1
9
,
d
o
i
:
1
0
.
1
0
0
7
/
9
7
8
-
981
-
13
-
6339
-
9
_
4
.
[
2
]
G
.
N
g
u
y
e
n
e
t
a
l
.
,
“
M
a
c
h
i
n
e
l
e
a
r
n
i
n
g
a
n
d
d
e
e
p
l
e
a
r
n
i
n
g
f
r
a
m
e
w
o
r
k
s
a
n
d
l
i
b
r
a
r
i
e
s
f
o
r
l
a
r
g
e
-
s
c
a
l
e
d
a
t
a
m
i
n
i
n
g
:
A
s
u
r
v
e
y
,
”
A
r
t
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
R
e
v
i
e
w
,
v
o
l
.
5
2
,
n
o
.
1
,
p
p
.
7
7
–
1
2
4
,
J
u
n
.
2
0
1
9
,
d
o
i
:
1
0
.
1
0
0
7
/
s
1
0
4
6
2
-
018
-
0
9
6
7
9
-
z.
[
3
]
C
.
S
h
o
r
t
e
n
a
n
d
T
.
M
.
K
h
o
s
h
g
o
f
t
a
a
r
,
“
A
s
u
r
v
e
y
o
n
i
m
a
g
e
d
a
t
a
a
u
g
m
e
n
t
a
t
i
o
n
f
o
r
d
e
e
p
l
e
a
r
n
i
n
g
,
”
J
o
u
r
n
a
l
o
f
B
i
g
D
a
t
a
,
v
o
l
.
6
,
n
o
.
1
,
D
e
c
.
2
0
1
9
,
d
o
i
:
1
0
.
1
1
8
6
/
s
4
0
5
3
7
-
0
1
9
-
0
1
9
7
-
0.
[
4
]
R
.
V
i
n
a
y
a
k
u
m
a
r
,
M
.
A
l
a
z
a
b
,
K
.
P
.
S
o
m
a
n
,
P
.
P
o
o
r
n
a
c
h
a
n
d
r
a
n
,
A
.
A
l
-
N
e
m
r
a
t
,
a
n
d
S
.
V
e
n
k
a
t
r
a
m
a
n
,
“
D
e
e
p
l
e
a
r
n
i
n
g
a
p
p
r
o
a
c
h
f
o
r
i
n
t
e
l
l
i
g
e
n
t
i
n
t
r
u
s
i
o
n
d
e
t
e
c
t
i
o
n
s
y
s
t
e
m
,
”
I
E
E
E
A
c
c
e
s
s
,
v
o
l
.
7
,
p
p
.
4
1
5
2
5
–
4
1
5
5
0
,
2
0
1
9
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
E
S
S
.
2
0
1
9
.
2
8
9
5
3
3
4
.
[
5
]
K
.
S
i
v
a
r
a
m
a
n
,
R
.
M
.
V
.
K
r
i
s
h
n
a
n
,
B
.
S
u
n
d
a
r
r
a
j
,
a
n
d
S
.
S
.
G
o
w
t
h
e
m
,
“
N
e
t
w
o
r
k
f
a
i
l
u
r
e
d
e
t
e
c
t
i
o
n
a
n
d
d
i
a
g
n
o
s
i
s
b
y
a
n
a
l
y
z
i
n
g
s
y
s
l
o
g
a
n
d
S
N
S
da
t
a
:
A
p
p
l
y
i
n
g
b
i
g
d
a
t
a
a
n
a
l
y
s
i
s
t
o
n
e
t
w
o
r
k
o
p
e
r
a
t
i
o
n
s
,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
I
n
n
o
v
a
t
i
v
e
T
e
c
h
n
o
l
o
g
y
a
n
d
E
x
p
l
o
r
i
n
g
E
n
g
i
n
e
e
r
i
n
g
,
v
o
l
.
8
,
n
o
.
9
S
3
,
p
p
.
8
8
3
–
8
8
7
,
A
u
g
.
2
0
1
9
,
d
o
i
:
1
0
.
3
5
9
4
0
/
i
j
i
t
e
e
.
I
3
1
8
7
.
0
7
8
9
S
3
1
9
.
[
6
]
A
.
D
.
D
w
i
v
e
d
i
,
G
.
S
r
i
v
a
s
t
a
v
a
,
S
.
D
h
a
r
,
a
n
d
R
.
S
i
n
g
h
,
“
A
d
e
c
e
n
t
r
a
l
i
z
e
d
p
r
i
v
a
c
y
-
p
r
e
s
e
r
v
i
n
g
h
e
a
l
t
h
c
a
r
e
b
l
o
c
k
c
h
a
i
n
f
o
r
I
o
T
,
”
S
e
n
s
o
r
s
,
v
o
l
.
1
9
,
n
o
.
2
,
J
a
n
.
2
0
1
9
,
d
o
i
:
1
0
.
3
3
9
0
/
s
1
9
0
2
0
3
2
6
.
[
7
]
F
.
A
l
-
T
u
r
j
m
a
n
,
H
.
Z
a
h
m
a
t
k
e
s
h
,
a
n
d
L
.
M
o
s
t
a
r
d
a
,
“
Q
u
a
n
t
i
f
y
i
n
g
u
n
c
e
r
t
a
i
n
t
y
i
n
i
n
t
e
r
n
e
t
o
f
m
e
d
i
c
a
l
t
h
i
n
g
s
a
n
d
b
i
g
-
d
a
t
a
s
e
r
v
i
c
e
s
u
s
i
n
g
i
n
t
e
l
l
i
g
e
n
c
e
a
n
d
d
e
e
p
l
e
a
r
n
i
n
g
,
”
I
E
E
E
A
c
c
e
s
s
,
v
o
l
.
7
,
p
p
.
1
1
5
7
4
9
–
1
1
5
7
5
9
,
2
0
1
9
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
E
S
S
.
2
0
1
9
.
2
9
3
1
6
3
7
.
[
8
]
S
.
R
.
E
l
-
B
e
l
t
a
g
y
a
n
d
M
.
A
.
A
b
d
a
l
l
a
h
,
“
E
x
p
l
o
r
i
n
g
r
e
t
r
i
e
v
a
l
a
u
g
m
e
n
t
e
d
g
e
n
e
r
a
t
i
o
n
i
n
A
r
a
b
i
c
,
”
P
r
o
c
e
d
i
a
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
,
v
o
l
.
2
4
4
,
p
p
.
2
9
6
–
3
0
7
,
2
0
2
4
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
p
r
o
c
s
.
2
0
2
4
.
1
0
.
2
0
3
.
[
9
]
H
.
A
b
d
e
l
a
z
i
m
,
M
.
T
h
a
r
w
a
t
,
a
n
d
A
.
M
o
h
a
m
e
d
,
“
S
e
m
a
n
t
i
c
e
m
b
e
d
d
i
n
g
s
f
o
r
A
r
a
b
i
c
R
e
t
r
i
e
v
a
l
A
u
g
m
e
n
t
e
d
G
e
n
e
r
a
t
i
o
n
(
A
R
A
G
)
,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
A
d
v
a
n
c
e
d
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
a
n
d
A
p
p
l
i
c
a
t
i
o
n
s
(
I
J
A
C
S
A
)
,
v
o
l
.
1
4
,
n
o
.
1
1
,
2
0
2
3
,
d
o
i
:
1
0
.
1
4
5
6
9
/
i
j
a
c
s
a
.
2
0
2
3
.
0
1
4
1
1
1
3
5
.
[
1
0
]
M
.
A
l
s
h
a
m
m
a
r
y
,
M
.
N
.
U
d
d
i
n
,
a
n
d
L
.
K
h
a
n
,
“
R
F
P
G
:
Q
u
e
s
t
i
o
n
-
a
n
s
w
e
r
i
n
g
f
r
o
m
l
o
w
-
r
e
s
o
u
r
c
e
l
a
n
g
u
a
g
e
(
A
r
a
b
i
c
)
t
e
x
t
s
u
s
i
n
g
f
a
c
t
u
a
l
l
y
a
w
a
r
e
R
A
G
,
”
2
0
2
4
I
E
E
E
1
0
t
h
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
C
o
l
l
a
b
o
r
a
t
i
o
n
a
n
d
I
n
t
e
r
n
e
t
C
o
m
p
u
t
i
n
g
(
C
I
C
)
.
I
E
E
E
,
p
p
.
1
0
7
–
1
1
6
,
O
c
t
.
2
8
,
2
0
2
4
,
d
o
i
:
1
0
.
1
1
0
9
/
c
i
c
6
2
2
4
1
.
2
0
2
4
.
0
0
0
2
3
.
[
1
1
]
K
.
S
.
J
a
f
a
r
,
A
.
A
.
M
o
h
a
m
m
a
d
,
A
.
A
.
I
s
s
a
,
a
n
d
A
.
V
.
P
a
n
o
v
,
“
A
u
t
o
m
a
t
i
n
g
t
h
e
s
e
a
r
c
h
f
o
r
l
e
g
a
l
i
n
f
o
r
m
a
t
i
o
n
i
n
A
r
a
b
i
c
:
A
n
o
v
e
l
a
p
p
r
o
a
c
h
t
o
d
o
c
u
m
e
n
t
r
e
t
r
i
e
v
a
l
,
”
R
u
s
s
i
a
n
T
e
c
h
n
o
l
o
g
i
c
a
l
J
o
u
r
n
a
l
,
v
o
l
.
1
2
,
n
o
.
5
,
p
p
.
7
–
1
6
,
O
c
t
.
2
0
2
4
,
d
o
i
:
1
0
.
3
2
3
6
2
/
2
5
0
0
-
3
1
6
x
-
2024
-
12
-
5
-
7
-
1.
.
Evaluation Warning : The document was created with Spire.PDF for Python.