I
AE
S
I
n
t
e
r
n
at
ion
al
Jou
r
n
al
of
Ar
t
if
icial
I
n
t
e
ll
ig
e
n
c
e
(
I
J
-
AI
)
Vol.
14,
No.
5:
Oc
tober
2025
,
pp.
4363
~
4370
I
S
S
N:
2252
-
8938
,
DO
I
:
10
.
11591/i
jai
.
v
14
.i
5
.
pp
43
63
-
4370
4363
Jou
r
n
al
h
omepage
:
ht
tp:
//
ij
ai
.
iaes
c
or
e
.
c
om
T
h
e
e
f
f
e
c
t
iv
e
n
e
ss
of
C
h
a
t
GPT
in
e
xt
r
ac
t
i
n
g
ar
c
h
it
e
c
t
u
r
al
p
at
t
e
r
n
s
an
d
t
ac
t
ic
s
Hi
n
d
M
il
h
e
m
,
Na
d
e
r
ah
Al
-
Jawabrah
,
Ragha
d
Abu
Wad
i
D
e
pa
r
tm
e
nt
of
I
nf
or
ma
ti
on
T
e
c
hnol
ogy,
F
a
c
ul
ty
of
P
r
in
c
e
A
l
-
H
us
s
e
in
bi
n A
bdul
la
h I
I
f
or
I
nf
or
ma
ti
on
T
e
c
hnol
ogy
,
H
a
s
he
mi
te
U
ni
ve
r
s
it
y,
Z
a
r
qa
,
J
or
da
n
Ar
t
icle
I
n
f
o
AB
S
T
RA
CT
A
r
ti
c
le
h
is
tor
y
:
R
e
c
e
ived
J
a
n
12,
2025
R
e
vis
e
d
J
ul
14,
2025
Ac
c
e
pted
Aug
6,
2025
T
h
i
s
w
o
r
k
i
n
v
es
t
i
g
at
es
the
p
o
t
e
n
t
i
al
of
Ch
a
t
G
PT
,
a
cu
t
t
i
n
g
-
ed
g
e
l
arg
e
l
an
g
u
a
g
e
mo
d
e
l
(L
L
M),
fo
r
s
o
ft
w
are
d
es
i
g
n
an
al
y
s
i
s
s
p
e
ci
fi
ca
l
l
y
in
d
et
ec
t
i
n
g
arch
i
t
ect
u
ral
p
at
t
er
n
s
an
d
t
ac
t
i
c
s
.
The
ev
al
u
at
i
o
n
i
n
v
o
l
v
e
s
co
mp
ari
n
g
Ch
at
G
PT
’s
p
erf
o
rman
ce
w
i
t
h
t
h
a
t
of
A
rch
i
e,
a
t
ra
d
i
t
i
o
n
a
l
E
c
l
i
p
s
e
p
l
u
g
i
n
d
es
i
g
n
ed
f
o
r
arch
i
t
ec
t
u
ra
l
an
a
l
y
s
i
s
.
T
h
e
s
t
u
d
y
u
s
e
s
t
h
e
s
o
u
rce
co
d
e
of
f
i
v
e
o
p
e
n
-
s
o
u
rce
s
o
ft
w
are
s
y
s
t
em
s
as
the
t
e
s
t
i
n
g
g
ro
u
n
d
.
Res
u
l
t
s
rev
ea
l
t
h
a
t
Ch
at
G
PT
ac
h
i
e
v
es
n
o
t
e
w
o
r
t
h
y
p
erf
o
rman
ce
in
b
o
t
h
p
at
t
ern
an
d
t
ac
t
i
c
d
et
ec
t
i
o
n
t
as
k
s
.
Sp
eci
f
i
cal
l
y
,
fo
r
p
at
t
ern
d
e
t
ect
i
o
n
,
Ch
at
G
PT
d
emo
n
s
t
rat
e
s
an
accu
racy
of
up
to
4
7
.
0
6
%
,
w
h
i
l
e
fo
r
t
act
i
c
d
et
e
ct
i
o
n
,
it
ach
i
e
v
es
a
p
rec
i
s
i
o
n
of
2
8
.
2
5
%
.
W
h
i
l
e
Ch
a
t
G
PT
’s
c
u
rren
t
cap
a
b
i
l
i
t
i
e
s
are
n
o
t
y
e
t
a
re
p
l
aceme
n
t
fo
r
s
p
ec
i
al
i
zed
t
o
o
l
s
l
i
k
e
A
rch
i
e,
it
o
ffer
s
s
i
g
n
i
fi
can
t
p
o
t
e
n
t
i
al
as
a
co
mp
l
emen
t
ary
t
o
o
l
in
arch
i
t
ec
t
u
ra
l
a
n
al
y
s
i
s
w
o
r
k
f
l
o
w
s
.
By
b
r
i
d
g
i
n
g
t
h
e
g
a
p
b
et
w
een
n
at
u
ral
l
an
g
u
a
g
e
u
n
d
er
s
t
a
n
d
i
n
g
an
d
s
o
ft
w
are
e
n
g
i
n
eer
i
n
g
,
Ch
at
G
PT
co
u
l
d
p
a
v
e
the
w
a
y
fo
r
mo
re
i
n
t
e
l
l
i
g
e
n
t
an
d
au
t
o
ma
t
ed
s
o
l
u
t
i
o
n
s
in
t
h
e
fi
el
d
.
H
o
w
ev
er,
a
k
ey
l
i
m
i
t
a
t
i
o
n
is
i
t
s
d
i
ff
i
cu
l
t
i
es
in
h
an
d
l
i
n
g
fo
u
n
d
at
i
o
n
al
or
t
rad
i
t
i
o
n
al
t
ac
t
i
c
s
,
res
u
l
t
i
n
g
in
a
l
o
w
er
d
et
ec
t
i
o
n
rat
e
in
cert
ai
n
areas
.
T
h
i
s
res
earch
co
n
t
r
i
b
u
t
e
s
v
al
u
ab
l
e
i
n
s
i
g
h
t
s
i
n
t
o
t
h
e
ap
p
l
i
cat
i
o
n
of
L
L
Ms
in
s
o
ft
w
are
e
n
g
i
n
eer
i
n
g
,
h
i
g
h
l
i
g
h
t
i
n
g
b
o
t
h
t
h
e
s
t
ren
g
t
h
s
a
n
d
the
l
i
m
i
t
a
t
i
o
n
s
of
Ch
at
G
PT
in
ad
d
res
s
i
n
g
co
m
p
l
e
x
arch
i
t
ec
t
u
ra
l
t
a
s
k
s
.
K
e
y
w
o
r
d
s
:
Ar
c
hit
e
c
tur
a
l
pa
tt
e
r
n
Ar
c
hit
e
c
tur
a
l
tac
ti
c
Ar
ti
f
icia
l
int
e
ll
igenc
e
C
ha
tGP
T
S
of
twa
r
e
e
nginee
r
ing
Th
i
s
is
an
o
p
en
a
c
ces
s
a
r
t
i
c
l
e
u
n
d
e
r
the
CC
BY
-
SA
l
i
ce
n
s
e.
C
or
r
e
s
pon
din
g
A
u
th
or
:
Hind
M
il
he
m
De
pa
r
tm
e
nt
of
I
nf
or
mat
ion
T
e
c
hnology
F
a
c
ult
y
of
P
r
ince
Al
-
Hus
s
e
in
bin
Abdullah
I
I
f
o
r
I
nf
or
mation
T
e
c
hnology
,
Ha
s
he
mi
te
Unive
r
s
it
y
Z
a
r
qa
,
J
or
da
n
E
mail:
hinda_is
@hu.
e
du
.
jo
1.
I
NT
RODU
C
T
I
ON
S
of
twa
r
e
a
r
c
hit
e
c
tur
e
plays
a
c
r
uc
ial
r
ole
in
de
ter
mi
ning
s
of
twa
r
e
s
ys
tems
’
maintaina
bil
it
y,
s
c
a
labili
ty,
a
nd
ove
r
a
l
l
qua
li
ty
.
It
e
nc
ompas
s
e
s
a
r
c
hit
e
c
tur
a
l
pa
tt
e
r
ns
,
tac
ti
c
s
,
a
nd
qua
li
ty
a
t
tr
ibut
e
s
.
Ar
c
hit
e
c
tur
a
l
pa
tt
e
r
ns
[
1]
a
r
e
r
e
us
a
ble,
high
-
leve
l
de
s
ign
s
olut
ions
that
pr
ovide
s
tr
uc
t
ur
e
d
a
ppr
o
a
c
he
s
to
or
ga
nizing
s
of
twa
r
e
s
ys
tems
,
of
ten
de
r
ived
f
r
om
pr
ove
n
be
s
t
pr
a
c
ti
c
e
s
in
f
r
a
mew
or
k
de
ve
lopm
e
nt
s
uc
h
as
br
oke
r
pa
tt
e
r
n
[
2
]
a
nd
laye
r
pa
tt
e
r
n
[
3]
.
T
a
c
ti
c
s
[
4]
a
r
e
de
s
ign
de
c
is
ions
that
inf
luenc
e
the
c
ont
r
ol
of
a
s
ys
tem’
s
qua
li
ty
a
tt
r
ibut
e
s
[
5
]
,
s
uc
h
as
pe
r
f
or
man
c
e
,
s
e
c
ur
it
y,
or
modi
f
iabili
ty
,
whic
h
c
oll
e
c
ti
ve
ly
d
e
f
ine
th
e
s
ys
tem’
s
s
tr
uc
tur
e
,
de
s
ign,
a
nd
be
ha
viour
.
I
de
nti
f
ying
thes
e
e
leme
nts
in
e
xis
ti
ng
c
ode
ba
s
e
s
is
c
r
i
ti
c
a
l
f
o
r
a
r
c
hit
e
c
tur
a
l
r
e
view
s
,
qua
l
it
y
a
s
s
e
s
s
ments
,
s
ys
te
m
re
-
e
nginee
r
ing,
a
nd
r
e
f
a
c
tor
ing
tas
ks
.
T
r
a
dit
io
na
ll
y,
thi
s
identif
ica
ti
on
pr
oc
e
s
s
r
e
li
e
s
he
a
vil
y
on
e
xpe
r
t
kno
wle
dge
a
nd
manua
l
a
na
lys
is
,
whic
h
can
be
ti
me
-
c
o
ns
umi
ng
a
nd
e
r
r
or
-
pr
one
,
s
uc
h
as
the
tr
a
dit
ional
tool
A
r
c
hie
[
6]
–
[
8
]
.
T
r
a
dit
ona
l
tool
s
li
ke
A
r
c
hie
de
p
e
nd
on
pr
e
de
f
ined
r
ules
a
nd
s
tatic
he
ur
is
ti
c
s
to
de
tec
t
pa
tt
e
r
ns
/t
a
c
ti
c
s
,
making
them
in
f
lexible
whe
n
a
na
lyzing
s
ys
tems
.
A
r
c
hie
a
ls
o
ha
s
a
li
m
it
a
ti
on
in
its
a
bil
it
y
to
int
e
r
p
r
e
t
c
ode
c
omm
e
nts
,
doc
umenta
ti
on
,
or
im
pli
c
it
de
s
ig
n
int
e
nt
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2252
-
8938
I
nt
J
Ar
ti
f
I
ntell
,
Vol.
14,
No.
5:
Oc
tober
2025
:
43
63
-
4370
4364
be
c
a
us
e
of
the
lac
k
of
na
tu
r
a
l
langua
ge
pr
oc
e
s
s
ing
(
NL
P
)
[
9]
.
Addi
ti
ona
ll
y,
thes
e
tool
s
a
ls
o
r
e
quir
e
manua
l
upda
tes
to
thei
r
r
ule
s
e
ts
to
a
dd
ne
w
pa
tt
e
r
ns
,
whic
h
can
be
time
-
c
ons
umi
ng
.
W
hil
e
A
r
c
hie
is
good
at
de
tec
ti
ng
tr
a
dit
ional
tac
ti
c
s
s
uc
h
as
Ke
r
be
r
os
a
n
d
a
uthentica
ti
on,
it
mi
s
s
e
s
moder
n
tac
ti
c
s
s
uc
h
as
c
ir
c
uit
br
e
a
ke
r
s
a
nd
r
e
tr
y
logi
c
,
whic
h
a
r
e
c
r
it
ica
l
tac
ti
c
s
in
c
loud
-
na
ti
ve
or
AI
-
dr
iven
s
ys
tems
.
All
of
thes
e
li
mi
tations
highl
ight
the
ne
e
d
f
o
r
mo
r
e
a
da
pti
ve
s
o
lut
ions
a
nd
mot
ivate
us
to
e
s
tablis
h
thi
s
wor
k.
R
e
c
e
nt
a
dva
nc
e
s
in
NL
P
[
9]
a
nd
the
e
mer
ge
nc
e
of
lar
g
e
langua
ge
models
(
L
L
M
)
[
10]
li
ke
C
ha
tGP
T
[
11]
ha
ve
ope
ne
d
ne
w
a
ve
nue
s
f
or
a
utom
a
ti
ng
s
of
t
wa
r
e
a
na
lys
is
tas
ks
.
W
hil
e
thes
e
tec
hnologi
e
s
ha
ve
a
ppli
e
d
to
dif
f
e
r
e
nt
a
r
e
a
s
s
uc
h
as
r
e
quir
e
ments
e
xtr
a
c
ti
on
[
12]
a
nd
bug
de
tec
ti
on
[
13]
,
their
a
ppli
c
a
ti
on
to
s
of
twa
r
e
a
r
c
hit
e
c
tur
e
a
na
lys
is
-
s
p
e
c
if
ica
ll
y
in
de
tec
ti
ng
pa
tt
e
r
ns
a
nd
tac
ti
c
s
-
r
e
mains
unde
r
e
xplor
e
d.
A
mo
r
e
c
ohe
s
ive
int
e
gr
a
ti
on
be
twe
e
n
NL
P
c
a
pa
bil
it
ies
a
nd
a
r
c
hit
e
c
tur
a
l
a
na
lys
is
c
ould
br
idge
thi
s
ga
p,
r
e
s
ult
ing
in
mor
e
int
e
ll
igent
a
nd
c
ontext
-
a
wa
r
e
a
utom
a
ti
on.
C
ha
tG
P
T
,
with
its
de
e
p
unde
r
s
tanding
of
human
langu
a
ge
a
nd
pr
e
-
tr
a
ined
knowle
dge
of
p
r
ogr
a
mm
ing
c
ons
tr
uc
ts
,
can
a
s
s
is
t
in
e
xtr
a
c
ti
ng
a
nd
r
e
a
s
oning
a
bout
a
r
c
h
it
e
c
tur
a
l
pa
tt
e
r
ns
,
tac
ti
c
s
,
a
nd
qua
li
ty
a
tt
r
ibut
e
s
dir
e
c
tl
y
f
r
om
the
s
our
c
e
c
ode
.
How
e
ve
r
,
the
e
f
f
e
c
ti
ve
ne
s
s
of
s
uc
h
langua
ge
models
in
pe
r
f
o
r
mi
ng
thes
e
s
pe
c
if
ic
tas
ks
r
e
mains
r
e
latively
une
xplo
r
e
d.
T
he
e
xplor
a
ti
on
of
C
ha
tGP
T
’
s
c
a
pa
bil
it
ies
e
xtends
a
c
r
os
s
v
a
r
ious
domains
.
F
or
ins
tanc
e
,
T
a
n
[
14]
highl
ight
s
its
a
bil
it
y
to
e
xtr
a
c
t
de
s
ign
c
onc
e
pts
f
r
om
na
r
r
a
ti
ve
s
,
s
howc
a
s
ing
it
i
s
tr
a
ns
f
or
mative
po
tential
in
c
r
e
a
ti
ve
f
ields
.
S
im
il
a
r
ly,
Gils
on
et
al.
[
15
]
e
mph
a
s
ize
NL
P
’
s
r
ole
in
identif
ying
qua
li
ty
a
tt
r
ibut
e
s
f
r
om
us
e
r
s
tor
ies
,
a
idi
ng
e
a
r
ly
a
r
c
hit
e
c
tur
a
l
de
c
is
ions
.
F
u
r
th
e
r
,
Da
s
et
al.
[
16]
s
tr
e
a
ml
ine
goa
l
modeling
p
r
oc
e
s
s
e
s
by
a
utom
a
ti
ng
the
e
xtr
a
c
ti
on
of
goa
ls
f
r
om
uns
tr
uc
tur
e
d
r
e
quir
e
ments
,
im
p
r
oving
s
take
holder
a
l
ignm
e
nt.
In
he
a
lt
hc
a
r
e
,
Hua
ng
et
al
.
[
17
]
de
mons
tr
a
te
C
ha
tGP
T
’
s
pr
of
icie
nc
y
in
c
li
nica
l
da
ta
e
xtr
a
c
ti
on,
outp
e
r
f
or
mi
ng
tr
a
dit
ional
methods
.
M
or
e
ove
r
,
S
un
et
al.
[
18]
le
ve
r
a
ge
C
ha
tGP
T
f
o
r
pha
r
mac
ovigi
lanc
e
e
ve
nt
e
xtr
a
c
ti
on,
a
nd
M
oha
jer
et
al
.
[
19]
r
e
ve
a
l
it
i
s
e
f
f
e
c
ti
ve
ne
s
s
in
s
tatic
a
na
lys
is
f
or
bug
de
tec
ti
on.
T
e
r
z
i
et
al
.
[
20
]
a
na
lyze
de
ve
loper
int
e
r
a
c
ti
ons
with
C
ha
tGP
T
’
s
c
ode
s
ugg
e
s
ti
ons
,
s
howing
im
pr
ove
d
outcome
s
with
r
e
f
ined
pr
om
pts
.
M
a
hmoudi
et
al.
[
21]
pr
opos
e
C
ha
tGP
T
-
ba
s
e
d
f
r
a
mew
or
ks
f
or
s
ys
tema
ti
c
r
e
view
s
,
while
P
r
a
gya
n
et
al.
[
22
]
highl
ight
its
potential
in
a
utom
a
ti
ng
us
e
c
a
s
e
e
xtr
a
c
ti
ons
.
Ahma
d
e
t
al.
[
2
3
]
e
x
a
mi
n
e
how
t
h
e
AI
c
a
n
a
s
s
i
s
t
s
o
f
t
wa
r
e
a
r
c
h
i
t
e
c
t
s
by
f
o
s
t
e
r
i
n
g
c
o
l
l
a
b
o
r
a
t
i
o
n
t
h
r
o
u
g
h
o
u
t
t
h
e
d
e
s
i
g
n
p
r
o
c
e
s
s
.
De
s
pit
e
thi
s
gr
owing
int
e
r
e
s
t,
li
mi
ted
wor
k
ha
s
f
oc
us
e
d
on
a
pplyi
ng
C
ha
tGP
T
to
the
c
onc
e
ptual
e
leme
nts
of
s
of
twa
r
e
a
r
c
hit
e
c
tur
e
,
s
uc
h
as
identif
ying
pa
tt
e
r
ns
a
nd
tac
ti
c
s
.
T
his
ga
p
r
e
pr
e
s
e
nts
an
oppor
tuni
ty
to
e
xtend
the
c
a
pa
bil
it
ies
of
AI
models
int
o
im
pa
c
tf
ul
a
r
e
a
s
,
c
ontr
ibut
ing
nove
l
ins
ight
s
to
the
f
ield.
T
his
pa
pe
r
a
im
s
to
e
va
luate
the
e
f
f
e
c
ti
ve
ne
s
s
of
C
ha
tGP
T
in
identif
ying
a
r
c
hit
e
c
tur
a
l
pa
tt
e
r
ns
a
nd
tac
ti
c
s
f
r
om
s
of
twa
r
e
s
ys
tems
’
s
our
c
e
c
ode
.
S
pe
c
if
ica
ll
y,
we
a
ddr
e
s
s
two
ke
y
r
e
s
e
a
r
c
h
que
s
ti
ons
:
R
Q1:
how
e
f
f
e
c
ti
ve
is
C
ha
tGP
T
in
e
xtr
a
c
ti
ng
a
r
c
hi
tec
tur
a
l
pa
tt
e
r
ns
f
r
om
s
of
twa
r
e
s
ys
tems
’
s
our
c
e
c
o
de
?
R
Q2:
how
e
f
f
e
c
ti
ve
is
C
ha
tGP
T
in
e
xtr
a
c
ti
ng
a
r
c
hi
tec
tur
a
l
tac
ti
c
s
f
r
om
s
of
twa
r
e
s
ys
tems
’
s
our
c
e
c
ode
?
By
a
ns
we
r
ing
thes
e
que
s
ti
ons
,
thi
s
r
e
s
e
a
r
c
h
s
e
e
ks
to
pr
ov
ide
ins
ight
s
int
o
the
c
a
pa
bil
it
ies
a
nd
li
mi
t
a
ti
ons
of
us
ing
C
ha
tGP
T
f
o
r
a
r
c
hit
e
c
tu
r
a
l
a
na
lys
is
,
potenti
a
ll
y
inf
or
mi
ng
the
de
s
ign
of
mor
e
int
e
ll
igent,
a
u
tom
a
ted
tool
s
f
or
s
of
twa
r
e
e
nginee
r
ing
tas
ks
.
Ou
r
c
ontr
ibu
ti
ons
to
thi
s
wor
k
a
r
e
:
−
C
onduc
ti
ng
e
xpe
r
im
e
nts
to
e
xtr
a
c
t
a
r
c
hit
e
c
tur
a
l
p
a
tt
e
r
ns
a
nd
tac
ti
c
s
f
r
om
the
s
our
c
e
c
ode
of
f
ive
op
e
n
-
s
our
c
e
s
ys
tems
us
ing
C
ha
tGP
T
whic
h
a
r
e
:
Apa
c
he
S
tor
m
[
24]
,
Apa
c
he
F
li
nk
[
25
]
,
Apa
c
he
S
pa
r
k
[
26]
,
Gr
a
dle
[
27]
,
a
nd
M
a
ve
n
[
28
]
.
−
P
e
r
f
or
mi
ng
a
c
ompar
a
ti
ve
a
na
lys
is
of
C
ha
tGP
T
's
pe
r
f
or
manc
e
a
ga
ins
t
Ar
c
hie,
a
tr
a
dit
ional
a
r
c
hit
e
c
t
ur
a
l
a
na
lys
is
tool
.
−
M
e
a
s
ur
ing
a
nd
e
va
luating
C
ha
tGP
T
's
pe
r
f
or
manc
e
thr
ough
p
r
e
c
is
ion,
r
e
c
a
ll
,
a
nd
a
c
c
ur
a
c
y
met
r
ics
.
−
Addr
e
s
s
ing
the
de
f
ined
r
e
s
e
a
r
c
h
que
s
ti
ons
by
a
na
l
yz
ing
the
e
xpe
r
im
e
ntal
r
e
s
ult
s
.
T
he
r
e
mainde
r
of
thi
s
pa
pe
r
is
o
r
ga
nize
d
as
f
oll
ows
:
s
e
c
ti
on
2
de
tails
the
r
e
s
e
a
r
c
h
methodology
e
mpl
oye
d.
Our
r
e
s
ult
s
a
nd
dis
c
us
s
ions
a
r
e
pr
e
s
e
nted
a
nd
a
na
lyze
d
in
s
e
c
ti
on
3
.
S
e
c
ti
on
4
a
ddr
e
s
s
e
s
potential
thr
e
a
ts
to
va
li
dit
y
.
F
inally
,
s
e
c
ti
on
5
c
onc
ludes
the
pa
pe
r
a
nd
outl
ines
dir
e
c
ti
ons
f
or
f
utur
e
r
e
s
e
a
r
c
h.
2.
M
E
T
HO
D
2
.
1.
Over
view
F
igur
e
1
pr
ov
ides
an
ove
r
view
of
the
pipeline
us
e
d
in
our
wor
k
.
T
he
pr
oc
e
s
s
is
s
tr
uc
tur
e
d
int
o
f
ou
r
main
s
teps
:
i)
I
nput
s
our
c
e
c
ode
:
‒
I
nputs
:
the
pipeline
be
gins
wi
th
two
types
of
inpu
t:
the
c
ompl
e
te
s
our
c
e
c
ode
(
z
ipped)
a
nd
s
e
lec
ted
c
ode
s
nippets
a
nd
‒
T
he
s
e
input
s
a
r
e
us
e
d
as
pr
ompt
s
f
or
C
ha
tGP
T
,
ini
ti
a
ti
ng
the
pr
oc
e
s
s
of
identi
f
ying
a
r
c
hit
e
c
tu
r
a
l
pa
tt
e
r
ns
a
nd
tac
ti
c
s
.
How
e
ve
r
,
only
the
c
ompl
e
te
s
our
c
e
c
ode
(
z
ipped)
is
us
e
d
as
input
f
o
r
A
r
c
hie.
ii)
E
xtr
a
c
t
pa
tt
e
r
ns
a
nd
tac
ti
c
s
:
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
Ar
ti
f
I
ntell
I
S
S
N:
2252
-
8938
T
he
e
ff
e
c
ti
v
e
ne
s
s
of
C
hatGP
T
in
e
x
tr
ac
ti
ng
ar
c
hit
e
c
tur
al
patt
e
r
ns
and
tactics
(
Hind
M
i
lhem
)
4365
‒
Ar
c
hie:
the
s
our
c
e
c
ode
unde
r
goe
s
pr
e
pr
oc
e
s
s
ing,
f
oll
owe
d
by
tr
a
ini
ng
,
a
nd
then
de
tec
ti
on
pr
ope
r
a
nd
thi
s
r
e
s
ult
s
in
two
output
s
:
e
xtr
a
c
ted
pa
tt
e
r
ns
a
nd
e
xtr
a
c
ted
tac
ti
c
s
.
‒
C
ha
tGP
T
:
the
input
is
pr
oc
e
s
s
e
d
thr
ough
pr
e
-
tr
a
in
e
d
unde
r
s
tanding,
a
na
lys
is
,
a
nd
r
e
c
ognit
ion
s
tage
s
a
nd
C
ha
tGP
T
pr
oduc
e
s
two
s
im
il
a
r
output
s
:
e
xt
r
a
c
ted
pa
tt
e
r
ns
a
nd
e
xtr
a
c
ted
tac
ti
c
s
.
iii)
C
ompar
e
r
e
s
ult
s
:
the
pa
tt
e
r
ns
a
nd
tac
ti
c
s
identif
ie
d
by
Ar
c
hie
a
nd
C
ha
tGP
T
a
r
e
c
ompar
e
d
to
a
na
ly
z
e
their
s
im
il
a
r
it
ies
,
dif
f
e
r
e
nc
e
s
,
a
nd
e
f
f
e
c
ti
ve
ne
s
s
in
i
de
nti
f
ying
thes
e
e
leme
nts
.
iv)
C
a
lcula
te
a
c
c
ur
a
c
y
:
the
f
inal
s
tep
invol
ve
s
c
a
lcula
t
ing
the
a
c
c
ur
a
c
y
of
the
output
s
f
r
om
bo
th
Ar
c
hie
a
nd
C
ha
tGP
T
to
e
va
luate
thei
r
pe
r
f
or
manc
e
in
e
xtr
a
c
t
ing
a
r
c
hit
e
c
tur
a
l
pa
tt
e
r
ns
a
nd
tac
ti
c
s
f
r
om
the
giv
e
n
s
our
c
e
c
ode
.
F
igur
e
1.
T
he
ove
r
view
of
our
wor
k
2.
2.
P
r
o
m
p
t
in
g
E
n
gin
e
e
r
in
g
S
t
r
at
e
gies
In
thi
s
s
tudy,
we
uti
li
z
e
d
s
pe
c
if
ic
pr
ompt
s
to
e
xtr
a
c
t
a
r
c
hit
e
c
tur
a
l
pa
tt
e
r
ns
a
nd
tac
ti
c
s
us
ing
C
ha
tGP
T
.
F
o
r
F
igur
e
s
2
a
nd
3,
the
e
nti
r
e
s
our
c
e
c
ode
f
il
e
of
each
s
ys
tem
wa
s
pr
ovided,
a
nd
C
ha
t
GPT
wa
s
tas
ke
d
with
identif
ying
pa
tt
e
r
ns
a
nd
tac
ti
c
s
f
r
om
t
he
f
ull
c
ode
ba
s
e
.
In
F
igur
e
4
,
we
s
uppli
e
d
s
e
lec
ted
s
nippets
of
the
c
ode
a
nd
r
e
que
s
ted
C
ha
tGP
T
to
e
xtr
a
c
t
pa
tt
e
r
ns
a
nd
tac
ti
c
s
ba
s
e
d
s
olely
on
thes
e
e
xc
e
r
pts
.
Additi
ona
ll
y,
we
a
s
ke
d
C
ha
tGP
T
to
e
xt
r
a
c
t
s
pe
c
if
ic
c
ode
s
nippets
f
r
om
the
s
ys
tem’
s
s
our
c
e
c
ode
a
nd
then
de
ter
mi
ne
the
a
r
c
hit
e
c
tur
a
l
pa
tt
e
r
ns
a
nd
tac
ti
c
s
ba
s
e
d
on
the
e
xt
r
a
c
ted
por
ti
ons
.
F
igur
e
4(
a
)
s
hows
the
pr
ompt
us
e
d
a
f
ter
pr
ovidi
ng
the
c
ode
s
nippets
,
whi
le
F
igu
r
e
4(
b)
pr
e
s
e
nts
the
pr
ompt
be
f
or
e
a
ny
s
nippets
we
r
e
given.
F
igur
e
2.
P
r
o
mpt
uti
li
z
ing
the
e
nt
ir
e
s
our
c
e
c
ode
to
identif
y
a
r
c
hit
e
c
tur
a
l
pa
tt
e
r
ns
F
igur
e
3.
P
r
o
mpt
uti
li
z
ing
the
e
nt
ir
e
s
our
c
e
c
ode
to
identif
y
a
r
c
hit
e
c
tur
a
l
tac
ti
c
s
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2252
-
8938
I
nt
J
Ar
ti
f
I
ntell
,
Vol.
14,
No.
5:
Oc
tober
2025
:
43
63
-
4370
4366
(
a
)
(
b)
F
igur
e
4.
P
r
o
mpt
us
ing
s
pe
c
if
ic
s
nippets
of
c
ode
:
(
a
)
a
f
ter
pr
ovidi
ng
s
nippets
of
c
ode
a
nd
(
b)
be
f
or
e
pr
ovidi
ng
s
nippt
e
s
of
c
ode
2.
3.
E
valu
at
io
n
m
e
t
r
ics
In
thi
s
s
tudy,
we
e
va
luate
the
pe
r
f
or
manc
e
of
C
ha
t
GPT
us
ing
r
e
c
a
ll
,
p
r
e
c
is
ion,
a
nd
a
c
c
ur
a
c
y
met
r
ics
to
a
s
s
e
s
s
the
e
f
f
e
c
ti
ve
ne
s
s
of
our
e
xpe
r
im
e
nts
.
F
or
thi
s
pur
pos
e
,
we
c
a
lcula
te
the
tr
ue
pos
it
ives
(
T
P
)
,
f
a
ls
e
pos
it
iv
es
(
F
P
)
,
a
nd
f
a
ls
e
ne
ga
ti
ve
s
(
F
N)
r
e
quir
e
d
f
or
thes
e
metr
ics
.
How
e
ve
r
,
we
do
not
c
omput
e
t
he
FN
as
a
ll
pa
tt
e
r
ns
a
nd
tac
ti
c
s
we
r
e
e
va
luate
d
a
nd
a
c
c
ount
e
d
f
or
dur
ing
de
tec
ti
on
.
S
pe
c
if
ica
ll
y:
−
T
P
:
the
number
of
pa
tt
e
r
ns
or
tac
ti
c
s
c
or
r
e
c
tl
y
iden
ti
f
ied
by
C
ha
tGP
T
or
Ar
c
hie.
−
FP:
the
nu
mber
of
pa
tt
e
r
ns
or
tac
ti
c
s
incor
r
e
c
tl
y
id
e
nti
f
ied
by
C
ha
tGP
T
or
Ar
c
hie.
−
F
N:
the
number
of
pa
tt
e
r
ns
or
tac
ti
c
s
mi
s
s
e
d
by
C
ha
tGP
T
or
Ar
c
hie.
3.
RE
S
UL
T
S
AND
DI
S
CU
S
S
I
ON
3.
1.
R
Q
1
:
h
o
w
e
f
f
e
c
t
i
v
e
is
C
h
a
t
G
P
T
in
e
x
t
r
a
c
t
i
n
g
a
r
c
h
i
t
e
c
t
u
r
a
l
p
a
t
t
e
r
n
s
f
r
o
m
s
o
f
t
w
a
r
e
s
y
s
t
e
m
s
s
o
u
r
c
e
c
o
d
e
?
To
a
dd
r
e
s
s
thi
s
que
s
ti
on,
we
uti
li
z
e
d
C
ha
tGP
T
to
identif
y
pa
tt
e
r
ns
f
r
om
the
c
ompl
e
te
c
ode
ba
s
e
a
nd
c
ompar
e
d
the
r
e
s
ult
s
with
thos
e
f
r
om
the
tr
a
dit
ion
a
l
tool
Ar
c
hie.
We
pr
e
s
e
nt
the
r
e
s
ult
s
of
pa
tt
e
r
n
a
nd
tac
ti
c
de
tec
ti
on
in
Apa
c
he
F
li
nk
us
ing
both
tool
s
.
T
a
b
l
e
1
s
u
m
ma
r
i
z
e
s
th
e
c
om
pa
r
is
o
n
of
pa
tt
e
r
n
de
tec
t
io
n
ou
tc
o
mes
b
e
t
we
e
n
A
r
c
h
ie
a
nd
C
ha
tG
P
T
,
wh
il
e
F
i
gu
r
e
5
il
lu
s
t
r
a
tes
th
e
p
e
r
c
e
nt
a
ge
of
p
a
t
te
r
ns
d
e
t
e
c
t
e
d
by
e
a
c
h
to
ol
.
As
obs
e
r
ve
d
,
C
ha
tG
P
T
a
c
hi
e
ve
d
a
pa
tt
e
r
n
d
e
t
e
c
ti
on
r
a
te
of
55
.
6
%
,
c
om
pa
r
e
d
to
A
r
c
hi
e
’
s
r
a
t
e
of
44
.
4%
.
T
a
ble
1.
C
ompar
is
on
r
e
s
ult
s
of
the
pa
tt
e
r
ns
de
tec
ti
on
f
or
Apa
c
he
F
li
nk
A
r
c
hi
te
c
tu
r
a
l
ta
c
ti
c
A
ls
o
known a
s
A
r
c
hi
e
C
ha
tG
P
T
P
ip
e
li
ne
pa
tt
e
r
n
S
tr
e
a
mi
ng
pi
pe
li
ne
❌
✔
M
a
s
te
r
-
s
la
ve
p
a
tt
e
r
n
❌
✔
L
a
ye
r
e
d
a
r
c
hi
te
c
tu
r
e
M
ul
ti
ti
e
r
a
r
c
hi
te
c
tu
r
e
❌
✔
E
ve
nt
-
dr
iv
e
n a
r
c
hi
te
c
tu
r
e
M
e
s
s
a
ge
-
dr
iv
e
n a
r
c
hi
te
c
tu
r
e
❌
✔
S
e
r
vi
c
e
c
ompone
nt
pa
tt
e
r
n
❌
✔
L
a
ye
r
s
T
ie
r
e
d
s
ys
t
e
m
✔
❌
B
r
oke
r
M
e
s
s
a
ge
br
oke
✔
❌
O
bs
e
r
ve
r
/
publ
is
h
-
s
ubs
c
r
ib
e
✔
❌
P
ip
e
s
a
nd f
il
te
r
s
✔
❌
S
ha
r
e
d
-
r
e
pos
it
or
y
C
omm
on
da
ta
r
e
pos
it
or
y
❌
❌
F
igur
e
5.
P
a
tt
e
r
ns
de
tec
ti
on
pe
r
c
e
ntage
s
f
or
C
ha
tG
P
T
a
nd
A
r
c
hie
f
o
r
Apa
c
he
F
li
nk
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
Ar
ti
f
I
ntell
I
S
S
N:
2252
-
8938
T
he
e
ff
e
c
ti
v
e
ne
s
s
of
C
hatGP
T
in
e
x
tr
ac
ti
ng
ar
c
hit
e
c
tur
al
patt
e
r
ns
and
tactics
(
Hind
M
i
lhem
)
4367
3.
2
.
R
Q
2
:
h
o
w
e
f
f
e
c
t
i
v
e
i
s
C
h
a
t
G
P
T
i
n
e
x
t
r
a
c
t
i
n
g
a
r
c
h
i
t
e
c
t
u
r
a
l
t
a
c
t
i
c
s
f
r
o
m
s
o
f
t
w
a
r
e
s
y
s
t
e
m
s
s
o
u
r
c
e
c
o
d
e
?
To
a
ddr
e
s
s
thi
s
que
s
ti
on,
we
us
e
d
C
ha
tGP
T
to
ide
nti
f
y
tac
ti
c
s
f
r
om
the
f
ul
l
c
ode
ba
s
e
a
nd
c
ompar
e
d
its
pe
r
f
or
manc
e
to
the
t
r
a
dit
ional
tool
Ar
c
hie
.
We
pr
e
s
e
nt
the
r
e
s
ult
s
of
tac
ti
c
de
tec
ti
on
in
Apa
c
he
F
li
nk
by
both
tool
s
a
r
e
p
r
e
s
e
nted.
T
a
ble
2
pr
ovides
a
c
o
mpar
is
on
of
the
tac
ti
c
s
de
tec
ted
by
Ar
c
hie
a
nd
C
ha
tGP
T
,
while
F
igur
e
6
il
lus
tr
a
tes
the
de
tec
ti
on
pe
r
c
e
ntage
s
.
As
s
hown,
C
ha
tGP
T
de
tec
ted
25.
0%
of
th
e
tac
ti
c
s
,
whe
r
e
a
s
Ar
c
hie
a
c
hieve
d
a
higher
de
tec
ti
on
r
a
te
of
75.
0%
.
T
his
is
e
xpe
c
ted,
as
Ar
c
hie
wa
s
s
p
e
c
if
ica
ll
y
de
s
igned
to
identif
y
t
r
a
dit
ional
tac
ti
c
s
,
whe
r
e
a
s
C
ha
tGP
T
is
mo
r
e
c
a
pa
ble
of
dis
c
ove
r
ing
moder
n
tac
ti
c
s
.
T
a
ble
2
.
T
he
c
ompa
r
is
on
r
e
s
ult
s
of
the
tac
ti
c
s
de
tec
ti
on
f
or
Apa
c
he
F
li
nk
A
r
c
hi
te
c
tu
r
a
l
ta
c
ti
c
A
ls
o
known a
s
A
r
c
hi
e
C
ha
tG
P
T
K
e
r
be
r
os
✔
✔
H
e
a
r
tb
e
a
t
✔
✔
P
in
g/
E
c
ho
C
onne
c
ti
vi
ty
pr
obe
✔
❌
E
xc
e
pt
io
n ha
ndl
in
g
E
r
r
or
ha
ndl
in
g/
f
a
ul
t
ha
ndl
in
g
✔
❌
A
ut
he
nt
ic
a
te
✔
❌
T
im
e
s
ta
mp
✔
❌
R
e
s
our
c
e
pooli
ng
R
e
s
our
c
e
s
ha
r
in
g
✔
❌
A
udi
t
tr
a
il
✔
❌
P
B
A
C
P
ol
ic
y
-
ba
s
e
d a
c
c
e
s
s
c
ont
r
ol
✔
❌
R
B
A
C
R
ol
e
-
ba
s
e
d a
c
c
e
s
s
c
ont
r
ol
✔
❌
R
e
s
our
c
e
s
c
he
dul
in
g
T
a
s
k
s
c
he
dul
in
g
✔
❌
S
e
s
s
io
n ma
na
ge
m
e
nt
✔
❌
L
oa
d
ba
la
nc
in
g
L
oa
d ma
na
ge
me
nt
✔
❌
R
e
s
ta
r
t
S
ys
te
m r
e
boot
✔
❌
T
im
e
-
out
✔
❌
C
a
nc
e
l
✔
❌
A
c
ti
ve
r
e
dunda
nc
y
D
a
ta
dupli
c
a
ti
on
✔
❌
C
he
c
kpoi
nt
✔
❌
R
e
tr
y
✔
❌
R
e
tr
y l
ogi
c
f
or
f
a
ul
t
to
le
r
a
nc
e
❌
❌
D
a
ta
pa
r
ti
ti
oni
ng f
or
s
c
a
la
bi
li
ty
❌
✔
R
e
s
our
c
e
pooli
ng f
or
e
f
f
ic
ie
nt
r
e
s
our
c
e
us
e
R
e
s
our
c
e
s
ha
r
in
g
❌
✔
C
ir
c
ui
t
br
e
a
ke
r
f
or
f
a
ul
t
is
ol
a
ti
on
❌
✔
F
igur
e
6.
T
a
c
ti
c
s
de
tec
ti
on
pe
r
c
e
ntage
s
f
or
C
ha
tGP
T
a
nd
A
r
c
hie
T
a
bles
3
a
nd
4
pr
e
s
e
nt
the
r
e
s
ult
s
f
or
T
P
,
tr
ue
ne
ga
ti
ve
(
TN
)
,
F
P
,
a
nd
FN
f
or
both
Ar
c
hie
a
nd
C
ha
tGP
T
.
C
ha
tGP
T
identif
ied
5
pa
tt
e
r
ns
a
nd
6
tac
ti
c
s
,
de
mons
tr
a
ti
ng
its
s
tr
e
ngth
in
de
tec
ti
ng
moder
n
a
r
c
hit
e
c
tur
a
l
pa
tt
e
r
ns
a
nd
tac
ti
c
s
,
s
uc
h
as
f
a
ult
-
tol
e
r
a
nc
e
s
tr
a
tegie
s
.
In
c
ontr
a
s
t,
Ar
c
hie
identif
ied
4
pa
tt
e
r
ns
a
nd
18
tac
ti
c
s
,
s
howc
a
s
ing
its
pr
of
icie
nc
y
in
r
e
c
og
nizing
tr
a
dit
ional
a
r
c
hit
e
c
tur
a
l
e
leme
nts
.
C
ha
tGP
T
mi
s
s
e
d
5
pa
tt
e
r
ns
a
nd
17
tac
ti
c
s
(
F
N)
,
unde
r
s
c
or
ing
its
li
mi
tations
in
identi
f
ying
f
ounda
ti
ona
l
or
t
r
a
dit
ional
tac
ti
c
s
.
On
the
other
ha
nd,
Ar
c
hie
mi
s
s
e
d
6
pa
tt
e
r
ns
a
nd
5
tac
ti
c
s
,
high
li
g
hti
ng
its
c
ha
ll
e
nge
s
in
de
tec
ti
ng
moder
n
or
nua
nc
e
d
tec
hniques
.
C
ha
tGP
T
a
ls
o
f
a
ls
e
ly
identif
ied
4
pa
tt
e
r
ns
a
nd
16
tac
ti
c
s
(
F
P
)
,
whe
r
e
a
s
Ar
c
hie
f
a
ls
e
ly
ident
if
ied
5
pa
tt
e
r
ns
a
nd
4
tac
ti
c
s
.
S
ince
a
ll
pa
tt
e
r
ns
a
nd
tac
t
ics
we
r
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2252
-
8938
I
nt
J
Ar
ti
f
I
ntell
,
Vol.
14,
No.
5:
Oc
tober
2025
:
43
63
-
4370
4368
e
va
luate
d,
TN
a
r
e
not
a
ppli
c
a
ble
in
thi
s
c
ontext.
T
he
r
e
s
ult
s
f
or
the
r
e
maining
s
ys
tems
will
be
a
va
il
a
ble
onli
ne
(
i.
e
.
,
a
tt
a
c
he
d
to
the
s
ubmi
s
s
ion
pa
pe
r
at
the
s
ubmi
s
s
ion
we
bs
it
e
)
.
T
a
ble
3
.
M
e
tr
ics
r
e
s
ult
s
of
Ar
c
hie
T
a
ble
4
.
M
e
tr
ics
r
e
s
ult
s
of
C
ha
tGP
T
A
r
c
hi
e
de
te
c
ti
on
me
tr
ic
s
A
r
c
hi
te
c
tu
r
a
l
pa
tt
e
r
ns
A
r
c
hi
te
c
tu
r
a
l
ta
c
ti
c
s
TP
4
18
TN
0
0
FP
5
4
FN
6
5
C
ha
tG
P
T
de
te
c
ti
on
me
tr
ic
s
A
r
c
hi
te
c
tu
r
a
l
pa
tt
e
r
ns
A
r
c
hi
te
c
tu
r
a
l
ta
c
ti
c
s
TP
5
6
TN
0
0
FP
4
16
FN
5
17
4.
T
HRE
AT
S
T
O
VA
L
I
D
T
Y
T
his
s
e
c
ti
on
outl
ines
the
potential
thr
e
a
ts
to
the
va
li
dit
y
of
the
f
indi
ngs
in
thi
s
s
tudy.
W
hil
e
the
r
e
s
e
a
r
c
h
e
xplor
e
s
the
c
a
pa
bil
it
ies
of
C
ha
tGP
T
in
a
r
c
hit
e
c
tur
a
l
a
na
lys
is
,
c
e
r
tain
li
mi
tations
c
ould
im
pa
c
t
the
r
obus
tnes
s
,
r
e
li
a
bil
it
y,
a
nd
ge
ne
r
a
li
z
a
bil
it
y
of
the
r
e
s
ult
s
.
T
he
s
e
th
r
e
a
ts
a
r
e
c
a
tegor
ize
d
in
to
thr
e
e
mai
n
types
:
c
ons
tr
uc
t
va
li
dit
y,
f
oc
us
ing
on
the
de
s
ign
a
nd
mea
s
ur
e
ment
of
the
s
tudy;
int
e
r
na
l
va
li
dit
y,
a
ddr
e
s
s
ing
f
a
c
tor
s
that
c
ould
inf
luenc
e
the
int
e
r
pr
e
tation
of
r
e
s
ult
s
;
a
nd
e
xter
na
l
va
li
dit
y,
c
onc
e
r
ning
the
a
ppli
c
a
bil
it
y
of
f
indi
ngs
to
br
oa
de
r
c
ontexts
.
E
a
c
h
c
a
tegor
y
hi
ghli
ghts
s
pe
c
if
ic
c
ha
ll
e
nge
s
a
nd
a
r
e
a
s
f
or
im
pr
ove
ment,
e
ns
ur
ing
a
ba
lanc
e
d
e
va
luation
of
the
s
tudy's
s
tr
e
ngths
a
nd
li
mi
tations
.
4.
1.
Cons
t
r
u
c
t
vali
d
it
y
One
of
the
thr
e
a
ts
of
thi
s
wor
k
is
that
the
s
tudy
us
e
s
pr
e
c
is
ion,
r
e
c
a
ll
,
a
nd
a
c
c
ur
a
c
y
to
e
va
luate
pe
r
f
or
manc
e
but
omi
ts
metr
ics
li
ke
F1
-
s
c
or
e
,
whic
h
c
ould
be
tt
e
r
ba
lanc
e
the
t
r
a
de
-
of
f
be
twe
e
n
pr
e
c
i
s
ion
a
nd
r
e
c
a
ll
.
We
m
it
igate
thi
s
thr
e
a
t
by
c
lea
r
ly
de
f
ine
the
c
a
lcula
ti
on
methods
f
or
each
metr
ic
a
nd
e
ns
ur
e
th
e
y
a
li
gn
with
s
tanda
r
d
p
r
a
c
ti
c
e
s
in
a
r
c
hit
e
c
tu
r
a
l
a
na
lys
is
.
Anothe
r
thr
e
a
t
is
that
the
e
f
f
e
c
ti
ve
ne
s
s
of
C
ha
tGP
T
he
a
vil
y
de
pe
nds
on
pr
ompt
qua
li
ty,
a
nd
va
r
iations
in
p
r
o
mpt
de
s
ign
mi
ght
inf
luenc
e
r
e
s
ult
s
.
T
he
lac
k
of
de
tailed
dis
c
us
s
ion
a
bout
pr
ompt
opti
mi
z
a
ti
on
c
ould
a
f
f
e
c
t
r
e
pr
oduc
ibi
li
ty.
We
mi
ti
ga
te
thi
s
by
us
e
a
s
tanda
r
dize
d
pr
ompt
e
va
luation
f
r
a
mew
or
k
to
e
ns
ur
e
c
ons
is
tenc
y
a
nd
r
e
pr
oduc
ibi
l
it
y.
T
he
a
ls
o
pa
pe
r
mi
gh
t
not
c
ove
r
a
ll
pos
s
ibl
e
a
r
c
hit
e
c
tur
a
l
pa
tt
e
r
ns
a
nd
tac
ti
c
s
c
ompr
e
he
ns
ively,
potentially
lea
ding
to
bias
e
d
r
e
s
ult
s
.
We
mi
ti
ga
te
thi
s
thr
e
a
t
by
c
ons
ult
domain
e
xpe
r
ts
to
e
ns
ur
e
t
he
s
e
lec
ted
pa
tt
e
r
ns
a
nd
tac
ti
c
s
r
e
pr
e
s
e
nt
a
c
ompr
e
he
ns
ive
a
nd
ba
lanc
e
d
s
ubs
e
t
of
a
r
c
hit
e
c
tur
a
l
e
leme
nts
.
4.
2.
I
n
t
e
r
n
al
va
li
d
it
y
O
ne
of
the
int
e
r
na
l
thr
e
a
ts
of
th
is
wor
k
is
that
t
he
c
hoice
of
f
ive
ope
n
-
s
our
c
e
s
ys
tems
mi
ght
no
t
ge
ne
r
a
li
z
e
to
other
types
of
s
of
twa
r
e
s
ys
tems
,
li
m
it
ing
the
s
c
ope
of
the
f
indi
ngs
.
We
m
it
igate
thi
s
t
hr
e
a
t
by
s
e
lec
ti
ng
one
pr
ojec
t
f
r
om
dif
f
e
r
e
nt
domain,
so
we
c
ove
r
mos
t
of
the
s
of
twa
r
e
e
nginee
r
ing
domains
.
T
he
method
f
or
manua
ll
y
ve
r
if
ying
TP
a
nd
FP
i
s
n’
t
e
xpli
c
it
ly
de
tailed
,
lea
ving
r
oom
f
o
r
s
ubjec
ti
vit
y
a
nd
potential
e
r
r
or
.
We
mi
ti
ga
te
thi
s
thr
e
a
t
by
judgi
ng
wha
t
a
r
e
T
P
a
nd
FP
.
4.
3.
E
xt
e
r
n
al
va
li
d
it
y
One
of
the
ge
ne
r
a
li
z
a
bil
it
y
th
r
e
a
ts
is
that
the
f
in
dings
a
r
e
ba
s
e
d
on
s
pe
c
if
ic
ope
n
-
s
our
c
e
pr
ojec
ts
,
a
nd
the
r
e
s
ult
s
may
not
be
a
ppli
c
a
ble
to
p
r
opr
ieta
r
y
or
les
s
-
s
tr
uc
tur
e
d
c
ode
ba
s
e
s
.
We
ha
ve
s
f
utu
r
e
wor
k
to
e
xpa
nd
the
s
tudy
to
include
pr
opr
ieta
r
y
s
ys
tems
a
nd
uns
tr
uc
tur
e
d
c
ode
ba
s
e
s
to
a
s
s
e
s
s
ge
n
e
r
a
li
z
a
bil
it
y.
Othe
r
thr
e
a
t
is
that
s
ince
the
s
tudy
e
va
luate
s
C
ha
tGP
T
(a
s
pe
c
if
ic
L
L
M
)
,
the
r
e
s
ult
s
may
not
ge
ne
r
a
li
z
e
to
other
L
L
M
s
or
AI
-
ba
s
e
d
tool
s
f
o
r
a
r
c
hit
e
c
tur
a
l
a
na
lys
is
.
We
ha
ve
a
nother
f
utur
e
wor
k
to
e
va
luate
the
pe
r
f
or
manc
e
of
other
L
L
M
s
a
nd
AI
-
ba
s
e
d
tool
s
to
pr
ovide
a
br
o
a
de
r
pe
r
s
pe
c
ti
ve
.
5.
CONC
L
USI
ON
T
his
wor
k
e
xplor
e
d
the
e
f
f
e
c
ti
ve
ne
s
s
of
C
ha
tGP
T
,
a
moder
n
LLM
,
in
identif
ying
a
r
c
hit
e
c
tur
a
l
pa
tt
e
r
ns
a
nd
tac
ti
c
s
withi
n
s
of
twa
r
e
s
ys
tems
,
c
ompar
ing
it
i
s
pe
r
f
or
manc
e
to
that
of
the
t
r
a
dit
io
na
l
tool
Ar
c
hie.
T
he
f
indi
ngs
highl
ight
the
unique
s
tr
e
ngt
hs
a
nd
li
mi
tations
of
both
tool
s
.
C
ha
tGP
T
de
mon
s
tr
a
ted
a
s
tr
ong
c
a
pa
bil
it
y
in
de
tec
ti
ng
moder
n
a
r
c
hit
e
c
tur
a
l
c
onc
e
pts
,
s
uc
h
as
f
a
ult
-
tol
e
r
a
nc
e
s
tr
a
tegie
s
,
r
e
f
le
c
ti
ng
its
a
bil
it
y
to
a
da
pt
to
e
volvi
ng
s
of
twa
r
e
p
r
a
c
ti
c
e
s
.
How
e
ve
r
,
it
s
tr
uggled
with
f
ounda
ti
ona
l
or
t
r
a
dit
iona
l
tac
ti
c
s
,
r
e
s
ult
ing
in
a
lowe
r
de
tec
ti
on
r
a
te
in
c
e
r
tain
a
r
e
a
s
.
In
c
ontr
a
s
t,
A
r
c
hie
e
xc
e
ll
e
d
in
identif
y
ing
t
r
a
dit
ional
a
r
c
hit
e
c
tur
a
l
e
leme
nts
but
s
howe
d
li
mi
tations
in
a
ddr
e
s
s
ing
moder
n,
nua
nc
e
d
tec
hniques
.
T
he
s
e
r
e
s
ult
s
unde
r
s
c
or
e
the
c
ompl
e
menta
r
y
na
tur
e
of
C
ha
tGP
T
a
nd
Ar
c
hie
in
s
of
twa
r
e
de
s
ign
a
na
lys
is
.
W
hil
e
Ar
c
hie
r
e
mains
highl
y
e
f
f
e
c
ti
ve
in
its
niche
,
C
ha
tGP
T
’
s
NL
P
c
a
pa
bil
it
ies
pr
ovide
s
igni
f
ica
nt
potential
f
o
r
e
xtending
a
r
c
hit
e
c
tur
a
l
a
na
lys
is
wor
kf
lows
,
e
s
pe
c
ially
in
c
o
ntexts
r
e
quir
ing
a
b
r
oa
de
r
unde
r
s
tanding
of
c
onte
mpor
a
r
y
pa
tt
e
r
ns
a
nd
tac
ti
c
s
.
F
utur
e
wor
k
c
ould
f
oc
us
on
e
nha
nc
ing
C
ha
tGP
T
’
s
tr
a
ini
ng
da
ta
to
im
pr
ove
its
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
Ar
ti
f
I
ntell
I
S
S
N:
2252
-
8938
T
he
e
ff
e
c
ti
v
e
ne
s
s
of
C
hatGP
T
in
e
x
tr
ac
ti
ng
ar
c
hit
e
c
tur
al
patt
e
r
ns
and
tactics
(
Hind
M
i
lhem
)
4369
r
e
c
ognit
ion
of
tr
a
dit
ional
tac
ti
c
s
,
as
we
ll
as
in
tegr
a
ti
ng
the
c
a
pa
bil
it
ies
of
both
tool
s
to
c
r
e
a
te
a
hybr
id
s
olut
ion.
S
uc
h
a
dva
nc
e
ments
c
ould
pa
ve
the
wa
y
f
or
mo
r
e
int
e
ll
igent,
a
utom
a
ted
s
ys
tems
that
b
r
idge
the
ga
p
be
twe
e
n
na
tur
a
l
langua
ge
unde
r
s
tanding
a
nd
s
of
t
wa
r
e
e
nginee
r
ing
tas
ks
,
u
lt
im
a
tely
c
ont
r
ibut
ing
to
im
pr
ove
d
s
of
twa
r
e
maintaina
bil
it
y
a
nd
s
c
a
labili
ty.
F
UN
DI
NG
I
NF
ORM
AT
I
ON
Author
s
s
tate
no
f
unding
invol
ve
d.
AU
T
HO
R
CONT
RI
B
U
T
I
ONS
S
T
AT
E
M
E
N
T
T
his
jour
na
l
us
e
s
the
C
ontr
ibut
o
r
R
oles
T
a
xo
nomy
(
C
R
e
diT
)
to
r
e
c
ognize
indi
vidual
a
uthor
c
ontr
ibut
ions
,
r
e
duc
e
a
utho
r
s
hip
dis
putes
,
a
nd
f
a
c
il
it
a
te
c
oll
a
bor
a
ti
on.
Nam
e
of
Au
t
h
or
C
M
So
Va
Fo
I
R
D
O
E
Vi
Su
P
Fu
Hind
M
il
he
m
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
Na
de
r
a
h
Al
-
J
a
wa
br
a
h
✓
✓
✓
✓
✓
✓
✓
✓
R
a
gha
d
Abu
W
a
di
✓
✓
✓
✓
✓
✓
✓
C
:
C
onc
e
pt
ua
li
z
a
ti
on
M
:
M
e
th
odol
ogy
So
:
So
f
twa
r
e
Va
:
Va
li
da
ti
on
Fo
:
Fo
r
ma
l
a
na
ly
s
is
I
:
I
nve
s
ti
ga
ti
on
R
:
R
e
s
our
c
e
s
D
:
D
a
ta
C
ur
a
ti
on
O
:
W
r
it
in
g
-
O
r
ig
in
a
l
D
r
a
f
t
E
:
W
r
it
in
g
-
R
e
vi
e
w
&
E
di
ti
ng
Vi
:
Vi
s
ua
li
z
a
ti
on
Su
:
Su
pe
r
vi
s
io
n
P
:
P
r
oj
e
c
t
a
dmi
ni
s
tr
a
ti
on
Fu
:
Fu
ndi
ng
a
c
qui
s
it
io
n
CONF
L
I
CT
OF
I
NT
E
RE
S
T
S
T
AT
E
M
E
N
T
T
he
a
uthor
s
de
c
lar
e
that
they
ha
ve
no
known
c
om
pe
ti
ng
f
inanc
ial
in
ter
e
s
ts
or
pe
r
s
ona
l
r
e
lations
hips
that
c
ould
ha
ve
a
ppe
a
r
e
d
to
inf
luenc
e
the
wor
k
r
e
p
or
ted
in
th
is
pa
pe
r
.
DA
T
A
AV
AI
L
A
B
I
L
I
T
Y
T
he
da
ta
that
s
uppor
t
the
f
indi
ngs
of
thi
s
s
tudy
a
r
e
ope
nly
a
va
il
a
ble
in
4
T
U
R
e
s
e
a
r
c
h
Da
ta
a
t
htt
ps
:/
/data
.
4tu.
nl/
pr
ivate
_da
tas
e
ts
/M
L
Z
Z
c
3B
r
6lM
HH
lG7vz
C
E
9r
c
a
_vE
li
vgI
NZ
jzC
__X35U
.
RE
F
E
RE
NC
E
S
[
1]
O
.
E
.
O
lu
kunl
e
a
nd
I
.
M
.
O
ye
r
in
de
,
“
A
r
e
vi
e
w
on
s
of
twa
r
e
a
r
c
hi
te
c
tu
r
a
l
pa
tt
e
r
ns
,”
G
lo
bal
Sc
ie
nt
if
ic
J
our
nal
s
,
vo
l.
9,
no.
8,
pp. 26
–
30, 2021.
[
2]
M
.
K
a
s
s
a
b,
M
.
M
a
z
z
a
r
a
,
J
.
L
e
e
,
a
nd
G
.
S
uc
c
i,
“
S
of
twa
r
e
a
r
c
hi
te
c
tu
r
a
l
pa
tt
e
r
ns
in
pr
a
c
ti
c
e
:
a
n
e
mpi
r
ic
a
l
s
tu
dy,”
I
nnov
at
io
ns
in
Sy
s
te
m
s
and Soft
w
a
r
e
E
ngi
ne
e
r
in
g
, vol
. 14, no. 4, pp. 263
–
271, De
c
. 2018, doi:
10.1007/s
11334
-
018
-
0319
-
4.
[
3]
Z
.
T
u,
“
R
e
s
e
a
r
c
h
on
th
e
a
ppl
ic
a
ti
on
of
la
ye
r
e
d
a
r
c
hi
te
c
tu
r
e
in
c
omput
e
r
s
of
twa
r
e
de
ve
lo
pme
nt
,
”
J
ou
r
nal
of
C
om
put
in
g
and
E
le
c
tr
oni
c
I
nf
or
m
at
io
n M
anage
m
e
nt
, vol
. 11, no. 3, pp. 34
–
38, Nov. 2023, doi:
10.54097/j
c
e
im
.v11i3.08.
[
4]
G
.
M
á
r
que
z
,
H
.
A
s
tu
di
ll
o,
a
nd
R
.
K
a
z
ma
n,
“
A
r
c
hi
t
e
c
tu
r
a
l
ta
c
ti
c
s
in
s
of
twa
r
e
a
r
c
hi
te
c
tu
r
e
:
a
s
ys
te
ma
ti
c
ma
ppi
ng
s
tu
dy,”
J
our
nal
of
Sy
s
te
m
s
and Soft
w
a
r
e
, vol
. 197, M
a
r
. 2023, doi:
10.1016/j
.j
s
s
.2022.111558.
[
5]
A
.
M
is
hr
a
,
Y
.
I
.
A
lz
oubi
,
a
nd
N
.
G
a
v
r
il
ovi
c
,
“
Q
ua
li
ty
a
tt
r
ib
ut
e
s
of
s
of
twa
r
e
a
r
c
hi
te
c
tu
r
e
in
io
t
-
ba
s
e
d
a
gr
ic
ul
tu
r
a
l
s
ys
te
ms
,”
Sm
ar
t
A
gr
ic
ul
tu
r
al
T
e
c
hnol
ogy
, vol
. 8, Aug. 2024, doi:
10.1016/j
.a
te
c
h.2024.100523.
[
6]
M
.
M
ir
a
khor
li
,
“
P
r
e
s
e
r
vi
ng
th
e
qua
li
ty
of
a
r
c
hi
te
c
tu
r
a
l
ta
c
ti
c
s
in
s
our
c
e
c
ode
,
”
M
.S
.
th
e
s
is
,
C
ol
le
g
e
of
C
omput
in
g
a
nd
D
ig
it
a
l
M
e
di
a
,
D
e
P
a
ul
U
ni
ve
r
s
it
y
,
C
hi
c
a
go
, U
ni
te
d S
ta
te
s
,
2014.
[
7]
M
. M
ir
a
khor
li
a
nd J
. C
.
-
H
ua
ng, “
D
e
te
c
ti
ng, t
r
a
c
in
g, a
nd
moni
t
or
in
g a
r
c
hi
te
c
tu
r
a
l
ta
c
ti
c
s
i
n c
ode
,”
I
E
E
E
T
r
ans
ac
ti
ons
on Soft
w
ar
e
E
ngi
ne
e
r
in
g
, vol
. 42, no. 3, pp. 205
–
220, 2016, doi:
10.1109/T
S
E
.2015.2479217
.
[
8]
M
.
M
ir
a
khor
li
,
A
.
F
a
khr
y,
A
.
G
r
e
c
hko,
M
.
W
ie
lo
c
h,
a
nd
J
.
C
.
-
H
ua
ng,
“
A
r
c
hi
e
:
a
to
ol
f
or
de
te
c
ti
ng,
moni
to
r
in
g,
a
nd
pr
e
s
e
r
vi
ng
a
r
c
hi
te
c
tu
r
a
ll
y
s
ig
ni
f
ic
a
nt
c
ode
,”
in
P
r
oc
e
e
di
ngs
of
th
e
A
C
M
S
I
G
SO
F
T
Sy
m
pos
iu
m
on
th
e
F
oundati
ons
of
Sof
tw
ar
e
E
ngi
ne
e
r
in
g
,
20
14, pp. 739
–
742
, doi
:
10.1145/2635868.26
61671.
[
9]
D
.
K
hur
a
na
,
A
.
K
ol
i,
K
.
K
ha
tt
e
r
,
a
nd
S
.
S
in
gh,
“
N
a
tu
r
a
l
la
n
gua
ge
pr
oc
e
s
s
in
g:
s
ta
te
of
th
e
a
r
t,
c
ur
r
e
nt
tr
e
nds
a
nd
c
h
a
ll
e
n
ge
s
,
”
M
ul
ti
m
e
di
a T
ool
s
and A
ppl
ic
at
io
ns
, vol
. 82, no. 3, pp. 3713
–
37
44, J
a
n. 2023, d
oi
:
10.1007/s
11042
-
022
-
13428
-
4.
[
10]
M
.
U
.
H
a
di
e
t
al
.
,
“
L
a
r
ge
la
ngua
ge
mode
ls
:
a
c
ompr
e
he
ns
i
ve
s
ur
ve
y
of
it
s
a
ppl
ic
a
ti
ons
,
c
ha
ll
e
nge
s
,
li
mi
ta
ti
ons
,
a
nd
f
u
tu
r
e
pr
os
pe
c
ts
,
”
T
e
c
hX
r
i
v
,
pp. 1
-
54,
N
ov. 2023
, doi
:
10.36227/t
e
c
hr
xi
v.23589741.
[
11]
R
.
I
s
la
m
a
nd
O
.
M
.
M
ous
hi
,
“
G
P
T
-
4o:
th
e
c
ut
ti
ng
-
e
dge
a
dva
nc
e
me
nt
in
mul
ti
moda
l
ll
m
,
”
T
e
c
hX
r
iv
,
pp.
1
-
6,
J
ul
.
2
024
,
doi
:
10.36227/t
e
c
hr
xi
v.171986596.65533294/
v1.
[
12]
L
.
Z
ha
o
e
t
al
.
,
“
N
a
tu
r
a
l
la
ngua
ge
pr
oc
e
s
s
in
g
f
or
r
e
qui
r
e
me
nt
s
e
ngi
ne
e
r
in
g:
a
s
ys
te
ma
ti
c
ma
ppi
ng
s
tu
dy,”
A
C
M
C
om
pu
ti
ng
Sur
v
e
y
s
, vol
. 54, no. 3, pp. 1
–
41, 2022, doi:
10.1145/3444689.
[
13]
R.
-
W
.
B
e
ll
o
a
nd
S
.
J
.
T
obi
,
“
S
of
twa
r
e
bugs
:
de
te
c
ti
on,
a
na
ly
s
is
a
nd
f
ix
in
g,”
SSR
N
E
le
c
tr
oni
c
J
our
nal
,
2
024,
doi
:
10.2139/s
s
r
n
.4662187.
[
14]
L
.
T
a
n,
“
U
s
in
g
C
ha
t
G
P
T
to
e
xt
r
a
c
t
de
s
ig
n
c
onc
e
pt
s
f
r
om
s
to
r
ie
s
,”
in
T
he
7t
h
I
nt
e
r
nat
io
nal
C
onf
e
r
e
nc
e
fo
r
D
e
s
ig
n
E
duc
at
io
n
R
e
s
e
ar
c
he
r
s
, 2024
,
pp. 1
-
11,
doi
:
10.21606/dr
s
lx
d2024.054.
[
15]
F
.
G
il
s
on,
M
.
G
a
ls
te
r
,
a
nd
F
.
G
e
or
is
,
“
E
xt
r
a
c
ti
ng
qua
li
ty
a
tt
r
ib
ut
e
s
f
r
om
us
e
r
s
to
r
ie
s
f
or
e
a
r
ly
a
r
c
hi
te
c
tu
r
e
de
c
i
s
io
n
ma
ki
ng
,
”
in
2019
I
E
E
E
I
nt
e
r
nat
io
nal
C
onf
e
r
e
nc
e
on
Sof
tw
a
r
e
A
r
c
hi
te
c
tu
r
e
C
om
pani
on
(
I
C
SA
-
C)
,
H
a
mbur
g,
G
e
r
ma
ny
,
2019,
pp.
129
–
136
,
doi
:
10.
1109/I
C
S
A
-
C
.2019.00031.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2252
-
8938
I
nt
J
Ar
ti
f
I
ntell
,
Vol.
14,
No.
5:
Oc
tober
2025
:
43
63
-
4370
4370
[
16]
S
.
D
a
s
,
N
.
D
e
b,
A
.
C
or
te
s
i,
a
nd
N
.
C
ha
ki
,
“
E
xt
r
a
c
ti
ng
goa
l
mo
de
ls
f
r
om
na
tu
r
a
l
la
ngua
ge
r
e
qui
r
e
me
nt
s
pe
c
if
ic
a
ti
ons
,”
J
our
na
l
of
Sy
s
te
m
s
and Soft
w
a
r
e
, vol
. 211, M
a
y 2024, doi:
10.1016/j
.j
s
s
.2
024.111981.
[
17]
J
.
H
ua
ng
e
t
al
.
,
“
A
c
r
it
ic
a
l
a
s
s
e
s
s
me
nt
of
us
in
g
C
h
a
t
G
P
T
f
or
e
xt
r
a
c
ti
ng
s
tr
uc
tu
r
e
d
da
ta
f
r
om
c
li
ni
c
a
l
not
e
s
,”
npj
D
ig
it
al
M
e
di
c
i
ne
,
vol
. 7, no. 1, M
a
y 2024, doi:
10.1038/s
41746
-
024
-
01079
-
8.
[
18]
Z
.
S
un,
G
.
P
e
r
gol
a
,
B
.
C
.
W
a
ll
a
c
e
,
a
nd
Y
.
H
e
,
“
L
e
ve
r
a
gi
ng
C
h
a
t
G
P
T
in
pha
r
ma
c
ovi
gi
la
nc
e
e
ve
nt
e
xt
r
a
c
ti
on:
a
n
e
mpi
r
ic
a
l
s
tu
dy,”
ar
X
iv
-
C
om
put
e
r
Sc
ie
nc
e
,
pp. 1
-
14,
F
e
b. 2024.
[
19]
M
.
M
.
M
oha
je
r
e
t
al
.
,
“
E
f
f
e
c
ti
ve
ne
s
s
of
C
ha
t
G
P
T
f
or
s
ta
ti
c
a
na
ly
s
is
:
how
f
a
r
a
r
e
w
e
?
,”
in
A
I
w
ar
e
2024
-
P
r
oc
e
e
di
ngs
of
th
e
1s
t
A
C
M
I
nt
e
r
nat
io
nal
C
onf
e
r
e
nc
e
on
A
I
-
P
ow
e
r
e
d
Sof
tw
ar
e
,
C
o
-
lo
c
at
e
d
w
it
h:
E
SE
C
/F
S
E
2024
,
2024,
pp.
151
–
160
,
doi
:
10.1145/3664646.36
64777.
[
20]
A
.
T
e
r
z
i,
S
.
B
ib
i,
N
.
T
s
it
s
im
ik
li
s
,
a
nd
P
.
A
ng
e
li
di
s
,
“
U
s
in
g
c
ode
f
r
om
C
ha
t
G
P
T
:
f
in
di
ng
pa
tt
e
r
ns
in
th
e
de
v
e
lo
pe
r
s
’
in
te
r
a
c
ti
on
w
it
h
C
ha
t
G
P
T
,”
R
e
us
e
and Soft
w
ar
e
Q
ual
it
y
(
I
C
SR
2024)
, pp. 137
–
152, 2024, doi:
10.1007/978
-
3
-
031
-
66459
-
5_9.
[
21]
H
.
M
a
hmoudi
,
D
.
C
h
a
ng,
H
.
L
e
e
,
N
.
G
ha
f
f
a
r
z
a
de
ga
n,
a
nd
M
.
S
.
J
a
la
li
,
“
A
c
r
it
ic
a
l
a
s
s
e
s
s
me
nt
of
la
r
ge
la
ngua
ge
mode
l
s
f
or
s
ys
te
ma
ti
c
r
e
vi
e
w
s
:
ut
il
iz
in
g
C
ha
t
G
P
T
f
or
c
ompl
e
x da
ta
e
xt
r
a
c
ti
on,”
SSR
N
E
le
c
tr
oni
c
J
our
nal
, 2024, doi:
10.2139/s
s
r
n.4797024.
[
22]
K
.
C
.
P
r
a
gya
n
, R
.
S
la
vi
n,
S
.
G
ha
na
va
ti
, T
.
B
r
e
a
ux,
a
nd
M
. B
. H
os
s
e
in
i,
“
A
n
a
n
a
ly
s
is
of
a
ut
oma
te
d
u
s
e
c
a
s
e
c
ompone
nt
e
xt
r
a
c
ti
on
f
r
om s
c
e
na
r
io
s
us
in
g
C
ha
t
G
P
T
,”
ar
x
iv
-
C
om
put
e
r
S
c
ie
nc
e
,
pp. 1
-
12,
A
ug. 2024.
[
23]
A
.
A
hma
d,
M
.
W
a
s
e
e
m,
P
.
L
ia
ng,
M
.
F
a
hmi
de
h,
M
.
S
.
A
kt
a
r
,
a
nd
T
.
M
ik
kone
n,
“
T
ow
a
r
ds
huma
n
-
bot
c
ol
la
bor
a
ti
ve
s
of
t
w
a
r
e
a
r
c
hi
te
c
ti
ng
w
it
h
c
ha
tg
pt
,”
in
P
r
oc
e
e
di
ngs
of
th
e
27t
h
I
nt
e
r
nat
io
nal
C
onf
e
r
e
nc
e
on
E
v
al
uat
io
n
and
A
s
s
e
s
s
m
e
nt
in
Sof
t
w
ar
e
E
ngi
ne
e
r
in
g
, 2023, pp. 279
–
285.
[
24]
M
.
H
.
I
qba
l
a
nd
T
.
R
.
S
oomr
o,
“
B
ig
da
ta
a
na
ly
s
i
s
:
a
pa
c
h
e
s
t
or
m
pe
r
s
pe
c
ti
ve
,”
I
nt
e
r
nat
io
nal
J
ou
r
nal
of
C
om
put
e
r
T
r
e
nds
and
T
e
c
hnol
ogy
, vol
. 19, no. 1, pp. 9
–
14, J
a
n. 2015, doi:
10.14445/
22312803/I
J
C
T
T
-
V
19P
103.
[
25]
A
.
K
a
ts
if
odi
mos
a
nd
S
.
S
c
he
lt
e
r
,
“
A
pa
c
he
f
li
nk:
s
tr
e
a
m
a
na
l
yt
ic
s
a
t
s
c
a
le
,”
in
2016
I
E
E
E
I
nt
e
r
nat
io
nal
C
onf
e
r
e
nc
e
on
C
lo
ud
E
ngi
ne
e
r
in
g W
or
k
s
hop (
I
C
2E
W
)
, I
E
E
E
, A
pr
. 2016, pp. 193
–
19
3
, doi
:
10.1109/I
C
2E
W
.2016.56.
[
26]
E
.
S
h
a
ik
h,
I
.
M
ohi
ud
di
n,
Y
.
A
l
uf
a
i
s
a
n
,
a
nd
I
.
N
a
hvi
,
“
A
p
a
c
h
e
s
p
a
r
k:
a
bi
g
d
a
ta
pr
o
c
e
s
s
in
g
e
n
gi
n
e
,
”
in
2
01
9
2n
d
I
E
E
E
M
id
dl
e
E
a
s
t
a
n
d
N
o
r
th
A
f
r
i
c
a
C
O
M
M
u
ni
c
at
i
on
s
C
o
nf
e
r
e
n
c
e
,
I
E
E
E
,
N
ov
.
20
19
,
p
p.
1
–
6
,
doi
:
10
.1
10
9/
M
E
N
A
C
O
M
M
4
66
66
.2
01
9.
89
88
54
1.
[
27]
T
. B
e
r
gl
und a
nd M
.
M
c
C
ul
lo
ugh,
B
ui
ld
in
g and te
s
ti
ng w
it
h gr
a
dl
e
:
U
nde
r
s
ta
ndi
ng ne
x
t
-
ge
ne
r
at
io
n buil
ds
. S
e
ba
s
to
pol
,
C
a
li
f
or
ni
a
:
O
'
R
e
il
ly
M
e
di
a
, 2011.
[
28]
M
.
J
e
s
ic
k
e
t
al
.
,
“
M
A
V
E
N
na
vi
ga
ti
on
ove
r
vi
e
w
,”
A
dv
anc
e
s
in
t
he
A
s
tr
onauti
c
al
Sc
ie
nc
e
s
:
Spac
e
fl
ig
ht
M
e
c
hani
c
s
2016
,
vol
.
158,
pp. 1235
–
1254, 2016.
B
I
OG
RA
P
HI
E
S
OF
AU
T
HO
RS
Hi
nd
M
i
l
hem
h
o
l
d
s
a
D
o
ct
o
r
of
S
o
ft
w
are
E
n
g
i
n
e
eri
n
g
d
e
g
ree
fro
m
O
t
t
aw
a
U
n
i
v
er
s
i
t
y
,
Can
ad
a
in
2
0
2
0
.
Sh
e
al
s
o
recei
v
e
d
h
er
B.
Sc.
an
d
M.
Sc.
(Co
mp
u
t
er
Sci
en
ce)
fro
m
Y
armo
u
k
U
n
i
v
er
s
i
t
y
,
J
o
r
d
an
in
2
0
0
7
a
n
d
2
0
0
9
,
res
p
ec
t
i
v
el
y
.
Sh
e
is
c
u
rren
t
l
y
an
a
s
s
i
s
t
an
t
p
ro
fe
s
s
o
r
at
D
ep
art
me
n
t
o
f
In
f
o
rma
t
i
o
n
in
Fac
u
l
t
y
o
f
IT
at
T
h
e
H
a
s
h
em
i
t
e
U
n
i
v
ers
i
t
y
,
Z
arq
a,
J
o
r
d
an
.
H
er
res
earc
h
i
n
c
l
u
d
es
s
o
ft
w
are
arch
i
t
ec
t
u
re
d
es
i
g
n
,
arch
i
t
ec
t
u
ra
l
p
at
t
ern
s
,
arch
i
t
ect
u
ral
t
ac
t
i
c
s
,
s
o
ft
w
are
ev
a
l
u
a
t
i
o
n
,
an
d
art
i
fi
ci
a
l
i
n
t
el
l
i
g
en
ce
(A
I)
.
Sh
e
p
u
b
l
i
s
h
ed
s
i
x
p
ap
er
s
in
i
n
t
er
n
at
i
o
n
al
j
o
u
rn
a
l
s
a
n
d
co
n
feren
ce
s
.
Sh
e
h
a
s
al
s
o
f
i
v
e
p
ap
er
s
u
n
d
er
p
ro
ce
s
s
i
n
g
-
w
ai
t
i
n
g
f
o
r
t
h
e
f
i
n
a
l
d
ec
i
s
i
o
n
of
t
h
ei
r
acce
p
t
a
n
ce.
Fro
m
2
0
1
9
t
i
l
l
n
o
w
,
s
h
e
is
w
o
r
k
i
n
g
as
a
rev
i
e
w
er
(PC
memb
er)
at
the
In
t
er
n
at
i
o
n
al
J
o
u
r
n
al
of
S
o
ft
w
are
E
n
g
i
n
eeri
n
g
an
d
K
n
o
w
l
ed
g
e
E
n
g
i
n
eeri
n
g
(I
J
SE
K
E
)
a
n
d
S
E
K
E
co
n
feren
ce.
S
h
e
can
b
e
co
n
t
ac
t
ed
at
emai
l
:
h
i
n
d
a
_
i
s
@
h
u
.
ed
u
.
j
o
.
N
a
der
a
h
A
l
-
J
a
w
a
br
a
h
h
o
l
d
s
a
D
o
c
t
o
r
of
So
f
t
w
ar
e
E
n
g
i
n
eer
i
n
g
d
e
g
ree
fro
m
Szeg
ed
U
n
i
v
er
s
i
t
y
,
in
2
0
2
1
.
Sh
e
al
s
o
recei
v
e
d
h
er
M.
Sc.
(So
ft
w
are
E
n
g
i
n
eer
i
n
g
)
fro
m
the
H
as
h
mi
t
e
U
n
i
v
ers
i
t
y
,
J
o
r
d
an
in
2
0
1
5
a
n
d
she
rece
i
v
e
d
h
er
B.
Sc.
(Co
mp
u
t
er
Sc
i
en
ce)
fr
o
m
Y
armo
u
k
U
n
i
v
ers
i
t
y
,
J
o
rd
an
in
2
0
0
4
.
Sh
e
is
cu
rren
t
l
y
an
as
s
i
s
t
a
n
t
p
r
o
fes
s
o
r
at
D
e
p
art
m
en
t
o
f
In
fo
rma
t
i
o
n
i
n
Facu
l
t
y
o
f
IT
at
The
H
as
h
emi
t
e
U
n
i
v
ers
i
t
y
,
Z
arq
a,
J
o
r
d
an
.
H
er
re
s
earch
i
n
c
l
u
d
es
s
o
f
t
w
are
t
e
s
t
i
n
g
,
art
i
f
i
ci
a
l
i
n
t
e
l
l
i
g
e
n
ce,
an
d
ma
ch
i
n
e
l
earn
i
n
g
.
Sh
e
can
b
e
co
n
t
act
e
d
at
emai
l
:
n
a
d
eram@
h
u
.
ed
u
.
j
o
.
R
a
g
ha
d
A
bu
Wa
d
i
h
o
l
d
s
h
er
B.
Sc.
d
eg
ree
in
Bu
s
i
n
e
s
s
I
n
fo
rma
t
i
o
n
T
ec
h
n
o
l
o
g
y
fro
m
the
H
a
s
h
em
i
t
e
U
n
i
v
er
s
i
t
y
,
J
o
r
d
a
n
in
2
0
2
4
.
Sh
e
is
cu
rren
t
l
y
w
o
r
k
i
n
g
as
a
res
earc
h
er
in
mach
i
n
e
l
earn
i
n
g
an
d
ar
t
i
f
i
ci
a
l
i
n
t
el
l
i
g
en
ce
t
o
p
i
c
s
.
Sh
e
can
b
e
co
n
t
act
e
d
at
emai
l
:
rag
h
a
d
.
ab
u
w
ad
i
@
g
mai
l
.
co
m
.
Evaluation Warning : The document was created with Spire.PDF for Python.