Indonesi
an
Journa
l
of El
ect
ri
cal Engineer
ing
an
d
Comp
ut
er
Scie
nce
Vo
l.
9
, No
.
2
,
Februa
ry
201
8
,
pp.
460
~
473
IS
S
N:
25
02
-
4752
,
DOI: 10
.11
591/
ijeecs
.
v9.i
2
.
pp
460
-
473
460
Journ
al h
om
e
page
:
http:
//
ia
es
core.c
om/j
ourn
als/i
ndex.
ph
p/ij
eecs
Effici
ency of Fl
at File D
atabase
Ap
proach i
n Data S
torage
and
Data Ext
raction f
or Big D
ata
Mohd
Kam
ir
Yu
s
of
1
,
Must
afa
Man
2
1
Univer
siti
Sultan Z
a
ina
l
Abidin
Kam
pus T
embila
,
22200
Besut
,
Te
ren
gg
anu, Ma
lay
s
ia
2
Univer
sit
i
Mal
a
y
si
a Te
r
engga
n
u,
21300
Kuala
Te
ren
gg
anu, Ter
engga
nu,
Mal
a
ysia
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
N
ov
7
, 2
01
7
Re
vised
Dec
9
,
201
7
Accepte
d
Ja
n
11
, 2
01
8
Big
dat
a
is
th
e
latest
industr
y
buz
zword
to
desc
ribe
la
rg
e
volume
of
struct
ure
d
and
u
nstruct
ure
d
da
ta
tha
t
ca
n
be
diff
i
cul
t
to
proc
ess
a
nd
ana
l
y
z
e.
Mos
t
of
orga
nizati
on
looki
ng
fo
r
the
best
appr
o
ac
h
to
m
an
age
and
anal
y
z
e
the
l
arg
e
volume
of
dat
a espec
i
all
y
in
m
aki
ng
a
de
ci
sion.
XM
L
and
JS
ON
are
c
hosen
b
y
m
an
y
orga
nizati
on
b
e
ca
use
of
powerf
ul
appr
oa
ch
duri
ng
ret
ri
eval
and
storage
pro
ce
ss
es.
How
ever,
the
se
appr
oa
c
hes,
the
exe
cu
tion
ti
m
e
for
ret
ri
evi
ng
l
arg
e
volume
of
dat
a
are
sti
ll
cons
ide
rab
l
y
ine
ff
ic
i
ent
due
t
o
seve
ral
f
ac
tors.
In
thi
s
cont
ri
buti
o
n,
thr
ee
d
at
ab
ase
s
appr
oa
che
s
name
l
y
Ext
ensible
Mark
up
La
ngu
age
(X
ML),
Java
Obje
ct
Nota
ti
on
(JS
ON
)
and
Fla
t
File
da
ta
b
ase
a
pproa
ch
wer
e
i
nvesti
gated
to
e
val
ua
te
the
ir
suita
bi
li
t
y
fo
r
handl
ing
thousa
nds
rec
ords
of
p
ubli
c
at
ion
d
ata.
The
resul
ts
show
ed
flat
f
i
le
is
the
best
choice
f
or
quer
y
re
tri
ev
i
ng
spee
d
and
CP
U usage
.
The
se
are
essential
to
cope
with
th
e
cha
ra
cteri
sti
cs
of
publi
ca
t
ion’
s
dat
a.
W
hil
st,
XM
L,
JS
O
N
and
Flat
File
d
a
ta
base
app
roa
ch
te
chnol
og
ie
s
ar
e
rel
a
ti
v
ely
new
to
dat
e
in
compari
son
to
the
rel
a
ti
ona
l
databa
se.
Inde
ed,
T
ext
File
Form
at
te
chno
lo
g
y
demons
tra
te
s
gr
ea
t
er
pot
ential
t
o
bec
om
e
a
k
e
y
d
ataba
se
tech
nolog
y
for
handl
ing
huge
d
at
a
due
to inc
r
eas
e
of
d
at
a
annu
all
y
.
Ke
yw
or
d
s
:
Data retrie
val
Flat
File
Form
at
JSON
XML
Copyright
©
201
8
Instit
ut
e
o
f Ad
vanc
ed
Engi
n
ee
r
ing
and
S
cienc
e
.
Al
l
rights re
serv
ed
.
Corres
pond
in
g
Aut
h
or
:
Mohd
Kam
ir Y
uso
f
,
Un
i
ver
sit
i S
ultan Zai
nal Abid
in
Kam
pu
s Tem
bila, 22200 B
es
ut,
Te
reng
ganu
,
Ma
la
ysi
a
.
1.
INTROD
U
CTION
Bi
g
data
or
bi
g
da
ta
analy
ti
c
ha
ve
bee
n
us
e
d
to
desc
ribe
t
he
data
set
s
an
d
a
naly
ti
cal
tech
ni
qu
e
s
in
app
li
cat
io
ns
t
ha
t
are
s
o
la
rg
e
and
c
om
plex
that
they
re
qu
ir
e
ad
va
nced
an
d
un
i
que
data
stora
ge,
m
anag
e
m
ent,
analy
sis
and
visu
al
iz
at
ion
a
nd
te
c
hnology
[1
]
.
Bi
g
data
al
so
ref
e
rs
to
to
ols/app
li
ca
ti
on
,
processe
s,
an
d
proce
dures
t
ha
t
al
low
orga
ni
zat
ion
s
t
o
c
re
at
e,
m
anipu
la
te
and
m
anag
e
ver
y
la
rg
e
da
ta
set
s
and
stora
ge
facil
it
ie
s.
Too
l
s
in
big
data
is
re
qu
ir
ed
in
or
der
to
ha
nd
le
the
iss
ues
i
n
big
data
s
uch
as
analy
s
is,
ca
pture,
dat
a
durati
on,
sh
a
ri
ng,
stora
ge,
tr
ansf
e
r,
vis
ualiz
at
ion
,
que
ryi
ng,
updatin
g,
and
in
form
at
i
on
pr
iva
cy
.
Most
of
orga
nizat
ion
s
or
in
dustrie
s
s
uch
a
s
healt
hc
are,
aca
dem
ic
publica
ti
on
s
,
e
tc
.
are
lo
ok
i
ng
for
the
best
appr
oac
h
or
m
et
hod
in
order
to
ha
nd
l
e
big
data.
F
or
instance
,
in
a
cadem
ic
pu
bli
cat
ion
s,
acco
rdi
ng
t
o
s
ociol
ogy
an
d
researc
h
a
rtic
le
,
a
nu
m
ber
of
r
eports
have
pointed
t
o
the
gr
owin
g
us
e
of
big
data
acr
os
s
e
conom
ic
sect
ors
an
d
it
s
po
te
ntial
to
bo
lst
er
pro
duct
ivit
y,
eff
ic
ie
ncy
an
d
gro
wt
h
[
2].
I
n
this
pap
e
r,
iss
ue
a
bout
ef
fici
enc
y
du
ri
ng
acce
ss
the
pu
bl
ic
at
ion
s
data
is
co
ns
ide
rab
le
inef
fici
ency.
The
e
ff
ic
ie
ncy
of
acce
ssi
ng
publica
ti
on
s
da
ta
is
relat
e
to
ho
w
data
is
stored
.
Bi
g
data
or
huge
data
m
us
t
be
store
d
us
i
ng
s
uitable
ap
proac
h.
I
n
trad
it
ion
al
appr
oach,
data
is
sto
red
us
in
g
relat
ion
al
da
ta
base.
By
us
i
ng
this
a
ppr
oa
ch,
t
he
data
ca
n
be
represe
nt
ed
i
n
a
ta
ble
fo
rm
.
Database
Ma
na
gem
ent
Syst
e
m
(D
BM
S)
is
us
ed
to
co
nt
ro
l
an
d
m
anipu
la
te
the
data
[3
]
[4
]
.
Howe
ver,
by
usi
ng
t
his
ap
pro
ach,
ti
m
e
to
f
et
ch
the
data
a
re
co
ns
ide
rab
ly
inef
fici
ency.
O
ne
of
t
he
s
olu
ti
on
to
handle
this
pro
blem
is
XML
appr
oach.
XM
L
is
an
e
m
erg
ing
sta
nd
a
rd
for
exch
a
ng
i
ng
re
pr
ese
ntati
on
over
the
In
te
r
net
[
5].
XML
is
widely
us
ed
to
store
and
m
anag
e
huge
of
data.
This
ap
proac
h
is
cho
sen
bec
ause
of
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci
IS
S
N:
25
02
-
4752
Eff
ic
ie
ncy o
f F
lat Fi
le
Data
base A
ppr
oa
c
h
i
n Da
t
a Stor
age
and D
ata Ext
r
action
…
(
Mo
hd K
am
ir
Yu
s
of
)
461
si
m
ple
synta
x,
easy
to
ge
nerat
e
and
pa
rse,
easy
to
de
bug,
extensi
bili
ty
,
et
c.
[6
]
.
T
his
a
ppr
oach
is
s
uc
cessf
ul
and cu
rr
e
ntly
used
by m
os
t of
industries s
uc
h as healt
h ca
re
,
educat
ion,
bus
iness, etc.
espe
ci
al
ly
inv
olv
es
with
huge
data.
A
second
data
base
app
r
oac
h
is
JSO
N
.
JSON
pro
vid
e
uniq
ue
str
eng
t
h
sim
il
ar
with
XML
ap
proac
h.
JSON
is
ch
os
e
n
beca
us
e
this
appr
oach
direc
tl
y
su
ppor
t
insi
de
Java
Scri
pt
and
is
be
st
su
it
ed
f
or
Ja
vaSc
rip
t
and
pro
vid
e
sig
nifi
cant
perf
or
m
ance
com
par
ed
to
XML
[7
]
.
JS
O
N
is
est
im
at
e
d
to
pa
rse
up
t
o
one
hu
ndre
d
tim
es
faster
tha
n
X
ML
in
m
od
ern
bro
wse.
J
SON
f
or
m
at
is
prov
e
n
m
or
e
po
werfu
l
c
om
par
e
to
XML
a
pp
ro
ac
h
in
te
rm
of
tim
e
t
o
fetc
h
or
retri
eve
data
from
database
[7
]
[
13]
[14].
H
owev
er,
aca
dem
icia
n
an
d
r
esearc
he
rs
sti
ll
lo
okin
g for the
b
est
database
appr
oach speci
al
ly
inv
olve
d
with
huge data
.
This
pa
pe
r
pr
opos
e
d
te
xt
fil
e
as
an
al
te
rnat
ive
database
appr
oach
c
om
par
ed
to
X
ML
and
J
SON
.
Publi
cat
ion
dat
aset
s
is
us
ed
for
exp
e
rim
ental
pur
po
ses
.
The
perform
ance
of
Flat
Fil
e
app
r
oa
ch
will
co
m
par
e
d
with
XML
an
d
JSO
N
a
ppr
oa
ch.
T
he
com
par
iso
ns
are
m
ade
fr
om
the
fo
ll
ow
i
ng
as
pects:
qu
e
ry
perfor
m
ance
and
CP
U
us
a
ge
f
or
data
retri
evin
g
process
.
The
rest
of
c
ontrib
utio
n
is
or
gan
iz
e
d
as
f
ollow
s:
Sect
io
n
2
gi
ves
the
relat
ed
w
orks.
Sect
ion
3
de
scribes
a
bout
the
ki
nd
s
of
da
ta
m
od
el
su
c
h
as
relat
ion
al
da
ta
base,
XML,
J
SON
and
Te
xt File
. S
ect
ion
4
discu
sses the thr
ee d
at
abase approa
ches concer
ne
d based
on expe
rim
ental
resu
lt
s an
d
our
e
xperienc
e
in
the
d
e
vel
opm
ent. F
inall
y,
a co
nclusi
on is g
ive
n
i
n
S
ect
i
on 5.
2.
RELATE
D
W
ORKS
Ba
sed
on
pa
st
researc
hes,
t
he
m
os
t
popu
la
r
appr
oach
c
ompare
d
to
relat
ion
al
database
i
s
XML
an
d
JSON.
XML
s
ta
nd
s
f
or
e
Xtensible
Ma
rk
-
up
Lan
gu
a
ge
a
sta
nd
a
rd
f
or
da
ta
exch
an
ge
issue
d
by
the
Worl
d
W
i
de
C
on
s
ort
ium
(W
3C)
i
n
1998
[
8].
XM
L
has
been
wi
dely
acce
pted
as
a
data
f
or
m
at
sta
nd
ar
d
for
data
intercha
nge
an
d
sto
rag
e
with
the
rap
i
d
de
vel
op
m
ent
of
inte
rn
et
an
d
web
s
erv
ic
e
[
9].
XM
L
appr
oach
es
have
been
im
ple
m
e
nted
for
cl
inic
al
data
sto
rag
e
[10].
T
his
te
c
hn
i
qu
e
is
ef
fec
ti
ve
to
m
anag
e
t
he
cl
inica
l
da
ta
and
trans
form
the
data
into
str
uc
ture
d
f
or
m
at
.
The
a
dvanta
ge
s
of
XML
ap
proac
h
for
cl
ini
cal
data
are
be
tt
er
in
te
rm
of
scal
abili
ty
,
flexibili
t
y
and
e
xtensi
bili
ty
.
Nati
ve
XML
appro
ac
h
al
s
o
has
bee
n
im
p
lem
ented
in
extern
al
and
distri
bu
te
d
dat
aba
se
[11].
T
he
purpose
of
native
XM
L
is
to
m
ini
mize
the
qu
e
ry
retrieval
s
pee
d
[
12
]
.
XML
a
ppr
oac
h
s
uccess
fu
l
t
o
ha
nd
le
hu
ge
da
ta
arou
nd
10
0000
rec
ords.
I
n
chem
ic
al
ind
ust
ry,
XML
al
s
o
us
e
d
for
inte
gr
at
io
n
of
c
hem
ic
a
l
data
[13].
T
he
im
ple
m
ent
at
ion
of
XM
L
ap
p
r
oac
h
be
cause
of
c
he
m
ist
ry
com
m
un
it
y
has
bee
n
slo
wer
to
ad
opt
the
I
nt
ern
et
as
a
ce
nt
ral
serv
ic
e
for
exc
hangin
g
in
form
ation
.
C
he
m
ic
a
l
data
in
volves
with
la
r
ge
nu
m
ber
of
data
fi
le
.
XML
ap
pro
ach
ca
n
im
pr
ove
the
ef
fici
en
cy
of
que
ry
pr
ocessin
g
wh
e
n
in
vo
l
ves
wit
h
the
la
r
ge
nu
m
ber
of
data
file
.
XML
is
im
ple
m
ented
to
ov
e
rco
m
e
the
inf
or
m
at
ion
sha
rin
g
each
oth
er
an
d
la
rg
e
nu
m
ber
of
databa
ses
is
su
es.
Th
rou
gh
XML
ap
proac
h,
diff
e
re
nt
syst
e
m
s
can
sh
a
r
e
and
exch
a
nge
the
i
nfor
m
at
ion
eas
il
y.
By
i
m
p
leme
ntati
on
of
X
ML
appr
oac
h
i
n
di
ff
e
ren
t
dom
ai
ns
,
XML
is
proven
to
handle
la
rge
nu
m
ber
of
da
ta
.
The
eff
ic
i
ency
of
qu
e
ry
processi
ng
us
ing
XML
is
effi
ci
ency
com
pa
red
t
o
relat
ion
al
data
base.
H
oweve
r
,
the
ef
fici
enc
y
of
qu
e
ry
processin
g
us
i
ng
XML
sti
ll
ca
n
i
m
pr
ove
by
us
ing
ano
t
her
a
ppr
o
ach
as
an
al
te
rn
at
ive
databa
se
appro
a
ch
.
Me
anwhil
e,
JS
ON
is
li
gh
t
we
igh
t
data
-
i
nter
change
form
at
is
easy
for
hum
ans
to
read
a
nd
wr
it
e,
and
f
or
m
achines
to
pa
rse
an
d
ge
ner
at
e
[
12]
.
No
wa
days,
m
or
e
and
m
or
e
da
ta
represe
nted
a
s
JSON
docum
ent.
J
SON
is
be
com
ing
t
he
universal
sta
ndar
d
data
form
at
for
the
represe
ntati
on
and
e
xc
hang
ing
the
i
nform
at
ion
.
JSON
appro
a
ch
is
m
or
e
po
w
er
ful
com
par
ed
to
XML
appr
oach.
JS
O
N
a
ppro
ac
h
ha
s
bee
n
im
ple
m
ented
a
nd
a
ble
to
handle
1000
rec
ords
t
o
25000
rec
ords.
Th
e
resu
lt
sho
ws
J
SON
ap
p
r
oac
h
is
po
we
rful
an
d
m
or
e
eff
ic
ie
nt
in
te
rm
of
stora
ge
an
d
que
r
y
retrieval
co
m
par
ed
to
XML
[
8][
15]
.
H
ow
e
ve
r,
t
he
resea
rc
her
s
sti
ll
loo
king
the
be
st
te
chn
i
qu
e
f
or
handling
huge
data.
In
t
his
pap
e
r,
te
xt
file
database
ap
pro
ach
is
intro
duc
ed
as
a
new
ap
proac
h
to
hand
le
and
m
anag
e
huge
data.
Te
xt
file
is
a
com
pu
te
r
f
il
e
that
on
ly
co
ntains
te
xt
a
nd
has
no
s
pecial
form
atting
suc
h
as
bo
l
d
te
xt
,
it
al
ic
te
xt,
i
m
ages,
et
c.
Text
file
i
s
si
m
ply,
that
way
te
xt
file
s
are
com
m
on
ly
us
ed
f
or
sto
r
age
of
in
form
at
ion
.
I
n
thi
s
pap
e
r
,
com
par
ison
wi
ll
m
ade
betwe
en
XML
an
d
J
SON
ap
proac
h
to
ha
ndle
huge
data
w
hich
i
s
m
or
e
than
50,
000.
This is im
po
rta
nt to sh
ows the
eff
ic
ie
nc
y o
f
J
SON a
pproach
for
ha
nd
li
ng
huge
d
at
a.
3.
TYPES
OF
D
ATAB
AS
E
M
ODEL
Four
ty
pe
of
da
ta
base
ap
proa
ches
are
re
p
res
ented
in
this
se
ct
ion
.
Cu
rr
e
ntly
,
m
os
t
of
data
so
urces
are
store
in
tra
diti
onal
databa
se
ap
proac
h
w
hich
i
s
cal
le
d
relat
ion
al
databa
se.
B
ecause
of
li
m
itati
on
this
ap
pr
oach,
m
any
researc
he
s
lo
ok
i
ng
an
al
te
rn
at
ive
dat
abase
a
ppr
oac
h.
Th
ree
al
te
r
na
ti
ve
ap
pro
ac
he
s
are
ide
ntifie
d
a
nd
perform
ance
am
on
g
them
are
com
par
ed
in
order
t
o
s
how
w
hich
one
i
s
bette
r
to
us
e
as
an
al
te
r
native
f
or
database
m
od
e
l.
They
are
X
ML,
JSON
an
d
TXT
.
Fig
ur
e
1
shows
t
he
di
agr
am
wh
ic
h
is
con
ta
ins
publ
ic
at
ion
data.
Ba
sed
on
Figu
re
1,
pu
b
li
cat
ion
data
co
m
ing
from
diff
eren
t
sou
rces
su
ch
as
arti
cl
e,
book,
in
procee
ding
,
m
ast
er th
esi
s (m
s
thesis),
proc
eedin
g,
we
bs
it
e/
URL (w
ww)
and P
hD thesis
(phdthesis
).
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2502
-
4752
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci,
Vo
l.
9
,
No.
2
,
Fe
bruary
2
01
8
:
4
60
–
4
7
3
462
Figure
1
.
Str
uc
ture of
P
ub
li
ca
ti
on
Data
These
data
s
ources
a
re
colle
ct
and
extract
to
relat
ion
al
database
a
ppr
oach.
Fig
ur
e
2
sh
ows
ho
w
publica
ti
on
da
ta
so
urce
in
s
tructu
red
data
form
at
is
ext
ract
an
d
sto
re
in
relat
ion
al
database
.
N
um
ber
of
record
s
are
store
d
in
relat
io
n
databa
se
is
arou
nd
50,
000
reco
r
ds
.
T
hes
e
reco
r
ds
are
sp
li
t
into
four
(4
)
segm
ents:
1,
00
0
rec
ords,
5,0
00
recor
ds
,
10,000
recor
ds
a
nd
50,
000
rec
ords.
Af
te
r
rec
ords
se
gm
enta
ti
on
is
done,
t
hese
rec
ords
f
ro
m
the
relat
ion
al
datab
ase
are
co
nv
e
rt
into
three
different
data
f
or
m
at
.
They
are
XML,
JSON
a
nd
TX
T
f
or
m
at
.
XML,
JS
O
N
an
d
TX
T
data
f
or
m
at
can
co
ns
i
der
e
d
as
an
al
te
rn
at
ive
a
ppr
oach
f
or
database
appr
oa
ch.
Figure
2.
P
ub
li
cat
ion
Data
are
Extract
i
nto
T
hr
ee
D
if
fer
e
nt
Fo
rm
at
Algorithm
is
design
e
d
in
ord
er
to
al
low
dat
a
fr
om
relat
ion
database
ap
proach
c
onver
te
d
into
XML,
JSON
a
nd
flat
file
(text
fo
rm
at
).
Sect
ion
3.
1
dem
on
strat
e
how
data
f
ro
m
the
relat
ion
al
database
is
co
nv
e
rted
into
XML,
JS
ON an
d flat
f
il
e (text
form
at
).
3.1.
The Rel
at
io
nal
D
at
ab
ase
Appro
ach
The
def
i
niti
on
about
relat
ion
a
l
database
is
a
data
a
bs
tract
io
n
that
prese
nts
the
data
in
a
database
as
a
set
of
ta
bles
[
16
]
.
Re
la
ti
on
al
data
is
com
plex,
it
m
i
m
ic
s
the
way
people
thin
k
by
groupin
g
sim
il
ar
ob
j
ect
s
tog
et
he
r
an
d
br
ea
king
do
w
n
com
plex
ob
j
ect
s
i
nto
sim
i
la
r
on
e
s.
T
ABLE
1
unti
l
TABLE
7
s
ho
ws
ho
w
publica
ti
on
data
is
stored
.
Ta
bles
that
con
ta
i
ns
the
pu
blica
ti
on
data
is
div
i
ded
int
o
two
pa
rt;
ro
w
a
nd
c
ol
um
n.
Colum
n
repres
ent att
rib
utes
nam
e and
r
ow
s
re
pr
ese
nt
nu
m
ber
of
data
(so
m
et
hin
g i
s call
ed
tup
le
s
).
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci
IS
S
N:
25
02
-
4752
Eff
ic
ie
ncy o
f F
lat Fi
le
Data
base A
ppr
oa
c
h
i
n Da
t
a Stor
age
and D
ata Ext
r
action
…
(
Mo
hd K
am
ir
Yu
s
of
)
463
Table
1.
Ar
ti
cl
e
Id
au
th
o
r
title
p
ag
es
y
ear
v
o
lu
m
e
Jo
u
rnal
u
rl
2
7
4
2
2
2
N.
Pr
ati
A Par
ti
al
Mod
el of
NP
with
E
.
1245
-
1253
1994
59
J. Sy
m
b
.
Log
.
d
b
/jo
u
rnals
/jsy
m
l/j
sy
m
l5
9
.ht
m
l
#
Prati94
2
7
4
2
2
4
J. Barkle
y
Ro
ss
er
Go
d
el
Theo
re
m
s f
o
r
No
n
-
Co
n
stru
ctiv
e
Log
ics.
129
-
1
3
7
1937
2
J. Sy
m
b
.
Log
.
d
b
/jo
u
rnals
/jsy
m
l/j
sy
m
l2
.ht
m
l#
Ro
ss
er37
2
9
6
0
2
7
An
d
reas
Dan
d
alis,
Vik
to
r
K.
Prasan
n
a
Ru
n
-
ti
m
e
p
erfo
r
m
an
ce
o
p
ti
m
iz
atio
n
o
f
an FPGA
-
b
ased
d
ed
u
ctio
n
en
g
in
e f
o
r
SAT
so
lv
ers.
547
-
5
6
2
2002
7
ACM
Tr
an
s.
Desig
n
Au
to
m
.
Electr.
S
y
st.
d
b
/jo
u
rnals
/to
d
aes/to
d
aes7
.ht
m
l
#
Dan
d
alisP0
2
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
Table
2.
B
ook
id
isb
n
au
th
o
r
title
Series
v
o
lu
m
e
p
u
b
lish
er
y
ear
u
rl
214
3
-
540
-
5
5
3
8
2
-
7
An
d
rew
Ch
eese
Parallel
E
x
ecut
io
n
o
f
Par
lo
g
Lecture
No
tes in
Co
m
p
u
ter
Science
586
Sp
ring
er
1992
-
219
3
-
540
-
1
2
2
8
2
-
6
Hein
z Ben
d
er
Ko
rr
ek
te
Zug
rif
f
e zu
v
erteilten Daten
Inf
o
r
m
atik
-
Fach
b
erichte
63
Sp
ring
er
1983
-
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
Table
3: Inpr
oc
eedin
g
id
au
th
o
r
title
Pag
es
y
ear
b
o
o
k
title
u
rl
3
3
8
3
9
8
Ro
salin
d
Bard
en
,
Su
san
Step
n
ey
Su
p
p
o
rt
f
o
r
Usin
g
Z.
255
-
280
1992
Z
Use
r
W
o
rks
h
o
p
d
b
/co
n
f
/zu
m
/zu
m
1
9
9
2
.ht
m
l#
Bard
en
S
9
2
3
3
8
4
0
5
Alf
S
m
ith
On
Recu
rsiv
e Fr
e
e
Ty
p
es in
Z.
3
-
39
1991
Z
Use
r
W
o
rks
h
o
p
d
b
/co
n
f
/zu
m
/zu
m
1
9
9
1
.ht
m
l#
S
m
ith
9
1
3
3
8
4
1
9
Dav
id
Gries
Equ
atio
n
al L
o
g
ic: A
Grea
t
Ped
ag
o
g
ical
Too
l f
o
r
Tea
508
-
509
1995
ZUM
508
-
5
0
9
:
:
:
:
:
:
:
:
:
:
:
:
:
:
Table
4.
Mst
he
sis
id
au
th
o
r
Title
y
ear
Sch
o
o
l
12
Tolg
a Yurek
Ef
f
icien
t View
M
a
in
ten
an
ce a
t Data
W
ar
eh
o
u
ses
.
1997
Un
iv
ersity
of
Calif
o
rnia at Santa
Barb
ara, Dep
art
m
14
Peter
Van
Ro
y
A Pr
o
lo
g
Co
m
p
iler
f
o
r
th
e PL
M
.
1984
Un
iv
ersity
of
Calif
o
rnia at
Berk
eley
15
Tatu Yln
en
Sh
ad
o
w Pagin
g
I
s Feasib
le.
1994
Helsin
k
i Univ
ersity
of
Techn
o
lo
g
y
,
Dep
a
rt
m
en
t of
C
:
:
:
:
:
:
:
:
:
:
Table
5.
Phdth
esi
s
id
ed
ito
r
Title
Year
Mon
th
Sch
o
o
l
1
Jo
an
n
J.
Ordille
Descripti
v
e Na
m
e
Services f
o
r
La
rge
Internets
.
1993
Un
iv
.
o
f
W
isco
n
sin
-
Madis
o
n
2
Francis
co
Rev
erbe
ll
Persisten
ce in Dist
ribu
ted
Object
Sy
ste
m
s: ORB
/O
D
B...
1996
Ap
ril
Un
iv
ersity
of
New
M
ex
ico
4
Diet
m
a
r
Se
ip
el
Deco
m
p
o
sitio
n
in Datab
ase and
Kn
o
wled
g
e
-
Bas
e
Sy
ste
m
s.
1989
Un
i W
u
rzbu
rg
:
:
:
:
:
:
:
:
:
:
:
:
Table
6.
Proce
edin
g
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2502
-
4752
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci,
Vo
l.
9
,
No.
2
,
Fe
bruary
2
01
8
:
4
60
–
4
7
3
464
id
ed
ito
r
title
b
o
o
k
title
Series
v
o
lu
m
e
p
u
b
lish
er
y
ear
isb
n
u
rl
13
30
Nav
een
Prakas
h
,
Co
lette
Ro
llan
d
,
Barb
ara
Pernici
Inf
o
r
m
atio
n
S
y
ste
m
Dev
elo
p
m
en
t
Proces
s, P
roceed
in
g
s
o
f
the IFI
P
W
G8
.1
W
o
rkin
g
Co
n
f
eren
ce
o
n
I
n
f
o
r
m
atio
n
Sy
ste
m
Develo
p
m
en
t
Proces
s, Co
m
o
,
I
ta
ly
,
1
-
3
Septe
m
b
er
,
1
9
9
3
Inf
o
r
m
ati
o
n
Sy
ste
m
Dev
elo
p
m
en
t
Proces
s
IFI
P
Tr
an
sactio
n
s
A
-
30
No
rth
-
Ho
llan
d
1993
0
-
444
-
8
1
5
9
4
-
5
d
b
/co
n
f
/
if
i
p
8
-
1
/if
ip
8
-
1
-
1
9
9
3
.htm
l
13
41
To
m
J.
v
an
W
ee
rt,
Ro
b
ert
Mun
ro
Inf
o
r
m
ati
cs an
d
T
h
e
Dig
ital Societ
y
:
So
cial,
Ethica
l and
Co
g
n
itiv
e I
ss
u
es,
IFI
P
TC3
/W
G3
.1
&
3
.2
Op
en
Co
n
f
erence on
So
cial,
Ethica
l and
Co
g
n
itiv
e I
ss
u
es o
n
Inf
o
r
m
ati
cs an
d
I
C
T,
Ju
ly
22
-
2
6
,
2
0
0
2
,
Do
rt
m
u
n
d
,
Ge
r
m
a
n
y
SECII
I
IFI
P
Co
n
f
erence
Proceed
in
g
s
244
Klu
wer
2003
1
-
4020
-
7363
-
1
d
b
/co
n
f
/if
i
p
3
-
1
/if
ip
3
-
1
-
2
0
0
2
.htm
l
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
Table
7: ww
w
id
ed
ito
r
title
Bo
o
k
title
y
ear
u
rl
2
Mar
y
F
.
Fe
rnan
d
ez,
Jo
n
ath
an
Ro
b
ie
XML
Quer
y
D
ata
Mod
el
-
2001
h
ttp
://www.w3.o
rg
/TR/q
u
ery
-
d
ata
m
o
d
el
3
Arno
n
Ro
sen
th
al
The Fu
tu
re
o
f
Clas
sic Data
Ad
m
in
istratio
n
:
Ob
jects
SW
E
E
1998
h
ttp
://www.
m
i
tre.
o
rg/s
u
p
p
o
rt/swee/rosen
th
al.ht
m
l
:
:
:
:
:
:
:
:
:
:
:
:
3.2.
X
M
L
A
p
proa
ch
XML
pro
vid
es
a
sta
nd
ar
d
for
the
sem
antic
m
anag
em
ent
of
data
.
It
is
a
fo
rm
al
m
et
a
-
l
angua
ge
facil
it
y
fo
r
def
i
ning
a
m
ark
up
la
ngua
ge.
The
basic
unit
in
an
XM
L
file
is
entit
y
or
c
hunk
t
hat
c
on
ta
i
ns
c
onte
nt
a
nd
m
ark
up.
Ma
ny
exc
el
le
nt
m
od
el
-
m
app
in
g
sche
m
as
hav
e
bee
n
pro
posed
f
or
stori
ng
a
nd
retrievin
g
XM
L
data
into/from
relat
i
on
al
databa
se
[
17
]
.
The
m
ark
up
desc
ribe
s
a
con
te
nt.
M
or
e
gen
e
rall
y,
m
ar
kup
c
onsist
s
of
ta
gs
,
at
tribu
te
s,
c
omm
ents,
an
d
pro
cessi
ng
i
ns
tr
uc
ti
on
s
f
or
the
c
on
te
nt.
I
n
a
sta
rt
ta
g,
the
nam
e
and
a
ny
ad
di
ti
on
al
inf
or
m
at
ion
are
su
rro
unde
d
by
the
“<”
and
“>”
char
act
e
rs.
Fig
ur
e
3
sh
ows
the
al
go
rithm
ho
w
data
fr
om
relat
ion
al
database is c
onve
rt
into XML
f
orm
at
.
Inpu
t
: T
ab
les n
a
m
es,
re
co
rds
O
utp
ut
: Data
set
(
A)
Steps
1.
Read
nu
m
b
e
r
o
f
ta
b
les
1
.1
Create
XM
L
tag
b
y
r
ep
resenti
n
g
table title,
x
1
.2
Read
r
ecord
s/tu
p
les, y
1
.3
Create ind
en
ts/attri
b
u
tes tag
by
repres
en
tin
g
attr
ib
u
te nam
e
1
.4
Ass
ig
n
eac
h
r
ecord
to each att
ribu
te ta
g
,
i
i
1
.5
Clo
se XML
tag o
f
x
1
.6
Rep
eat step
1.1
un
til end
o
f
r
ecord
s
2.
Rep
eat step
1 u
n
til end
of
tables
3.
Ass
ig
n
A
=
<x
>
<y
i
>i
i
</y
i
>
<y
i+1
>i
i
</
y
i+1
>
:
<y
n
+1
>i
i
</y
n
+1
>
</x
>
4.
Disp
lay
data s
et A
Figure
3
.
Algor
it
h
m
(
Re
la
ti
onal
D
at
abase
to
XML
F
or
m
at
)
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci
IS
S
N:
25
02
-
4752
Eff
ic
ie
ncy o
f F
lat Fi
le
Data
base A
ppr
oa
c
h
i
n Da
t
a Stor
age
and D
ata Ext
r
action
…
(
Mo
hd K
am
ir
Yu
s
of
)
465
Af
te
r
al
gorith
m
in
Fig
ur
e
3
is
co
nv
e
rt
in
pro
gr
am
m
ing
c
od
e
,
the
n
e
xec
ution
is
occur,
syst
e
m
will
pr
od
uce
XML
file
as
rep
rese
nted
in
Fi
gure
4.
Sim
il
ar
ly
,
an
end
ta
g
consi
sts
of
the
ta
g
nam
e
su
rroun
ded
by
the
“<
/”
and
“>”.
XML
is
case
sensitiv
e
so
sta
rt
an
d
en
d
ta
g
nam
es
m
us
t
m
at
ch
exactl
y.
Figu
r
e
3
shows
how
th
e
publica
ti
on d
at
a is re
pr
ese
nte
d
in
X
ML
form
at
.
<rec
o
rd>
<article>
<id
>2
7
4
2
2
2
</id
>
<au
th
o
r>N.
Pr
ati</
au
th
o
r>
<title>A
Partial
M
o
d
el of
NP
with
E
.
</title>
<p
ag
es>1
2
4
5
-
1
2
5
3
</p
ag
es>
<y
e
a
r>19
9
4
</y
ear
>
<v
o
lu
m
e>5
9
</v
o
lu
m
e>
<jo
u
rnal>J
.
Sy
m
b
.
L
o
g
.</jou
rnal>
<u
rl>db
/jo
u
rnals
/j
sy
m
l/js
y
m
l5
9
.ht
m
l
#
Prati94
</u
rl>
</article>
:
:
<b
o
o
k
>
<id
>2
1
1
</id
>
<isb
n
>3
-
540
-
6
0
0
5
8
-
2
</isb
n
>
<au
th
o
r>M
arco Ca
d
o
li</au
th
o
r>
<title>T
ractable R
easo
n
in
g
in
Artif
icial I
n
tellig
en
ce</tit
le>
<series>L
ectu
re
N
o
tes in
Co
m
p
u
ter
Scien
ce</series
>
<v
o
lu
m
e>9
4
1
</v
o
lu
m
e>
<p
u
b
lish
er>Sprin
g
er</pu
b
lish
er>
<y
ea
r>19
9
5
</y
ear
>
<u
rl>.
..
</u
rl>
</b
o
o
k
>
:
:
<in
p
roceed
in
g
>
<id
>3
3
8
3
9
6
</id
>
<au
th
o
r>Reg
in
e L
aleau, A
m
el
Ma
m
m
a
r<
/au
th
o
r>
<title>A
Gen
eric
P
roces
s to
Ref
in
e a
B Sp
ecif
icatio
n
into
</title>
<p
ag
es>2
2
-
4
1
</p
a
g
es>
<y
ea
r>20
0
0
</y
ear
>
<b
o
o
k
title>ZB</b
o
o
k
title>
<u
rl>db
/co
n
f
/zu
m
/
zb
2
0
0
0
.ht
m
l#
L
ale
au
M00
</u
rl>
</in
p
roceed
in
g
>
:
:
<
m
sth
esis
>
<id
>1
1
</id
>
<au
th
o
r>Ku
rt
P.
Brown
</au
th
o
r>
<title>PR
PL:
A
D
atab
ase W
o
rklo
ad
Specif
icatio
n
L
an
g
u
ag
e,
v
1
.3.</title>
<y
ea
r>19
9
2
</y
ear
>
<sch
o
o
l>Un
iv
.
o
f
W
isco
n
sin
-
Madis
o
n
</sch
o
o
l>
</
m
sth
esis
>
:
:
<p
h
d
th
esis
>
<id
>1
</id
>
<ed
ito
r>Jo
an
n
J.
O
rdille</ed
ito
r>
<title>Desc
riptiv
e
Na
m
e
Ser
v
ices
f
o
r
Lar
g
e
I
n
ternets
.
</
title>
<y
ea
r>19
9
3
</y
ear
>
<
m
o
n
th
>..
.</
m
o
n
th
>
<sch
o
o
l>Un
iv
.
o
f
W
isco
n
sin
-
Madis
o
n
</sch
o
o
l>
</p
h
d
th
esis
>
:
:
<p
roceed
in
g
>
<id
>1
3
2
5
</id
>
<ed
ito
r>E
len
Balk
a,
Rich
ard S
m
ith
</
ed
ito
r>
<title>
W
o
m
an
,
W
o
rk an
d
Co
m
p
u
teri
za
tio
n
: Ch
arting
a
Co
u
rse to th
e Futu
re,
IFI
P
T
C9
/
W
G9
.1
Sev
en
th
Internatio
n
al Co
n
fer
en
ce on
W
o
m
an
,
W
o
rk an
d
Co
m
p
u
terization
,
Ju
n
e 8
-
1
1
,
2
0
0
0
,
Van
co
u
v
er,
B
ritish
Co
lu
m
b
i
a,
Can
ad
a</title>
<b
o
o
k
title>W
o
m
a
n
,
W
o
rk an
d
Co
m
p
u
terization
</b
o
o
k
ti
tle>
<series>I
FIP
Co
n
f
erence P
roceed
in
g
s</s
eries>
<v
o
lu
m
e>1
7
2
</v
o
lu
m
e>
<p
u
b
lish
er>Kluwer</pu
b
lish
er>
<y
ea
r>20
0
0
</y
ear
>
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2502
-
4752
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci,
Vo
l.
9
,
No.
2
,
Fe
bruary
2
01
8
:
4
60
–
4
7
3
466
<isb
n
>0
-
7
9
2
3
-
7
8
6
4
-
4
</isb
n
>
<u
rl>db
/co
n
f
/i
f
ip
9
-
1
/if
ip
9
-
1
-
2
0
0
0
.ht
m
l
</u
rl>
</p
roceed
in
g
>
:
:
<www>
<id
>1
</id
>
<ed
ito
r>.
..
</ed
ito
r
>
<title>Jav
a
L
an
g
u
ag
e Ho
m
e
Page</ti
tle>
<b
o
o
k
title>..
.</bo
o
k
title>
<y
ea
r>.
.
.</
y
ear
>
<u
rl>http
://jav
a.su
n
.co
m
/</u
rl>
</www>
</record>
:
:
:
Figure
4.
Tree
Re
pr
ese
ntati
on
of
P
ub
li
cat
io
n
XML
3.3.
JS
O
N Ap
proa
ch
In
this
ap
proac
h,
data
is
repre
sented
in
ar
ray
fo
rm
at
.
JSO
N
is
bu
il
t
on
two
struct
ur
es
.
T
he
first
is
a
colle
ct
ion
of
nam
e/
value
of
pairs.
I
n
va
riou
s
la
ngua
ge,
this
is
reali
z
ed
as
an
obje
ct
,
reco
r
d,
str
uctu
re,
dicti
on
a
ry,
has
h
ta
ble,
k
ey
e
d
l
ist
, o
r
ass
ociat
e arr
ay
.
Th
e
sec
ond
is a
n orde
r
ed
li
st o
f value
s.
I
n
m
os
t l
anguag
e
,
this
is
reali
zed
as
an
ar
ray,
l
ist
or
se
qu
e
nc
e.
Each
obj
ect
beg
i
ns
with
“
{“
and
e
nds
w
it
h
“}”.
A
rr
ay
is
an
order
e
d
colle
ct
ion
of v
al
ue
s.
An
ar
ray beg
i
n wit
h
“[” an
d
e
nd
s
with “]”
. Meanwhil
e, a v
al
ue
can
be
a string
in
double
qu
otes,
or
a
nu
m
ber
,
or
t
ru
e
or
false
,
or
a
n
obj
ect
or
an
ar
ray.
Fi
gure
5
represe
nts
al
gorithm
how
t
o
conve
rt
relat
io
nal
datab
ase
to
JSON
ap
proa
ch.
In
t
his
al
gorithm
,
two
input
are
re
quire
d
wh
ic
h
is
ta
ble
nam
e
and
data/
tu
ple
in
each
ta
ble.
Var
ia
ble
of
x
is
assig
n
to
ta
bl
e
nam
e
and
vari
able
y
represe
nted
t
o
tu
ple
f
or
eac
h
ta
ble. D
at
a
set
B are
represe
nted
as
a JSO
N
f
il
e after e
xecu
t
e this al
gorit
hm
.
Inpu
t
: Tables
nam
es
, r
ec
ords
Out
p
ut
: Data set
B
St
eps
1.
Re
ad
nu
m
ber
of table
s,
x
1.1
Assign ta
ble na
m
e to v
a
riabl
e x
1.2
Re
ad data
r
ec
ord/tu
ple
1.3
Assign
data
re
cord/t
uple
to
y
1.4
Com
bin
e v
aria
ble of
x
a
nd y
as b
el
ow
:
-
{x:[y
i
, y
i
+1
, y
n
+1
]}
1.5
Re
peat ste
p 1
.2 to ste
p 1.4
un
t
il
en
d o
f
rec
ord
s
2.
Re
pe
at
step
1 u
ntil
en
d o
f
ta
bles
3.
Assign B =
{x: [
y
i
, y
i+1
, …
, y
n+1
]}
4.
Disp
la
y
data se
t (B)
Figure
5.
Al
gorithm
(
Re
la
ti
on
al
D
at
abase
to
JSON
F
orm
at
)
JSON
file
is
produce
d
as
a
Figure
6
afte
r
ex
ecute
al
gorith
m
in
Figu
re
4.
JSON
file
is
si
m
ple
wh
ic
h
each
data/
recor
d
se
par
at
e
by li
ne.
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci
IS
S
N:
25
02
-
4752
Eff
ic
ie
ncy o
f F
lat Fi
le
Data
base A
ppr
oa
c
h
i
n Da
t
a Stor
age
and D
ata Ext
r
action
…
(
Mo
hd K
am
ir
Yu
s
of
)
467
{"arti
cl
e":
[{"id":"
274222","a
ut
hor":"N. Prati
","ti
tl
e":
"A
Par
ti
al
Mod
el
of
NP
with
E.","pages":
"
1245
-
12
53
"
,"ye
ar":"1
994",
"
volum
e"
:"
59
","
j
our
nal":
"J.Sym
b.
Lo
g.
","
ur
l"
:"
db
\
/jo
ur
nals
\
/jsym
l
\
/jsy
m
l59
.h
t
m
l#Pr
at
i94
"}]
,
"boo
k":[{"id":
"211","is
bn
":
"
3
-
540
-
60
058
-
2"
,"autho
r":"M
arco C
ad
oli","t
itle"
:"
Tracta
ble
Re
ason
i
ng in Artific
ia
l I
ntell
igence","
series
":
"Lect
ur
e
No
t
es
in C
om
pu
te
r
Scie
nce","
vo
l
um
e":
"94
1",
"
pu
blisher":"S
pri
nger"
,"ye
ar":"1
995",
"
ur
l"
:"
..."
}]
,
"i
npro
cee
ding"
:[{"i
d":"3
3839
6",
"a
uthor":"R
egine L
al
eau,
Am
el
Ma
m
m
ar
","ti
tl
e":
"A
Gen
eric
Pr
oc
ess t
o
Re
fi
ne
a B
S
pecific
at
ion
int
o",
"
pa
ges":"2
2
-
41
"
,"ye
ar":"2
000"
,"bo
ok
ti
tl
e":
"ZB
","ur
l"
:"
db
\
/c
onf
\
/z
um
\
/zb
20
00.h
tm
l#Lal
eauM0
0"}],
"
m
sthesis"
:[{"i
d":"1
1",
"a
utho
r":"Kurt P
. Bro
wn
"
,"ti
tl
e":
"P
RPL:
A
Datab
ase
Wo
r
klo
a
d
Sp
eci
ficat
io
n
L
angua
ge,
v1.
3.","y
ear":"
1992
","sch
oo
l"
:"
U
ni
v.
of
W
isc
ons
in
-
Ma
dison"}]
"phdthe
sis":
[{"id":"
1",
"e
ditor"
:"
Jo
an
n
J
. Or
di
ll
e","t
i
tl
e":
"D
escripti
ve Nam
e Ser
vices
for L
arg
e
In
te
r
nets."
,"ye
ar":"1
993",
"m
on
t
h":"...
"
,"schoo
l"
:"
U
niv.
of
W
i
sco
ns
in
-
Ma
dison"}]
,
"procee
ding":[
{"i
d":"132
5",
"
editor":"El
en
Ba
lka, Ric
ha
rd Sm
i
th"
,"ti
tl
e":"
W
om
an,
Wor
k
a
nd
Com
pu
te
rizat
ion
: C
har
ti
ng
a
Course t
o
the
F
uture,
IFIP TC
9
\
/
W
G
9.1 Sev
enth I
ntern
at
i
onal
Confere
nce
on
Wo
m
an,
Wor
k an
d
C
om
pu
te
rizat
ion
, Ju
ne 8
-
11, 200
0,
Va
nc
ouve
r,
B
riti
sh
Colum
bia, Can
ada","
bookti
tl
e
":
"Wo
m
an,
Wo
r
k
a
nd Com
pu
te
rizat
ion
"
,"s
eries":
"IFIP
Confere
nce
Proceedi
ngs","
volum
e"
:"
17
2"
,"
publishe
r":"Kl
uw
e
r",
"y
ear":"
2000","isb
n":"
0
-
7923
-
7864
-
4",
"
ur
l"
:"
db
\
/c
on
f
\
/i
fip9
-
1
\
/i
fip9
-
1
-
2000
.h
tm
l"
}
]
,
"www":[{"i
d":
"1","e
ditor
":
"..
.","ti
tl
e":
"Java Langu
a
ge H
om
e
Page","
bookti
tl
e":
"...
","y
ear"
:
"...
","
ur
l"
:"
http
:
\
/
\
/java.
s
un.c
om
\
/"
}
:
:
:
Figure
6.
P
ub
li
cat
ion
Data
in
JSON
F
orm
at
3.4.
Flat Fi
le
A
p
pr
oa
c
h
In
this
a
ppr
oac
h,
data
is
re
pr
e
sented
in
flat
fi
le
(text
form
at
)
.
Flat
file
a
re
t
ext
file
s
st
or
e
d
in
c
om
pu
te
r
sci
ence.
Data
in
flat
file
is
si
m
ple
and
can
ported
t
o
a
ny
pro
gr
am
.
The
basic
cha
racter
ist
ic
s
of
a
flat
file
are
that
data
are
store
d
as
plain
t
ex
t,
eve
n
the
num
ber
are
plain
te
xt,
an
d
tha
t
each
li
ne
of
t
he
file
co
ntain
s
on
e
record
or
case
in
the
data
set
.
Each
li
ne
a
flat
file
,
sever
al
con
ta
in
t
he
valu
es
for
the
dif
fe
ren
t
va
riables
in
th
e
data
set
.
Fiel
ds
within
a
recor
d
are
sepa
rate
d
by
a
sp
eci
al
char
act
e
r,
or
de
lim
it
er.
Each
li
ne
after
the
he
ader
consi
sts
of
tw
o
fiel
ds
se
par
a
te
d
by
a
col
on
(the
c
ha
racter
“:
”
is
the
delim
it
er)
.
Alte
r
na
ti
vely
,
we
ca
n
us
e
d
“wh
it
e
sp
ace”
(o
ne
or
m
or
e
sp
ace
ta
bs
)
a
s
the
delim
it
e
r.
Fig
ur
e
7
show
the
al
gorit
hm
ho
w
data
f
r
om
relat
ion
al
database is c
onve
rt
ed
int
o flat
f
il
e (text
form
at
).
Inpu
t
: Tables
nam
es
, r
ec
ords
Out
p
ut
: Data set
C
St
eps
1.
Re
ad nu
m
ber
of table
s,
x
2.
Assign ta
ble na
m
e to v
a
riabl
e x
2.1
Re
ad
rec
ords/t
up
le
s
in
t
he
ta
bl
e
2.2
Assign
rec
ord/tup
le
t
o
y
2.3
Assign
x
a
nd y
to
z
Z = {
x,
y
i
, y
i+1
,
y
n+1
}
2.4
Re
peat ste
p 2
.1 to ste
p 2.3
un
t
il
en
d o
f
rec
ord
in x
3.
Assign
var
ia
bl
e m
M = {
x,
y
i
, y
i+
1
, y
n+1
, … x,
y
i
,
y
i+1
, y
n+1
, …,
x, y
i
, y
i+1
, y
n+1
}
4.
Re
peat ste
p 1
unti
l 3 un
ti
l e
n
d
of table
s
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2502
-
4752
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci,
Vo
l.
9
,
No.
2
,
Fe
bruary
2
01
8
:
4
60
–
4
7
3
468
5.
Assign M t
o
C
6.
Disp
la
y
data se
t (C)
Figure
7
.
Al
gorithm
(
Re
la
ti
on
al
D
at
abase
to
XML
F
or
m
at
)
Figure
8
s
how
s
li
st
of
data
r
epr
ese
nted
in
f
la
t
file
(text
f
orm
at
).
These
da
ta
are
extr
act
ed
f
r
om
or
igin
al
data
so
urces
which
is st
or
e
in
relat
ion
al
database
appr
oach.
Ar
ti
cl
e,29
5331
,
Em
il
ia
Me
nd
es,
Nile
Mosley
,
Steve
Couns
el
l,
W
e
b
Me
tric
s
-
Esti
m
a
ti
ng
Desig
n
and
A
uthori
ng
Ef
f
or
t.
,
50
-
57,
2001,
8,
I
EEE
Mul
ti
Me
dia,
db
/
j
ou
rn
al
s/i
ee
e
m
m
/i
eee
m
m
8
.h
tm
l#Me
nd
es
MC
01
Ar
ti
cl
e,29
5332
,
A
ndreas
V
ogel
,
Bri
g
it
te
Kerher
ve,
G
re
gor
von
Boc
hm
ann
,
Ja
n
Gecsei,
Distrib
uted
Mult
i
m
edia
and
Q
OS
:
A
Surv
ey
.
,
10
-
19,
1995,
2,
IEE
E
Mult
iM
edia,
db
/
j
ou
rn
al
s/i
ee
e
m
m
/i
eee
m
m
2
.h
tm
l#Vo
gel
K
BG95
Ar
ti
cl
e,29
5333
,
Fo
r
ouza
n
G
ol
sh
ani,
F
ro
m
Mult
i
m
edia
T
oo
ls
to
Ar
ti
sti
c
Con
te
nt.
,
1,
2002,
9
,
IEEE
M
ulti
Me
dia,
db
/
j
ou
rn
al
s/i
eee
m
m
/i
eee
m
m
9.
ht
m
l#Go
l
sh
a
ni02c
Ar
ti
cl
e,29
5334
,
Arn
d
Stei
nm
et
z,
Me
dia
an
d
Dista
nce:
A
L
earn
i
ng
E
xperi
ence.,
8
-
10,
20
01,
8,
IEEE
M
ulti
Me
dia,
db
/
j
ou
rn
al
s/i
eee
m
m
/i
eee
m
m
8.
ht
m
l#Stei
nm
e
tz
01
Ar
ti
cl
e,29
5335
,
Ri
ccard
o
Le
on
a
r
di,
Piera
n
gelo
Mi
glio
rati
,
Sem
antic
In
dex
i
ng
of
Mul
tim
edi
a
Do
c
um
ents.,
44
-
51,
20
02,
9,
IEE
E
Mu
lt
iM
edia
,
db
/
j
ou
rn
al
s/i
ee
e
m
m
/i
eee
m
m
9
.h
tm
l#Leon
a
rdi
M02
Ar
ti
cl
e,29
5336
,
Step
han
e
Val
ente,
Jea
n
-
L
uc
Dugelay
,
Face
Track
i
ng
a
nd
Re
al
ist
ic
An
i
m
at
ion
s
for
Tel
eco
m
m
un
ic
ant
Cl
on
es.
,
34
-
43,
2000,
7,
IE
EE
Mult
iM
edia,
db
/
j
ou
rn
al
s/i
ee
e
m
m
/i
eee
m
m
7
.h
tm
l#ValenteD00
Ar
ti
cl
e,29
5337
, V
ip
ul K
as
hya
p,
Am
i
t P. S
he
th,
Buil
di
ng
S
uc
cessf
ul H
um
an
-
Ce
ntere
d
Sys
tem
s.
,
102
-
10
3,
2001,
8
,
I
E
EE M
ulti
Me
dia,
db
/
j
our
nals/i
eee
m
m
/i
e
ee
m
m
8.
htm
l#Kash
ya
pS0
1
Ar
ti
cl
e,2
9533
8,
Sorel
Re
ism
an,
Ta
king
St
ock
of
the
W
eb.
,
4,
1997,
4,
IEEE
M
ulti
Me
dia,
db
/
j
ou
rn
al
s/i
ee
e
m
m
/i
eee
m
m
4
.h
tm
l#Rei
s
m
an
97
Ar
ti
cl
e,29
5339
,
Ro
nnie
T.
A
pteke
r,
J
am
es
A.
Fishe
r,
Val
entin
S
.
Kisim
ov,
Hanoc
h
N
ei
sh
los,
Vide
o
Acce
ptabili
ty
and
Fr
am
e
R
at
e.,
32
-
40,
199
5,
2,
IEEE
Mu
lt
iM
edia,
db
/
j
ou
rn
al
s/i
ee
e
m
m
/i
eee
m
m
2
.h
tm
l#Ap
te
ke
r
FKN95
Ar
ti
cl
e,29
5340
,
Jan
Gecsei,
Ad
a
ptati
on
i
n
Distrib
uted
M
ultim
edia
Syste
m
s.,
58
-
66,
1997,
4,
IEEE
M
ulti
Me
dia,
db
/
j
ou
rn
al
s/i
eee
m
m
/i
eee
m
m
4.
ht
m
l#Ge
csei
97
:
:
:
:
Figure.
8
:
Publ
ic
at
ion
Data
in
TXT
Form
at
4.
E
X
PERI
MEN
TAL RES
UL
TS
In
this
sect
io
n,
we
evaluate
t
he
pe
rfor
m
ance
of
the
acce
ss
ing
the
data
from
XML
and
JSON.
F
ou
r
diff
e
re
nt
qu
e
ri
es
are
us
e
d
in
exp
e
rim
ents.
The
syst
e
m
s
are
bu
il
d
us
in
g
a
per
s
onal
com
pu
te
r
eq
uippe
d
with
2.40G
Hz
I
ntel®
Core
™
i7
-
5500U
CP
U,
8.00
GB
RA
M
and
a
250
GB
so
li
d
-
sta
te
dr
i
ve.
T
he
op
erati
ng
syst
e
m
is
M
ic
ro
soft
W
i
ndow
s
10.
T
he
data
base
im
ple
m
en
ti
ng
the
XML
database
(a
ppr
oach
I)
us
i
ng
X
-
Pat
h
for qu
e
ryi
ng
pur
poses a
nd JS
ON d
at
a
base
(
appr
oach I
I)
.
We
us
e
be
nchm
ark
dataset
DBLP
[
18]
.
The
va
riat
ion
in
qu
e
ry
tim
e
w
it
h
the
siz
e
of
the
database
is
al
s
o
stud
ie
d.
For
e
ach
of
tw
o
dat
abase
a
ppr
oac
hes,
the
ti
m
e
t
o
qu
e
ry
a
nd
C
PU
us
a
ge
with
va
ryi
ng
com
pl
exity
sp
eci
fied
a
bove
is
m
easur
ed
with
data
bases
con
ta
ini
ng
10
00,
50
00,
10,
000
a
nd
50,
000
records
res
pec
ti
vely
.
Fo
r
qu
e
ry
retri
eval,
at
each
set
ti
ng
,
the
query
is
m
ade
fo
r
10
tim
es
to
cal
culat
e
the
aver
ag
e
tim
e
and
sta
ndar
d
dev
ia
ti
on
[10].
The
disc
us
sio
n
is
based
on
t
wo
e
xp
e
rim
ent
s
in
the
databa
ses
dev
el
op
m
ent
and
their
a
ppli
cat
ion
f
or
the sto
rag
e
of s
tructu
red data
, fr
om
the p
ers
pe
ct
ives of test
data, e
ff
ic
ie
nc
y and scal
abili
ty
, and
e
xtensi
bi
li
t
y.
4.1.
Te
st
D
ata
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci
IS
S
N:
25
02
-
4752
Eff
ic
ie
ncy o
f F
lat Fi
le
Data
base A
ppr
oa
c
h
i
n Da
t
a Stor
age
and D
ata Ext
r
action
…
(
Mo
hd K
am
ir
Yu
s
of
)
469
The
perform
ance
of
tw
o
data
base
a
ppro
ac
he
s
is
evaluated
by
us
in
g
ben
c
hm
ark
dataset
DBLP.
T
he
data
con
ta
i
n
50,
000
rec
ords
.
TABLE
8
s
ho
ws
the
queries
with
dif
fer
e
nt
com
plexity
and
TABLE
9
shows
th
e
qu
e
ries c
onstruc
te
d
in t
he SQ
L stat
e
m
ent.
Table
8: Queri
e
s w
it
h
Dif
fer
e
nt Com
plexity
[6]
Qu
e
ry
Qu
e
ry
descr
i
ption
I
List
o
ut all
t
he URLs
wh
ic
h b
egin wit
h
t
he
“
db
/
j
ou
rn
al
s”
pa
th
II
List
o
ut all
t
he t
it
le
s o
f
the
m
a
ste
r
thesis
w
hich
c
on
ta
in
s the
“D
at
a”
keyw
ord
III
List
the tit
le
s of inpr
ocee
ding
wh
e
re t
he
a
uthor
is “Re
gi
ne
L
al
eau,
Ma
m
m
a
r”
IV
Count the
num
ber o
f phd t
hes
is p
ub
li
sh
e
d
i
n ea
ch
ye
ar
Table
9.
Q
uer
i
es
Co
ns
tr
ucted
in S
QL
Com
m
ands
Qu
e
ry
Qu
e
ry
descr
i
ption
I
Sele
ct
*
from
u
rl where
text li
ke
‘%db/j
our
na
ls/
%’
II
Sele
ct
*
f
ro
m
ti
tl
e w
he
re text l
ike ‘
%
Data%’
III
Sele
ct
ti
tl
e fr
om
inp
r
ocee
ding
wh
e
re a
uthor
=’Regine
Lale
au,
Mam
m
ar’
IV
Sele
ct
co
unt(i
d), year
fro
m
p
hdthesis
gro
up
by
yea
r
4.2.
Data Ex
tr
act
i
on
(
X
M
L,
JS
ON an
d Fl
at
Fil
e (t
e
xt
f
or
mat))
I
n
this
sect
i
on,
data
from
relat
ion
al
database
are
e
xtract
an
d
c
onve
rt
int
o
three
dif
fer
e
nt
data
f
orm
at
.
The
data
siz
e
f
or
each
f
or
m
at
are
represe
nted
in
in
KB.
T
ABLE
10
unti
l
TABLE
13
s
how
the
data
si
ze
an
d
perform
ance
qu
ery
retrie
val
in
three
diff
e
re
nt
f
orm
at
wh
ic
h
are
XML
,
JS
ON
a
nd
Flat
Fi
le
(text
fo
rm
at
).
Data
are
sp
li
t
into
4:
-
10
00
rec
ords,
5000
rec
ords,
20,00
0
rec
ords
a
nd
50,
000
rec
ords.
T
he
n,
the
se
rec
ords
ar
e
conve
rt
into
diff
e
ren
t
data
f
orm
at
.
Ba
sed
on
T
ABLE
10
un
ti
l
TABL
E
13,
Flat
Fil
e
(text
f
or
m
at
)
fo
r
m
at
is
sm
a
ll
er
co
m
par
ed
to
XML
a
nd
JS
O
N.
T
ha
t
way,
tim
e
to
data
retrieval
al
so
sh
ows
fl
at
file
in
te
xt
form
a
t
faster c
om
par
e
d
to
X
ML
and
JSON.
Table
10.
Qu
e
r
y per
form
ance o
f
the t
hr
ee
ap
proac
hes on da
ta
base
with
dif
fer
e
nt size:
Qu
ery I
Ap
p
roach
Da
tab
ase I
m
p
le
m
e
n
tatio
n
Mean
±
SD (
m
s)
–
Qu
ery I
Size
(KB)
Ti
m
e
(
m
s)
Size
(KB)
Ti
m
e
(
m
s)
Size
(KB)
Ti
m
e
(
m
s)
Size
(KB)
Ti
m
e
(
m
s)
I
XML
339
1
4
.41
±
0
.28
1712
7
5
.30
±
0
.42
3405
1
3
6
.66
±
1
.44
1
6
3
0
0
6
8
5
.67
±
3
.12
II
JSON
249
1
0
.41
±
0
.35
1266
4
0
.35
±
0
.21
2515
8
0
.89
±
0
.43
1
1
8
4
4
3
9
7
.64
±
2
.23
III
TXT
185
8
.18
±
0
.14
948
3
8
.56
±
0
.77
1890
7
6
.35
±
0
.25
8868
3
7
0
.48
±
0
.26
Table
11.
Qu
e
r
y per
form
ance o
f
the t
hr
ee
ap
proac
hes on da
ta
base
with
dif
fer
e
nt size:
Qu
ery I
I
Ap
p
roach
Databas
e I
m
p
le
m
e
n
tatio
n
Mean
±
SD (
m
s)
–
Qu
ery II
Size
(KB)
Ti
m
e
(
m
s)
Size
(KB)
Ti
m
e
(
m
s)
Size
(KB)
Ti
m
e
(
m
s)
Size
(KB)
Ti
m
e
(
m
s)
I
XML
339
1
1
.36
±
0
.31
1712
4
0
.56
±
0
.25
3405
7
3
.51
±
0
.32
1
6
3
0
0
3
3
3
.28
±
3
.11
II
JSON
249
8
.41
±
0
.32
1266
3
5
.36
±
0
.25
2515
6
6
.48
±
0
.23
3
1
1
8
4
4
3
0
9
.75
±
1
.26
III
TXT
185
8
.24
±
0
.07
948
2
3
.54
±
0
.28
1890
4
5
.57
±
0
.20
8868
2
2
3
.94
±
0
.64
Table
12.
Qu
e
r
y per
form
ance o
f
the t
hr
ee
ap
proac
hes on da
ta
base
with
dif
fer
e
nt size:
Qu
ery I
II
Ap
p
roach
Databas
e
I
m
p
le
m
en
tatio
n
Mean
±
SD (
m
s)
–
Qu
ery III
Size
(KB)
Ti
m
e
(
m
s)
Size
(KB)
Ti
m
e
(
m
s)
Size
(KB)
Ti
m
e
(
m
s)
Size
(KB)
Ti
m
e
(
m
s)
I
XML
339
8
.56
±
0
.28
1712
4
1
.53
±
0
.26
3405
8
5
.52
±
1
.07
1
6
3
0
0
3
8
3
.48
±
1
.84
II
JSON
249
8
.14
±
0
.10
1266
3
5
.14
±
0
.53
2515
6
5
.42
±
0
.99
1
1
8
4
4
3
6
2
.82
±
0
.89
III
TXT
185
7
.92
948
2
5
.51
1890
5
0
.73
±
8868
2
6
8
.42
Evaluation Warning : The document was created with Spire.PDF for Python.