TELKOM
NIKA Indonesia
n
Journal of
Electrical En
gineering
Vol.12, No.1, Jan
uary 20
14
, pp. 741 ~
746
DOI: http://dx.doi.org/10.11591/telkomni
ka.v12i1.3350
741
Re
cei
v
ed Ma
y 27, 201
3; Revi
sed
Jul
y
1
0
, 2013; Acce
pted Augu
st 25, 2013
Semantic Representation of Complex Resource
Requests for Service-o
r
iented Architecture
Xing Wang*,
Chan
g
w
a
n
g Liu
Schoo
l of Softw
a
r
e, Na
n
y
an
g
Normal Un
iver
sit
y
163
8, W
o
lon
g
Roa
d
, Nan
y
a
n
g
, 4730
61, He
nan Prov
inc
e
, PR Chi
na, 00
8
6
-37
7
-63
5
1
3
4
6
6
*corres
pon
di
ng
author, e-mai
l
: hn
w
a
ng
xi
ng
@
126.com
A
b
st
r
a
ct
Many
open
dist
ributed systems across I
n
ternet such
as thos
e in grid c
o
m
p
uting and
e-Comm
erc
e
involv
e the r
e
q
uestin
g
, all
o
cat
i
on
and
mai
n
te
nanc
e of sort
s
of resourc
e
s. T
he d
i
scovery
of larg
e a
m
o
unt
of
resourc
e
s i
n
d
i
fferent sites is
an i
m
porta
nt is
sue fo
r th
e d
e
s
ign
of thes
e s
ystems. T
h
e
b
o
o
m
i
ng s
e
man
t
ic
W
eb techno
lo
gy provid
es a
suitabl
e infra
s
tructure
for the pu
blis
hin
g
,
requesti
ng a
nd match
m
aki
ng of
resourc
e
s. T
h
is paper
prese
n
t
s a gener
ic re
prese
n
tatio
n
for quantifi
ed res
ource re
qu
esti
ng w
i
th Seman
t
ic
W
eb. It allow
s
the repres
entat
ion of
co
mplex
resource
desc
r
iptio
n
s such a
s
contain
m
ent
hier
archi
e
s an
d
disjo
i
nt constra
i
nts betw
een th
em. A
mode
l-theor
etic se
ma
ntics for matc
h
m
ak
in
g w
i
th countab
le reso
ur
ces
is give
n for this repres
entati
on.
A constrai
nt-base
d
tech
niq
ue for the
match
m
aki
ng
check w
i
th such
repres
entati
on is
desi
gne
d.
Ke
y
w
ords
: qu
antifie
d reso
urce, semantic
W
eb, resource
match
m
aki
ng, Service-
orie
nte
d
Architecture
Copy
right
©
2014 In
stitu
t
e o
f
Ad
van
ced
En
g
i
n
eerin
g and
Scien
ce. All
rig
h
t
s reser
ve
d
.
1. Introduc
tion
A lot of di
stri
buted
appli
c
a
t
ions
acro
ss
Inte
rnet i
n
volve the
req
u
e
s
ting, all
o
cation a
n
d
maintena
nce
of many so
rts of and
la
rge
amount of
re
sou
r
ces i
n
di
f
f
erent
sites. I
n
e-Comm
erce,
for example,
a cu
stome
r
may issue a
requ
est to
a
sho
p
for a q
u
antity of goods. A travel a
gent
may boo
k a
numbe
r of
airline tickets from an
airli
n
e
agent
and
a
numbe
r of
ap
artment
s fro
m
a
hotel a
gent. I
n
the fiel
d of
grid
co
mputin
g, task
s may
requi
re
for
different type
s
o
f
com
putation
a
l
resou
r
ces of
ce
rtain
am
ounts,
su
ch
as
co
m
pute
r
s,
thei
r
me
morie
s
and
disk spa
c
e, and
band
width wi
th networks. Most of these Intern
et appli
c
ation
s
involve intera
ction
s
betwe
en
hetero
gen
eo
us informatio
n sou
r
ces a
n
d
agent
s in
open environm
ents, in whi
c
h the probl
e
m
o
f
interop
e
rability between th
e hetero
gen
e
ous
sou
r
ces i
s
a big issu
e.
Semantic Web [1-3] is a booming technology
to achi
eve
semantic-l
evel interoperability
based on X
M
L. It was
motivated to have info
rm
ation so
urce
s ma
chin
e-u
nderstan
dabl
e and
agent-sh
a
rabl
e by mean
s
of annot
ating
their conten
t with comm
on data m
o
d
e
l and
sha
r
e
d
ontology. Se
mantic
We
b i
s
e
s
pe
cially
suitable fo
r th
e
task of resou
r
ce
di
scovery
acro
ss Intern
et.
First, ontolo
g
y
technology
provide
s
a mean
s to
co
nce
p
tuali
z
e a
nd mana
ge
different so
rt
s o
f
resou
r
ces, a
nd to spe
c
ify reso
urce ad
vertis
em
ents
and re
que
sts. Second,
the employmen
t
of
publi
c
ly sta
n
dardi
ze
d
se
mantic W
eb
spe
c
ification
s
help
s
to a
c
hieve inte
rop
e
rability fo
r t
he
intera
ction be
tween resou
r
ce re
que
ste
r
s, providers an
d bro
k
e
r
s.
The main
concern of t
h
is pa
per i
s
t
he re
pre
s
entatio
n for quantified-reso
urce
matchm
aki
n
g
between
re
sou
r
ce advertisement
s an
d resource
reque
sts.
Qu
antified re
so
urce
requ
estin
g
is
mostly investi
gated in the field of
grid
co
mputing [2, 3, 4, 5], where
a
s few
wo
rks is
kno
w
n
ab
out
quantified
-
resource
match
m
akin
g in
th
e
co
ntext of e
-
Comm
erce
al
though
it
sho
u
ld
have mo
re ex
tensive ap
pli
c
ation
s
in the
area
and m
a
nifest mo
re complex form
s. Our work th
us
mainly focu
s on two exten
s
ion
s
: one is
to allow
to advertise summ
arized re
so
urce de
scriptio
ns,
anothe
r is to allow mo
re e
x
pressive qu
erie
s for qu
an
tified reso
urces.
2. Resou
r
ce
s, Resou
r
ce
Adv
e
rtiseme
n
ts, and Res
ource
Requ
e
s
ts
2.1. Resou
r
c
e
s
The term
“resource
” is exte
nsively and f
r
eely
use
d
in i
n
formatio
n field without a
widely-
accepted accurate definit
ion. We view reso
urces as
anything that
is of
certain degrees of
utility
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 2302-4046
TELKOM
NIKA
Vol. 12, No
. 1, Janua
ry 2014: 741 – 7
4
6
742
and ca
pa
city to some com
peting pr
ocesse
s. In the fields of co
mp
uter scien
c
e
s
espe
cially grid
comp
uting, typical resou
r
ces in
clud
e co
mputers,
memories
, CPU
time, dis
k
s
,
printers
, network
band
width, o
r
even
programs
and
dat
a source
s. I
n
e-Com
m
erce, typical re
sou
r
ces in
clu
de
variou
s
so
rts of goo
ds, t
r
affics, e
nergy su
pplie
s, hu
man
re
sou
r
ces, a
nd et
c.
Re
sou
r
ces a
s
a
whol
e ca
n be
classified
alo
ng differe
nt dimensi
o
n
s
a
c
cordin
g to feature
s
such as
if a reso
urce i
s
con
s
um
ptive, divisible, and
sharable.
For
th
e cont
ext
of
this pa
per, we are only
con
c
ern
ed with cla
s
sification ba
se
d
on
the
ways th
ey a
r
e represent
ed, adverti
se
d and
que
ri
ed. Since
a
resou
r
ce can be
either an
individual, o
r
a colle
ction
of individual
s, or a
n
am
ount of sub
s
tance
s
o
r
en
ergy, mainly
we
disting
u
ish reso
urce
s bet
wee
n
re
so
urce el
ement
s whi
c
h a
r
e i
ndividual
re
source item
s,
and
resou
r
ce po
rt
ions whi
c
h
m
a
y co
ntain ot
her
re
sou
r
ce
s. A resource
portio
n
i
s
eit
her
co
untabl
e
in
that it con
s
ist
s
of a finite
set of resou
r
ce individual
s,
or un
co
unta
b
le such as
water
and fu
el in
that they are con
s
id
ere
d
to be contin
uou
sly divisibl
e.
Re
sou
r
ce po
rtions are mai
n
con
c
e
r
n of this
pape
r.
2.2. Resou
r
c
e
Adv
e
rtisements
To allo
w re
source di
scov
ery acro
ss
Web,
we a
s
sume an o
p
e
n
archite
c
ture in whi
c
h
resou
r
ce o
w
ners a
d
vertise their re
so
urces in
a publi
c
resource adverti
se
ment
ba
se, and
r
e
so
ur
ce
re
qu
e
s
ter
s
iss
ue r
e
s
o
ur
ce
re
qu
e
s
ts
to
the
reso
urce
adve
r
tisem
ent ba
se for avail
abil
i
ty.
It is impracti
cal to registe
r
all the resou
r
ce
items in th
e resou
r
ce a
d
vertise
m
ent
base wh
en the
quantitie
s of
re
sou
r
ces
are
so m
a
n
y
. Rather it
is rea
s
on
a
b
le to allo
w a summa
ri
zed
advertiseme
n
t
for ea
ch typ
e
of re
so
urce
s. For
in
stan
ce, a
re
sou
r
ce adve
r
tisem
ent ba
se mi
g
h
t
advertise that
there a
r
e
50
comp
uters i
n
a LAN
rath
er than li
st e
a
ch
of them.
Furthe
rmo
r
e
we
claim that it is useful to al
low multi-vie
w
descr
iption
s and hie
r
a
r
chical de
script
ions in re
so
u
r
ce
advertiseme
n
t
s.
For a
n
exam
ple of multi
-
view d
e
scriptio
n
, it
might be
advertised th
at a lab
o
rato
ry has
5
serve
r
s a
nd,
at the
same
time, 20
co
mp
uters in
sta
lle
d with
Unix. T
hey are multi
-
view d
e
scri
ption
in that they describ
e the same resource re
po
si
tory with different capa
cit
i
es. Hie
r
arch
ical
descri
p
tion
s involve the rep
r
e
s
entatio
n of in
clusiv
e relation
s betwe
en different re
so
urce
repo
sito
rie
s
and re
sou
r
ce
s cap
a
citie
s
.
An
exam
ple
of hie
r
a
r
chi
c
al
de
scriptio
ns:
“Computi
n
g
Cente
r
ha
s 2
labs, one la
b has 4
0
PC-486
s, the other ha
s 30 P
C
-5
86
s”. It is our obje
c
tive to
extend the existing ap
pro
a
c
h with
su
ch
multi-view
d
e
s
cription
s an
d hiera
r
chi
c
al
descriptio
n
s.
2.3. Resou
r
c
e
Requ
ests
While
co
mpl
e
x re
sou
r
ces are
commo
n in e
-
Com
m
erce, the i
s
sue
ha
s n
o
t bee
n
addresse
d in
existing
g
r
i
d
-o
riente
d
re
sou
r
ce
req
u
e
st la
ngu
age
s [2,
3]. Althoug
h
compl
e
x
resou
r
ces
co
uld be
rep
r
e
s
ente
d
a
s
co
mpositio
n of
atomic o
n
e
s
with logi
c
co
nne
ctives, e.
g.,
usin
g logical conj
un
ction to expre
ss two portion of
reso
urce a
s
a whole such as “9 PCs an
d 2
workstation
s
”. Such appro
a
ch may cau
s
e confu
s
ion
when two p
o
rtion
s
of re
sou
r
ces a
r
e
not
disjoi
nt. For instan
ce, “3
profe
s
sors a
nd 2 fe
male
teach
e
rs” ma
y denote a set of 3, 4, or 5
teach
e
rs d
e
p
endin
g
on th
e
numb
e
r of fe
male p
r
ofe
s
sors in the
set.
Sometime
s such
de
scripti
on
need
s to be
clarifie
d with
clea
re
r altern
atives su
ch
a
s
“3 p
r
ofe
s
so
rs plu
s
, in ad
dition, 2 female
teach
e
rs” o
r
“3 professo
rs includi
ng 2
wome
n ”
whi
c
h imply re
sp
ectively the use of exclu
s
i
v
e-
joining and inclusion bet
ween reso
urce portions. Below is a more
complex example illustrating
the usa
ge of reso
urce excl
usive-j
o
inin
g and in
clu
s
ion:
CS dep
artme
n
t of Beijing
Institute of T
e
ch
nol
o
g
y(BIT) might
sele
ct a group
of seni
or
schola
r
s a
s
the do
ctorial t
hesi
s
-defen
se committ
ee
membe
r
s fo
r A PhD stude
nt who
s
e the
s
is i
s
about the co
mbination of grid an
d age
nt. The
requi
reme
nts for the com
m
ittee members mi
ght
be sp
ecifie
d based on u
n
i
v
ersity-p
olicy
as follo
ws:
(a)
There must b
e
7 schola
r
s
who a
r
e all compute
r
-sci
e
n
ce p
r
ofe
s
sors in Beijing.
(b)
At least 4 of them mu
st be out of BIT.
(c)
At least 3 are
experts in g
r
i
d
(d)
At least 3 are
experts in ag
ent
(e)
In addition, a
se
creta
r
y for t
he defe
n
se s
hould
be sele
cted
who m
u
st be d
epa
rtm
ent teacher
with PhD de
g
r
ee in comp
uter scien
c
e.
This
huma
n
resou
r
ce req
u
irem
ent sho
w
s
ho
w a
co
mplex re
so
urce
req
u
e
s
t could b
e
comp
osed of simple
r one
s with
joi
n
ing, exclu
s
ive
jo
i
n
ing, an
d in
clu
s
ion. B
o
th (a) and
(e)
sh
ou
ld
be inclu
ded
but they sho
u
ld be disj
oin
t. Groups
co
rrespon
ding t
o
(b), (c), an
d (d) may no
t b
e
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Sem
antic Re
pre
s
entatio
n of Com
p
lex Reso
urce
Req
uest
s
for Service-o
r
ie
nted
… (Xing Wan
g
)
743
disjoi
nt, and
all these 3 g
r
oup
s are in
cl
uded i
n
gr
ou
p co
rrespon
d
i
ng to (a). La
ter we will
sh
ow
how
su
ch re
q
uest
s
wo
uld b
e
formulate
d
in our represe
n
tation.
3. Complex
Reso
urce
Re
presen
ta
tion
Based o
n
Semantic
We
b
3.1. Resou
r
c
e
Ontology
In our frame
w
ork, differe
nt form
s of reso
u
r
ce
s,
incl
udin
g
res
our
ce r
e
posit
o
r
ie
s,
re
sou
r
c
e
p
o
r
tio
n
s
,
an
d r
e
s
o
urc
e
items
,
ar
e
u
n
i
for
m
ly mo
de
le
d
as resource
obje
c
ts. T
he
rea
s
on
is to g
a
in
rep
r
e
s
entatio
nal u
n
iformity
and
si
mplicit
y for rea
s
oni
ng
with the
h
i
era
r
chical rel
a
tion. First,
we
as
sume
a
r
oot
cl
as
s
R
e
so
ur
ceO
b
je
ct
f
o
r
all
th
e resource
obje
c
ts, a
n
d
its
2
sub
c
l
a
sse
s
Re
sou
r
ceEle
m
ent an
d Re
sou
r
cePortio
n. In cla
s
s
Reso
urcePo
rtion 2
role
s i
n
clud
e an
d di
sjoint
are d
e
fined
whi
c
h d
enot
e re
spe
c
tivel
y
the contai
nment an
d
disjoi
nt relati
on bet
ween
two
resou
r
ces. In descri
p
tion
-lo
g
ic style the
s
e are written
as a
s
:
R
e
s
o
ur
ce
Eleme
n
t
Re
so
urc
e
Obj
e
ct
Re
sou
r
cePort
ion
R
e
sour
ceObjec
t
(
includ
e Resou
r
ce
Obje
ct)
(
disjoint
Re
sou
r
c
e
Obj
e
ct
)
For o
u
r p
u
rp
ose
of qua
ntified re
sou
r
ce
ma
tchm
aki
n
g, cla
ss
QtPortion a
r
e
especi
a
lly
defined
whi
c
h inhe
rits
Re
sou
r
cePort
ion and
ad
ditionally def
ines
2 rol
e
s qu
antity and
element
Cla
s
s whi
c
h respe
c
tively denot
e ho
w many
and
what typ
e
of re
so
urce
s elem
ents a
r
e
decl
a
re
d.
QtPortion
Re
sou
r
cePort
ion
(=
1 quantity Number)
(
1 elem
e
n
tCla
ss
Cla
s
s)
Here the valu
e of attribute element
Cla
s
s is in
itself a descri
p
tion
-lo
g
ic cl
ass con
s
tru
c
tor
whi
c
h mu
st be a sub
c
la
ss of Re
so
u
r
ceEl
emen
s. QtPortion is
divided into two sub
c
la
sses
DQtPortio
n
for discrete p
o
rtion
s
and
CQtPortio
n
for co
ntinuo
u
s
portio
n
s. Reso
urceElem
ent is
also divide
d into two sub
c
la
sses DRe
s
ou
rc
eEleme
nt and CResourceElem
en
t. In
addition
to
these
re
so
urce-related
co
n
c
ept
s,
the
ont
ology al
so i
n
clud
es a
sse
rt
ions re
gardin
g
the
pro
perti
es
of these
con
c
ept
s. Fo
r ex
ample,
“Fo
r
QtPortion r1 and QtPortio
n
r2,
if the el
ementCl
ass
of r1
element
Cla
s
s of r2
are di
sjoint, then d
i
sjoint(r1,
r2
) is true” Thi
s
might be re
pre
s
ente
d
as
a
RuleM
L
rul
e
in the logic lay
e
r of sem
anti
c
We
b infra
s
t
r
uctu
re.xx
3.2. Repre
s
e
n
ta
tion of Q
u
antifie
d
Re
source
Adv
e
rtisemen
t
w
i
th RDF
In our frame
w
ork, a
re
so
urce a
d
vertisem
ent b
a
se
decl
a
re
s a
set of re
sou
r
ce obje
c
t
instan
ce
s lin
ked
with
role
inclu
de. A reso
urce
adv
ertise
ment b
a
se i
s
rep
r
e
s
ented a
s
a set of
RDF
stateme
n
ts whi
c
h a
r
e
subje
c
t-p
r
e
d
i
c
ate-obje
c
t tri
p
les.
(1)
University BIT has 1
00 cla
s
sroo
ms ;
(2)
70 of (1) a
r
e
multi-medi
a e
nable
d
.
(3)
40 of (1) a
r
e l
a
rge o
n
e
s
tha
t
can hold 2
0
0
stude
nts;
(4)
50 of (1) a
r
e
middle on
es t
hat can h
o
ld
100 stu
dent
s;
(5)
10 of (1) a
r
e
small that ho
ld 50 stud
ent
s;
(6)
All large cl
assro
o
m
s
are m
u
lti-medi
a en
abled;
For such adv
ertise
ment, p
a
rt of predi
cat
e
-form
RDF statements a
r
e
as follows:
advertise( r0): isa(r0, DQtP
ortion
); eleme
n
tCla
ss(r0, Cl
assro
o
m)
; qu
antity(r0, 100
);
isa(r1, DQtPo
r
tion); in
clud
e
(r0, r1); elem
ent
Cla
s
s(r1, Media
C
la
ssro
om); qua
ntity(r1, 70);
isa(r2, DQtPo
r
tion); in
clud
e
(r0, r2); elem
ent
Cla
s
s(r2, Larg
e
Cl
assro
o
m); qua
ntity(r2, 40);
…
Whe
r
e M
edia
C
la
ssroo
m
is assume
d to
be defin
e
d
in
the ontology
as the
sub
c
l
a
sse
s
of
Cla
s
sroo
m a
nd su
bsume
s
Large
Cla
s
sroom3.
3 Reso
urce re
que
st spe
c
ification
s
.
3.3. Resou
r
c
e
Requ
est S
p
ecifica
tions
While
re
sou
r
ce a
d
v
e
rt
is
e
m
ent
s
spe
c
if
ies a
set
of
re
sou
r
c
e
in
st
an
ce
s,
a r
e
so
ur
ce
requ
est
sp
ecifies a
pattern of resou
r
ce obj
ect
s
tha
t
is to
be m
a
tche
d a
gain
s
t the
de
clare
d
resou
r
ce adv
ertise
ment
s. As pattern re
sou
r
ce
reque
st gen
erali
z
e
s
re
so
urce a
d
vertise
m
ent by
introdu
cin
g
p
a
ttern varia
b
l
e
s (p
refixed
with ‘?’ in bel
ow) a
s
well a
s
co
nst
r
aints
betwe
en the
m
.
For exampl
e, the requ
est of
example (5
) in
se
ction 2.3
can b
e
formul
ated as follo
ws:
Req
u
e
s
t (?X, ?Y, ?Z1,?Z2, ?Z3):
disjoi
nt(?X,?
Y
);isa(?X,DQ
t
Portion);qu
a
n
tity(?X,7);
element
Cla
s
s(?X,Schol
ar[
m
ajor:
c
omp
u
ter-scien
c
e,
title:profess
o
r, loc
a
tion: Beijing] );
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 2302-4046
TELKOM
NIKA
Vol. 12, No
. 1, Janua
ry 2014: 741 – 7
4
6
744
isa(?Y,Tea
ch
er);in
stitute(?
Y
, bit); departm
ent(?Y,cs_
dept); deg
re
e
(
?Y, phd_
cs);
inclu
de(?X, ?Z1); isa
(
?Z
1,DQtPortio
n
);q
uantity(
?Z1, 4
)
; element
Cla
s
(?
Z1, Sch
o
lar[institute
bit]) ;
inclu
de(?X,?
Z2); isa
(
?Z
2,DQtPortio
n
);
q
uantity(?Z2, 3
)
; element
Cla
s
s(?Z
2,
Sc
holar[expertis
e
: grid]
)
;
inclu
de(?X, ?Z3); isa
(
?Z
3, DqtPortio
n
);q
uantity(?Z3,3
)
;
element
Cla
s
s(?Z3,S
c
hola
r
[
e
xpertise: ag
ent])
In the requ
e
s
t spe
c
ificati
on, RDF
-
tripl
e
s
are
written a
s
bin
a
ry
predi
cate
form, and
a
frame-li
ke
syntax is adopt
ed to denote
a spe
c
ia
li
zati
on of cla
ss
wi
th role co
nst
r
aints.
4. A Semanti
c
Model for
Reso
urce M
a
tchma
k
ing
The p
r
obl
em
of quantifie
d re
sou
r
ce
matchm
aki
n
g
with ou
r re
pre
s
entatio
n
can
be
formulate
d
a
s
follo
ws: Given a
re
so
urce adve
r
tise
m
ent ba
se
sp
e
c
ified in
form
as
presente
d
in
se
ction 3.2, a
nd a resource req
u
e
s
t sp
ecified in
fo
rm as p
r
e
s
ent
ed in sectio
n
3.3, how
can
we
deci
de if the
requ
est i
s
satisfied
with th
e re
sou
r
ce a
d
vertise
m
ent
s a
s
a
whol
e, i.e., if the sorts
and the amo
unts of re
sou
r
ce
s
spe
c
ifie
d in a resou
r
ce re
que
st is available in
the colle
ction
of
resou
r
ces sp
ecified in a reso
urce adve
r
tisem
ent
base? To clea
rly
define the problem, a formal
sema
ntics for the represen
tation is ne
ce
ssary.
Defini
tion 1:
A reso
urce m
a
tchma
k
in
g specifi
c
ation i
s
a triple (O1,
O2, A, Q) wh
ere
O1 is a
n
onto
l
ogy, called b
a
se o
n
tology,
which
co
nsi
s
ts of a hiera
r
chy of first-order
cla
s
ses
together
with their re
sp
ecti
ve roles;
O2 i
s
an
onto
l
ogy ba
sed
o
n
O1
co
nsi
s
ti
ng of a
hie
r
a
r
chy of
se
con
d
-o
rde
r
cla
s
ses
with
root
DQtPortio
n
, whi
c
h has
rol
e
s eleme
n
tCl
a
ss,
qu
antit
y, disj
oint an
d
inclu
s
io
n a
s
descri
bed
in
previou
s
secti
on.
A
is an advertisement
ba
se
form
ed as advertise
(r): Tr whi
c
h publ
ish re
sou
r
ce r with
a
RDF
descri
p
tion d
enoting its hi
era
r
chical co
mpositio
n wit
h
role in
clu
s
io
n .
Q is a res
our
ce re
que
st formed a
s
req
u
e
st(X
):
with a finite s
e
t X o
f
res
o
urce variables
and a finite set of const
r
ai
nts Cx betwe
en the variabl
es.
The follo
wing
que
stion is,
given a resou
r
ce
matchma
k
ing
sp
ecifi
c
a
t
ion and a
n
al
locatio
n
of it, what do
es m
ean
by “The resou
r
ce
req
uest i
s
sa
tisfiable with
the
re
so
urce advertiseme
n
t
”.
A semanti
c
formali
z
ation of
our qua
ntified re
sou
r
ce re
pre
s
entatio
n is thus n
e
cessary.
Definition 2.
Given a reso
urce m
a
tchma
k
in
g specifi
c
ation
R=(O1, O2, A, Q), an
interpretation
of R is
a triple I =
(
U, E,
[.]),
where U is
a s
e
t of individuals
,
E
U is
the s
e
t of all
individual
s of resource items, [.]I is a ma
pping
from any expression i
n
R to a set-t
heoretic
c
o
ns
truc
t over U such that
1)
For a
c
l
ass
name
c
in O1, [c
] I
power(U), e
s
pe
cially
[ResourceEl
ement] I = E;
for an
role
r
in O1, [r] I
powe
r
(U
U)
.
2)
For any cl
ass c, sub
c
la
ss
c1 of
c, and in
stan
ce a of c
in R, [c1]I
[c
]I ;
[a] I
[c
] I.
3)
The co
nventi
onal de
scripti
on logi
c con
s
tructo
rs
a
s
well as sub
s
u
m
ption rel
a
tio
n
in O1 are
the same a
s
t
hose of conv
entional d
e
scription logi
c;
4) [DQtPortion]I
=
po
we
r(E
);
[quantit
y]I is
a fun
c
tion i
n
power2
(
E)
N,
su
ch
that
for
any x
power(E), [qu
antity]I(x) = |x
|, i.
e., the nu
mber of
elem
ents i
n
x; [ele
mentCla
s
s]I i
s
a
fun
c
tion
in po
we
r2(E
)
power2(E),
s
u
c
h
that, for any x, y
p
o
we
r(E
) , (x
,
y
)
[element
Clas
s
]I iff x
y ; [inc
lude]I
po
we
r2
(E)
p
o
w
e
r
2
(E),
su
ch
th
at for
any
x
,
y
p
o
w
e
r
(
E
), (x
, y
)
[inc
lude] I iff x
y ; [dis
joint]I
power2
(
E)
pow
e
r
2
(
E
)
,
such that
for
any x, y
po
we
r(E),
(x
, y
)
[dis
joint] I i
ff x
y =
5) For
A
advertis
e(r): Tr , [A]
I =
{[r]I } s
u
c
h
that [Tr]I is true}
6) For
Q
?
reques
t(X
1
, …, Xn) : C, [Q] I
powe
r
2
(
E) and
[Q] I =
{
[X1]
I
,
V
…
[Xn]
I,V | for all valuation V of
variabl
es {X
1, …, Xn} su
ch that [C]
I,V is
true}
With this int
e
rp
retation,
we
can d
e
fine some
se
mantic p
r
op
erties
of a
resou
r
ce
matchm
aki
n
g
spe
c
ificatio
n
.
First, an ad
vertisem
ent must refle
c
t the true
cont
ainment relat
i
on
betwe
en two
portion
s of re
sou
r
ces.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Sem
antic Re
pre
s
entatio
n of Com
p
lex Reso
urce
Req
uest
s
for Service-o
r
ie
nted
… (Xing Wan
g
)
745
Defini
tion 3
. Let R=(O1,
O2, A, Q) be
a resou
r
ce matchm
aki
n
g
spe
c
ificatio
n
,
I be an
interp
retation
of R. I is inadmi
ssi
ble
with re
sp
ect
to A iff [A]I is und
efine
d
; otherwise
I is
admissibl
e
wi
th re
sp
ect to
A. A is
i
n
valid iff all interpretations
of
R
is
inadmiss
ible with res
p
ec
t
to
A; otherwi
se
A is valid.An invalid re
sou
r
ce adve
r
tise
ment descript
i
on is illegal b
e
ca
use it make
s
no sen
s
e. It i
s
imp
o
rtant to
be a
b
le to
ch
eck the
vali
d
ness via
synt
actic inferen
c
e. An immedi
ate
observation i
s
that if A contains a
n
incl
ude-cl
a
u
se DQtPortion[qu
antity:
n1, elementCl
ass:
c1]
inclu
de DQt Portion [qua
ntity: n2,
element Cla
s
s: c2], and
n1
< n2
or
c1
c2 =
then A is
invalid.
Defini
tion 4
:
Let R=(O1, O
2
, A, Q) be
a
valid re
sou
r
ce matchma
k
i
ng spe
c
ificati
on, I be
an a
d
missibl
e
interpretation of
R. Q
is sa
tis
f
ied
with A in I iff there exis
t
s
x
[Q] I s
u
c
h
that
x
[A]I.
Q is
s
a
tis
f
ied with A iff for all interpretation I of
R, Q is sati
sfied with A in I. Q
is unsatisfiable
with A iff for a
ll interpretation I of R, Q is
not s
a
tis
f
ied with A in I.
We th
us e
s
ta
blish
ed a
se
mantic
acco
u
n
t for
the
satisfactio
n
of reso
urce re
qu
est with
resou
r
ce adv
ertise
ment
s.
5. Implementation and
Ap
plication
5.1. Resou
r
c
e
Matc
hmaki
ng as Objec
t
Cons
traint
Satisfa
c
tion
To imple
m
en
t the matchm
akin
g bet
wee
n
a
com
p
lex
resou
r
ce req
uest
and
a reso
urce
advertiseme
n
t, we take the matchma
k
ing p
r
o
b
le
m
as on
e of obje
c
t con
s
traint satisfa
c
ti
on
(O
CS)[8-10]. The variabl
e
s
of an OCS
are re
sou
r
ce variable
s
in the resource re
que
st which
rang
ed ove
r
instan
ce
s o
f
Dqt Portion
;
the con
s
tra
i
nts a
r
e role
con
s
tr
aints
in the resource
requ
est. The
domain
s
of the co
nstraint variable
s
con
s
ist of DqtPo
r
tion instan
ce
s gen
erate
d
by
joining finite numbe
r of sub-p
o
rtio
ns o
f
reso
ur
ce po
rtions in th
e reso
urce adve
r
tisem
ents. F
o
r
the allo
cation
to be op
era
b
le, we
stipul
ate that
all the sub-po
rtio
ns a
r
e from
among
a
set
of
mutually dis
j
oint res
o
urc
e
portions
. To mak
e
t
he id
ea
clea
re
r, we gi
ve the followi
ng definition:
Defini
tion 5:
Let R=(O
1, O2, A, Q) be a valid resou
r
ce matchm
aki
ng sp
ecifi
c
ati
on. VQ
and CQ are
resp
ectively the resou
r
ce variable
se
t and
query co
nst
r
aint of Q. And A quota out of
A is a set of pairs
={
s1/r1, …, sn/rn }h
ere r1, …,rn a
r
e
node
s in A, whi
c
h satisfie
d followin
g
conditions:
(
a
)
s
1
, …
,
s
n
ar
e r
e
s
p
ec
tive
ly s
u
b-
p
o
r
t
io
ns
o
f
r1, …,rn in that includ
e(ri
, si) hold
s
for
each i.;
(b)
s1, …, sn a
r
e
mutually disj
oint ,i.e., disjoint(si, sj
) hold
s
for ea
ch i a
nd j.
(c)
the quantity of si is determi
ned.
For ea
ch
sub
s
et R of { s1,
…, sn }, let JR be a ne
w in
stan
ce of Dqt
P
ortion by joi
n
ing all
the resou
r
ce portion
s of R
in following
way:
(1)
the quantity value of JR is the sum of
th
ose of all the
resou
r
ce porti
ons of R
(2)
the element
Class value of JR is the
DL
-uni
on of tho
s
e of all the resou
r
ce po
rtio
ns of R
(3)
the set of incl
ude value
s
of
JR is
R
(4)
the set of disj
oint values of
JR is the inte
rse
c
tion of th
ose of all re
source po
rtion
s
of R
An assig
n
me
nt of Q
with q
uota
i
s
a
m
appin
g
whi
c
h m
a
p
s
e
a
ch re
so
urce va
riable
in
Q to a subset S of {
s1, …, sn }.
is an allocation of A to Q iff
when each re
so
urce variabl
e X
in Q i
s
re
pla
c
ed
in
CQ
b
y
J
[X], the i
n
stantiate
d
con
s
trai
nt is
satisfie
d
with
A a
s
d
e
fine
d in
definition 4.
A resource
matchm
aki
n
g
algorith
m
ba
sed
on thi
s
i
dea thu
s
n
e
e
d
to find on
e
or mo
re
mutually disjo
i
nt sub-po
rtio
ns of
adverti
sed resou
r
ce portion
s that
satisfie
d the con
s
trai
nt of the
requ
est. The
con
s
traint
-so
l
ving algorith
m
is cu
rrently under d
e
velo
pment.
5.2. Application Back
gro
und
The re
se
arch aims at reso
urce man
ageme
n
t in an ongoi
ng
multi-ag
ent education
manag
eme
n
t system for
co
llege. The mu
lti-agent sy
stem con
s
i
s
ts o
f
two set of agents [11]. O
ne
is a set of reso
urce ag
e
n
ts , such as es
tate ag
e
n
ts, human reso
urce ag
e
n
ts, and textboo
k
agent
s, which
provide
se
rvice
s
of
re
so
urce re
que
sting
,
boo
king, an
d allo
catio
n
.
The
other set of
agent
s are ta
sk
age
nts, such
as
dep
artment cle
r
ks
,
whi
c
h p
e
rfo
r
m task pla
n
n
i
ng, sche
duli
ng,
monitori
ng an
d executio
n. The reque
sti
ng and
allo
ca
tion of re
sou
r
ce a
r
e imp
o
rt
ant part
s
in the
intera
ction
be
tween
the ta
sk ag
ents an
d re
so
urce
a
gents [12]. Despite
the
diversity of va
ri
ous
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 2302-4046
TELKOM
NIKA
Vol. 12, No
. 1, Janua
ry 2014: 741 – 7
4
6
746
sort
s of re
so
urces, the b
ehavio
rs of the re
sou
r
ce agent
s are q
u
ite similar.
Thus a g
ene
ric
frame
w
ork for reso
urce mo
deling i
s
ne
ce
ssary.
6. Conclusio
n
In this pape
r
we p
r
opo
se
d
a rep
r
e
s
entat
ion for qu
anti
f
ied re
sou
r
ce
matchma
k
in
g with a
numbe
r of n
o
v
el feature
s
.
First it all
o
ws the re
prese
n
tation of com
p
lex re
sou
r
ce
req
u
e
s
ts a
n
d
advertiseme
n
t
s with qua
ntified resou
r
ce
quota,
containme
n
t hiera
r
chie
s
and di
sjointn
e
ss
con
s
trai
nts. T
h
is
enha
nce
the flexibility and
expre
s
si
veness
of th
e re
present
at
ion.
To give an
accurate d
e
finition of th
e
reso
urce
matchmaki
ng
wi
th
su
ch
representation,
a
se
mantic theo
ry is
establi
s
h
ed. Secon
d
it is sema
ntic-We
b
-o
riente
d
in that the re
p
r
e
s
entatio
n follows co
nventi
ons
of RDF
and
sema
ntic
Web
ontolog
y. In addition,
the re
sou
r
ce
-servi
cing
architectu
re
with
summ
ari
z
ed
reso
urce adve
r
tisem
ent rep
o
sitory c
oop
e
r
ating with
re
sou
r
ce-req
ue
sting ag
ents i
s
in line with th
e spirit of se
mantic
Web a
nd is
suitable
for wide
ran
g
e
of e-comme
rce a
ppli
c
atio
ns.
The futu
re
work in
clu
de t
he
developm
ent
of
efficient algorithm
s for the matchmaki
n
g
with this
repres
entation.
Ackn
o
w
l
e
dg
ement
This
wo
rk i
s
funded
by the Foun
dation
and F
r
ontie
r Tech
nolo
g
y Re
sea
r
ch Project of
Hen
an Provin
ce (Gra
nt No.
1223
0041
042
6).
Referen
ces
[1]
T
Berners-Lee,
J H
e
n
d
ler,
an
d O L
a
ssi
la. T
he S
e
ma
ntic
W
eb.
Scie
ntific
Americ
an
.
20
12; 2
84(
5): 3
4
-
43.
[2]
Bock C, Gru
n
i
nger
M. PSL:
A Sema
ntic D
o
mai
n
for F
l
o
w
Mod
e
ls. S
oftware
and
Syste
m
s M
o
d
e
li
n
g
Journ
a
l
. 20
12; 8(6): 29-3
8
.
[3]
Liu C a
nd F
o
ster I. A Constraint Lang
ua
ge A
ppro
a
ch
to Grid Reso
urce Se
lectio
n. Chica
g
o
: Univers
i
t
y
of Chica
go. 20
11: 60-8
1
.
[4]
C Liu, L Y
a
n
g
, I F
o
ster, and D Ang
u
lo.
D
e
s
i
gn
and
eval
ua
tion of a r
e
sou
r
ce selecti
on fr
amew
ork
. I
n
Proceedings
of the E
l
ev
enth
IEEE Inter
national S
y
mposium
on
High-Performance Distributed
Comp
uting, Ed
inb
u
rgh, Scottl
and, 20
12.
[5]
Hon
g
sud
a
T
a
n
g
mun
a
runk
it,
Stefan D
e
cker,
Carl
Kess
elm
an:
Ontol
ogy-B
ased
Res
ourc
e
Match
i
ng
i
n
the Grid
-
T
he Grid Meets the
Semantic W
e
b
.
Inter
nation
a
l Semantic
W
eb
Confer
ence 20
09:
706-
72
1
[6]
Sean
Bec
h
h
o
fer, et.al.
OW
L W
eb Ontol
o
g
y
La
ngu
ag
e
Refer
ence.
Avail
a
b
l
e
at
http://www.
w
3.
org/TR/owl-re
f/. accessed on Jan 2009.
[7]
I Horrocks, PF Patel-Sc
hneider, and F van Harm
elen.
Review
in
g th
e desi
gn of D
A
ML+
OIL: An
ontol
ogy
la
ngu
age
for th
e se
ma
ntic w
e
b
. In
Proc. of the 18th Nat. Conf. on
Artificial Inte
ll
ige
n
ce (AAA
I
200
2), 200
2.
[8]
T
he OW
L Services C
oal
ition. OW
L-S: Se
mantic M
a
rkup for W
eb Servic
es. Availa
ble
a
t
http://www
.daml.or
g/servic
es/owl-s/1.0/owl
-s.html, access
ed on Ja
n 20
1
3
.
[9]
L Li and I Horrocks.
A softw
are fra
m
ew
ork f
o
r
match
m
akin
g b
a
se
d o
n
s
e
ma
ntic w
e
b
te
chno
logy.
In
Proceedings of
the
T
w
elfth Inte
rnational World Wide
Web Conference (WWW
2003), 2003: 111-
120.
[10]
K S
y
c
a
ra, S
W
i
do, M Kl
usch,
and
J L
u
. L
a
rk
s: D
y
namic
ma
tchmakin
g am
o
ng
hetero
g
e
n
e
ous s
o
ft
w
a
r
e
age
nts in c
y
ber
space.
Intern
ation
a
l of w
eb se
rvice an
d co
mp
uting
2
0
0
8
; 18(
9): 23-30.
[11]
F
G
F
adel, MS
F
o
x, M Gru
n
i
nger M.
A G
e
neric E
n
terpris
e
Res
ourc
e
Ontolo
gy.
Proce
edi
ngs
of th
e
T
h
ird W
o
rkshop on En
ab
lin
g T
e
chnolo
g
ie
s - Infrastr
uctures for Col
l
a
borativ
e Enter
p
rises, W
e
st
Virgin
ia U
n
iver
sit
y
. 20
11: 83-
92.
[12]
R Raman. Ma
tchmakin
g F
r
ame
w
orks for Distr
ibut
ed Re
source Ma
nag
ement. Comp
uter Scienc
e
:
Univers
i
t
y
of Wisconsi
n
. 20
10:
24-32.
Evaluation Warning : The document was created with Spire.PDF for Python.