TELKOM
NIKA
, Vol.13, No
.2, June 20
15
, pp. 654 ~ 6
6
0
ISSN: 1693-6
930,
accredited
A
by DIKTI, De
cree No: 58/DIK
T
I/Kep/2013
DOI
:
10.12928/TELKOMNIKA.v13i2.649
654
Re
cei
v
ed O
c
t
ober 1, 20
14;
Revi
se
d Ja
n
uary 22, 201
5
;
Accepte
d
Febru
a
ry 14, 2
015
Determining Trust Scope Attributes Using
Goodne
ss of Fit Test: A Survey
Titin Pramiy
ati*
1
, Iping Supriana
2
, A
y
u Pur
w
arianti
3
ST
EI-Institiut T
e
kno
l
og
i
Ban
d
ung
Jl. Ganesha
10
, Bandun
g, Ind
ones
ia
*Corres
p
o
ndi
n
g
author, e-ma
i
l
: titin.harson
o
@
gmai
l.com
1
, iping@stei.itb.ac.id
2
, ay
u@st
ei.itb.ac.id
3
A
b
st
r
a
ct
Indon
esi
an, as
one of the
co
untries w
i
th hi
g
h
nu
mb
er of in
te
rnet users h
a
s
the potenti
a
l
to serve
as the plac
e w
i
th great info
rmati
on res
our
ces. Ho
w
e
ver, these resourc
e
s must be ac
compa
n
ie
d by th
e
avail
a
b
ility of dep
en
dab
le
in
format
i
on. Inf
o
rmati
o
n
trustw
orthin
ess ca
n
be
obta
i
ne
d
b
y
assess
ing
t
he
co
n
f
i
d
e
n
c
e
l
e
ve
l
(tru
st) o
f
the
so
u
r
ce
o
f
in
fo
rma
ti
on
. Th
i
s
ca
n
b
e
d
e
t
e
rmi
n
e
d
b
y
usi
n
g
tru
s
t scope
attributes. He
n
c
e, in this stud
y, w
e
intende
d to est
abl
ish the
trust scope attributes by
mea
n
s of utili
z
i
ng t
h
e
ones
cont
ain
e
d
in t
he
User Pr
ofile
prov
ide
d
by soc
i
al
med
i
a; in
this c
a
se
F
a
cebo
ok, Go
ogl
e+
, T
w
itter,
an
d
Link
edi
n. W
e
carried
o
u
t th
e res
earch
by
con
ducti
ng f
our sta
ges
na
me
ly
dat
a c
o
l
l
e
ction,
attribut
es
grou
pin
g
, attrib
ute selecti
on, and surv
eys. T
he data co
ll
ected ori
g
in
ated fr
om th
e
User P
r
ofile co
ntents of
the 4 s
o
cia
l
me
dia r
e
se
arche
d
:
F
a
cebo
ok, T
w
itter, G
oogle+
, and
Link
edIn.
A survey w
a
s t
hen
distri
buted
t
o
257 ra
nd
o
m
ly
selecte
d
resp
o
nde
nts (divi
d
e
d
into tw
o
clus
ters: civilia
ns and
mi
litary off
i
cers) to seek
for
their o
p
in
io
ns i
n
terms
of w
hat attributes w
e
re co
ns
id
ere
d
to be cruc
ia
l i
n
defi
n
in
g the
beli
e
va
bil
i
ty of
an
infor
m
ati
on so
urce. Ch
i-squ
a
r
e Good
ness
o
f
fit T
e
st
w
a
s cond
ucted to c
o
mp
are
observ
e
d data w
i
th
dat
a
w
e
w
ould exp
e
c
t to obtain. T
h
e results of the
research
s
ugg
ested that ther
e w
a
s
simi
lar j
udg
ment in ter
m
s
of dictating so
urce of infor
m
ation trustw
orthin
ess c
hose
n
by the resear
ch partici
pants
w
i
th the attrib
utes
provi
ded by tru
s
t scope categ
o
ry. In this research, both
c
i
v
ilia
ns an
d milit
ary officer clus
ters concurre
ntly
perce
ived
that
educ
atio
nal
ba
ckgrou
nd
w
a
s
the most de
pe
nda
ble
attrib
ut
e. T
hey s
ubs
e
que
ntly i
n
d
i
cat
e
d
that the
plac
e
w
here a
pers
o
n stud
ies,
occ
u
patio
n, an
d
pla
c
e of w
o
rk w
e
r
e
ess
entia
l attri
butes to
ens
ur
e a
source of infor
m
ati
on trustw
orthiness.
Ke
y
w
ords
:
Re
ferral T
r
ust, F
u
nction
al T
r
ust, T
r
ust Scope, T
r
ust Scope Attributes, Goo
dne
ss of F
i
t
T
e
st
Chi-Sq
uar
e
1. Introduc
tion
No
wad
a
ys, p
eople in
Ja
karta are am
ong the larg
est intern
et use
r
s–e
s
pe
ci
ally on
Twitter- in
Asia an
d a
r
e
nu
mber 4 i
n
the
wo
rld.
T
h
is fact p
r
ove
s
th
at Indon
esi
a
has ab
und
ant
of
informatio
n and potential i
n
format
io
n source
s used
in decisi
on
makin
g
. Tru
s
t information
is
obtaine
d ba
sed on level of
trust and
rep
u
tation of
the informatio
n source
s. The
r
e are
some trust
model
s d
e
veloped to
dete
r
mine tru
s
t le
vel, such a
s
trust m
odel t
hat ca
n h
e
lp
use
r
s to a
s
sess
the trustwort
h
ine
ss of an
applicatio
n [1]; to deter
mine trust leve
l of internet use
r
s [2]; an
d to
dictate t
r
u
s
t l
e
vel of the
p
eer to reserv
e servi
c
es [3]
;
to dete
r
min
e
tru
s
t level
and
rep
u
tatio
n
of
teammate wit
hout kn
owi
n
g
the person to
be sele
cted [
4
].
Tru
s
t informa
t
ion is used to deci
de tru
s
t level of the
media dist
ri
buting inform
ation or
so
cial netwo
rking, su
ch as
Facebo
ok
,
Twitter
,
Go
ogl
e+
,
Li
nked In
and so on. Users of
the so
cial
netwo
rki
ng site
may cre
a
te
a pe
rso
nal
p
r
of
ile,
excha
nge
m
e
ssag
es, i
n
cluding
autom
atic
notification
s
whe
n
thei
r p
r
ofile is
upd
ated with
ne
w conte
n
t
from other us
ers [
5
], therefo
r
e,
the
profile is o
ne
way to kno
w
l
edge
sha
r
ing
from one
syst
em to anothe
r [6].
User profile i
s
u
s
e
d
in
several
re
se
arch
s
fo
r the
pu
rpose of i
denti
f
ying and
ma
tching
a
person. Alisa
and Gordon
(2005
) used
user p
r
ofile
to integrate
contextual inf
o
rmatio
n abo
ut
mobile u
s
e
r
s
and devi
c
e
s
i
n
their e
n
viro
ntment [7
]. They used u
s
e
r
location a
s
t
he main
drive
s
for the
co
ntext-awa
r
e
information, be
ca
use
lo
cation
-
a
wa
re i
n
form
ation
serv
i
c
e
s
are
services
that
provide the u
s
er
with the i
n
formatio
n se
t that is
related to their cu
rre
nt pos
ition. Results of this
resea
r
ch, the
y
have ide
n
tified the
co
nte
x
tual eleme
n
t need
ed to
d
e
scrib
e
the
u
s
er p
r
ofile a
n
d
spe
c
ified thei
r definition in
the prop
osed
Re
sou
r
ch De
scription F
r
a
m
ewo
r
k (RDF).
Elie Raad et.al (201
0) u
s
e
d
use
r
profile
to so
lve the
probl
em of matchin
g
use
r
profile i
n
its glo
bality
by providi
n
g
a
suitabl
e
matchin
g
frame
w
ork
ab
le to
co
ns
id
er a
ll th
e pr
o
f
ile
’
s
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Determ
inin
g Tru
s
t Scop
e Attributes Usi
ng Goo
dne
ss of Fit Tes
t: A
Surv
ey
(Titin Pramiy
ati)
655
attributes.
Th
e frame
w
o
r
k i
s
abl
e to di
scover the
bigg
est po
ssible
n
u
mbe
r
of p
r
of
iles that
refe
r
to
the same p
h
ysical user tha
t
existi
ng app
roche
s
are un
able to dete
c
t [8].
Olga Peled
et.al (2013
) use
d
a sup
e
r
vised le
arni
ng method t
o
match u
s
e
r
profile
s
accro
s
s multi
p
le Online So
cial Net
w
o
r
ks, this
method is ba
sed on
machi
ne lea
r
ning the
c
hniq
ues
that use
a v
a
riety of fe
ature
s
extracte
d from
a
u
s
er’s profile as well
as
th
eir frien
d
’s p
r
ofi
l
e.
Re
sult of the
re
sea
r
ch is
high mat
c
hin
g
perf
o
rma
n
ce
wh
en
the method wa
s evaluated usi
n
g
real
-life data colle
cted fro
m
two OSNs, Facebo
ok
a
nd Xing. The
high re
sult is eviden
ce that
use
r
identification ba
sed o
n
web p
r
ofile
s is
con
c
eptu
a
lly and pra
c
t
i
cally po
ssi
ble [9].
Ho
wever, the
validity and trust
w
o
r
thine
s
s of
the information are often time questionable,
becau
se, the mecha
n
ism
to determine
informatio
n
trust level is n
o
t provided y
e
t. Therefore, a
new mo
del is needed to a
c
commo
date
the need
s to
ensure the b
e
lievability of the information
distrib
u
ted in
so
cial medi
a.
We inte
nded
to establi
s
h
an inform
ation tru
s
t mod
e
l usin
g tru
s
t
level of informatio
n
source [10], utilizing feedback [11
], trust level
[12],
interaction-
based [13],
context informati
o
n
[14] and rep
u
t
ation of informat
ion so
urce [15] param
eters.
Ho
wever in
this
sp
ecifi
c
p
aper we
will
only
di
scuss
sou
r
ce
of inf
o
rmatio
n trust level
whi
c
h i
s
dete
r
mine
ba
sed
on tru
s
t sco
p
e
and fe
edb
a
ck
given by
others u
s
e
r
s.
Tru
s
t scop
e
is
retrieve
d fro
m
the result of inform
ation mai
n
attri
bute mat
c
hin
g
proces
s
wi
th the source of
informatio
n
context that ex
ists i
n
User P
r
ofile
. Thi
s
paper
will
discuss ho
w to figure t
r
ust scope
of information
source
s u
s
in
g referral tru
s
t
and function
al trust app
ro
ach [16], and
survey.
2. Res
earc
h
Method
In this
re
sea
r
ch, the
meth
od u
s
e
d
to d
e
termin
e attri
bute tru
s
t
scope
wa
s
bro
k
en
int
o
several processe
s. They
are: data collectin
g a
ttributes g
r
ou
pi
ng, sele
cting
attributes,
and
conducting survey. Social
media
utilized in this research are
Fa
ceb
ook
,
Twitter
, and
Go
ogle
+
.
Data collecti
on pro
c
e
s
s wa
s con
d
u
c
ted by colle
cting all attributes in
User Profile
provide
d
by
the
so
cial
me
dia
cho
s
e
n
in
this
re
se
arch
. The
next ste
p
was ad
mini
sterin
g
attribu
t
es
grou
ping. Thi
s
g
r
ou
ping
m
e
thod wa
s co
mpleted by
mean
s of l
o
o
k
ing
at fun
c
ti
on
simila
rities of
each attribut
es. Sele
ction
attributes proce
s
s was
d
one by o
b
se
rving comp
atibility betwee
n
attributes wit
h
two
catego
ries:
refe
rr
al trust
(tru
st which
is
built based on kn
owle
dge
of
t
he
use
r
s) and
fu
nction
al tru
s
t
(trust that i
s
establi
s
hed based on one’
s ability
to sol
v
e certain task).
Furthe
rmo
r
e,
su
rvey di
stri
bution
proce
s
s was
co
ndu
c
t
e
d
b
y
c
r
ea
tin
g
s
e
lf-a
dmin
is
ter
e
d
sur
v
e
y
que
stion
s
, and distri
bute
d
dire
ctly
to resp
ond
ent
s. We also use
d
survey
agency service
(Lem
bag
a Surve
y
Mu
da In
done
sia/LSMI
)
.
To en
su
re
whether
Tru
s
t Scope
attribu
t
es can b
e
u
s
ed to di
ctate
trust
w
orthi
n
e
ss
of the
sou
r
ce of
in
formation, we
rando
mly survey
e
d
2
5
7
pa
rticip
ant
s: 10
0
colle
ge
stude
nts,
50
employees f
r
om finance domain, 23 m
edical doct
ors, 15 universi
t
y prof
essors, and 69 military
officers. The
underlyin
g reason
s why we sele
cted the partici
pa
nts we
re a
s
follows. It wa
s
assume
d tha
t
these indivi
dual
s in as
much
as thei
r age (above
17 years ol
d), edu
cation
al
backg
rou
nd,
job resp
on
sibilitie
s or job dem
a
n
d
s
, and their work ethi
cs can judg
e
the
trust
w
orthi
n
e
ss of a certai
n informatio
n be
tter compa
r
ed to othe
r segment
s of populatio
n.
We then divi
ded the pa
rticipa
n
ts into 2
cluste
rs: civil
i
ans (110 p
a
rticipant
s) an
d
military
(69
pa
rticip
a
n
ts).
We
assumed
that
civilians an
d
military offic
e
rs
differ
s
i
gnifi
cantly in te
rm
s of
trusting a
n
informatio
n. They are tra
i
ned to ve
rify the information sy
ste
m
atically and
very
c
a
refully to ens
ure safety.
In this
surve
y
we ad
mini
stere
d
chi-sq
uare
Go
odne
ss
of fit Test
. We g
ene
ra
ted two
hypothe
se
s for this te
sting
;
there are n
u
ll hypothesi
s
(
H
0
) an
d alternative hypoth
e
si
s (
H
a
):
H
0
:
p
1
= p
2
=
...=
p
n
= 1/
n
H
a
:
there is a
proba
bility g
r
eater than 1 /
n
Null hyp
o
the
s
is me
an
s ea
ch attribute
s
h
a
ve sa
me p
r
obability, and
altern
ative h
y
pothesi
s
mean
s at l
e
a
s
t on
e attrib
u
t
e ha
s p
r
ob
a
b
ility great
e
r
that others
attributes o
r
m
ean valu
e. Chi-
squ
a
re t
e
st
(
X
2
test) use
d
to hypothesi
s
test, that is
a test to co
mpare ob
servation freque
ncy
with expe
ctan
cy freque
ncy.
We used foll
owin
g formul
a to obtain
X
2
value
;
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 13, No. 2, June 20
15 : 654 – 66
0
656
(1)
We counte
d
use
s
chi-Sq
u
a
re
tabl
e
to
get
X
2
table
according to
degree of freedom
(df)
=(
r-1
)(
k
-
1)
and sig
n
ifica
n
ce level
α
. Null hypothe
sis reje
cted if chi-sq
ua
re va
lue gre
a
ter th
an
chi-sq
ua
re ta
ble (X
2
value > X
2
table).
3. Resul
t
s
and
Discus
s
ion
3.1. Analy
s
is
Based
on th
e re
sult
s gott
en from
the
data
colle
ctio
n process
We did
on G
o
ogle
+
, in
User
Profile
se
ction, We
found
that
the
r
e we
re attrib
utes group
s named
People
,
Story
,
Work
,
Educatio
n
,
Plac
es
,
B
a
si
c I
n
f
o
rm
at
ion
,
Links
, and
Contact Inform
ation
. Each
grou
p co
ntai
ns
attributes th
a
t
match its
g
r
oup
ch
aract
e
risti
c
(
ill
ustrated in Ta
ble
1.). The tabl
e de
scribe
s t
he
attributes exi
s
ts in
Use
r
Profile on Goo
g
l
e+.
Table 1. Attributes in
User Profile on Go
ogle
+
Grou
ps Attributes
People
In m
y
Circles
I in the
y
Circles
Stor
y Tagline
Introduction
Bragging
Right
Work Occupation
Skil
ls
Emplo
y
ment
Grou
ps Attributes
Education School
Name
Major or Field of
Stud
y
Start
y
e
ar
End
Y
e
ar
Curre
nt
Description of Co
urses
Places Cit
y
Name
Curre
nt
Basic Information
Gende
r
Looking
for
Birthda
y
Relationship
Other
Names
Links Other
P
r
ofile
Contributo
r
to
Links
Contact Inform
ation
Home
Wor
k
User Profile Attributes on
Face
boo
k (I
ndon
esi
an Versi
o
n
)
, con
s
i
s
ts of severa
l grou
ps,
they are:
wo
rk a
nd ed
ucation
,
Plac
es You
’
v
e
Liv
e
d
,
Conta
c
t Inform
ation, Basi
c Inform
ation,
Fam
ily
and
Relationships
,
About You
and some
addition
al informatio
n su
ch as:
F
r
iend
,
Application
,
Grou
p
,
Photo
s
as
see
n
in Table 2.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Determ
inin
g Tru
s
t Scop
e Attributes Usi
ng Goo
dne
ss of Fit Tes
t: A
Surv
ey
(Titin Pramiy
ati)
657
Table 2. Attributes in
User Profile on Fa
ceb
o
o
k
Grou
ps Attributes
Work & Education
Workplace
Professional
Skill
College
High
School
Places Y
ou’ve Li
ved
Hometo
w
n
Other Places Lived
Contact Inform
ation
Email
Mobile
phones
Address
Other
Accounts
Website
Basic Information
Birth Date
Birth
Y
e
a
r
Gende
r
Religious
View
s
Language
Y
o
ur
Political Vie
w
s
Famil
y
& Relatio
n
ships
Relationships
Famil
y
Membe
r
About Y
o
u
About Y
o
u
Favorite
Quotes
Friends
Friend’s
Name
Grou
p
Grou
p’s
Name
Photos
Applications Application’s
Na
me
Twitter Use
r
Profile has the followin
g
attributes:
Nam
e
,
Use
r
nam
e
,
Bio
,
Web
s
ite
,
Tweetin
g Sin
c
e
an
d
Lo
ca
tion
. An account hold
e
r
can
write a
n
y
kinds
of information a
b
out
him/herself o
n
Bio
attribut
e. Ho
weve
r, the content
s
were f
illed i
n
variou
s
kind
s of format
s. Such
as the foll
owi
ng:
Bio: inter fam
ily, unjan
i fam
ily, TAB fa
m
ily, and
m
u
slim
fa
m
i
ly. @addi
cted
to
chem
istry
(S
ource:
Twitter
,
a
c
count h
o
lder @agiit
_tiiga),
o
r
Bio
which i
s
si
mply create
d
as
follows
:
Wali
kota Bandun
g 2013
-20
1
8
(S
ource:
Twitter
, acco
unt hol
der
@rid
wa
n
k
amil
).
The data o
b
tained fro
m
the data coll
ect
i
on pro
c
e
s
s
wa
s then an
a
l
yzed to se
e attribute
simila
rities in
the th
ree
social
medi
a
cho
s
e
n
. Fo
r
example, th
e
one
s we
se
e on
Situs Web
attribute on
F
a
ce
boo
k
,
We
bsite
attribute
on
Twitter
an
d
Links
attrib
ute on
Goo
g
l
e
+
.
Based
on th
e
profile
gro
u
p
i
ng p
r
ocess
result
of th
e three
social
media
cho
s
e
n
, it wa
s
found that th
ere
were 6 profile gro
ups t
hat give information ab
out
soci
al ci
rcl
e
s
,
the profile
,
job
,
location
,
ba
si
c inf
o
rm
at
ion
, and
ed
ucational ba
ckgro
und
of a
n
a
c
cou
n
t hol
der.
Table
3.
sho
w
s
the profile g
r
o
ups that have
been creat
e
d
base
d
on the
similarity attributes
.
Table 3. Prof
ile grou
ps
Social Media
Grou
ps
Google+
Facebook
T
w
itter
About
Friend
G
r
o
up
Bio
Google+
Facebook
T
w
itter
Story
About Y
o
u
Bio
Google+
Facebook
Wor
k
Education
Work & Education
Google+
Facebook
T
w
itter
Basic Information
Basic Information
Bio
Google+
Facebook
T
w
itter
Places
Places
Location
Google+
Facebook
T
w
itter
Links
Contact Inform
ation
Contact Inform
ation
Name
Username
Website
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 13, No. 2, June 20
15 : 654 – 66
0
658
After groupi
n
g
the data, the nex
t process co
ndu
cted
wa
s ch
oo
si
ng
suitable attri
bute that
match
referral trus
t
cate
gory, as seen in
Table 4.
Table 4. Refe
rral tru
s
t attri
butes
Grou
ps Attributes
W
o
rk Occupation
Skil
ls
Education
Field of Study
Description of courses
W
o
rk & Educatio
n
W
o
rkplace
College
Professional
Skill
About You
Bio
The attribute
sele
ction p
r
o
c
e
ss
re
sults u
s
ing
func
tional trus
t
categ
o
ry ca
n be
se
en in th
e
followin
g
Tabl
e 5.
Table 5. Fun
c
tional trust att
r
ibute
s
Grou
ps Attributes
Work & Education
Workplace
College
Professional
Skill
Bio
Work Occupation
S
k
i
ll
Emplo
y
ment
Education
Major or field of s
t
ud
y
Description of Co
urses
All the attribu
t
es obtai
ned,
we
re
com
b
in
ed into
one
b
i
g group
call
e
d
Tru
s
t S
c
op
e
Tru
s
t
that will be used to determi
ne source
of i
n
formation trust level.
Table 6. shows t
he newly formed
grou
p of attributes (Trust
Scope attri
b
u
t
es):
Tabel 6. Trust scop
e
Attributes
Grou
ps Attributes
Trust Scope
Education
School
Name
Major or Field of
Stud
y
Workplace
Emplo
y
ment
Occupation
Professional Skill
Skil
ls
Interested in
Communit
y
3.2. Sur
v
e
y
Table
7. belo
w
de
scri
be
s the ob
se
rvatio
n freq
uen
cy o
f
110 p
a
rti
c
ip
ants
(civilia
ns). In thi
s
table we u
s
e
d
10 attibutes, which i
s
Atr1 for Educat
ion attribute,
Atr2 for Scho
ol Name
attribute,
Atr3 for
Maj
o
r o
r
Fi
eld
o
f
Studi attrib
ute, At
r4 for
Wo
rkpl
ace at
ttibute, Atr5 for Empl
oyme
nt
attribute, Atr6 for Occupati
on attribute, Atr7 for
Profession
al Skill, Atr8 for Skills attribute, Atr9
for Intere
sted
in attribute, and Atr10 for
Comm
unity attribute.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Determ
inin
g Tru
s
t Scop
e Attributes Usi
ng Goo
dne
ss of Fit Tes
t: A
Surv
ey
(Titin Pramiy
ati)
659
Table 7. Ob
servation Freq
uen
cy
Atr1
Atr2
Atr3
Atr4
Atr5
Atr6
Atr7
Atr8
Atr9
Atr10
∑
1
1 1 0 0 1 0 3 2 0
5
13
2
2 0 2 0 0 5 4 0 0
3
16
3
1 2 0 2 1 0 5 0 1
1
13
4
1 0 2 2 2 2 0 0 2
0
11
5
2 2 1 3 4 5 8 9 7 17
58
6
1 2 1 4 2 3
12
5
23
31
84
7
4 2 3 2 3
14
14
11
25
16
94
8
2
3
5 11 2 13
14
20
23
4
97
9
4
8
4 13 4 16
20
19 5
2
95
10
9
12 14 14 20 24
8
10
4
8
123
11
7 10 7 21
22
10 5
8
5
2
97
12
15 6 19
21
19 7
4
6
3
5
105
13
7 13
32 9 10 5
2
6
6
6
96
14
13
37 9
4 10 3
4 10 3
2
95
15
41
12
11 4 10 3
7
4
3
8
103
∑
110 110 110 110 110 110 110 110 110
110
1100
Based
on
ob
servatio
n fre
quen
cy data
see
n
in
Tabl
e 7, expe
ctat
ion fre
que
ncy
in ea
ch
cell wa
s cal
c
ulate
d
by mean
s
of co
unting
the
mean. T
he
mean val
ue
wa
s obtai
ne
d by
multiplying th
e sum
of o
b
servation
freq
u
ency i
n
e
a
ch
row a
nd th
e
sum of
ob
se
rvation frequ
en
cy
in each col
u
mn divided
by the sum
of obse
r
va
tio
n
frequ
en
cy in each col
u
mn. Expectat
i
on
freque
ncy val
ues
can b
e
seen in Tabl
e 8.
Table 8. Expectan
c
y frequ
ency
Rank
Atr1
Atr2
Atr3
Atr4
Atr5
Atr6
Atr7
Atr8
Atr9
Atr10
1
1,3 1,3 1,3 1,3 1,3 1,3 1,3
1,3 1,3 1,3
2
1,6 1,6 1,6 1,6 1,6 1,6 1,6
1,6 1,6 1,6
3
1,3 1,3 1,3 1,3 1,3 1,3 1,3
1,3 1,3 1,3
4
1,1 1,1 1,1 1,1 1,1 1,1 1,1
1,1 1,1 1,1
5
5,8 5,8 5,8 5,8 5,8 5,8 5,8
5,8 5,8 5,8
6
8,4 8,4 8,4 8,4 8,4 8,4 8,4
8,4 8,4 8,4
Rank
Atr1
Atr2
Atr3
Atr4
Atr5
Atr6
Atr7
Atr8
Atr9
Atr10
7
9,4 9,4 9,4 9,4 9,4 9,4 9,4
9,4 9,4 9,4
8
9,7 9,7 9,7 9,7 9,7 9,7 9,7
9,7 9,7 9,7
9
9,5 9,5 9,5 9,5 9,5 9,5 9,5
9,5 9,5 9,5
10
12,3 12,3 12,3 12,3 12,3 12,3 12,3
12,3 12,3 12,3
11
9,7 9,7 9,7 9,7 9,7 9,7 9,7
9,7 9,7 9,7
12
10,5 10,5 10,5 10,5 10,5 10,5 10,5
10,5 10,5 10,5
13
9,6 9,6 9,6 9,6 9,6 9,6 9,6
9,6
9,6
9,6
14
9,5 9,5 9,5 9,5 9,5 9,5 9,5
9,5
9,5
9,5
15
10,3 10,3 10,3 10,3 10,3 10,3 10,3
10,3
10,3
10,3
T
otal
110 110 110 110 110 110 110
110
110
110
3.3. Discuss
ion
The d
a
ta coll
ection
process gave
u
s
inf
o
rmatio
n that
Facebo
ok
,
Twitter
a
nd
G
oogle
+
provide
attrib
utes that inform peopl
e abo
ut t
he profile
of an acco
unt
use
r
. For ex
ample,
Go
ogl
e+
sup
p
lie
s
Work
, and
Education
attrib
utes: me
an
whil
e, Twitter
uses
Bio
attrib
ute that cont
ains
variou
s kin
d
s of information about an a
c
count u
s
er:
and Fa
ceb
o
o
k
utilize
s
wo
rk and ed
ucation
attribute as th
e mean
s to give personal i
n
formatio
n ab
out the accou
n
t user.
Furthe
rmo
r
e,
survey
re
su
lts
su
gge
ste
d
that there
wa
s simil
a
r j
udgme
n
t in term
s of
dictating
so
u
r
ce
of info
rm
ation tru
s
two
r
thine
s
s cho
s
en by the
re
sea
r
ch p
a
rticipants
with t
h
e
attributes pro
v
ided by tru
s
t
scope
categ
o
ry. The
Goo
dne
ss
of Fit
Test u
s
in
g ch
i-sq
ua
re g
a
ve
us
t
he re
sult
t
h
a
t
chi-
squ
a
re
(
X
2
) value wa
s 765,9
588,
and chi-sq
uare (X
2
) table
was 15
3,197
9
(X
2
value
≥
X
2
table), the
r
efo
r
e H
0
wa
s
reje
cted. Thi
s
ga
ve us i
n
form
a
t
ion that the
resp
on
se
s we
re
not ho
moge
n
eou
s. In oth
e
r
words,
ou
r
partici
pant
s g
a
ve different
judgme
n
t in
evaluating
th
e
each attribute
.
For in
stan
ce
, based on
the chi-squ
a
re te
st, it wa
s discove
r
ed that ed
ucatio
nal
backg
rou
nd ranked the hig
hest in
the ci
vilians clu
s
te
r’s judgm
ent. From the sca
l
e of 1 – 15, the
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 13, No. 2, June 20
15 : 654 – 66
0
660
partici
pant
s g
a
ve the high
est sco
r
e (1
5
)
to educ
atio
nal ba
ckgro
u
nd attribute. Followed by, the
context of wh
ere the edu
cation take
s pl
ace (14);
pla
c
e whe
r
e the person
works/type of job (
13);
and the
co
m
m
unity the p
e
rson
belon
g
s
to (6). T
h
i
s
give
s u
s
confiden
ce th
at trust
scop
e
attributes
can
be use
d
to determin
e
the level of
trust towa
rd certai
n
source of inf
o
rmatio
n.
Moreover, by
means of using
the
sam
e
testing met
hods,
we found that the military
officers’
clu
s
ter
con
c
u
r
red
that edu
cati
onal b
a
ckg
r
o
und
wa
s the
most
signifi
cant attri
bute
to
ensure the
believability of certai
n source of
inform
ation
(15).
T
h
ey assigned
10 for contex
t of
whe
r
e the ed
ucatio
n take
s place; pla
c
e
whe
r
e the pe
rson
works (1
0); and o
c
cup
a
tion (10
)
.
4. Conclu
sion
Based o
n
the
rese
arch
we
have co
ndu
ct
ed two majo
r
con
c
lu
sio
n
s h
a
ve been d
r
a
w
n:
1.
Chi-sq
ua
re test admini
s
tered informe
d
that
the hypothesi
s
wa
s rej
e
cted. Which
means that
each parti
cip
ants surveye
d
had differe
nt opinion
in
judging tru
s
t
sco
pe attrib
utes. In this
research bot
h civilians
and military offi
cer cl
usters chose that educa
tional background
was
the mos
t
trus
twor
thy attribute.
2.
Tru
s
t scope
has b
een fo
und to be o
ne of effect
if alternatives to determin
e
sou
r
ce of
informatio
n trustworthin
ess.
Our
re
com
m
endatio
n for
future
re
sea
r
ch i
n
thi
s
do
main i
s
to
a
nalyze
the
correlation
betwe
en attri
butes sele
ction
with a
c
co
unt u
s
er
’
s
ba
ckgro
und, su
ch as
edu
cat
i
on
b
a
ckg
r
ou
nd,
rea
s
on
s to use so
cial medi
a, and the intensity in usi
n
g so
cial medi
a
.
Referen
ces
[1]
J Mat
y
sie
w
i
cz. Cons
umer trus
t – challe
ng
e for e-hea
lthcare.
Mana
ge
me
nt
. 200
3: 337
–3
42
.
[2]
J
T
i
an, J Li.
A
T
r
ust Doma
in-
B
ased R
e
so
urce Selecti
on M
ode
l for Multi-A
gen
t. 20
09: 28
0–2
86.
[3]
J Gong, J Chen, H Deng, J W
ang.
A T
r
ust Model Co
mbi
n
in
g Rep
u
tatio
n
and Cr
ede
nti
a
l
. WASE Int.
Conf. Inf. Eng
.
200
9;
200
8: 63
5–6
38.
[4]
QX Bo Z
,
Yang X.
T
r
ust and
Reputati
on b
a
sed Mo
del S
e
lecti
on Mech
anis
m
for Deci
sion-
mak
i
n
g
.
Secon
d
Int. Conf. Net
w
o
r
ks
Secur. W
i
rel. Co
mmun. T
r
ust.
Comp
ut. 2010:
14–1
7.
[5]
K Curran, S Morrison, SM
Caul
e
y
. Goog
l
e
+
vs F
a
cebook
: T
he Compariso
n
.
T
e
lk
omnik
a
. 201
2;
10(2): 37
9–
388
.
[6]
N Karna, I Su
pria
na, U Ma
ul
idevi. Inte
lli
gen
t Interface for a Kno
w
l
e
dg
e-b
a
sed S
y
stem.
Te
lkom
n
i
ka
.
201
4; 12(4).
[7]
A Devli GJ.
L
o
catio
n
-Aw
a
re
Informatio
n
S
e
rvices
usi
ng
User Profi
l
e
Matchin
g
. 8th International
Confer
ence
on
T
e
lecomu
nicat
i
ons-C
onT
el. 2005: 32
7–
33
4.
[8]
E Raad, R C
hbe
ir, A Dipa
nda.
User Pr
ofile Match
i
ng
in Socia
l
Ne
tw
orks
. 13th Internati
o
n
a
l
Confer
ence
on
Net
w
o
r
k-Bas
e
d Information S
y
stem. 20
1; 97
8-0-7
695-
41
67
-9: 297–
30
4.
[9]
O Peled, M Fire, L Rokach
, Y Elovici. Entit
y
Matchin
g
in O
n
lin
e Soc
i
al
N
e
t
w
o
r
ks.
SocialCom
. 201
3;
53: 339
–3
44.
[10]
Y Gil, V Ratnakar.
T
r
usting In
formati
on S
our
ces One C
i
ti
z
e
n at a T
i
me
. Proceeding First Int. Semant.
W
eb Conf
.
2
002
.
[11]
JM Jiuju
n
C, Yuli
an W
,
Ming L, Antti YJ,
Kuifei Y.
A Ne
w
T
r
ust
Mechanis
m
Bas
ed
On Gravitatio
n
Mode
l of Rep
u
tation Va
lu
e In Socia
l
Netw
ork
. Proc. IC-BNMT
.
2010: 103
5–
103
9.
[12]
V
T
undju
ngsar
i, JE Isti
y
a
nto, E W
i
narko, R W
a
rdo
y
o.
A Reputati
on b
a
sed T
r
ust Mode
l to See
k
Judg
ment in P
a
rticip
atory Group D
e
cisi
on
Makin
g
. Int. Conf. Distrib. Frame
w
. Multimed. Appl
.
201
0.
[13]
K Jung, Y Lee
.
Autono
mic T
r
ust Extraction for T
r
ustw
orth
y Service Disc
o
very in Urb
an
Co
mp
uting
.
Eighth IEEE Int. Conf. Dependabl
e, Auton. Secur. Comput
.
200
9
;
978
–0
–7
695: 50
2–
50
7.
[14]
R Ne
isse, M
W
egdam, M V
an S. T
r
ust
w
o
r
th
iness
an
d Q
ualit
y
of C
onte
x
t Informati
on.
IEEE
. 200
8;
978
–0
–76
95: 1
925
–1
931.
[15]
S Javanm
ardi,
CV Lo
pes. M
ode
lin
g T
r
us
t in Co
lla
bor
ativ
e Informatio
n
S
y
stems.
Evo
l
ution (
N
. Y).
200
7.
[16]
K T
h
irunara
y
a
n
, P Ananth
a
r
a
m, CA Hens
o
n
, AP S
heth.
Some T
r
ust Issues in S
o
ci
al
Net
w
orks a
n
d
Sensor N
e
t
w
or
ks.
IEEE
. 2010
; 978–1
–4
24
4: 573
–5
80.
Evaluation Warning : The document was created with Spire.PDF for Python.