TELKOM
NIKA
, Vol.14, No
.4, Dece
mbe
r
2016, pp. 15
98~160
7
ISSN: 1693-6
930,
accredited
A
by DIKTI, De
cree No: 58/DIK
T
I/Kep/2013
DOI
:
10.12928/TELKOMNIKA.v14i4.4521
1598
Re
cei
v
ed Au
gust 6, 201
6; Re
vised Novem
ber 15, 20
16; Accepted
No
vem
ber 3
0
,
2016
The Strategies for Quorum Satisfaction in Host-to-Host
Meeting Scheduling Negotiation
Rani Meg
asa
ri*
1
, Kuspri
y
a
nto
2
, Emir M. Husni
3
, Dw
i
H. Widy
antoro
4
1
Departem
en P
end
idik
an Ilmu
Komputer F
P
M
I
PA UPI, Jl. Setiabu
dhi
229 B
and
un
g, Indon
esia
1,2,
3,4
ST
EI Institut
T
e
knolo
g
i B
and
un
g,
Jl. Ganeca 1
0
Ban
d
u
ng, Indo
nesi
a
*Corres
p
o
ndi
n
g
author, e-ma
i
l
: megasar
i@u
p
i.ed
u
1
, kuspri
yanto@
ya
ho
o.c
o
m
2
, ehusni
@
y
aho
o.com
3
,
d
w
i@stei.itb.ac.id
4
A
b
st
r
a
ct
T
h
is p
a
p
e
r pr
o
poses
tw
o stra
tegies
for s
a
tis
f
yi
ng th
e
qu
orum of tw
o c
o
ll
i
d
in
g
me
etin
gs
throug
h
host-to-host
n
egoti
a
tio
n
sch
eme. T
h
e
stra
tegy is
to
l
e
t a me
mber attend
th
e
other
me
etin
g u
n
d
e
r
the
cond
ition th
at the gro
up d
e
cis
i
on re
gar
din
g
the sche
d
u
l
e is
not chan
ge
d and
me
etin
g q
uoru
m
is fulfi
l
l
ed,
na
me
ly the
un
assig
n
m
ent str
a
tegy. An
ot
her
strategy is to
s
ubs
titute
th
e
a
b
sent perso
nn
el me
mber in o
r
der
to kee
p
th
e
nu
mb
er
of atten
d
ees
ab
ove t
he
quor
u
m
, n
a
mel
y
the s
ubstituti
on strate
gy. T
h
is p
aper
a
d
a
p
ts
a
mec
h
a
n
is
m d
e
sig
n
ap
proac
h, w
h
ich is Clarke T
a
x
Me
ch
an
i
s
m
,
in
o
r
d
e
r
to
im
pl
em
en
t i
n
ce
n
t
i
v
e
compati
b
il
ity a
nd
ind
i
vid
u
a
l
ra
tiona
lity pr
inci
p
l
e i
n
m
eetin
g s
c
hed
uli
ng. By
usin
g the
u
nas
sign
ment strat
egy
and s
ubstituti
o
n
strategy, the
me
et
in
gs can
still be
hel
d si
mu
ltan
eous
ly a
ccordi
ng to the
sched
ule w
i
th
out
the nee
d for resche
dul
in
g. This pa
per sho
w
s the simula
t
i
on res
u
lt of using the strate
gies w
i
thin so
me
scenar
ios. It d
e
monstrates
that th
e
n
u
m
be
r of
me
eting
failur
e
s c
ause
d
by
unsatisfi
ed
qu
oru
m
c
a
n
b
e
reduc
ed w
i
th h
o
st-to-host ne
g
o
tiatio
n.
Ke
y
w
ords
:
me
etin
g sch
ed
ulin
g, co
nflict h
and
lin
g, ne
goti
a
tion strate
gi
e
s
,
Clarke
Tax mec
h
a
n
is
m, meetin
g
quor
u
m
Copy
right
©
2016 Un
ive
r
sita
s Ah
mad
Dah
l
an
. All rig
h
t
s r
ese
rved
.
1. Introduc
tion
Meeting
sche
duling i
s
incl
u
ded un
de
r the are
a
of pro
b
lem optimi
z
ation whi
c
h p
e
rtain
s
to
uncertainty i
n
sche
dulin
g p
r
oble
m
. An
o
p
timal me
etin
g sch
edul
e in
fluences me
e
t
ing value,
bu
t it
may red
u
ce
some
co
st
s [1]. Instead of
focu
sing
on
the use of m
andato
r
y de
ci
sion to
sel
e
ct
a
timeslot fo
r a
meeting
sch
edule, m
u
ch
resea
r
ch h
a
s bee
n d
one
on n
egotiatio
n
of p
e
rson
n
e
l
prefe
r
en
ce
s to rea
c
h an
optimal sche
dule [2-4
]. Negotiation i
s
processe
d throu
gh several
iteration
s
bet
wee
n
perso
n
nel or pe
rson
nel-to
-
pe
rs
on
nel (P2P) in
orde
r to find a mutual time for
all pe
rsonnel
. The ne
goti
a
tion pe
rmits perso
nnel t
o
upd
ate th
eir p
r
efe
r
en
ces, but
with
no
guarantee
for their true
av
ailability [2, 3]
. The
un
ce
rta
i
nty of pe
rson
nel avail
ability may lea
d
to
a
meeting
sche
duling fail
ure
cau
s
e
d
by th
e veto po
we
r of a p
e
rson
n
e
l memb
er who is ab
se
nt
or
can
c
el
s th
e
meeting
[4]. In order to
manag
e th
e
confli
ct cau
s
ed by
ne
w
meeting
sch
e
dule
s
whi
c
h ca
n ap
pear ove
r
time, designi
ng
negotiatio
n
st
rategie
s
to avoid meeting sche
duling fail
ure
become
s
a challen
g
ing ta
sk.
Meeting
sche
duling
ca
n b
e
viewe
d
a
s
a co
mp
lex d
e
ci
sion m
a
ki
ng proble
m
i
n
artificial
intelligen
ce d
o
main, pa
rticularly in me
chani
sm de
sig
n
pro
b
lem [5]
.
Mecha
n
ism
desi
gn ap
pro
a
ch
has be
en u
s
e
d
in meeting sched
uling to
optimize t
he meeting of so
cial welfa
r
e g
oals [9-11]. Thi
s
approa
ch i
s
based o
n
two pri
n
cipl
es,
namely in
cen
t
ive comp
atib
ility and indiv
i
dual rationali
t
y.
Incentive compatibility
repres
ent
s personnel prefer
ences in bi
dding m
e
chanism
s, whil
st
individual
rationality con
s
i
ders the p
o
ssibility of per
son
nel ab
se
n
c
e an
d calcul
ates its effe
ct on
grou
p d
e
ci
si
on. Cla
r
ke T
a
x Mechani
sm is o
ne
sh
ot voting me
cha
n
ism
that
pro
p
o
s
e
s
a
n
equilibrium
strategy in bi
dding
the value. The mechanism i
s
no
n-mani
pulabl
e. However,
si
nce
the person
nel
have no cont
rol over thei
r role in
the me
eting afterwards, the mech
anism n
eed
s
to
be improved.
Due to the p
o
ssibility of chang
eable p
e
r
so
nnel p
r
efe
r
en
ce
s, this study employs a host-
to-ho
s
t sche
me to obtain
personn
el a
v
ailabilit
y information di
re
ctly.
Every perso
nnel me
m
ber
can
only b
e
sched
uled
by one
ho
st at
one time
an
d
interp
reted
b
y
other
ho
sts as
unavail
a
b
l
e
person
nel
at
that time. Fu
rtherm
o
re, thi
s
schem
e al
so en
able
s
a
host to
ne
got
iate with
oth
e
r
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
The Strategie
s
for Qu
orum
Satisfaction i
n
Ho
st-to-Ho
st Meeting Schedul
i
ng… (Rani Mega
sa
ri)
1599
host
s
in
o
r
de
r to
co
ntinue
assig
n
ing
or una
ssigni
ng
the p
e
rso
n
n
e
l mem
ber,
with o
r
witho
u
t a
sub
s
titute, wh
ich is expl
ain
ed in the next sectio
n.
This
study d
e
termin
es m
eeting failure
by
measu
r
in
g quorum satisfactio
n
. The
existing
meeting
sch
edulin
g ne
go
tiation is
co
ndu
cted in
several time
s in o
r
de
r to
gain
a me
etin
g
sched
ule tha
t
meets the quorum. Nev
e
rthele
s
s, the quorum co
mplian
c
e do
es not hap
p
en
onward
s
be
cause of the dynamic
of p
e
rsonal
sche
dule. This p
a
per p
r
op
ose
s
two negotiat
i
on
strategi
es in
host-to
-h
ost
scheme,
na
mely un
sche
duled
meetin
g ho
st a
n
d
sche
dule
d
me
etin
g
host. The negotiation ai
m
s
to
deci
de
who
will unassi
gn or assign
per
sonnel, in
orde
r to keep the
sched
ule bet
wee
n
two co
nflicted meeti
ngs. Th
e
ne
gotiation stra
tegies a
r
e a
dapted fro
m
the
mech
ani
sm d
e
sig
n
app
roa
c
h, parti
cula
rl
y Clarke Tax Mech
ani
sm (CTm
).
2. Adap
ting Mecha
n
ism Design
Mech
ani
sm d
e
sig
n
or
kno
w
n a
s
reve
rse game the
o
ry is con
c
e
r
ne
d with the m
e
thod to
impleme
n
t a good sy
ste
m
in a wide
range of so
lu
tions that involve multiple self-inte
r
e
s
ted
agent
s. In meeting sch
e
d
u
ling, me
cha
n
ism d
e
si
gn
attempts to
optimize g
r
oup de
ci
sion
in
sele
cting
the
de
sire
d time
to hol
d a
m
eeting,
whe
n
ea
ch
pe
rson
nel me
mbe
r
has hi
s/her o
w
n
prefe
r
en
ce
s.
Mech
ani
sm d
e
sig
n
is
com
m
only used i
n
au
ction
s
, in
particula
r to
bid an
outco
me
by a value.
Assu
ming th
at outcome
s are
po
ssi
ble times for person
nel mem
ber (
) to
hold a
meeti
ng in a
meeti
ng pe
riod
1
,
2
,..
, the bid
d
ing v
a
lue that i
s
called p
e
rson
n
e
l
membe
r
utility not only rep
r
esents
pe
rso
nnel p
r
efe
r
en
ce
s, but al
so
their availa
bili
ty to attend the
meeting in
e
a
ch tim
e
slot.
High
utility denote
s
pe
rs
o
nnel’
s
hig
h
p
o
ssibility to come; conversely,
lowe
r utility in
dicate
s
so
me
co
nflicts with
pe
rs
onal
sch
edule
that m
a
y cau
s
e
pe
rsonnel
ab
se
nce.
By con
s
ide
r
i
ng the
incen
t
ive comp
atible p
r
in
ci
ple,
this
study t
o
talize
s
personnel
memb
er’
s
utilities which
are in th
e ra
nge of
zero t
o
nine
with n
i
ne a
s
the m
o
st p
r
eferabl
e time and vi
ce
versa. The total utility (
), numbe
r of person
nel (
), and personn
el membe
r
’s util
ity in
timeslot
j
(
,
), is denoted
in (1).
∑
,
(1)
The
sele
cted
timeslot i
s
ba
sed
on g
r
o
u
p
deci
s
io
n o
r
social
welfa
r
e.
The
sele
cted
timeslot
(
j*
) o
r
a meeting tim
e
is the timesl
ot which ha
s the maximum
total utility defined in (2
).
∗
a
r
g
∑
,
(2)
This
study
ca
lculate
s
th
e p
i
vot value for
each me
mbe
r
by u
s
in
g
CT
m in o
r
d
e
r to
satisfy
the individual
rationality p
r
inci
ple. CT
m cal
c
ul
ate
s
the maximu
m total utility with perso
n
nel
absen
ce for e
a
ch me
mbe
r
and ea
ch tim
e
slot, su
ch a
s
(3).
,
∑
,
(3)
Whe
n
the me
mber’
s
ab
sen
c
e doe
s not chang
e
∗
as a meeting time,
then the member’
s
pivot is
rep
r
e
s
ente
d
by ze
ro
valu
e. Pivot valu
e de
scrib
e
s t
he m
e
mbe
r
’s pivotal p
r
e
s
ence in
a
me
eting
whi
c
h influen
ce
s the su
ccess of
the meeting or in
previou
s
work calle
d veto powe
r
[6]. T
h
e
absen
ce of a member
wh
o has n
o
ze
ro pivot
value may cause the meeting
sched
ule to be
delayed or canceled and resch
eduli
ng become
s
n
e
cessary.
Re
scheduli
ng me
e
t
ing ca
used
by
the ab
sen
c
e
of pivotal me
mber is
a
co
mmon
wa
y,
but the d
e
ci
sion is ba
se
d
on a
su
bje
c
tive
value. CTm
sho
w
s the p
e
rspe
ctive to explain t
he pivot of a member th
roug
h
the pivot value.
Pivot
value is the difference of
maximum total utility except
in timeslot
and total
personnel’
s
utility except
in
∗
as defined i
n
(4).
Pi
v
o
t
=
∑
,
,
∗
0
,
∑
,
,
∗
(4)
In this s
t
udy, pivot value desc
rib
e
s the
i
n
fluen
ce
of o
ne m
e
mbe
r
’
s
absen
ce
on
a
group
de
ci
si
on.
Therefore, th
e meeting
scheduli
ng ne
e
d
s to up
date
all person
n
e
l’s pivot if the deci
s
io
n is to
una
ssi
gn the
membe
r
. Tab
l
e 1 sho
w
s t
he example of pivot calculati
on.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 14, No. 4, Dece
mb
er 201
6 : 1598 – 160
7
1600
Table 1. Pivot Calculation
Personnel
Personnel’s Utilit
y
in
T
o
tal Utility
except
in
T
o
tal Utility
Except
in
∗
Max
i
mum total utility
except
in
Pivot
9 8
3
0
5
8
20
*22
20
9
20
22
2
0 5
8
5
7
17
*23
17
15
7
23
23
0
2 7
8
6
2
15
*21
17
14
12
21
21
0
6 8
6
9
0
11
*20
19
11
14
20
20
0
T
o
tal utility
in
17 *28
25
20
14
Game the
o
ry
tells abo
ut the domin
ant
strat
egy of
each playe
r
; therefo
r
e, n
o
other
choi
ce f
r
om o
t
her pl
ayers can ma
ke
one
player
get worse. Mea
n
while, me
chani
sm d
e
si
gn
u
s
e
s
the
syste
m
’s perspe
c
tive, desi
gning
a system
that
ca
n ma
ke pl
aye
r
s
reveal th
ei
r true
valuati
o
n
as
a do
mina
nt strate
gy [5
]. Since p
e
rsonnel
ca
n lie
or
manip
u
lat
e
their utility value, this
st
udy
use
s
th
e CT
m to gath
e
r t
he tru
e
p
e
rso
nnel p
r
ef
e
r
en
ce
s. Howeve
r, the utility up
date may
persist
in some
con
d
i
tions, su
ch a
s
the invitation of a hi
gh pri
o
rity meeting
whi
c
h is requ
ested late
r.
Table 2 sho
w
s the mani
pulation valu
e in
by
by
updatin
g 8 with 6. The
cal
c
ulatio
n result sho
w
s
that the ma
nipulati
on
do
es not
cha
n
ge or influ
e
n
c
e the pivot
of
. The out
com
e
de
sired by
CTm i
s
n
o
t c
ontrolle
d by p
e
rson to
pe
rson but
depe
n
d
s o
n
the
so
cial w
e
l
f
are
. The red
numbe
rs in Table 2 are up
dated value
s
from Table 1.
Table 2. Utility Update by Membe
r
1
Personnel
Personnel’s Utilit
y
in
T
o
tal Utility
except
in
T
o
tal Utility
Except
in
∗
Max
i
mum total utility
except
in
Pivot
9 6
3
0
5
8
20
*22
2
0
9
20
22
2
0 5
8
5
7
17
*21
17
15
7
21
21
0
2 7
8
6
2
15
*19
17
14
12
19
19
0
6 8
6
9
0
11
18
*19
1
1
1
4
18
19
1
T
o
tal utility
in
17 *26
25
20
14
This pa
per a
dapts tax val
ue in
o
r
igin
a
l
CTm
in
o
r
der to d
e
si
g
n
pe
rson
nel’
s
pivotal
influen
ce on
gro
up de
cision.
Fu
rthermore, pivo
t
values are u
s
ed
in
negot
iation
strateg
i
e
s
prop
osed in t
h
is
study whi
c
h a
r
e
com
p
atible with
th
e ho
st-to-ho
st sch
eme. Th
e strate
gie
s
are
defined
as t
he un
assig
n
m
ent strateg
y
and sub
s
titution strateg
y
. The de
sig
n
of neg
otiation
strategi
es ai
ms to sati
sfy meeting qu
orum.
2.1. Unassig
n
ment Str
a
te
g
y
In the real meeting sche
d
u
ling, a meeti
ng can
still be held on schedul
e, although it is
not attended
by all membe
r
s. Conve
r
sely, the negot
ia
tion made
by other
schem
es i
s
a cy
cle
of
con
s
trai
nts or prefere
n
ce relaxation
s un
til all t
he me
mbers meet a
t
the same available time. This
study rai
s
e
s
t
he qu
orum v
a
riabl
e a
s
the
con
s
e
quen
ce to determin
e
meeting
su
ccess d
ue to
the
possibility of a conflict. In this stu
d
y, the amount
of perso
nnel avail
ability is nego
tiable as lon
g
as
it
sat
i
sf
ie
s t
h
e quo
rum.
A
c
cor
d
ingly
,
b
o
t
h
co
lli
ding
m
eeting
s
, whi
c
h are sch
edu
led me
eting
and
unsch
edule
d
meeting, ca
n still be held
si
multaneo
usly
.
Since the piv
o
t value in CTm indicat
e
s
person
nel infl
uen
ce on g
r
o
up deci
s
io
n, the use
of this metho
d
in meeting
sched
uling e
nable
s
a p
e
rsonnel m
e
mbe
r
’s a
b
sen
c
e whose pivot valu
e
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
The Strategie
s
for Qu
orum
Satisfaction i
n
Ho
st-to-Ho
st Meeting Schedul
i
ng… (Rani Mega
sa
ri)
1601
is
zero. The
strategy is call
ed the
una
ssi
gnment
strategy. This pa
p
e
r d
e
fine
s m
eeting failu
re
as
the co
ndition
whe
n
a m
e
eting do
es n
o
t meet
qu
orum an
d ne
e
d
s to
be
re
sche
dule
d
. Firstly,
meeting failure is caused by the pi
votal
personnel unavailability; t
herefore, the social
welfare or
grou
p de
ci
sio
n
sh
ould b
e
moved to an
o
t
her time
slot. Secon
d
ly, it is cau
s
ed by t
he un
availabi
lity
of person
nel
with ze
ro pi
vot but cau
s
i
ng t
he num
b
e
r of attende
es to not me
et the quoru
m
.
Based
on
Ta
ble 1, if pe
rsonnel
is in
confli
ct ca
used by
a ne
w meeting assi
gnment,
then his/
her absence in
the sc
hed
u
l
ed meetin
g
will ca
use
sched
uling
meeting fai
l
ure.
Mean
while, whe
n
pe
rson
nel
is in
conflict, the sy
stem
will
check the quorum before
una
ssi
gning t
he memb
er.
2.2. Substitu
tion Stra
tegy
The utilizatio
n of pivot val
ue in CTm p
r
ompt
s this
st
udy to propo
se an
other n
egotiatio
n
strategy, whi
c
h is the su
bst
i
tution strateg
y
. T
he idea comes from a
dding p
e
rson
nel scen
ario
s as
descri
bed in
Table 3.
is an
additional m
e
mbe
r
to sub
s
titute
, thus the utility in
sele
ct
ed t
i
me
(
) take
s the same value a
s
that of
.
Table 3. Pivot Recalculatio
n for Addition
al Membe
r
Personnel
Personnel’s Utilit
y
in
T
o
tal Utility
except
in
T
o
tal Utility
Except
in
∗
Max
i
mum total utility
except
in
Pivot
9 8
3
0
5
8
*28
22
20
9
28
28
0
0 5
8
5
7
17
*31
17
15
7
31
31
0
2 7
8
6
2
15
*29
17
14
12
29
29
0
6 8
6
9
0
11
*28
19
11
14
28
28
0
0 8
0
0
0
17
*28
25
20
14
28
28
0
T
o
tal utility
in
17 *36
25
20
14
I
n
case of
conflict, the system sug
g
e
s
ts to add personn
el
. CTm
recalculation sho
w
s
that
wi
ll be free b
e
cause of the zero pivot an
d
accordingly
can b
e
una
ssi
gned
to attend
oth
e
r m
eeting.
This
stra
tegy
is
used in
o
r
de
r to
hold t
he
schedul
e
and
avoid re
sche
duling. Sub
s
t
i
tution strate
gy is su
ch
a
sec
ond
-tier
strategy
to p
r
event meeti
n
g
s
from failure. Finding
a proper
su
bstitu
te is t
he ne
xt challeng
e, beginni
ng
with re
se
arch i
n
comm
unity detection p
r
obl
em [7] and si
milarity
detection probl
em [8] about personnel p
r
ofile.
3. Host-to-Host-Objec
tiv
e
s
The personnel’s utility is gathered by using
host
-to-host scheme.
Th
is part expl
ains the
use of
CTm i
n
a ho
st-to-h
o
st meetin
g sche
duling
ne
gotiation, whi
c
h in
clud
es in
put, pro
c
e
ss,
and
output. There
are thre
e intention
s
of assemb
li
ng me
eting sche
dul
ing in a cent
ralized man
n
e
r
and exe
c
utin
g host-t
o
-h
ost
negotiation t
o
handl
e mee
t
ing confli
cts.
3.1. Complete Information
Meeting
sch
edulin
g re
se
arch is
relate
d to resea
r
ch on Group
ware
Cale
nda
r System
(G
CS) [14
-
1
6
]. GCS
coll
ects pe
rsonn
el sche
dul
e
s
in orde
r to
arrang
e ev
ery ne
w m
e
eting
without a
n
y conflict. Ho
we
ver, GCS d
e
velopme
n
t
had
a usability problem. Thi
s
study examine
s
the probl
em by using thre
e perspe
c
tive
s, namel
y pe
rso
nnel a
s
in
dividual
s, meeting host wit
h
so
cial focus,
and technol
o
g
y [9], as illustrated in Figu
re 1.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 14, No. 4, Dece
mb
er 201
6 : 1598 – 160
7
1602
Figure 1. GCS’ Problem
1.
To avoid
sch
edule
confli
ct, each m
e
mb
er of
the p
e
rsonnel m
u
st
share
his
or h
e
r sch
edul
e to
the meetin
g
host th
rou
g
h
GCS,
whi
c
h
unfortu
natel
y raises priv
acy i
s
sue
(e
dge 1
)
. Thi
s
probl
em ha
s
been me
ntion
ed by previou
s
re
sea
r
ch [5, 6].
2.
In a so
cially-centere
d
pe
rspective, a ho
st meet
ing
ha
s a difficulty i
n
determining
prio
rity level
of personn
el attendan
ce b
e
ca
use the utility
val
ue wh
ich re
pre
s
e
n
ts perso
nnel
prefe
r
en
ce i
s
cryptic
(ed
ge
2) [10]. In addition, person
nel
availabilit
y can ch
ang
e
any time due to a higher
prio
rity of afte
r meeting
sch
edule.
3.
Every organi
zation may h
a
ve its own
GCS.
Since
every memb
er of the person
nel can
b
e
invited
by
several org
ani
zat
i
ons, s/he mu
st
s
hare h
e
r/
his
sched
ule
repe
atedly, which
ca
use
s
relu
ctan
ce
of usi
ng
GCS
(edge
3).
The
n
, to get
pe
rson
nel
sche
d
u
le
compl
e
tel
y
, a meetin
g
host d
uplicates oth
e
r h
o
st
s to its
syste
m
and n
egot
i
a
te with its p
e
rsonn
el [11]. Ho
wever, th
i
s
method i
s
n
o
t
effective be
cau
s
e
perso
n
nel utilit
y in meeting i
s
still fluctuating
until the du
e
time.
The thre
e poi
nts mentio
ne
d in Figu
re 1
descr
i
b
e
s
the main p
r
obl
em of GCS,
whi
c
h i
s
incom
p
lete in
formation. T
h
e inform
ation
about p
e
rso
nnel
sched
ul
e may be lef
t
blank
or
n
o
t
blan
k but cry
p
tic. Blank uti
lities are eith
er be
cau
s
e o
f
the perso
nn
el relu
ctan
ce to share or th
eir
unknown situations
regardi
ng the future
sched
ule. On the other hand,
crypti
c utility comes from
their
difficulties to reveal t
heir convenient utility.
Without complete
inform
ation, optimal
schedule
is difficult to achieve.
This
s
t
udy at
tempts
to find the s
o
lution
of inco
mpl
e
te inform
ation problem i
n
meeting
scheduling.
When personnel utilit
y deli
v
ered
by each member
of
the personnel is
suspected to
be the ro
ot proble
m
, assig
n
ment inform
ation from
ea
ch me
eting h
o
st ca
n be g
a
there
d
as
a
n
alternative scheme to get
the personnel member’
s
utility information. This
scheme is named
host-to
-h
ost schem
e and
h
a
s bee
n
m
ent
ioned
in pr
evi
ous work [1
2, 13]. Th
e p
e
rson
nel m
e
mb
er
become
s
p
a
ssive
by onl
y receiving t
he me
eting
i
n
vitation, whi
l
st a m
eeting
host i
nput
s
the
meeting
re
qu
ests which a
r
e d
e
livere
d
by syst
em
a
s
p
e
rson
al
sche
dule
s
if
GCS m
eets
the
approp
riate timeslot.
Ho
st-to-ho
st
meeting
sche
duli
ng involves multiple
meetings
from multiple
organizations
su
ch a
s
sho
w
n by the
previous
wo
rk
[14, 15]. GCS detect
s
m
eeting
confli
cts if the invited
person
nel ha
ve been sch
e
duled by a p
r
evious me
et
ing. Assuming
that every meeting tend
s
to
invite person
nel u
s
ing thi
s
G
C
S sche
me, meeting
sched
uling
can
use com
p
lete informa
t
ion
without th
e ri
sk of p
e
rson
nel relu
ctan
ce to
u
s
e
G
C
S. Theref
ore,
perso
nnel
o
n
ly get me
eting
invitation that is free f
r
om
confli
ct. By the
time the
new m
eetin
gs em
erge,
every ho
st can
comm
uni
cate
to each othe
r to handle an
y conflict and
negotiate thei
r con
s
traints.
3.2. Stable Scheduling
Personn
el p
r
eferen
ce
of
meeting
dep
end
s o
n
the
existing
situa
t
ion. The
con
f
lict may
come
over ti
me and influ
e
n
ce the
pe
rso
nnel ava
ila
bili
ty in meeting sched
uling. E
v
en though th
e
CTm
has been proved to be a
non-manipul
able mechani
sm
[
16
], utility is not only
about
person
nel preferen
ce, bu
t also perso
nnel ava
ilabi
lity or conflict. Since the prob
ability of
schedule
conflicts emerges in un
certainty, a meeting schedule
w
ill never be fixed until the
due
time.
This stu
d
y d
e
sig
n
s a m
e
eting
sched
u
ling
to pro
d
u
c
e a
m
o
re
stable sched
u
ling
by
adaptin
g piv
o
t value fro
m
the CT
m
in a ho
st-t
o-ho
st ne
got
iation. The
aim is to
a
v
oid
resch
eduli
ng
whe
n
confli
cts occu
r after a meet
ing ha
s bee
n sched
uled and p
ubl
ishe
d. There
are
some
strategi
es introdu
ce
d
in this p
ape
r that
can
be
cho
s
e
n
for eit
her u
n
sch
edu
led meetin
g o
r
sched
uled m
eeting.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
The Strategie
s
for Qu
orum
Satisfaction i
n
Ho
st-to-Ho
st Meeting Schedul
i
ng… (Rani Mega
sa
ri)
1603
3.3. Win-
w
i
n
Solution
The ne
gotiati
on strategie
s
prop
osed in
this
pap
er a
r
e inspi
r
e
d
by the confli
ct handli
n
g
mode [12]. To implement the strate
gie
s
, this pape
r ra
ise
s
a variabl
e in meeting sched
uling wi
t
h
the ho
st-to-h
o
st ne
gotiatio
n
sc
hem
e, n
a
mely quo
ru
m. This vari
a
b
le is atta
ch
ed to a me
eting
role, whi
c
h i
s
defined b
e
lo
w:
(5)
Quorum o
r
minimum
nu
mber
of atte
ndee
s i
s
u
s
u
a
lly use
d
to
define m
eeti
ng succe
s
s
whe
n
a
meeti
ng h
a
s
bee
n
held. Q
uorum
is n
o
t u
s
ed
to define
a m
eeting
sched
ule which i
s
not
su
ccessful ye
t. In a host
-
to
-ho
s
t sch
e
me
, quorum vari
able
can
be
use
d
in m
eet
ing sch
eduli
n
g
negotiatio
n
proce
s
s to rea
c
h a win-win
solution w
hen
a confli
ct occurs a
s
sho
w
n
in Figure 2.
Figure 2. Meeting Sche
dul
ing with Host
-to-Ho
st Ne
go
tiation Schem
e Flowcha
r
t
Neg
o
tiation i
s
tri
gge
red
b
y
perso
nal
sche
dul
e
confl
i
ct. The
confl
i
ct is dete
c
te
d after
meeting schedule is selected.
The syst
em will check the pers
onnel availabilit
y based on t
heir
person
a
l cale
ndar to d
e
termine the me
mber
w
ho m
a
y not attend the meetin
g. Furthe
rmo
r
e
,
confli
ct handli
ng is called n
egotiation
wh
ose p
r
o
c
ed
ure is explain
e
d
below:
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 14, No. 4, Dece
mb
er 201
6 : 1598 – 160
7
1604
Procedur
e
Host-to
-
Host Negotiation
Input:
m
a
yNotAttend, unsche
dule
d
Mee
t
ing, sch
edul
edMeetin
g
Outpu
t:
atten
dee_
um
, atte
ndee
_sm
nm
gets nu
mbers of
m
a
yNotAttend
pe
rso
nnel
na_um
get
s numb
e
rs of attendee
s in
unsch
edule
d
M
eeting
attendee
_um
gets a li
st of attendee
s in
unsch
edul
e
d
Meeting
na_
sm
gets numb
e
rs of attendee
s in
sched
uledM
e
e
ting
attendee
_sm
gets a li
st of attendee
s in
sched
uled
Meeting
quo_
um
gets
unsch
edul
edMeetin
g
qu
orum
quo_
sm
gets
sched
uled
Meeting
qu
orum
for
i=1, i
<
=
nm
,
i++
pivot_um
get
s pivot
may
N
otAttend(i)
in
u
n
sch
e
duledM
eeting
if
pivot_um
= 0 and
n
a_u
m
- 1 >
quo_
um
attendee
_um
u
p
d
a
te attendee,
una
ssi
gn me
mber
m
a
yNot
Attend
(i)
in
un
sched
uledM
e
e
ting
na_um
na_um
-1;
else
pivot
_
sm
gets pi
vot
m
a
yNotAttend(i
)
in
sch
edule
d
Meetin
g
if
pivot_sm
= 0 an
d
na_
sm
- 1 >
quo
_sm
a
ttendee_
sm
update att
ende
e, una
ssign
m
a
yNotAt
tend
(i)
in
sche
dule
d
Meeting
n
a_sm
na_
sm
-1;
else
subtituted_um
find subtituted
m
a
yNotAttend(i
)
in
un
sche
dule
d
Mee
t
ing
i
f
subtituted
_u
m
exis
ts
attendee
_
um
update attendee
_um,
subtitute
m
a
yNotAttend
(i)
with oth
e
r personnel
else
subtituted_sm
fin
d
subtituted
may
N
otAttend(i)
in
sched
u
l
edMeetin
g
if
subti
t
uted_sm
exis
ts
atte
ndee
_
sm
update atten
d
ee_
sm, sub
s
t
i
tute member
may
N
otAttend
(i) with oth
e
r perso
nnel
end
e
nd
end
end
end
Neg
o
tiation strategie
s
sele
cted by un
schedul
ed meet
ing or
sched
u
l
ed meeting i
s
ba
sed
on the
meeti
ng’s statu
s
. T
he
strategi
es
are
de
si
gn
ed
to keep
the
numbe
r
of m
eeting
attend
ees
satisfying the
meeting qu
orum.
4. Simulation Resul
t
s
Since
this st
udy u
s
e
s
a
differe
nt schem
e
of n
e
gotiation, a
new me
asu
r
ement i
s
proposed to
define meeting sch
eduling success. T
he objective
is to fulfill meeting quorum;
otherwise, the meeting wil
l
fail and need to be resch
edule
d
as a
new me
eting requ
est. Mee
t
ing
failure (F
) is d
e
fined a
s
(2
).
∈
,
if
numberof
att
endees
q
uorum
,
if
numberof
at
t
endees
(6)
Meeting failu
re measurem
ent simulatio
n
in this stud
y takes 4 sce
nario
s, whi
c
h
are the
combi
nation
of
different calend
ar den
si
ties
a
nd
quo
rums
as de
picted in Fi
gu
re
3.a., 3.b., 3.c.,
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
The Strategie
s
for Qu
orum
Satisfaction i
n
Ho
st-to-Ho
st Meeting Schedul
i
ng… (Rani Mega
sa
ri)
1605
and 3.d.
Cal
enda
r de
nsity
is the o
c
cup
ancy of
p
e
rsonnel
sche
du
le in the
ran
g
e
s of 2
0
%-5
0
%
and 20%-80. Mean
while, q
uoru
m
or the
numbe
r of mi
nimum attend
ees in ea
ch
role used in this
study i
s
bet
ween
50% a
n
d
80%. Every
meeting
i
n
vites between 5
-
25 pe
rso
nnel
memb
ers
a
n
d
a
perio
d take
s
14 days
with 8 times a day
. This simu
l
a
tion aims to show the pe
rfo
r
man
c
e of ho
st-
to-ho
s
t negoti
a
tion (H2H) compa
r
ed to p
e
rsonn
el-t
o
-
p
e
rsonn
el neg
otiation (P2P) which d
o
e
s
not
use q
uorum a
nd req
u
ires m
eeti
ng a
c
cept
ance by all personn
el.
Figure 3
(
a)
a
nd Fig
u
re
3(c) sho
w
almo
st t
he sam
e
n
u
mbe
r
of fail
ed me
eting
s
by usin
g
P2P negotiati
on ca
used by the same ca
lenda
r den
sit
y
setting of 20%-50%. Similar rate i
s
also
sho
w
n
in Fi
g
u
re
3(b)
and
Figure 3
(
d
)
, b
u
t with
a
hig
h
e
r
n
u
mb
er of failed
me
etin
gs ca
used by
a
den
ser calen
dar,
50%-8
0
%
. Since th
e
neg
otiation
doe
s n
o
t u
s
e
the q
u
o
r
um,
the
re
sult of
P2P
negotiatio
n
i
s
n
o
t influe
nce
d
by
qu
orum
setting
. In anoth
e
r ca
se,
H2
H neg
otiation
ha
s
decrea
s
e
d
th
e num
be
r of
failed me
etin
gs i
n
P2P
ne
gotiation. Th
e re
sult
of
H2H
neg
otiatio
n
is
influen
ced m
o
re by cal
end
ar den
sity an
d followe
d by quorum setting.
Figure 3(a
)
. Numb
er of fai
l
ed meeting
s
with
20%-5
0% cal
enda
r den
sity and 50% qu
orum
Figure 3(b
)
. Numb
er of fai
l
ed meeting
s
with
20%-8
0% cal
enda
r den
sity and 50% qu
orum
Figure 3(c). Numbe
r
of failed meeting
s
wi
th
20%-5
0% cal
enda
r den
sity and 80% qu
orum
Figure 3(d
)
. Numb
er of fai
l
ed meeting
s
with
20%-8
0% cal
enda
r den
sity and 80% qu
orum
The si
mulation is built for
multiple meeti
ngs from diff
erent o
r
g
ani
zation environ
ments.
The n
u
mbe
r
of meeting
failure
s in
crea
ses a
s
th
e
n
u
m
ber
of me
etings incre
a
se
s. Figu
re
4
sh
ows
the simulatio
n
results wit
h
the additio
nal cate
gory
of 50-80%
calend
ar de
nsity. The meeting
failure
i
s
an a
v
erage of
30 iteration simu
lations.
Th
e simulation aim
s
to qu
antify the influe
nce
of
different co
m
p
lexity of the
environ
ment
by ca
len
dar d
ensity and q
u
o
rum
setting.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 14, No. 4, Dece
mb
er 201
6 : 1598 – 160
7
1606
Figure 4. Meeting Failu
re
with Some Scenari
o
s
Based
on
Fi
gure
4, the
meeting
sch
edulin
g with
H2
H ne
gotiat
i
on ha
s fe
we
r failed
meeting
s
. Its number of
meeting failures is a
ffecte
d
by meeting quorum
with increa
se
s
bel
ow
30% for three
sce
nari
o
s. O
n
the other h
and, meet
ing
failure
s incre
a
se for a
ppro
x
imately 30% for
different cale
ndar
den
sity scena
rio
s
. Since q
u
o
r
um
setting relate
s to the una
ssignm
ent stra
tegy
and calen
dar
den
sity relate
s to the sub
s
titution
strate
g
y
, acco
rdin
gly the H2
H neg
otiation ca
n b
e
optimize
d
by
setting the i
n
vited person
n
e
l memb
er
o
r
by optimizin
g the search
for su
bstitute
d
person
nel.
Since q
u
o
r
u
m
is set by a ho
st, meeting failure ca
n be avoi
de
d by usi
ng t
he loo
s
e
quorum. Ho
wever,
cale
n
dar d
e
n
s
ity whi
c
h si
gnifi
cantly influen
ce
s the failu
re of a mee
t
ing
depe
nd
s on
other m
eetin
gs. The
r
efo
r
e, negotiatio
n
strate
gie
s
are im
porta
nt to manag
e
the
confli
ct. This study descri
bes the u
s
e
of negotia
tio
n
strate
gie
s
to examine their u
s
ability
in
different scal
ability. Meeting sche
dulin
g without
ne
gotiation in simulation ref
e
rs to the ot
her
scheme n
e
g
o
tiations. Since this
stu
d
y uses
th
e host-to
-ho
s
t negotiatio
n
sch
eme, the
comp
ari
s
o
n
to other
sch
e
m
es mu
st be
elabo
rated.
T
he future works
will explain
about it.
5. Conclusio
n
T
h
is s
t
ud
y e
x
p
l
a
i
ns
th
e use
o
f
th
e un
ass
i
g
n
m
ent
stra
tegy and
sub
s
titution
strat
egy in
a
host-to
-h
ost
negotiatio
n
scheme for sche
dulin
g
a meeting. The variab
le of quoru
m
is
con
s
e
que
ntly propo
sed
to
gain
wi
n-wi
n solution
be
tween
conflicted me
etings; therefo
r
e, n
o
t
every co
nflict
can
cau
s
e
meeting failu
re. Based o
n
the sim
u
lation
results, cale
ndar
den
sity has
signifi
cant inf
l
uen
ce on
b
o
th meeting
scheduli
ng
without ne
go
tiation and
with neg
otiation.
Mean
while, changi
ng the q
uoru
m
co
nst
r
aint only in
fluences m
eetin
g sched
uling
with neg
otiation
becau
se in
meeting
sch
edulin
g with
o
u
t negotiati
o
n
one
co
nflict immediately
cau
s
e
s
m
e
e
t
ing
failure.
Ackn
o
w
l
e
dg
ement
This research
is funded by
Ministry of Re
sea
r
ch, Tech
nology an
d Hi
gher Ed
ucation of
the Rep
ubli
c
of Indone
sia. The autho
rs woul
d
like to than
k them for providing the
funding.
Referen
ces
[1]
P Marshal
l, RI W
h
itfield, A Du
ffy
, M Haffe
y
,
S W
a
ter. A new
mo
de
l for hig
h
valu
e meeti
n
gs. 2015.
[2]
S Junuzovic, P
De
w
a
n.
Towar
d
s self-optim
i
z
i
ng c
o
llaborativ
e system
s
. Pro
c
eed
ing
of th
e
ACM 2
0
1
2
confere
n
ce o
n
Comp
uter Sup
ported C
o
o
per
ative W
o
rk. 20
12.
[3]
B Marcino
w
s
k
i.
F
l
exi
b
l
e
meeti
ng sche
d
u
lin
g. 201
4.
[4]
EM Shakshuk
i
,
SM Hossain
. A persona
l meetin
g sche
duli
ng
age
nt.
Person
al a
n
d
ubiq
u
ito
u
s
computi
ng.
20
14;
18(4): 9
09-
922.
[5]
H Le
e. Yo
ur ti
me a
nd m
y
tim
e
: a tem
pora
l
appr
oach
to
gr
oup
w
a
re
cal
e
n
dar s
y
stems.
I
n
formation &
Mana
ge
me
nt.
200
3;
40(3):1
5
9
-16
4
.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
The Strategie
s
for Qu
orum
Satisfaction i
n
Ho
st-to-Ho
st Meeting Schedul
i
ng… (Rani Mega
sa
ri)
1607
[6]
G Leshe
d, P Seng
ers.
I lie to myse
lf that i h
a
ve free
do
m in
my ow
n sche
dul
e: prod
uctivi
ty tools an
d
exper
ienc
es of
busyn
ess
. Pro
c
eed
ings
of th
e SIGCHI Con
f
er
ence
on
Hu
man F
a
ctors i
n
Comp
uting
S
y
stems. 20
11
.
[7]
E Cra
w
f
o
rd, M Veloso.
Mecha
n
is
m des
i
gn for mu
lti-a
gent meeti
n
g
schedu
lin
g i
n
clu
d
in
g time
prefere
n
ces, a
v
aila
bi
lit
y, and
valu
e of prese
n
ce
. IEEE/WIC
/
ACM Inter
national Conf
erenc
e
.
2004.
[8]
SJ Russe
ll, P
Norvig, JF C
a
nn
y, JM Mal
i
k, DD Ed
w
a
rds.
Artificial i
n
tel
lig
ence: a m
o
d
e
r
n
ap
proac
h.
Ne
w
Jers
e
y
: Pr
entice h
a
ll. 2
0
0
3
: 679-6
88.
[9]
E Ephrati, G Zlotkin, JS Ros
ensch
ein.
A non-manipulable
m
e
eting scheduling system
. Procee
din
g
s
of the 13th inte
rnatio
nal
w
o
rks
hop o
n
dist
ri
bu
ted artificia
l
int
e
lli
ge
nce. 19
94
.
[10]
J Cook. Mecha
n
ism Desi
gn a
nd Veto Mec
h
a
n
isms for Sequ
entia
l Meetin
g
Sched
uli
ng. 20
07.
[11]
A Grubshtei
n, A Meise
l
s.
Cost of coo
p
e
ratio
n
for sc
hed
uli
ng
meet
ings
. Intelligent Distributed
Computing III.
Sprin
ger. 2009: 227-236.
[12]
QW
Lishuo
Zhang.
A
Communit
y
Detection Algor
ithm Based on NSGA-II.
TELKOMNIKA
T
e
leco
mmunic
a
tion C
o
mputi
n
g Electron
ics a
nd Co
ntrol.
20
16; 14(3A): 2
8
8
-29
6
.
[13]
YHYN Ju
n Li.
A Similar
i
t
y
Detectio
n Met
hod B
a
se
d on
Distanc
e Mat
r
ix M
ode
l
w
i
th
Ro
w
-
C
o
lum
n
Order pen
alt
y
F
a
ctor.
Bulletin
of Electrical E
ngi
neer
in
g and
Informatics.
2
014; 3(4).
[14]
J Grudin. Grou
p
w
ar
e an
d soc
i
al
d
y
nam
ics: e
i
ght ch
all
e
n
ges
for deve
l
o
pers
.
Commun
i
cati
ons of th
e
ACM.
1994; 3
7
(
1): 92-10
5.
[15]
SP McKech
nie
,
JE Beatt
y
,
et
al. C
ontemp
o
r
a
r
y
ca
len
dar
mana
geme
n
t: E
x
plori
n
g
the
inte
rsections
o
f
grou
p
w
ar
e an
d
person
a
l c
a
le
ndars.
Ma
nag
e
m
e
n
t revu
e, Socio-
econ
o
m
ic
Studies.
2
0
1
5
; 26(3): 18
5-
202.
[16]
L
Pal
en.
Soc
i
al, indiv
i
dual
and te
chnological issues for gr
oupware calendar system
s
. Pro
c
eed
ings
o
f
the SIGCHI conferenc
e on H
u
man
factors in computing s
ystems:
the CHI is the limit. 19
99.
[17]
F
W
ang. Adap
tive meetin
g s
c
hed
uli
ng for l
a
rge-sc
ale
dist
ribute
d
gro
u
p
w
are.
BT
techno
logy j
our
nal.
200
3; 21(4): 13
8-14
5.
[18]
B Marcino
w
s
k
i.
F
l
exi
b
l
e
meeti
ng sche
d
u
lin
g. 201
4.
[19]
R Meg
a
sari, K
Kuspri
ya
nto,
EM Husni,
DH
W
i
d
y
antor
o. T
o
w
a
rds
host-to-host me
etin
g
sched
uli
n
g
neg
otiati
on.
Internati
ona
l Jour
nal of Adva
nc
es in Intelligent Informatics.
20
15; 1(1): 23-2
9
.
[20]
PJ Modi, M V
e
los
o
.
Bu
mpi
n
g strategi
es fo
r the
multi
a
g
e
n
t agre
e
m
e
n
t prob
le
m
. Proc
eed
ings
of the
fourth intern
ati
ona
l joi
n
t confe
r
ence o
n
Auto
n
o
mous a
g
e
n
ts and mu
ltiag
ent
s
y
stems. 200
5
.
[21]
O Korjus, et al.
Meetin
g sche
d
u
lin
g ass
i
stan
t, automatic sch
edu
lin
g bet
w
e
e
n
heter
oge
ne
o
u
s cale
nd
ar
s
y
stems. 20
12.
Evaluation Warning : The document was created with Spire.PDF for Python.