Quantitative evaluation system for multiagent grouping and method thereof
View Patent ↗Disclosed is a quantitative evaluation system and method for multiagent grouping, which enables agents to be grouped in such a manner as to ensure efficient agent cooperation. The system comprises: a plurality of agents in a network-connected multiagent system, each of which agents has quantitative information on a predicate that is a type of information to be produced or consumed, and produces or consumes one or more types of information; an adjusting means for receiving the quantitative information from each of the agents, performing quantitative evaluation thereof, and then grouping the plurality of agents into one or more communities; and a message broker means for transferring the information gathered from the plurality of agents to the adjusting means.
1. A quantitative evaluation system for multiagent grouping, comprising:
a processor and a memory;
a plurality of agents in a network-connected multiagent system, each of which agents has quantitative information on a predicate that is a type of information to be produced or consumed, and produces or consumes one or more types of information, wherein the quantitative information includes an RQ (Request Quality) value corresponding to quality of information on the predicate and a TR (Time Resolution) value corresponding to quantity of information created for the predicate;
an adjusting means for receiving the quantitative information from each of the agents, performing quantitative evaluation of the received quantitative information after all of the quantitative information from the plurality of agents is received by using any one of UoP (Utility of Predicate), UoA (Utility of Agent) and UoC (Utility of Community), estimating one or more communities for the plurality of agents according to a result of the quantitative evaluation, and then grouping the plurality of agents into the one or more communities; and
a message broker means for transferring the information gathered from the plurality of agents to the adjusting means.
2. The system of claim 1 , wherein the adjusting means carries out grouping of the plurality of agents by executing calculation of the following equation:
CO
global
=
{
w
❘
max
(
∑
k
=
w
UoC
w
·
Cwg
w
)
}
where CO global denotes a universal set of all the communities within the system, w denotes a string representing an agent set sequence in which a multiplication of UoC and Cwg is maximized, k denotes a constant, UoC w denotes the UoC of community w, and Cwg w denotes the weight which community w has.
3. The system of claim 2 , wherein the Cwg is calculated by executing the following equation:
Cwg
c
=
∑
j
=
Mco
c
∑
k
=
all
predicate
rq
k
j
·
tr
k
j
∑
j
=
Mco
global
∑
k
=
all
predicate
rq
k
j
tr
k
j
where j denotes a variable, Mco c denotes a set of agents belonging to community c, rq k j denotes an RQ value for the predicate of agent j, and tr k j denotes a TR value for the predicate of agent j.
4. The system of claim 3 , wherein the UoCw is calculated by executing the following equation:
UoC
c
=
∑
k
=
Mco
w
UoA
k
c
·
Awg
k
c
where c denotes a variable representing an ordinal number of community c, UoA k c denotes the UoA of agent k belonging to community c, and Awg k c denotes an influence which agent k exercises in community c.
5. The system of claim 4 , wherein the Awg k c is calculated by executing the following equation:
Awg
a
c
=
∑
k
=
all
predicate
rq
k
a
·
tr
k
a
∑
j
=
Mco
c
∑
k
=
all
predicate
rq
k
j
·
tr
k
j
where a denotes a variable, rq k a denotes an RQ value for the predicate of agent a, and tr k a denotes a TR value for the predicate of agent a.
6. The system of claim 5 , wherein the UoA k c is calculated by executing the following equation:
UoA
k
c
=
(
∑
k
=
all
predicate
UoP
k
a
,
c
∑
k
=
all
predicate
rq
k
a
·
tr
k
a
×
100
+
100
)
2
where UoP a,c k denotes the UoP which agent a belonging to community c satisfies for predicate k.
7. The system of claim 6 , wherein the UoP a,c k is calculated by executing the following equation:
UoP
p
a
,
c
=
{
∑
k
=
MtrId
p
c
-
rq
p
k
·
tr
p
k
(
if
tr
p
a
>
0
,
∑
k
=
MtrId
p
c
-
rq
p
k
·
tr
p
k
≥
rq
p
a
·
tr
p
a
)
∑
k
=
MtrId
p
c
-
rq
p
k
·
tr
p
k
-
(
rq
p
a
·
tr
p
a
∑
k
=
MtrId
p
c
-
rq
p
a
·
tr
p
a
)
(
if
tr
p
a
>
0
,
∑
k
=
MtrId
p
c
-
rq
p
k
·
tr
p
k
≥
rq
p
a
·
tr
p
a
)
rq
p
a
·
tr
p
a
-
(
∑
k
=
MtrId
p
c
+
rq
p
k
·
tr
p
k
-
rq
p
a
·
tr
p
a
)
(
if
tr
p
a
<
0
,
∑
k
=
MtrId
p
+
rq
p
k
·
tr
p
k
≥
rq
p
a
·
tr
p
a
)
∑
k
=
MtrId
p
c
+
rq
p
k
·
tr
p
k
-
(
rq
p
a
·
tr
p
a
-
∑
k
=
MtrId
p
c
+
rq
p
k
·
tr
p
k
)
(
if
tr
p
a
<
0
,
∑
k
=
MtrId
p
c
+
rq
p
k
·
tr
p
k
<
rq
p
a
·
tr
p
a
)
0
(
if
rq
p
a
·
tr
p
a
=
0
)
where
MtrId p c+ denotes a set of agents belonging to community c, which have TR values with sign (+) for predicate p, and MtrId p c− denotes a set of agents belonging to community c, which have TR values with sign (−) for predicate p.
8. The system of claim 7 , the sign (+) meaning production, and the sign (−) means consumption.
9. A quantitative evaluation method for grouping a plurality of agents in a network-connected multiagent system, the method comprising the steps of:
(a) gathering, by each of the agents, quantitative information on a predicate that is a type of information to be produced or consumed, wherein the quantitative information includes an RO (Request Quality) value corresponding to quality of information on the predicate and a TR (Time Resolution) value corresponding to quantity of information created for the predicate:
(b) transmitting, by each of the agents, the gathered quantitative information to a message broker means;
(c) transferring, by the message broker means, the quantitative information to an adjusting means;
(d) receiving, by the adjusting means, the quantitative information;
(e) performing, by the adjusting means, quantitative evaluation of the quantitative information after all of the quantitative information from the plurality of agents is received by using any one of UoP (Utility of Predicate), UoA (Utility of Agent) and UoC (Utility of Community);
(f) estimating, by the adjusting means, one or more community organizations for the plurality of agents according to a result of the quantitative evaluation by the adjusting means, and then grouping the plurality of agents into the one or more community organizations;
(g) determining, by the adjusting means, if a current community organization for one of the plurality of agents is appropriate; and
(h) performing, by the adjusting means, maintenance or update for the current community organization of the one of the plurality of agents based on a result of the determination.
10. The method of claim 9 , wherein the adjusting means carries out grouping of the plurality of agents by executing calculation of the following equation:
CO
global
=
{
w
❘
max
(
∑
k
=
w
UoC
w
·
Cwg
w
)
}
where, CO global denotes a universal set of all the communities within the system, w denotes a string representing an agent set sequence in which a multiplication of UoC and Cwg is maximized, k denotes a constant, UoC w denotes the UoC of community w, and Cwg w denotes the weight which community w has.
11. The method of claim 10 , wherein the Cwg is calculated by executing the following equation:
Cwg
c
=
∑
j
=
Mco
c
∑
k
=
all
predicate
rq
k
j
·
tr
k
j
∑
j
=
Mco
global
∑
k
=
all
predicate
rq
k
j
tr
k
j
where j denotes a variable, Mco c denotes a set of agents belonging to community c, rq k j denotes an RQ value for the predicate of agent j, and tr k j denotes a TR value for the predicate of agent j.
12. The method of claim 11 , wherein the UoCw is calculated by executing the following equation:
UoC
c
=
∑
k
=
Mco
w
UoA
k
c
·
Awg
k
c
where c denotes a variable representing an ordinal number of community c, UoA k c denotes the UoA of agent k belonging to community c, and Awg k c denotes an influence which agent k exercises in community c.
13. The method of claim 12 , wherein the Awg k c is calculated by executing the following equation:
Awg
a
c
=
∑
k
=
all
predicate
rq
k
a
·
tr
k
a
∑
j
=
Mco
c
∑
k
=
all
predicate
rq
k
i
·
tr
k
j
where a denotes a variable, rq k a denotes an RQ value for the predicate of agent a, and tr k a denotes a TR value for the predicate of agent a.
14. The method of claim 13 , wherein the UoA k c is calculated by executing the following equation:
UoA
k
c
=
(
∑
k
=
all
predicate
UoP
k
a
,
c
∑
k
=
all
predicate
rq
k
a
·
tr
k
a
×
100
+
100
)
2
where UoP a,c k denotes the UoP that agent a belonging to community c satisfies for predicate k.
15. The method of claim 14 , wherein the UoP a,c k is calculated by executing the following equation:
UoP
p
a
,
c
=
{
∑
k
=
MtrId
p
c
-
rq
p
k
·
tr
p
k
(
if
tr
p
a
>
0
,
∑
k
=
MtrId
p
c
-
rq
p
k
·
tr
p
k
≥
rq
p
a
·
tr
p
a
)
∑
k
=
MtrId
p
c
-
rq
p
k
·
tr
p
k
-
(
rq
p
a
·
tr
p
a
∑
k
=
MtrId
p
c
-
rq
p
a
·
tr
p
a
)
(
if
tr
p
a
>
0
,
∑
k
=
MtrId
p
c
-
rq
p
k
·
tr
p
k
≥
rq
p
a
·
tr
p
a
)
rq
p
a
·
tr
p
a
-
(
∑
k
=
MtrId
p
c
+
rq
p
k
·
tr
p
k
-
rq
p
a
·
tr
p
a
)
(
if
tr
p
a
<
0
,
∑
k
=
MtrId
p
+
rq
p
k
·
tr
p
k
≥
rq
p
a
·
tr
p
a
)
∑
k
=
MtrId
p
c
+
rq
p
k
·
tr
p
k
-
(
rq
p
a
·
tr
p
a
-
∑
k
=
MtrId
p
c
+
rq
p
k
·
tr
p
k
)
(
if
tr
p
a
<
0
,
∑
k
=
MtrId
p
c
+
rq
p
k
·
tr
p
k
<
rq
p
a
·
tr
p
a
)
0
(
if
rq
p
a
·
tr
p
a
=
0
)
where
MtrId p c+ denotes a set of agents belonging to community c, which have TR values with sign (+) for predicate p, and MtrId p c− denotes a set of agents belonging to community c, which have TR values with sign (−) for predicate p.
16. The method of claim 15 , wherein the sign (+) means production, and the sign (−) means consumption.