Community Computing - Advanced Design Approaches to Emerging Software Systems - page 215

environment and the feedback from other
members. In AGDRSCOM, the idea of
adaptive cooperation is introduced. In
adaptive cooperation, the function struc-
ture of a cooperative member agent has
skills, where a skill is represented with a
five-element tuple: Skill = <Activity, Pre-
Processing, Programming, Action-Set,
Post-Processing> . In this representation,
Activity is the basic action when a skill is
executed. Pre-Processing is the processing
of information required by a programming
task prior to execution. Programming can
be a rule set or state transfer figure and
is referred to when the skill is executed.
Action-Set is the possible action set, and
Post-Processing is a result or post-process-
ing of Programming. In this model, the
function structure of the adaptation of co-
operation was introduced, but the detailed
method of adaptation was not proposed. In
addition, cooperation was also represent-
ed as a programming element in the skill
description.

mental cooperation process and is statical-
ly established according to the scheduling
plan during the design of the application
system. A recipe is defined by possible
plans, plan steps, and sub-activities in
these plan steps.

CONCLUSION

Our ultimate goal is to design and develop the
multi-agent systems which provide complex and
dynamic ubiquitous services through coopera-
tion among existing agents. As mentioned above,
existing research is not suitable for our goal. In
this chapter, we therefore redefined community
computing as a new computing paradigm in which
services are provided by cooperation among given
smart objects. In order to make the meaning of
community computing concrete, we compared
our work with other related work, proposed an
overall concept, and defined terminology to help
with understanding. In order to actualize commu-
nity computing, we first proposed the community
computing models and a development process.
As an early version of the community computing
model, we proposed a simple community comput-
ing model where a community has the necessary
roles, goals, and code-like cooperation protocol.
However, this model has no cooperation model
and no conflict resolution scheme. Therefore, we
proposed an improved model, the static community
situation based community computing model.
In this model, we employed the static commu-
nity situation based cooperation model, which
is a limited model that assumes certainty of the
community situation and members' cooperative
behavior. In addition, we also analyze conflicts
in community computing systems and propose
policies to resolve those conflicts. In order to
examine feasibility of our community computing
model and development process, we developed
two small systems and presented the simulation
results of several scenarios.

• Cooperation model of MAPFS (Perez,

2004): MAPFS (MultiAgent Parallel File
System) is a parallel file system integrated
with a multi-agent system responsible for
information retrieval. In the cooperation
model of MAPFS, cooperation is achieved
using shared plans, where plans contain
precise instructions or actions for achiev-
ing such objectives. Thus, a cooperation
process is also procedural and is described
by actions and instructions.

• IMCAC (Guo, 2006): In 2006, Guo pro-

posed hybrid cooperation using recipes,
policies, and advertisements and imple-
mented the idea of hybrid cooperation in
IMCAC (Infrastructure for Managing and
Controlling Agents' Cooperation). In this
model, a policy is the obligations and re-
strictions that agents should meet, and ad-
vertisements are the records of interests of
other agents. A recipe represents the funda-