=Paper=
{{Paper
|id=Vol-494/paper-42
|storemode=property
|title=Engineering Development of Agents using the Cooperative Behaviour of their Components
|pdfUrl=https://ceur-ws.org/Vol-494/masspaper5.pdf
|volume=Vol-494
|dblpUrl=https://dblp.org/rec/conf/mallow/BonjeanBG09
}}
==Engineering Development of Agents using the Cooperative Behaviour of their Components==
https://ceur-ws.org/Vol-494/masspaper5.pdf
Engineering Development of Agents using the
Cooperative Behaviour of their Components
Noélie Bonjean, Carole Bernon, Pierre Glize
Institut de Recherche en Informatique de Toulouse
University of Toulouse III
118 route de Narbonne, 31062 Toulouse cedex 9, France
Email: Firstname.Name@irit.fr
Abstract—The objective of the work undertaken here is to Section II presents the context for this study and Section
endow an agent-oriented methodology (such as ADELFE) with III positions it according to existing works. The simulation
a semi-automatic tool for helping designers when drawing up adopted as a base for the proposed approach is presented in
the agents composing an adaptive multi-agent system (AMAS).
This tool acts as a guide for enabling designers to influence the Section IV. This enables expounding how developing such a
emergent global behaviour of an AMAS by acting on the local simulation may be guided in Section V before concluding with
behaviour of its cooperative agents. The preliminary approach some prospects.
proposed in this article can be seen as a feasibility study aiming
at developing a textual guide by considering the principles of II. C ONTEXT
the AMAS theory. Simulation of the behaviour of healthy and
cancerous cells is used as a base for this study. The aim of this study is to facilitate the design of agents
composing a specific type of MAS, adaptive ones. This section
I. I NTRODUCTION introduces the concepts to which this study is related.
Multi-Agent Systems (MAS) are a recognised paradigm
A. AMAS Theory and Cooperative Agents
for designing and implementing complex applications and
several agent-oriented methodologies were proposed to guide Adaptive multi-agent systems at the heart of this study apply
engineers in such a task [1] [2]. However when complexity self-organisation principles according to the AMAS theory
prevents designers from discovering an a priori algorithm, detailed in [3]. According to it, for designing a system whose
designing MAS may also be a complex problem and new functionality is adequate with what is expected by the designer,
approaches may be adopted, such as bottom-up ones. Agents it is sufficient to drawing this system up with parts which have
composing a MAS are identified, their behaviour and inter- a cooperative attitude. This cooperative attitude means always
actions defined to let them build the actual organisation of trying to avoid, otherwise repair, situations that are judged,
this MAS and let the global collective behaviour emerge from from the own point of view of an agent, as non cooperative. An
these interactions. This approach is adopted when building agent does this by changing its relationships with other agents.
Adaptive Multi-Agent Systems (AMAS) in which every agent This also changes the internal organisation of the multi-agent
has a cooperative attitude [3]. However this emergence at the system it belongs to and as a result transforms the collective
macro-level does not prevent engineers from having difficulties function the system is performing, making it adaptive.
for finding the right micro-level cooperative behaviours and A behavioural model of a cooperative agent was proposed
helping them is still an issue. ADELFE1 was proposed as a and used during the microMega2 project which aim was to
guide dedicated for designing AMAS, but it has still some model and simulate the behaviour of a unicellular microor-
lacks. Therefore, additional guidelines and tools have to be ganism [4]. Adopting such a model has firstly simplified the
provided for enriching it. visualisation of the different parts composing the behaviour of
The objective of the work undertaken here is then to endow a cooperative agent (see Fig. 1). Indeed, this model separates
ADELFE with a semi-automatic tool for helping engineers the nominal behaviour of a cooperative agent from its adaptive
when drawing up the agents composing an AMAS. This (or cooperative) one, this latter being itself broken into tuning,
tool acts as a guide for enabling designers to influence the reorganisation and evolution behaviours. A designer may then
emerging global behaviour of an AMAS by acting on the local work on each part of the agent’s behaviour almost indepen-
behaviour of its cooperative agents. The approach proposed dently and implement and test them in a gradual way. These
in this article is still a preliminary one. It can be seen as behaviours may be described as follows:
a feasibility study aiming at developing a textual guide to • the nominal behaviour represents the basic behaviour of
facilitate designing agents. This textual guide is developed by an agent, what it does for achieving its local function
studying an application related to the biological domain and without necessarily coping with Non Cooperative Situa-
simulating the behaviour of healthy and cancerous cells. tions (NCS),
1 Atelier de Développement de Logiciels à Fonctionnalité Emergente 2 National ANR-funded project, 2005-2008
of agents by including some actions that enable an agent
to prevent or repair non cooperative situations it will en-
counter. Furthermore, he is also capable of separating the
nominal behaviour from the adaptive one and implements
an agent according to the behavioural model presented
above.
• In the second case, this designer does not really know
how to identify non cooperative situations for the agent
he is designing, or this identification is incomplete. Fur-
thermore, he does not know how to define the actions this
agent has to perform for staying in a cooperative state.
The nominal behaviour he designs essentially consists in
Fig. 1: Behavioural model of a cooperative agent. the basic behaviour of this agent and does not concern
the aspects related to its cooperative attitude. Such a de-
signer encounters difficulties for finding the behavioural
• the adaptive behaviour, added on top of the nominal one, dichotomy the model lays down.
aims at dealing with these cooperative failures in three In both cases finding every non cooperative situation an agent
different ways: may encounter is not guaranteed, let alone defining every
– by trying to adjust the values of the parameters used action required for removing such situations. Theoretically this
during the nominal behaviour (tuning behaviour), would result in the inability of agents to collectively achieve a
– by changing its relationships with others for trying functionally adequate global function: what emerges from their
to solve dead-ends (reorganisation behaviour), interactions will not suit the designer’s expectations. Ideally
– and finally by self-removing or creating other agents ADELFE should help designers for verifying these points
if NCS still remain (evolution behaviour). and/or complete the agent design they made.
A. Simulating for Designing?
B. ADELFE and Living Design
In 2002, using simulation for AOSE and designing agents
For assisting engineers when designing AMAS, ADELFE was a challenge as underlined in [5]. For a few years now,
was proposed. During the design phase, a specific activity is steps are made in this direction with several research works
provided as a guide for designing agents: every non coop- using simulation in the AOSE domain [6]–[8].
erative situation has to be identified and every preventing or
Some works were made in this sense for enriching ADELFE
repair action has also to be defined. This textual guide is still
as mentioned above. Simulation has been used for automat-
insufficient because currently, nothing in ADELFE guarantees
ically detecting NCS in an AMAS prototype [9]. Simulation
that this identification is the proper and complete one. It is
has also be used for making agents self-adjust their be-
therefore necessary to enrich ADELFE with a better tool.
haviour by making their behavioural rules self-reorganise [10].
Ideally this automatic (or semi-automatic) tool would enable Although both approaches had conclusive results for rather
to develop the adequate behaviour for a cooperative agent, the simple applications, they nevertheless have some drawbacks.
very one that would allow achieving the functional adequacy The kind of agents that was taken into account (which are
of the AMAS this agent belongs to. This tool should also situated in an environment and communicate in an indirect
take into account the benefits brought by the behavioural manner through this environment) and the underlying use of
model presented in the previous section by enabling designers the SeSAm platform [11] (need to become familiar with it,
to separately act on the different parts composing such a performance problems when a great number of rather complex
behaviour. agents are simulated) may be considered as limitations. Fur-
The following section presents this issue and gives insight thermore enabling a designer to make a prototype of his system
into existing works that have links with this one. before really implementing it is interesting; however, our main
III. T HE P ROBLEM aim is rather to make a first step towards enriching ADELFE
with a “good practice guide” before (semi-)automating this
Considering the model given in Fig. 1, each part of the guide through an appropriate tool.
behaviour of an agent has an action on the other one. A NCS
occurring during the nominal behaviour may trigger a repair B. The Adopted Approach
action that will be performed during the adaptive behaviour. The issue here is therefore to study how general principles
Depending on the designer’s degree of familiarity with the may be extracted from the features of AMAS, and more
AMAS concepts, the behaviour of an agent may be devised especially those of cooperative agents, in order to deliver this
in two ways: guide.
• The designer succeeds in identifying non cooperative An AMAS simulating a behaviour is used as the base for
situations. He certainly designs the nominal behaviour this study. A designer of MAS, who is considered as being
unfamiliar with the way of implementing AMAS, follows
ADELFE and designs agents for achieving a first functional
version of this system. This version, probably only based
on the nominal part of agents, will likely not be the proper
one. Therefore, the second step is to try and improve this
implementation by acting on the different parts of the adaptive
behaviour an agent may have.
IV. C ASE S TUDY: S IMULATING C ELLS
To stay in line with the application domain of microMega,
and because biological phenomena are complex and dynamic,
a simulation inspired from the biological world was chosen. Fig. 2: Ratio of cancerous cells over time.
Actually, simulation and modelling in biology are a very
active field of research, notably when cells are involved [12].
Mathematical models, based on differential equations e.g., may identified : healthy cells and cancerous ones. Molecules are
be used [13] or less classical tools such as Petri nets [14], moving in an autonomous way, have also a limited perception
cellular automata [15] or neural networks. For a few years of their environment but they do not have a local goal and
now, MAS are also contributing to this effort, mainly because are not considered as agents. Message passing simulates the
they are able to scale and to model specific properties in a more absorption of a paracrine molecule by a cell and the related
comprehensible way than mathematical models [16]. Among communication. Healthy cells may divide, mutate, die, signal
these works, some are more precisely interested in simulating molecules, or detect and absorb molecules. Cancerous cells do
the behaviour of microorganisms or cells [17]–[19]. not mutate and die.
This section therefore describes the features of the target
application and how it was implemented. B. Collective Nominal Behaviour Obtained
The designer of this application is regarded as being able to
A. Description of the Simulation deal with only the nominal behaviour of agents. As a result, the
In a living organism, mitosis enables division of a eukaryotic nominal behaviour of cells was implemented and the adaptive
cell into two daughter cells and apoptosis makes such a behaviour is what the guide has to help finding.
healthy cell degrade and die. Mutations may also occur (due To see how the collective behaviour obtained for this MAS
to radiations, viruses, genetic predispositions, etc.) by copying fits a user’s needs, the ratio R of cancerous cells was studied.
C
errors of genetic material during division. Most of the time, R = C+H where C is the number of cancerous cells and
mutations are not harmful for the cell or are repaired by H, the number of healthy ones. The curve obtained for R is
internal mechanisms. However, some changes may lead to shown on the right part of Fig. 2.
malignant cells leading to cancer. Among other properties, a A healthy cell has a given probability of mutation and
cancerous cell is not able to repair DNA anymore, divides proliferates only if some free space is near it. On the contrary,
in an unchecked way and does not die. Cells “communicate” a cancerous cell tries to push healthy ones if it does not have
with their environment by endocytosis, which enables them available space around it and divides itself if it succeeds. As a
to absorb material from outside, or exocytosis, by secreting result, a slow increase of the number of cancerous cells can be
material to the extracellular environment. Molecules which seen on the curve at the beginning of the simulation because
may be released are moving through the interstitial tissue. few mutations occur yet. However, once several malignant
Some of them play a role in the paracrine signalling by cells appear, they proliferate in a faster way than healthy ones,
carrying information from a cell to another one [20]. Some also because they do not die, and invade the tissue with time.
molecules also supply cells with energy. In both cases, moving Let us suppose now that this collective behaviour is not the
make them degrade with time. one expected by the end-user of this simulated system, e.g., a
According to this simplified biological context, the aim biologist. This user would like to obtain a curve with another
of the considered application is to simulate the behaviour shape; for instance, one that would express a more regular
of healthy and cancerous cells. The development process of appearance of cancerous cells which is shown on the left part
ADELFE was applied for analysing the requirements, verify- of Fig. 2. This shape has no biological reality and was chosen
ing the AMAS adequacy and identifying the agents involved in only for studying how the engineer could be guided to adapt
this MAS, and designing their nominal behaviour with respect the local behaviour of cells in order to obtain the new expected
to the basic biological knowledge previously presented. Due collective behaviour.
to lack of space and because the objective of the paper is not
to focus on it, only some details are given. V. G UIDING THE S IMULATION
Because cells are autonomous, have a local goal (survive for Actually, changing the behaviour of an agent may be
healthy ones and proliferate for malignant ones), have only a done by changing the basic function it performs, its nominal
partial view of their environment, two types of agents were behaviour. However, this choice is contrary to the hypothesis
we previously made; only the adaptive behaviour has to be TABLE I: Influence table for healthy cell-agents.
built or modified. This means modifying one or all the parts Param/Action A1 A2 A3 A4 A5 A6 A7
of this behaviour: tuning, reorganisation and evolution. Since P1 + --
this study is a first step towards our goal, only the tuning part P2 +++ ++ ++ ++ -
will be reviewed. P3 + +++ -- -- +++
A. Act on the Tuning Behaviour of Agents P4 -- +++ +++
In general, the tuning behaviour of an agent has to be P5 ++ -
modified in order to get the expected global behaviour (the P6 - ++
expected shape for the studied curve, e.g.). Modifying the P7 + -
value of the parameters that are used by an agent is going P8 + -
to modify how the nominal behaviour, which uses these very P9 + -
parameters, performs and therefore is going to influence the P10 ++
global behaviour of its collective. The issue lies in identifying P11 + +++
the right parameters that have to be calibrated in order to get P12 +++
the proper behaviour. ...
The approach proposed is to study first how the parameters
of an agent may promote its possible actions. Which actions
have to be influenced (fostered or encouraged) in order to get • P7 - Energy cost for signalling a paracrine molecule,
the expected behaviour is then determined. Since a parame- • P8 - Energy cost for absorbing a molecule of any type,
ter may influence one or several actions, these relationships • P9 - Energy cost for signalling a “simple” molecule,
have to be gradually propagated for determining the proper • P10 - Apoptosis signals: signals which tend to influence
parameter to change in order to modify the actions previously death when overpopulation occurs,
identified. • P11 - Sensitivity threshold to paracrine: expresses the
ability of a cell to react to paracrine molecules,
B. Relate Parameters and Actions of Agents • P12 - Mutation rate: number of cells that may mutate
By leaning on the application domain knowledge, the de- during a lifecycle.
signer has to establish a table relating the actions agents may For each (parameter, action) pair, two criteria are repre-
perform and the parameters these agents use. Actually, this sented in these influence tables:
table should be done before beginning implementation, it is
• In which direction (increase/decrease) the given parame-
also a means to see how agents behave and to reflect on what
ter may vary for promoting the action. This is expressed
they perform according to what they have to do.
by + or −.
Such tables were built for healthy cell agents (see Table
• The influence of the given parameter on the action. This
I) and cancerous ones (see Table II). In this application, 18
is expressed by the number of symbols + or − used.
parameters may influence the 7 actions of healthy cells:
For example, the parameter P1-Energy of a cell has to
• A1 - Proliferate,
increase for promoting the action A1-Proliferate of a healthy
• A2 - Mutate,
cell and this influence on A1 is high because three + are used.
• A3 - Die,
P1 has also an impact on the action A7-Signal a “simple”
• A4 - Absorb a paracrine molecule,
molecule. In order to encourage a healthy cell to make this
• A5 - Signal a paracrine molecule,
action, P1 has to be decreased and this influence is low because
• A6 - Absorb a “simple” molecule,
only one − is used.
• A7 - Signal a “simple” molecule.
Once these tables designed, a reasoning has to be done in
or the 5 ones of cancerous cells (A1’, A4’, A5’, A6’ and A7’). order to modify the behaviour of cell-agents.
Due to lack of space, out of the 18 parameters, 12 parameters
are considered in the tables built, those that are only required C. Propagate Influences
for the coming reasoning: A parameter may influence several actions (P1 influencing
• P1 - Lifetime of a cell: lifetime in the means case, A1 and A7, e.g.), and by modifying it, the related actions will
• P2 - Energy of a cell: internal energy a cell has for living, also be promoted or discouraged, depending on the impact
• P3 - Molecule concentration in the environment: number of this parameter. Therefore before modifying a parameter,
of molecules a cell may perceive, studying how its influences are propagated from an action to
• P4 - Environmental conditions: occupied spaces sur- another one is required. The influence tables previously built
rounding a healthy cell and free ones for a cancerous will be a help for this study.
one, For trying to obtain a more linear curve for R, starting from
• P5 - Proliferation speed: expressed in simulation steps, Fig. 2, R has to increase when R is below the straight line
• P6 - Energy cost for proliferating: energy required for which represents its expected value and it has to decrease when
this action, it is above this line.
TABLE II: Influence table for cancerous cell-agents. to reflect their importance, a new formula may be used:
Parameter/Action A1’ A4’ A5’ A6’ A7’ (a×N CC +b×N HC)/(a×N CC +b×N HC +c×AS).
P1 Furthermore, increasing this parameter would also pro-
P2 +++ ++ ++ - mote, with the same importance, the actions A2-Mutate
P3 + +++ -- -- +++ and A3-Die and this goes in the right direction.
P4 ++ 3) Promote “A3-Die” for Healthy Cells: This action is
P5 ++ influenced by P1-Lifetime of a cell and P4-Environmental
P6 -
conditions which were studied before. Some other parameters
are also involved (from P5 to P10). P10-Apoptosis signals
P7 -
influence solely this action and acting on them may be
P8 -
easy. Energy costs (from P6 to P9) could be increased for
P9 -
promoting A3-Die. However, by studying propagation, these
P11 +++
costs have also an impact on other actions such as absorption
...
and signalling of molecules, and this could modify A2-Mutate;
on the other hand, A1-Proliferate could also be deserved, and
so on. Not modifying these parameters seems to be a good
In a first step, R is going to be increased. Since R is choice to try and avoid a too great influence on the system.
a fraction, it increases when its numerator does and/or its 4) Promote “A2-Mutate” for Healthy Cells: Among the
denominator decreases. Since C cannot decrease, C has to parameters that were not examined before, P12-Mutation rate
increase or S has to decrease. To make C increase and S has no influence on any other actions than A2, therefore it may
decrease, the action A1-Proliferate has to be promoted for be modified as needed. Adjusting P11-Sensitivity threshold
cancerous cells, A1’-Proliferate has to be repressed for healthy would be propagated on A4-Absorb paracrine without any
ones and A2-Mutate as well as A3-Die have to be promoted harmful effect on the system.
for healthy cells. Once this propagation studied, some parameters may vary
Each action has to be analysed depending on the rela- while others cannot. According to these new conditions, mod-
tionships it has with the parameters and how this influence ification of some of the potential modifiable parameters has
propagates. then to be done.
1) Promote “A1’-Proliferate” for Cancerous Cells: Pro-
D. Tune Parameters
liferation of cancerous or healthy cells are promoted by
the same parameters except P4-Environmental conditions. Table III shows the parameters that may vary (in bold font)
Therefore, promoting A1’ by playing on P1, P2, P3, P5 or not (normal font) for promoting (+ before the name of an
or P6 will also promote A1, and A1 has to be deserved. action) or deserving (−) actions of healthy or cancerous cells
These parameters cannot be modified. The P4 parameter (their specific parameters or actions are distinguished from
represents the available space around a cell and is calculated those of healthy cells by a ’ following their name).
by (N CC + N HC + AS) − (N CC + N HC) where N CC Parameters for which propagation is not opposite to the
is the number of neighbouring cancerous cells, N HC is the expected result are first modified. P1, P4, P11 and P12 are
number of neighbouring healthy cells and AS is the number chosen, especially because they have the highest influence.
of available spaces around the cell. This formula should give The formula related to P4 is thus modified for making can-
the highest possible result for promoting A1’. cerous cells have a higher impact and becomes (2 × N CC +
2) Promote “A1-Proliferate” for Healthy Cells: As above, N HC)/(2 × N HC + N CC + AS).
some parameters cannot be changed. However other parame- E. Verify the Impact of Tuning
ters have an impact on A1 and do not influence A1’: Once the above reasoning done and parameters modified in
• P1-Lifetime of a cell has a low influence on A1. Deserving the simulation code, a new kind of curve for R was obtained
it should be obtained by decreasing P1 and this would (see Fig. 3). The first part of this curve is more linear than
strongly promote A3-Die which is the expected effect. previously and gets closer to a straight line. However, its
• P4-Environmental conditions has a high influence on second part (from step 800) does not properly fit expectations.
A1. It represents the occupied spaces around an healthy Therefore, the same reasoning has to be done in order to
cell. P4 is increased by promoting A1/A1’ for each kind make the top curve as linear as possible by decreasing R
of cells or by deserving A3. The former alternative is without cancelling the work done previously. An opposite op-
contrary to the expected effect for healthy cells and eration has to be done: deserve A1’-Proliferate for cancerous
redundant for what was done above for cancerous ones. cells, A2-Mutate and A3-Die for healthy ones, and promote A1-
The latter alternative is the opposite of the expected Proliferate for healthy cells. Since the benefits brought by the
effect. Another solution would be to compute the value previous adjustment have to remain, the parameters that may
of P4 in a different way. The current formula is (N CC + be modified are those that were still untouched. In this case,
N HC)/(N HC +N CC +AS). By adding the respective only P4’ was not modified and the value of environmental
coefficients a, b and c to N CC, N HC and AS in order conditions for cancerous cells is then turned down.
TABLE III: Influence table for making R increase.
Parameter/Action − A1 + A2 + A3 + A1’
P1 - ---
P2 --- ++ +++
P3 - +
P4 +++ +++ +++
P4’ ++
P5 -- - ++
P6 + + -
P7 +
P8 +
Fig. 3: New ratio obtained after the first tuning.
P9 +
P10 ++
P11 +
P12 +++
F. Observe the Collective Behaviour Obtained
After these two adjustments, a typical final curve obtained
is shown in Fig. 4. It is more linear and looks like the one
end-users expected. Therefore the collective behaviour of the
cells is more in line with the expected one. As expected, the
shape of the curve shows that cell-agents have modified their
behaviour in the right direction.
Fig. 4: Final ratio obtained.
It is also worth observing that the collective behaviour
before and after modifications made on the tuning behaviour of
cells does not give the same spatial distribution for cancerous
cells, as shown in Fig. 5. In this figure, healthy cells are empty
hexagons, cancerous ones are black hexagons and paracrine
molecules are dots between cells. To simulate an “infinite”
tissue, cells on opposite borders are neighbours. These snap-
shots were taken when around 10 cancerous cells appeared.
They show that cancerous cells tend to be more scattered in
the simulated tissue when only the nominal behaviour was
concerned. After tuning, they tend to form clusters because
these cancerous cells can mainly be created at the frontiers of Fig. 5: Example of distribution obtained before (left) and after
clusters, and this prevents their exponential development. Of (right) tuning.
course as time goes by, the number of cancerous cells increases
and their distribution tends to be more dense in both cases.
However the initial disparity is consistent with the results and guide” will explain in ADELFE) allows the determina-
the aim of obtaining a straight curve. tion of the relevant adjustments to do.
Furthermore, this disparity and clusterisation was obtained
without reasoning on this macro-level; only by studying re- VI. C ONCLUSION
lationships between parameters and actions inside the agents. In this article a first step was made towards integrating
Here also, there is emergence of a phenomenon at the macro- a methodological guide into ADELFE for helping engineers
level which was controlled by cooperation at the micro-level. when designing the behaviour of cooperative agents. This
More precisely, there are three levels: preliminary textual guide may be summed up as follows. The
1) The macro-level which corresponds to the cell tissue. designer has first to devise the nominal behaviour of agents
It is used to observe the global behaviour and to enter involved in the AMAS he has to engineer. Depending on
feedbacks according to the end-user wishes. the feedbacks given by end-users, he has then to act on the
2) The meso-level constituted by the individual cells, where adaptive behaviour of these agents to improve the collective
ad hoc feedbacks arrive according to their type (cancer- behaviour of this AMAS. For the time being, only the tuning
ous or not). part of this adaptive behaviour was studied. The designer has
3) The micro-level corresponding to the cell components, then to establish relationships between parameters involved in
where cooperative “negotiations” (as the “good practice the AMAS and actions agents may do (and this may be done
before, as a help for finding the nominal behaviour also). Then, [4] C. Bernon, D. Capera, and J.-P. Mano, “Engineering Self-Modeling
these relationships have to be quantified and studied, especially Systems: Application to Biology,” in International Workshop on Engi-
neering Societies in the Agents World, ser. LNCS, A. Artikis, G. Picard,
by propagating effects a parameter may have on different and L. Vercouter, Eds., no. 5485. Springer, 2009, pp. 236–251.
actions (promote or deserve them). This enables finding which [5] A. Uhrmacher, “Simulation for Agent-Oriented Software Engineering,”
and how some parameters have to be changed for positively in First International Conference on Grand Challenges for Modelling
and Simulation, San Antonio, Texas, W. Lunceford and E. Page, Eds.
influencing the collective behaviour towards what is expected San Diego: SCS, 2002.
by end-users. [6] G. Fortino, A. Garro, W. Russo, R. Caico, M. Cossentino, and F. Ter-
A MAS simulating cancerous cell proliferation in a tissue mine, “Simulation-Driven Development of Multi-Agent Systems,” in
Workshop on Multi-Agent Systems and Simulation, Palermo, Italia, 2006.
was used for grounding this demonstration. Although it cannot [7] M. Cossentino, G. Fortino, A. Garro, S. Mascillaro, and W. Russo,
be considered as having a strong biological reality, it accounts “PASSIM : A Simulation-based Process for the Development of Multi-
for an interesting complex use case, at least for tuning, Agent Systems,” Int. Journal of Agent-Oriented Software Engineering,
vol. 2, no. 2, pp. 132–170, 2008.
because it has several interrelated actions and parameters. The [8] L. Gardelli, M. Viroli, M. Casadei, and A. Omicini, “Designing Self-
proposed approach was applied on this cell simulation for organising Environments with Agents and Artefacts: A Simulation-
playing on the evolution of the ratio of cancerous cells over driven Approach,” Int. J. Agent-Oriented Softw. Eng., vol. 2, no. 2, pp.
171–195, 2008.
time and trying to influence the shape of the related curve. [9] C. Bernon, M.-P. Gleizes, and G. Picard, “Enhancing Self-Organising
By changing some parameters used by healthy and cancerous Emergent Systems Design with Simulation,” in International Workshop
cells, this curve was actually changed in the right direction. on Engineering Societies in the Agents World (ESAW), Dublin, ser.
LNCS, G. O’Hare, M. O’Grady, A. Ricci, and O. Dikenelli, Eds., vol.
As a consequence the distribution of cells in the tissue was 4457. Springer-Verlag, September 2007, pp. 284–299.
also modified which shows that emergence of a phenomenon [10] S. Lemouzy, C. Bernon, and M.-P. Gleizes, “Living Design: Simulation
at the global level may be influenced by changes in the local for Self-Designing Agents,” in European Simulation and Modelling
Conference (ESM), Malte, J. Sklenar, A. Tanguy, C. Bertelle, and
behaviours. G. Fortino, Eds. Eurosis, October 2007, pp. 432–436.
However, a lot of work has to be done yet. First, it [11] F. Klügl, R. Herrler, and O. Oechslein, “From Simulated to Real
is necessary to study how reorganisation and evolution be- Environments: How to use SeSAm for Software Development,” in
Multiagent System Technologies - 1st German Conference MATES, ser.
haviours could be used for guiding the engineer, then enriching LNAI, M. Schillo, Ed., vol. 2831. Springer-Verlag, 2003, pp. 13–24.
ADELFE with this (still textual) guide could be done, once [12] T. Fisher and T. Henziger, “Executable Cell Biology,” Nature Biotech-
formalised. The last step would be to automate this guide by nology, vol. 25, no. 11, pp. 1239–1249, November 2007.
[13] P. Mendes, “GEPASI: A Software Package for Modelling the Dynamics,
implementing the related tool as an AMAS. Indeed, discover- Steady States and Control of Biochemical and other Systems,” Computer
ing the parameters and actions to adjust in a complex adaptive App. in the Biosciences, vol. 9, no. 5, pp. 563–571, October 1993.
system clearly corresponds to a specific agentification level, [14] C. Li, Q. W. Ge, M. Nakata, H. Matsuno, and S. Miyano, “Modelling
and Simulation of Signal Transductions in an Apoptosis Pathway by
not directly required by the end-user problem. In this tool, using Timed Petri Nets,” J. Biosci., vol. 32, pp. 113–127, 2007.
for at least the tuning part, actions and parameters would be [15] H. Silva and M. Martins, “A Cellular Automata Model for Cell Differ-
considered as cooperative agents. Their collective goal would entiation,” Physica A, no. 322, pp. 555–566, 2003.
[16] E. Merelli, G. Armano, N. Cannata, F. Corradini, M. d’Inverno,
be to find the right parameters to modify depending on the A. Doms, P. Lord, A. Martin, L. Milanesi, S. Moller, M. Schroeder,
feedbacks given by end-users which would be considered as and M. Luck, “Agents in bioinformatics, Computational and Systems
NCS that these agents should avoid and solve. Furthermore, Biology,” Briefings in Bioinformatics, vol. 8, no. 1, pp. 45–59, 2006.
[17] M. d’Inverno and R. Saunders, “Agent-Based Modelling of Stem
all the results observable by an end-user would have to be Cell Self-organisation in a Niche,” in Engineering Self-Organising
agentified (e.g. the curve shown on Fig. 2 would be an Systems: Methodologies and Applications, ser. LNCS, S. Brueckner,
agent in order to reason cooperatively on function R when G. Di Marzo Serugendo, A. Karageorgos, and R. Nagpal, Eds., vol.
3464. Springer, 2004, pp. 52–68.
an external feedback occurs). Consequently, engineering semi- [18] L. Dib, N. Bonnet, Z. Guessoum, and M. Laskri, “Multi-agent System
automatically the development of a complex adaptive system Simulating Tumoral Cells Migration,” in 18th Australian Joint Confer-
must be AMAS-compliant, even when this complex system ence on Artificial Intelligence. Springer Verlag, December 2005, pp.
624–632.
itself is not AMAS-designed. [19] K. Webb and T. White, “Cell Modeling with Reusable Agent-based
Consequently this tool is not ADELFE-dependent and could Formalisms,” Applied Intelligence, vol. 24, no. 2, pp. 169–181, April
be included in any methodology devoted to the development 2006.
[20] T. J. Weber, R. W. Siegel, L. M. Markillie, W. B. Chrisler, X. C. Lei, and
of complex adaptive systems. Nevertheless, a lot of work N. H. Colburn, “A Paracrine Signal Mediates the Cell Transformation
and tests are required before considering such a kind of Response to Low Dose Gamma Radiation in JB6 Cells,” Molecular
deployment. Carcinogenesis, no. 43, pp. 31–37, 2005.
R EFERENCES
[1] F. Bergenti, M.-P. Gleizes, and F. Zambonelli, Methodologies and
Software Engineering for Agent Systems. Kluwer Publishing, 2004.
[2] B. Henderson-Sellers and P. Giorgini, Agent-Oriented Methodologies.
Idea Group Pub, June 2005.
[3] D. Capera, J.-P. Georgé, M.-P. Gleizes, and P. Glize, “The AMAS Theory
for Complex Problem Solving Based on Self-organizing Cooperative
Agents,” in 12th IEEE International Workshops on Enabling Technolo-
gies, Infrastructure for Collaborative Enterprises, Linz, Austria. IEEE
Computer Society, June 2003, pp. 383–388.