A Modeling Environment for Normative Multi-Agent Systems
Emmanuel S. S. Freire, Robert M. Rocha Júnior and Mariela I. Cortés
GESSI – Grupo de Engenharia de Software e Sistemas Inteligentes, Computer Science Department,
Universidade Estadual do Ceará (UECE), Fortaleza, Ceará, Brazil
Keywords: Normative Multi-Agent Systems, Norms, NorMAS-ML, MAS-ML Tool, CASE.
Abstract: Tools to support the development of Multi-Agent Systems are essential to promote the effective utilization
of the approach in the industry. More specifically, an adequate support for the modeling activity is useful to
increase the productivity and contributes to the reduction of possible errors that would only be detected in
execution time. This paper presents the evolution of MAS-ML tool to support the modeling of normative
multi-agent systems modeled in NorMAS-ML language.
1 INTRODUCTION
Like others paradigms, agent-oriented development
requires adequate techniques for exploring its
benefits and properties in the construction and
maintenance of such software (Zambonelli et al.,
2001). In this scenario, tools to support the activities
of engineering (CASE) (Sommerville, 2007) become
decisive factors for the choice and effective use of
available methods and techniques. In particular,
modeling tools contribute to increase the
productivity of developers and ensuring good
formation models for construction.
In Multi-Agent System (MAS), agents are
inserted in environments where they interact with
others agents to achieve their goals. In complex
scenarios, the agent's behavior is governed by
norms, which aims to govern the agent actions
through the concepts of obligation, permission and
prohibition (Figueiredo and Silva, 2010).
Norms can be defined both modeling time as at
runtime. However, the definition in modeling time
carries a deeper and more comprehensive
understanding of the structure, as well as an overview
that allows observing the behavior of the system,
before being executed. Therefore, the use of a
language that allows the modeling of norms along
with entities in MAS is fundamental.
The modeling language NorMAS-ML (Normative
Multi-Agent System Modeling Language) (Freire et
al., 2012) represents an extension of MAS-ML (Silva
et al., 2007) that brings the necessary mechanisms to
support the modeling of the main entities found in
MAS along with the static properties of norms. MAS-
ML has a modeling environment, MAS-ML tool
(Feijó, 2012), that allows modeling all static diagrams
and the dynamic diagram of sequence originally
defined in the language. However, in the current
version, the tool does not support the norm concepts
introduced in NorMAS-ML, as well the static
diagram for modeling norms and their properties.
This paper aims to present the evolution of
MAS-ML tool in order to model the norm diagram.
The paper is structured as follows: Section 2
presents related works. The language NorMAS-ML
and the modeling environment MAS-ML tool are
described in Section 3. Section 4 discusses the
evolution of MAS-ML tool to enable the modeling
the norm diagram. A case study illustrating the use
of the tool is presented in Section 5. Finally,
conclusions and future works are discussed in
Section 6.
2 RELATED WORK
In this section, we analyze the modeling tools
candidate to extension for supporting NorMAS-ML,
i.e., incorporating the adequate support for modeling
of norms along with the entities of the MAS.
VisualAgent (De Maria et al., 2005) is a software
development environment that aims support the
developers in the specification, project and
development of MAS. The tool is based in the MAS-
ML metamodel (Silva et al., 2007) and proposes a
model driven approach for development of MAS
451
S. S. Freire E., M. Rocha Júnior R. and I. Cortés M..
A Modeling Environment for Normative Multi-Agent Systems.
DOI: 10.5220/0004438604510458
In Proceedings of the 15th International Conference on Enterprise Information Systems (ICEIS-2013), pages 451-458
ISBN: 978-989-8565-59-4
Copyright
c
2013 SCITEPRESS (Science and Technology Publications, Lda.)
using the ASF (Agent Society Framework) (Silva,
Cortés and Lucena, 2004), that represents the main
point of the approach. However, the tool does not
enable the checking of the models in relation to
MAS-ML metamodel, neither the modeling of the
normative concepts.
NormML Tool Kit (Figueiredo, 2011) is a
modeling environment based in the NormML
language that supports the mechanisms for model
checking and checking for conflicts between norms.
The environment is composed for two plugins for
Eclipse framework (Eclipse Platform,
www.eclipse.org): NormML Editor and NormML
Conflict Checker. However, only the modeling of
the normative elements is supported, thus the
modeling of the typical entities of MAS and their
properties is not foreseen.
Finally, MAS-ML tool (Feijó, 2012) is a
modeling environment that allows the modeling of
MAS according to the specification of the MAS-ML
metamodel. This tool allows the modeling of the
static diagrams and the dynamic sequence diagram
defined in the language. It was developed as an
Eclipse plugin through model driven approach and
provides the mechanisms for model checking.
However, the tool does not allow the complete
modeling of the statics elements that compose a
norm: only the modeling of permission and
obligation norms for agents that are associated with
agent roles, sub-organizations and organizations are
supported.
Considering that MAS-ML tool and NorMAS-
ML are based in MAS-ML metamodel, this makes it
more suitable to be evolved in order to allow
modeling of the norm diagram along with the model
checking.
3 BACKGROUD
3.1 NorMAS-ML
NorMAS-ML (Freire et al., 2012) is a modeling
language based in UML that allow the modeling of the
all static elements that compose a norm (Figueiredo,
2011) along with the typical entities of MAS. This
language is result of the MAS-ML extension (Silva et
al., 2007).
The NorMAS-ML metamodel (Figure 1) was
defined through the creation of the new metaclasses
and relationships in MAS-ML metamodel in order to
represent the following static normative elements
defined by Figueiredo (2011):
Deontic Concepts: the deontic logic refers to the logic
of requests, commands, rules, laws, moral principles
and judgments (Meyer and Wieringa, 1993). In
MAS, such concepts have been used to describe the
constraints for the behavior of agents in the form of
obligations (what the agent must execute),
permissions (what the agent can execute) and
prohibitions (what the agent cannot execute).
Involved Entities: considering that the norms are
defined to restrict the entities behavior, the
identification of the related entities is essential. The
norm may regulate the behavior of individuals (for
example, a particular agent, or an agent while playing
a particular role), or the behavior of a group of
individuals (for example, all agents playing a
particular role, groups of agents, groups of agents
playing roles or all agents in the system).
Actions: once a norm is set to restrict the behavior of
the entities, it is important the clear specification of
the actions that are being regulated. Such actions may
be communication, usually represented by sending
and receiving a message, or non-communicative
actions (such as access and modify a resource, get in
an organization, move to another environment, etc.).
Activation Constraints: a norm have a period of time
in which its restrictions must be fulfilled, but only
when this norm, is active. Norms may be activated by
a constraint or a set of constraints that can be: the
execution of actions, the definition of specific time
intervals (before, after or in between), the reaching of
system states or temporal aspects (such as dates) and
also the activation/deactivation of other norm and
fulfillment/violation of a norm.
Sanctions: when a norm is violated the entity may
suffer a punishment, and when a norm is fulfilled, the
entity involved may receive a reward. Rewards and
punishments are referred to as sanctions and should
be related to the norm specification.
Context: the norms are usually defined in a
determined context that determines its application
area. The norm may, for example, be described in the
context of a specific environment and must be filled
only by agents in execution in the environment.
Similarly, a norm can be defined in the context of an
organization and fulfilled only by the agents that play
a role in the organization.
3.2 MAS-ML Tool
MAS-ML tool (Feijó, 2012) is a modeling
environment that allows the modeling of MAS.
Through this environment, developers can work with
the concepts of problem domain, while using
explicitly concepts defined in the solution domain, in
this case the concepts and abstractions for the
ICEIS2013-15thInternationalConferenceonEnterpriseInformationSystems
452
paradigm of MAS.
Currently, the tool allows the modeling of all
static diagrams: (i) Class (Farias et al., 2009), (ii)
organization (Gonçalves et al., 2011) and (iii) role
(Feijó, 2012). In addition, only the dynamic diagram
of sequence is supported by the tool (Feijó, 2012).
The environment was developed as a plugin of
the Eclipse platform (Eclipse Platform,
www.eclipse.org, 2012). This implies that users can
work with modeling of MAS while making use of
the resources offered by the Eclipse platform. Since
many agent platforms are implemented in Java, such
as Jade (Bellifemine et al., 2007), Jadex (Pokahr et
al., 2003), Jason (Bordini et al., 2007), the use of the
Eclipse platform also facilitates a possible code
generation within the same development
environment.
4 EVOLVING MAS-ML TOOL
The version of the tool (Feijó, 2012) used as the
basis for its evolution provides the modeling of all
static diagrams defined in MAS-ML: diagram of
classes, organization and roles. In other hand, the
norm diagram from NorMAS-ML need to be defined
in MAS-ML tool. Additionally, with the inclusion of
norms in the tool, it is necessary to incorporate new
rules that allow checking of conflicts in the
definition of norms.
The strategy adopted to evolve the tool follows a
model-driven approach, in which the metamodel of
NorMAS-ML is used as a start point (Figure 1). In
following are described in summary form, the six
steps undertaken in the process of evolution of the
tool:
Domain Model Extension: were included
metaclasses and stereotypes defined in NorMAS-
ML, as well as the necessary adjustments in
existing metaclasses.
Graphic Model Extension: The new metaclasses
of NorMAS-ML (Figure 1) were added in the
Graph Model Definition and its relationships with
other entities were created. Moreover, adaptations
in existing metaclasses been incorporated.
Tool Model Extension: new elements were
created in the palette of the tool to create norms
and their relationships required to build the norm
diagram.
Graphic Model Definition Extension: at this
stage, were created compartments for the
representation of norms and their relationships
(Figure 2), beyond the representation of nodes and
Figure 1: NorMAS-ML Metamodel (Freire et al., 2012).
AModelingEnvironmentforNormativeMulti-AgentSystems
453
connections from the previous step. Additionally,
the graphic element of the entity AgentRoleClass
was changed due to the removal of stereotypes
<<duty>> and <<right>>. Similarly, the graphic
element of OrganizationClass was changed due to
the removal of the stereotype <<axiom>>.
Mapping Model Extension: This model is created
from a combination of the models defined in the
previous steps, including validation rules to
verificate the correctness of the models generated
by constraints in OCL (OCL, 2012).
Finally, following the generative approach
(Czarnecki and Eisenecker, 2000), the code of the
tool is generated based on the extended platform
specific model. Figure 3 shows the overview of the
diagram editor for norms.
5 CASE STUDY
To illustrate the use of the new version of MAS-ML
tool presented in Section 4, we used the Conference.
Management System. This case study has been used
by several authors, as Figueiredo (2011), Zambonelli
et al. (2001), Dignum (2004) and Harmon et al.
(2008), to illustrate their approaches.
The conference management systems are used to
select the papers to be published in a scientific
event. For this, authors should submit your papers
before a certain date (deadline), from which the
evaluators (at least three) initiate the review process.
After the end of review period, event organizers
must disclose the results.
Figure 2: Graphic elements for entities: (i) Norm, (ii) AgentRoleClass and (iii) OrganizationClass, and the relationships:
(iv) Context, (v) Restrict, (vi) Reward and (vii) Punishment.
Figure 3: Overview of the norm diagram implemented in MAS-ML tool.
ICEIS2013-15thInternationalConferenceonEnterpriseInformationSystems
454
Authors can: (i) submit your papers (full or
short) until the submission date, (ii) check the status
of your paper and (iii) view review of the evaluators.
Evaluators may: (i) submit papers and (ii) evaluate
the papers that were listed by event organizers. The
organizers may (i) extend the submission period of
papers, (ii) choose the evaluators and (iii) disclose
the results of the review.
5.1 Identification of the System's
Entities
In the Conference Management environment we can
identify the main organization Conference and the
agent type: user agent, that can play the roles: author
speaker, organizer, conference chair, website manager
and reviewer. These roles are defined by main
organization along with object role submitted. The
instances of this role are played by instances of the
class Paper, which has two subclasses: ShortPaper and
FullPaper. Figure 4 shows the class diagram,
specifying the relationships between classes and
environments. Figure 5 illustrates the role diagram
identifying the roles played by agents and objects in the
organization Conference. Figure 6 shows the
organization diagram referring to Conference. Thus, we
define the classes of the organization along with main
classes of agent, object, and agent role and object role.
5.2 Norms for System
For the conference management system, the
following norms are defined:
N1: The organizers are prohibited from submit
papers.
N2: Reviewers are allowed to submit papers.
N3: The reviewers are prohibited from reviewing
their own papers.
N4 (Punishment for violation of N3): The
reviewers that violate N3 should have their role
canceled.
N5 (Punishment for violation of N3): The president
of the conference is obliged to drop paper.
Figure 7 shows the modeling of the norms N1 and
N2. In addition, Figure 8 shows the modeling of the
norms N3, N4 and N5.
Figure 4: Class Diagram.
Figure 5: Role Diagram related to organization Conference.
6 CONCLUSIONS AND FUTURE
WORKS
This paper presents the evolution of MAS-ML tool
in order to support the modeling of normative multi-
agent systems through the norm diagram defined by
Freire et al. (2012) along with the model checking
based on the NorMAS-ML metamodel. Thus,
structural errors can be detected in advance, at
design time, ensuring good formation of the models
generated within the tool.
The approach to the evolution of MAS-ML tool
is model-based where, graphic elements of the entity
Norm and its relationships Context, Restrict and
AModelingEnvironmentforNormativeMulti-AgentSystems
455
Figure 6: Organization Diagram represents the main organization Conference.
Figure 7: Modeling of norms N1 and N2 presented through the norm diagram.
Sanction (Reward and Punishment), defined in
NorMAS-ML, were incorporated into the tool, and
the entities AgentRoleClass and OrganizationClass
had their graphic elements changed as the new
definition in NorMAS-ML. Thus, in its new version,
the tool supports the modeling of the entities that
compose the normative multi-agent systems and
their properties.
As future work, the checking of conflicts
between norms can be cited. For this, the approach
ICEIS2013-15thInternationalConferenceonEnterpriseInformationSystems
456
Figure 8: Modeling of norms N3, N4 and N5 presented through the norm diagram.
proposed by Figueiredo (2011) may be used. In
addition, the code generation from diagrams
generated by the tool can also be cited. For this task,
a model-based approach similar to that proposed by
Lopes et al. (2012) may be used.
REFERENCES
Bellifemine, F. L.; Caire, G.; Greenwood, D. (2007).
Developing Multi-Agent Systems with JADE. [S.l.]:
Wiley (Wiley Series in Agent Technology).
Bordini, R. H.; Wooldridge, M.; Hübner, J. F. (2007).
Programming Multi-Agent Systems in AgentSpeak
using Jason (Wiley Series in Agent Technology), John
Wiley & Sons.
Czarnecki, K.; Eisenecker, U. (2000). Generative
Programming - Methods, Tools, and Applications,
Adison-Wesley, June 2000.
De Maria, B. A.; Silva, V. T.; Lucena, C.J.P.; Choren, R.
(2005). VisualAgent: A Software Development
Environment for Multi-Agent Systems. Proceedings of
the 19º Simpósio Brasileiro de Engenharia de
Software (SBES 2005), Brazil.
Dignum, V. (2004). A model for organizational
interaction: based on agents, founded in logic. PhD
dissertation, Universiteit Utrecht, SIKS dissertation
series 2004-1.
Eclipse Platform (2012), http://www.eclipse.org/, acessado
em 20 de Março de 2012.
Farias, K.; Oliveira, K.; Nunes, I.; Silva, V. T.; Lucena,
C.J.P. (2009). MAS-ML Tool: Um Ambiente de
Modelagem de Sistemas Multi-Agentes, V Workshop
on Software Engineering for Agent-oriented Systems
(SEAS 2009), Fortaleza, Brasil, pp. 1-12.
Feijó, A. R. (2012). Evolução da Ferramenta MAS- ML
tool para a Modelagem dos Diagramas de Papéis e
Sequência. Monografia. Fortaleza: UECE,
Departamento de Computação.
Figueiredo, K. (2011). Modeling and Validation Norms in
Multi-Agents Systems. Dissertação de Mestrado.
Niterói: UFF, Instituto de Computação.
Figueiredo, K.; Silva, V. T. (2010). NormML: A Modeling
Language to Model Norms. In: 1st Workshop on
Autonomous Software Systems. Salvador, Brazil.
Freire, E. S. S.; Cortés, M. I.; Gonçalves, E. J. T.; Lopes,
Y. S. (2012). A Modeling Language for Normative
Multi-Agent Systems. In: 13th International
Workshop on Agent-Oriented Software Engineering
(AOSE@AAMAS), Valencia (Spain).
AModelingEnvironmentforNormativeMulti-AgentSystems
457
Gonçalves, E. J. T.; Farias, K.; Cortés, M. I.; Feijó, A. R.;
Oliveira, F. R.; Silva, V. T. (2011). MAS-ML TOOL
- A Modeling Environment for Multi-agent Systems,
In: The International Conference on Enterprise
Information Systems (ICEIS 2011), Beijing, China.
Harmon, S. J.; Deloach, S. A. (2008). Trace-based
Specification of Law in Guidance Policies for
Multiagent Systems. Engineering Societies in the
Agents World VIII, Springer-Verlag Berlin,
Heidelberg.
Jennings, N. R. (1996). Coordination Techniques for DAI.
In: Foundations of distributed artificial intelligence.
Editores Greg O‟hare; Nicholas Jennings. John Wiley
and Sons.
Lopes, Y. S. ; Gonçalves, E. J. T. ; Cortés, M. I. ; Freire,
E. S. S. (2012). A MDA Approach Using MAS-ML
2.0 and JAMDER. In: 13th International Workshop on
Agent-Oriented Software Engineering
(AOSE@AAMAS), 2012, Valencia (Spain).
OCL (2012), disponível em:<http://www.eclipse.org/
modeling/mdt/?project=ocl>, acessado em 20 de
Março de 2012.
Pokahr, A.; Braubach, L.; Lamersdorf, W. (2003).
Jadex: Implementing a BDI-Infrastructure for JADE
Agents. EXP - In Search of Innovation (Special Issue
on JADE), vol. 3, no. 3 , Telecom Italia Lab, Turin,
Italy, S. 76-85.
Silva, V. T.; Choren, R.; Lucena, C. J. P. de (2007).
MAS-ML: A Multi-Agent System Modeling
Language. Conference on Object Oriented
Programming Systems Languages and Applications
(OOPSLA); In: Companion of the 18th annual ACM
SIGPLAN Conference on Object-oriented
programming, systems, languages, and applications;
Anaheim, CA, USA, ACM Press, pp. 304- 305.
Silva, V.; Cortés, M.; Lucena, C. (2004). An Object-
Oriented Framework for Implementing Agent
Societies. MCC32/04. Technical Report, PUC-Rio.
Rio de Janeiro, Brazil.
Sommerville, I. (2007). Engenharia de Software. 8 ed. São
Paulo: Pearson Addison-Wesley.
Meyer, J. J.; Wieringa, R. J. (1993). Deontic logic in
computer science: normative system specification.
Deontic logic in computer science: normative system
specification, John Wiley and Sons Ltd. Chichester,
UK.
Zambonelli, F.; Jennings, N. R.; Wooldridge, M. J. (2001).
Organizational Rules as an Abstraction for the
Analysis and Design of Multi -Agent Systems. In:
International Journal of Software Engineering and
Knowledge Engineering, Volume 11, Number 3.
ICEIS2013-15thInternationalConferenceonEnterpriseInformationSystems
458
