MAS Ontology: Ontology for Multiagent Systems
Felipe Cordeiro
1
, Vera Maria B. Werneck
1
, Neide dos Santos
1
and Luiz Marcio Cysneiros
2
1
Universidade do Estado do Rio de Janeiro, Master Program in Computational Science, Rio de Janeiro, Brazil
2
York University, School of Information Technology, Toronto, Canada
Keywords: Agent-based Software Engin
eering, Domain Ontology, Comparison of Agent-oriented Methodologies.
Abstract: This work describes the Multiagent Systems (MAS) Ontology to assist in the development of multi-agent
system using different methodologies. The MAS Ontology consists of fragmenting agent-oriented
methodologies following an ontology approach based on the best aspects of four prominent AOSE
methodologies and Guardian Angel exemplar that identify the strengths, weaknesses, commonalities and
differences. In this paper, we present a brief explanation of Multiagent methodologies and the step-by-step
process to describe the agent-based systems domain and how it can be represented. Given the numerous works
in the literature about MAS methodologies, our aim is to help select the best and more appropriate properties
to be used in Multiagent Systems development.
1 INTRODUCTION
In the past two decades, the agent technology
approach has been considered as a new paradigm for
developing complex systems. This approach has
attracted an increasing amount of interest from the
research community and has demonstrated its
potential in many fields, such as: (i) working with
different types of distributed devices (e.g., sensor
networks, mobile phones, and personal computers),
(ii) enabling various types of communication and data
exchange (e.g., audio and video), and (iii) ability to
dynamically adapt to the ever changing requirements
and dynamic operating environment (Munroe et al.,
2006), (Pěchouček and Mařík, 2008), (Dam and
Winikoff, 2013).
Agent-oriented systems must be built in terms of
autonomous task-oriented entities. They need to be
organized to interact (cooperate, coordinate and
negotiate) with one another. To adopt the agents’
perspective requires a new set of tools to support
software development (Cernuzzi, Cossentino and
Zambonelli, 2005).
Currently, we are faced with a multitude of
different frameworks, some of them even supported
by tools. However, very few methodologies are broad
enough to support the whole software development
life cycle or to support the complexity of developing
such systems. Years ago, Luck, Mcburney and Preist
(2003) stated: "One of the most fundamental
obstacles to large-scale take-up of agent technology
is the lack of mature software development
methodologies for agent-based system".
In this work, we assume "methodology" as a set
of phases that a practitioner must go through to design
an agent-based system. We see a methodology as
being composed of general concepts (deals with the
question of whether a methodology adheres to the
basic notions of agents and multiagent systems),
specific concepts (underlying one particular
capability or a characteristic), notation (symbols used
to represent elements), modeling techniques (set of
models that depict a system at different levels of
abstraction and different aspects of the system),
process (development aspect) and pragmatics
(practical implementation aspects) (Sturm and
Shehory, 2004).
The main goal of this paper is to provide an
ontology structure for selecting the best and more
appropriate artefacts to be used to develop one
particular Multiagent Systems, the MAS Ontology. It
is motivated by a large number of existing approaches
and supported by our experience in using some of
them. It is important to notice that this work does not
claim nor intends to be complete. It is expected to be
a first approach that will be perfected overtime but
that will yet be of importance to help developers to
use the best each of the current four (Gaia, MaSE,
Prometheus and Tropos) methodologies covered in
this work has to offer.
The main goal of the ontology proposed in this
work is to capture and facilitate the reuse of
536
Cordeiro, F., Werneck, V., Santos, N. and Cysneiros, L.
MAS Ontology: Ontology for Multiagent Systems.
In Proceedings of the 18th International Conference on Enterprise Information Systems (ICEIS 2016) - Volume 1, pages 536-543
ISBN: 978-989-758-187-8
Copyright
c
2016 by SCITEPRESS Science and Technology Publications, Lda. All rights reserved
knowledge gained through the evaluation of several
MAS methodologies based on more than 20 projects
developed with different methodologies using
Guardian Angel (GA) Exemplar proposed by Yu and
Cysneiros. However, to complement the ontology we
added experiences extracted from other well know
evaluation studies from the literature. We populated
the ontology with the four methodologies because
they were evaluated by GA, Sturm and Shehory
(2004, 2014) and Dam and Winikoff (2004, 2013).
2 AGENT-ORIENTED
METHODOLOGIES
Cernuzzi et.al. (2005) suggests a clean and
disciplined approach to analyzing, designing and
developing multiagent systems, using specific
methodologies and techniques by means of notations,
diagrams and tools to support the development.
We assume that each method has strengths and
weakness, and these characteristics may influence the
use of one methodology over another for one specific
project. To validate the MAS Ontology we used four
methodologies, namely, Gaia (Zambonelli, Jennings
and Wooldridge, 2003), (Wooldridge, Jennings and
Kinny, 2000), MaSE (Deloach, 2001), (Deloach,
2004), Prometheus (Padgham and Winikoff, 2002),
(Padgham and Winikoff, 2003) and Tropos (Bresciani
et al, 2004).
Jennings and Wooldridge proposed Gaia in 1999.
It was extended and modified by Zambonelli in 2000
(Wooldridge, Jennings and Kinny, 2000), finally
Zambonelli, Jennings and Wooldridge presented a
stable version in 2003 (Zambonelli, Jennings and
Wooldridge, 2003). Unlike many other
methodologies, Gaia starts from modelling
requirements. Later it guides developers to a well-
defined design for the multiagent system, that way
programmers can easily model and implement it,
while dealing with the characteristics of complex and
open multiagent systems.
MaSE methodology is heavily based on UML and
RUP. It is divided into seven phases: capturing goals,
applying use cases, refining roles, creating agent
classes, constructing conversations, assembling agent
classes and system design (Deloach, 2001), (Deloach,
2004).
Prometheus is an iterative methodology covering
the complete software engineering process while
aiming at the development of intelligent agents (in
particular BDI agents). The concepts applied are
goals, beliefs, plans, and events, resulting in a
specification that can be implemented with JACK
(Coburn, 2000). Prometheus covers three phases: the
system specification, architectural design phase,
detailed design phase (Padgham and Winikoff, 2002),
(Padgham and Winikoff, 2003).
Tropos relies on the notion that an agent is based
on goals and tasks adopted by the i* framework (Yu,
2009) and offers supports to applications, particularly
for the development of BDI agents and the agent
platform JACK. (Coburn, 2000). Tropos consists of
four phases: early requirements, late requirements,
architecture design, detailed design and
implementation (Bresciani et al, 2004), (Tropos,
2014), (Coburn, 2000).
3 EVALUATION OF AGENT
ORIENTED METHODOLOGIES
Several evaluations of agent orientated methodologies
have been published (Dam and Winikoff, 2014),
(Sturm and Shehory, 2014), (Dam, 2003), (Dam and
Winikoff, 2004), (Tran and Low, 2005), (Elamy and
Far, 2008), (Iglesias, Garijo and González, 1999),
(Cernuzzi, Rossi and Plata, 2002), (Sure, Staab and
Studer, 2002). Sturm and Shehory (2004, 2014), and
Dam and Winikoff (2004, 2013), (Dam, 2003) were the
most cited works in the MAS area.
Sturm and Shehory (2004), proposed a
framework for quantitative and qualitative evaluation
of MAS methodologies (Gaia, MaSE and Tropos). It
explores the following aspects: concepts, properties,
notations and modeling techniques, process and
pragmatics. Dam and Winikoff (2004, 2013), (Dam,
2003) illustrate the strengths and weaknesses of
MaSE, Prometheus and Tropos methodologies
through an attribute-based evaluation process.
The Guardian Angel (GA) Exemplar proposed by
Yu and Cysneiros (Yu and Cysneiros, 2002) defines
a set of questions to evaluate the behaviour of MAS
methodologies and is expressed in terms of a set of
numbered scenarios. The GA is an easily
comprehended open system that provides automated
support to assess patients with chronic diseases
through a set of “guardian angel” software agents.
The GA exemplar is a complete solution, with a
practical, real and significant enough example, to test
and verify how the methodology behaves in close-to-
real situations. The primary concern of the exemplar
is to highlight the strengths, weaknesses and
potentials of each methodology justified by the
artefacts (work products) that can be used to answer
the methodology questions.
MAS Ontology: Ontology for Multiagent Systems
537
We chose to use the GA exemplar as it was the
only one we found in the literature that proposes
complex situations that can be used empirically to
evaluate different methodologies that go beyond toy
problems.
4 DOMAIN THEORY:
AGENT-ORIENTED
METHODOLOGIES
In order to define a Domain Theory for Agent-
Oriented Methodologies, we have compiled the
knowledge gathered from papers on AOSE
methodologies listed in section 2 (Dam and Winikoff,
2013), (Luck, Mcburney and Preist, 2003), (Sturm
and Shehory, 2014), (Tran and Low, 2005), (Elamy
and Far, 2008) together with the results from our
experience using the Guardian Angel exemplar over
the past 10 years.
While building the MAS Ontology, we tried to
answer the following research questions: (i) in what
situations is a methodology or method fragment best
applied?; (ii) which instruments are used to define the
methodological questions from GA and from the
works of (Dam and Winikoff, 2013), (Luck,
McBurney and Preist, 2003), (Sturm and Shehory,
2014), (Tran and Low, 2005), (Elamy and Far, 2008)
(iii) what are the general concepts of agents that a
MAS methodology should support?; (iv) what are the
specific concepts of agents that a MAS methodology
can support?; (v) what are the notations and modeling
techniques found in the methodology?; (vi) what are
the support resources offered by the methodology?
4.1 Approach
In the ontology, we assembled the knowledge
generated by using the GA exemplar pertinent to four
different methodologies (Gaia, MaSE, Prometheus
and Tropos). We organized the knowledge and
experiences gained by signaling which work product
is responsible for a certain task when answering the
questions listed above while applying the exemplar to
each of the aforementioned methodologies.
4.2 MethodBase GA: Experience
Modeling with GA
The MethodBase GA is the knowledge base that
compiles the work done over many years by MSc and
last year undergrad Computer Science students.
During this time, these students modeled multiagent
systems using methodologies such as Gaia, MaSE,
Prometheus and Tropos and relying on scenarios
proposed in Guardian Angel exemplar. After
modeling the solutions, the students answered the
methodological issues in accordance with strengths,
neutral or weakness, as seen in figure 1.
Figure 1: Methodbase GA.
4.3 Evaluated MethodBase: Evaluating
Methodologies
The Evaluated MethodBase is the knowledge base
built based on the work of Sturm and Dam (Sturm and
Shehory, 2014), (Dam and Winikoff, 2013), as
illustrated in figure 2. The proposed Evaluated
MethodBase includes the following concepts:
General Concepts of MAS, Specific Concepts of
MAS, Notations, Modeling Techniques, Process and
Pragmatics (practical aspects).
The ranking of values ranges from 0 to 6, where 0
represents cases where a certain characteristic is not
applicable, 1
Refers to but not detailed, 2 Limited, 3 Neutral, 4
Small issues, 5 Minor deficiencies
and 6 is the ideal
efficiency. This was an adaptation of (Sturm and
Shehory, 2004), (Dam and Winikoff, 2002) using the
databases Methodbased GA and Evaluated
Methodbase.
5 ONTOLOGY LEARNING FROM
EXPERIENCE
Many definitions of ontology can be found in the
literature. However, Sure (Sure, Staab and Studer,
2002) provides a simple and comprehensive
definition: "An ontology is a formal and explicit
specification of a shared conceptualization". In this
definition "formal" means readable by computers;
"explicit specification" refers to concepts, properties,
relations, functions, constraints, axioms, explicitly
defined; "shared" means consensual knowledge, and
"conceptualization" refers to an abstract model of
ICEIS 2016 - 18th International Conference on Enterprise Information Systems
538
some phenomenon in the world real. The ontology
built in this work was based on a middle-out strategy
(Uschold and King, 1995), in which concepts were
generalized and specialized.
Figure 2: Evaluated Methodbase.
Building on identify concepts (terms) that can
provide short assertive sentences, we developed an
ontology based on the knowledge acquired from both
databases Methodbased GA and Evaluated
Methodbase.
The ontology development is defined through
seven stages: Ontology Specification, Knowledge
Acquisition, Conceptualization, Formalization,
Integration, Implementation and Evaluation.
The Ontology Specification is used to prepare a
document using natural language, containing
information such as the primary ontology goal and its
other purposes.
Knowledge Acquisition focuses on possible
sources of knowledge. In this survey the GA
experiences were used in order to manage the data
collected, analyzed and categorized according to their
degree of strength, weakness or neutrality.
Conceptualization focuses on structuring the
domain knowledge into a conceptual model and was
based on the acquired vocabulary in the previous
phases, in order to describe the problems and their
possible solutions
In the Formalization, the concepts are now
formally written through OWL. The Protégé tool
version 4.3 (Protege, 2000) was used, and the first
preliminary version of the ontology was generated. At
this stage, a taxonomy that shows the processes of a
multiagent system is available.
The Integration stage obtains the representative
experimental ontology from the Guardian Angel
exemplar and is re-evaluated to better address the
domain of multiagent systems.
At this stage, other studies on the comparison of
methodologies are integrated. (Sturm and Shehory,
2004).
The Implementation used the Pellet, a Protégé
plugin to automatically check the ontology
consistency and also takes into account the
experience of validating the data, as well as
establishing the comparable relationship between the
Methodbase GA and Evaluated Methodbase values,
classes and attributes. Each phase of the ontology is
related to models, tables or charts, which serve to
guide the building process of the MAS Ontology, here
defined as products, as seen in figure 3.
Figure 3: MAS Ontology.
MAS Ontology: Ontology for Multiagent Systems
539
6 MAS ONTOLOGY AND
RESULTS
The MAS ontology has three main classes:
Methodology, Phases and Work_Products.
The class Methodology focuses on the
methodologies that are the study objects (Tropos,
ADELFE, Gaia, Prometheus, MaSE, and
MESSAGE).
The Phases class combines the characteristics
essential to multiagent systems (General Concepts,
Modeling Techniques, Notation, Pragmatics, Process
and Specific Concepts). Each class has a set of
attributes associated with it. (e.g. General Concepts
attributes such as: Autonomy, Reactiveness,
Sociality, Proactiveness, Reasoning, Mobility).
The class Work_Products lists the necessary
artifacts to build a multiagent system.
Figure 3 shows a simplified MAS ontology. The
Phase class is associated with the Methodology class.
In this relationship, subclasses of Phase are related to
subclasses of Methodology. The class
Work_Products is also listed to illustrate the artefacts
in Figure 4. It is important to determine which
attributes from the Phases class might be associated
with corresponding work products. For example, in
Figure 5 the subclass Tropos_Products has two
phases: Tropos_Analysis and Tropos_Design. Each
subclass has its own subclasses. Tropos analysis is
composed of Actors Diagram and Reasoning
Diagram. Tropos Design consist of Extended Actors
Diagram, Table of Actors and Capabilities, Table of
Agents, Agents Interaction Diagram, Tasks or Plans
Diagram and Capabilities Diagram.
6.1 Schematic Model
In order to facilitate understanding the domain of
multiagent systems by ontological representation, a
Schematic model of MAS Ontology (Figure 6) was
developed to illustrate the relationship between
classes, attributes and expected values.
In Figure 6 the schematic forms are described as
follows:
Ellipse Form - Classes or Subclasses
Rectangle Form - Attributes
Dotted Ellipse - Value types for attributes
Dotted Rectangle Form - Class properties
Figure 4: Work Products Detailed.
Figure 5: Tropos Work Products.
ICEIS 2016 - 18th International Conference on Enterprise Information Systems
540
Figure 6: Schematic model of MAS Ontology.
6.2 Examples
For MAS Ontology population the individuals were
separated into two groups: (i) representing specific
values of GA attributes (Yu and Cysneiros, 2002)
and (ii) representing the set of attributes that make
up a methodology in the evaluation comparison
papers (Sturm and Shehory, 2004, 2014), (Dam and
Winikoff, 2002, 2013). Thereby the query may
return a particular specific situation or a
methodology.
Figure 7a represents an unsuccessful search
carried out in plugin DL Query (Protege, 2000),
where the attribute Mobility (The quality or state of
being mobile) was defined as value 4, and no
individual was found. Figure 8b represents a
successful search done in plugin DL Query, where
the attribute Mobility was defined as value 3
In this scenario, three methodologies were
found (figure 7b). Figure 8 shows the associated
work products (e.g. Mobility Tropos and Mobility
Prometheus) obtained from a refined search for
Tropos and Prometheus.
Figure 7.a: Unsuccessful Search.
Figure 7.b: Successful Search.
MAS Ontology: Ontology for Multiagent Systems
541
Figure 8: Associated Work Products.
7 CORRELATED WORKS
Several works addressing the evaluation of agent
orientation methodologies have been published
(Dam and Winikoff, 2002), (Sturm and Shehory,
2004), (Dam, 2003), (Dam and Winikoff, 2004),
(Tran and Low, 2005), (Elamy and Far, 2008),
(Iglesias, Garijo and González, 1999), (Cernuzzi,
Rossi and Plata, 2002), (Sure, Staab and Studer,
2002), (Casare et al, 2014). They consist of
quantitative and qualitative evaluation framework
based on checklists of certain properties, qualities,
attributes or characteristics of the methodology and
some simple problems.
Tran and Low (2005, 2005a) compared ten
methodologies (Gaia, Tropos MAS
CommonKADS, Prometheus, Passi, ADELFE,
MaSE, RAP, MESSAGE, Ingenius). They used a
criteria checklist that was developed to assess the
resources of the chosen methodologies, covering the
process, techniques and model stages.
Cernuzzi and Zambonelli (2011) used the
multivalued statistical method for quantitative
evaluation of profiles. The goal was to present the
potential profile analysis in the comparison process
for the evaluation of methodologies, searching for
similar evaluations to confirm the results.
Our study differs from similar works by
proposing the use of a knowledge base where the
knowledge is expressed and organized as an
ontology. The ontology can guide the developer to
select fragments of methodologies that best fit the
multiagent system under development. It allows for
queries to be made that can help developers to
customize their development process. It helps them
to search for where the methodologies best fit their
needs considering the specific project at hand.
On another level, it also helps researchers
further developing these methodologies to easily
compare where their approaches fall behind when
compared to other existing methodologies and
therefore, where they should invest more effort to
develop furthere their methodologies.
8 CONCLUSIONS
As a result of the fast dissemination of MAS
methodologies, deciding what methodology to use
in a project is a complex task. Many frameworks
and toolkits are provided, but they do not always
offer support to assist developers in choosing the
best or most appropriate methodology to handle the
project at hand. This paper proposes an ontology-
based support to help developers faced with the
need to use agent-oriented properties to develop
software. The ontology was created based on the
experience gathered by applying the Guardian
Angel exemplar in four agent-oriented software
engineering methodologies, as well as adding the
knowledge obtained from the results from Sturn and
Dam (2004) and Dam (2003). The knowledge base
provided in this ontology can assist developers to
use these methodologies and also to choose better
the adequate artifacts for a particular domain.
The MAS Ontology approach focuses on being
a facilitator for developing a MAS process, as it
concentrates on relationships between the principles
of software engineering evaluation and experience.
Furthermore, it can be extended to suit the
particularities of other AOSE methodologies and
other studies based on statistics, as in (Iglesias,
Garijo and González, 1999).
Future works will address a systematic
validation of the Ontology using case studies where
different groups of randomly selected students will
be asked to develop solutions to a specific problem.
Some students will use the ontology, and another set
of students will have to develop the solution using
pre-determined methodology. Final results will
then be compared.
REFERENCES
Bresciani, P., Perini, A., Giorgini, P., Giunchiglia, F., &
Mylopoulos, J. (2004). Tropos: An agent-oriented
software development methodology. Autonomous
Agents and Multi-Agent Systems, 8(3), 203-236.
Casare, S. J., Brandão, A. A., Guessoum, Z., & Sichman,
J. S. (2014). Method Framework: a situational
approach for organization-centered. MAS.
Autonomous agents and multi-agent systems, 28(3),
430-473.
Cernuzzi, L., & Zambonelli, F. (2011). Improving
comparative analysis for the evaluation of AOSE
methodologies. International Journal of Agent-
Oriented Software Engineering, 4(4), 331-352.
Cernuzzi, L., Cossentino, M., & Zambonelli, F. (2005).
Process models for agent-based development.
ICEIS 2016 - 18th International Conference on Enterprise Information Systems
542
Engineering Applications of Artificial Intelligence,
18(2), 205-222.
Cernuzzi, L., Rossi, G., & Plata, L. (2002, November). On
the evaluation of agent oriented modeling methods. In
Proceedings of Agent Oriented Methodology
Workshop, Seattle, (Vol. 29).
Coburn, M. (2000). Jack intelligent agents: User guide
version 2.0. AOS Pty Ltd..
Cossentino, M., Fortino, G., Garro, A., Mascillaro, S., &
Russo, W. (2008). PASSIM: a simulation-based
process for the development of multi-agent systems.
International Journal of Agent-Oriented Software
Engineering, 2(2), 132-170.
Cossentino, M., Gaglio, S., Galland, S., Gaud, N., Hilaire,
V., Koukam, A., & Seidita, V. (2009). A MAS
metamodel-driven approach to process fragments
selection. In Agent-Oriented Software Engineering IX
(pp. 86-100). Springer Berlin Heidelberg.
Dam, K. H. (2003). Evaluating and comparing agent-
oriented software engineering methodologies
(Doctoral dissertation, School of Computer Science
and Information Technology, RMIT University,
Australia).
Dam H. K., Winikoff M., 2013. Towards a next-
generation AOSE methodology. Science of Computer
Programming, v. 78, n. 6, p. 684-694.
Dam, K. H., & Winikoff, M. (2004, January). Comparing
agent-oriented methodologies. In Agent-Oriented
Information Systems (pp. 78-93). Springer Berlin
Heidelberg.
Deloach, S. (2004). The MaSE methodology.
Methodologies and software engineering for agent
systems, 107-125..
DeLoach, S. A. (2001). Analysis and Design using MaSE
and agentTool. Air force inst of tech wright-patterson
afb oh school of engineering and management.
Elamy, A. H. H., & Far, B. (2008). On the evaluation of
agent-oriented software engineering methodologies: a
statistical approach. In Agent-Oriented Information
Systems IV (pp. 105-122). Springer Berlin
Heidelberg.
Iglesias, C. A., Garijo, M., & González, J. C. (1999). A
survey of agent-oriented methodologies. In Intelligent
Agents V: Agents Theories, Architectures, and
Languages (pp. 317-330). Springer Berlin
Heidelberg.
Luck, M., McBurney, P., & Preist, C. (2003). Agent
technology: enabling next generation computing (a
roadmap for agent based computing).
AgentLink/University of Southampton.
Munroe S. et al., 2006. Crossing the agent technology
chasm: Lessons, experiences and challenges in
commercial applications of agents. The Knowledge
Engineering Review, v. 21, n. 04, p. 345-392.
OMG Group. (2008). Software & Systems Process
Engineering Meta-Model Specification, at
http://www.omg.org/spec/SPEM/2.0/
Padgham, L., & Winikoff, M. (2002, November).
Prometheus: A pragmatic methodology for
engineering intelligent agents. In Proceedings of the
OOPSLA 2002 Workshop on Agent-Oriented
Methodologies (pp. 97-108).
Padgham, L., & Winikoff, M. (2003). Prometheus: A
methodology for developing intelligent agents. In
Agent-oriented software engineering III (pp. 174-
185). Springer Berlin Heidelberg.
Pěchouček M., Mařík V., 2008. Industrial deployment of
multiagent technologies: review and selected case
studies. Autonomous Agents and Multiagent Systems,
v. 17, n. 3, p. 397-431.
PROTEGE (2000). The Protege Project. at
http://protege.stanford.edu.
Sturm, A., & Shehory, O. (2004, January). A framework
for evaluating agent-oriented methodologies. In
Agent-Oriented Information Systems (pp. 94-109).
Springer Berlin Heidelberg.
Sturm, A., & Shehory, O. (2014, January). The landscape
of agent-oriented methodologies. In Agent-Oriented
Software Engineering (pp. 137-154). Springer Berlin
Heidelberg.
Sure, Y., Staab, S., & Studer, R. (2002). Methodology for
development and employment of ontology based
knowledge management applications. ACM SIGMOD
Record, 31(4), 18-23.
Tran, Q. N. N., & Low, G. C. (2005). Comparison of ten
agent-oriented methodologies. Agent-oriented
methodologies, 341-367.
Tran, Q. N. N., Low, G., & Williams, M. A. (2005). A
preliminary comparative feature analysis of multi-
agent systems development methodologies. In Agent-
Oriented Information Systems II (pp. 157-168).
Springer Berlin Heidelberg.
Uschold, M., & King, M. (1995). Towards a methodology
for building ontologies (pp. 15-30). Edinburgh:
Artificial Intelligence Applications Institute,
University of Edinburgh.
Wooldridge, M., Jennings, N. R., & Kinny, D. (2000).
The Gaia methodology for agent-oriented analysis
and design. Autonomous Agents and multi-agent
systems, 3(3), 285-312.
Yu, Eric (2009). Social Modeling and i*. In Conceptual
Modeling: Foundations and Applications (pp. 99-
121). Springer Berlin Heidelberg.
Yu, E., & Cysneiros, L. M. (2002, May). Agent-oriented
methodologies-towards a challenge exemplar. In
Proc of the 4 Intl. Bi-Conference Workshop on AOIS,
Toronto (Vol. 151).
Zambonelli, F., Jennings, N. R., & Wooldridge, M.
(2003). Developing multiagent systems: The Gaia
methodology. ACM Transactions on Software
Engineering and Methodology (TOSEM), 12(3), 317-
370.
MAS Ontology: Ontology for Multiagent Systems
543