MODELING OF OPEN NORMATIVE MULTIAGENT SYSTEMS

Carolina Howard Felicíssimo*, Carlos José Pereira de Lucena* and Jean-Pierre Briot**

* DI/PUC-Rio, Rua Marquês de São Vicente 225, 22453-900, RJ, Brazil

** LIPVI/ParisVI, Avenue du Président Kennedy, 75016, Paris, France

Keywords: Normative Open Multiagent Systems, Electronic Agent-Based Organizations/Institutions.

Abstract: A major challenge in the research of multiagent systems (MAS) is the design and implementation of open

MAS in which norms can be effectively applied to their agents and easily managed. These tasks are arduous

because norms are usually written for general purposes, hindering a more precise regulation. The motivation

for this research came forth from the need to resolve this challenge, providing an approach applicable in

open systems. In such systems, heterogeneity and autonomy rule out any assumption concerning the way

third-party entities are implemented and behaved. This paper summarizes the result of a study done on solu-

tions for the modeling of MAS. That study motivates the development of our DynaCROM approach.

1 INTRODUCTION

Three main observations summarize the starting

point of this research. Firstly, “autonomous agents

and MAS represent a new way of analyzing,

designing and implementing complex software

systems” (Jennings, 1998). Those systems are

usually formed by rich social interactions, i.e., by

agents cooperating, coordinating and/or negotiating

(Jennings, 2001).

Secondly, with the Web evolving towards a Se-

mantic Web (Berners-Lee et al., 2001), it is believed

that available information will be presented in a

meaningful way for allowing not only humans to

process its content, but also (software) agents. In this

scenario, agents will be able to migrate among MAS

in order to obtain resources and/or services not

found in their original systems. Thus, if one main

contribution of Semantic Web can be singled out, it

has to be openness. Openness will permit new types

of applications for MAS, as ubiquitous systems

(Weiser, 1991), in which dynamicity, due to inter-

nal/external events, is a key characteristic.

Thirdly, considering that MAS will be open in

nature, norms play a central role in the social phe-

nomena occurring in the MAS field, which is mov-

ing more and more from the individual, cognitive

focused agent models to models of socially situated

agents. In normative MAS (NMAS), the main posed

question is: “How to ensure efficiency at the level of

MAS whilst respecting individual autonomy?” (Boel-

la, 2006). NMAS as an area of research has become

a major issue in the MAS field and it can be situated

at the intersection of normative systems and MAS.

Following these three main observations, it is be-

lieved that upcoming information systems will be

implemented as open MAS formed by several goal-

oriented problem-solving entities.

Openness has consequences for the design, im-

plementation and use of information, requiring novel

modeling primitives and methods in order to make a

MAS a real application. Solutions for open MAS

must deal with issues inherent to open environments,

namely: heterogeneity of agents; trust and accounta-

bility; exception handling (detection, prevention and

recovery from failures that may jeopardize the glob-

al operation of the system); and, societal change

(capability of accommodating structural changes)

(Dignum et al., 2007; Al-Muhtadi et al., 2003).

A very dynamic, open and distributed domain –

like the Semantic Web and applications for ubiquit-

ous computing, both that can be implemented by

MAS – is always subject to unanticipated events

(Hewitt, 1991), caused by malicious agents that do

not conform to recommendations of correct and

incorrect behaviors. This risk imposes the necessity

for regulatory mechanisms for preventing undesira-

ble actions to happen and, consequently, to inspire

trust for the members of the system.

This paper presents a study that focuses on solu-

tions for the modeling of MAS. The study means to

be an overview of the existing works on multiagent

organisations and normative multiagent systems.

The result of that study motivates the development

of our DynaCROM approach (meaning Dyna

mic

540

Felicíssimo C., Lucena C. and Briot J. (2009).

MODELING OF OPEN NORMATIVE MULTIAGENT SYSTEMS.

In Proceedings of the International Conference on Agents and Artiﬁcial Intelligence, pages 540-546

DOI: 10.5220/0001663705400546

 SciTePress

ontextual Regulation Information Provision in

Open MAS). From the individual agents’ perspec-

tive, DynaCROM is an information mechanism that

makes application agents aware of the norms they

are bound to at a given moment. From the system

developers’ perspective, DynaCROM is a methodol-

ogy that operationalizes regulative norms in open

MAS, enabling developers to embody abstract

norms with domain values.

Further details about specific aspects of Dyna-

CROM can be found in more specialized publica-

tions. In (Felicíssimo et al., 2008b), the guidelines to

operationalize regulative norms in NMAS by using

DynaCROM are presented. Following those guide-

lines, concrete norms are reached from abstract ones,

operationalized in a NMAS. In (Felicíssimo et al.,

2008a), the details about how DynaCROM contex-

tualizes norms in a NMAS, from the perspectives of

individual agents and the system developer, are

provided. In (Felicíssimo et al., 2008c), a case study

from the television domain and, in (Felicíssimo et

al., 2007), a case study from the domain of multina-

tional corporations are presented.

The remainder of this paper is organized as fol-

lows. Section 2 gives an overview of the foundations

upon the modeling of MAS. Basic theories and re-

lated research fields are analyzed in order to provide

readers with a better understanding about the con-

cepts and ideas described in this paper. Section 3

discusses current solutions for MAS. Section 4 fina-

lizes the paper by presenting our conclusions.

2 MODELING OF MAS

Traditional modeling of MAS often assumes an

individualistic perspective in the sense that agents

are considered autonomous entities, pursuing their

own individual goals based on their own beliefs and

capabilities. Even in this perspective, global behav-

iour emerges from individual interactions and, there-

fore, the modeling has to be expanded to consider

not only an agent-centric view, but also societal and

organizational-centric views. Furthermore, the

overall problem of analyzing the social, legal, eco-

nomic and technological dimensions of an agent

organization is not normally considered when, ide-

ally, it should be resolved.

Agent-centered approaches can be useful for

closed systems, composed of a small number of

agents, but they fail to design open systems

(Rodríguez-Aguliar, 2001; Esteva et al., 2003). For

instance, in critical applications such as those within

business, environments or government agencies

(hospitals, police, justice, etc.), the structural charac-

teristics of the domain have to be incorporated. That

is, the design of an agent society must also consider

organizational characteristics such as stability over

time, some level of predictability, commitment to

aims and strategies, and so on.

The idea of modeling MAS as organizations was

early proposed by (Gasser et al., 1987; Pattison et

al., 1987; Corkill and Lesser, 1983; Werner, 1987)

and it is still a major issue in the MAS research

field, especially in applications on the areas of Ser-

vice Oriented Computing, Grid Computing and

Ambient Intelligence. Recently, the subject of MAS

design from the organizational perspective has been

mainly discussed in the COIN workshop (COIN,

URL), which has been held yearly since 2005, as a

dual event co-located within large international

conferences of the area in different geographic re-

gions.

Even with this research effort, organizational ap-

proaches have not been a common use in MAS,

which is usually seen as a pure aggregation of

agents. The fact that organizational approaches have

not been effectively adopted suggests that some

work still needs to be done in providing better tools

for the design and implementation of MAS in which

intrinsic characteristics of the application domain

(e.g., society structure) can be considered. Moreo-

ver, this necessity increases when considering open

systems from particular ‘cultures’ (i.e., “the predo-

minating attitudes and behavior that characterize

the functioning of a group or organization” (EDic-

tionary, URL)).

In the next subsections, two major research lines

for the modeling of MAS will be presented and,

then, discussed. The first research line proposes the

modeling of MAS based on organizations. The

second one proposes the modeling of MAS based on

the electronic institutional aspects of organizations.

By ‘electronic institutional’ aspects, the authors

mean an organization restricted through the defini-

tions of all the following: related roles, common

language, valid interactions and set of norms.

2.1 Electronic Agent-Based

Organizations

The definition of the organization term usually va-

ries between two meanings for MAS researchers. In

the first meaning, an organization is often unders-

tood as an entity with identity that represents (not

identical) groups of agents. In the second meaning,

an organization is often understood as constraints

(structures, norms and patterns) found in a social

context that shapes the actions and interactions of

agents (Coutinho et al., 2005).

MODELING OF OPEN NORMATIVE MULTIAGENT SYSTEMS

541

The first sense of an organization comes from an

administrative/economic point of view: organiza-

tions are like enterprises that perform some service

or produce some goods. The second sense comes

from a sociological point of view: an organization is

better called the social organization implicitly or

explicitly present in a society, community or groups

of agents that shape the interactions among agents.

These two meanings of an organization are not

mutually exclusive; the second meaning is more

general than the first one. Thus, it is natural to say

that every organization (first meaning) has a social

organization (second meaning), but the opposite is

not always true – every social organization (second

meaning) does not always give rise to an organiza-

tion (first meaning).

Considering the case that every organization has

a social organization, the latter is materialized in the

first one by the specification of the structure and

objectives of the system. Thus, a social organization

is envisioned by the organization as a whole and by

describing the activity of the system as realized by

the individual agents (Vázquez-Salceda et al., 2005).

In this sense, the organizational dimension covers

both the organization and the agent perspectives in

the design of agent societies.

The work on MAS modeling based on the orga-

nizational dimension mainly started with the emer-

gence of the HarmonIA (Vázquez-Salceda and Dig-

num, 2003) and OperA (Dignum, 2004) formal

frameworks. HarmonIA provides the way to model

especially highly regulated electronic organizations

from the abstract level, where norms are usually

defined, to the final protocols and procedures that

implement those norms. The HarmonIA framework

also incorporates ontologies to describe and connect

different levels of norms.

OperA is a formal specification framework that

focuses on the organizational dimension, properly

modeling not only organizational structures in an

agent society (that structures the global behavior of

the society), but also the aims and behavior of the

agents from the agent perspective. The framework

also explicitly provides a solution for ontological

descriptions of agent interactions.

In (Vázquez-Salceda et al., 2005), the O

rganiza-

tional Model for Normative Institutions (OMNI)

framework is presented, resulting from the combina-

tion of some aspects of the HarmonIA and OperA

frameworks. The OMNI framework focuses on the

organization dimension (that also structures the

global behavior of the society), on the behavior of

the agents from the agent perspective, on agent inte-

ractions and on a normative structure that is separate

from the agents that will populate the MAS.

In order to support the development of closed-

systems and open, flexible environments, OMNI

presents a rigid specification of its structure, defin-

ing particular fields for the description of scenes,

roles and groups of roles. There are no normative

aspects further than the ones for organizations, roles,

group of roles, agent interactions and agents (only

norms for roles, group of roles, scene and transition

can be specified). The organization entity is not

explicitly present. An organization is formed by

listing all its institutional roles (e.g., managers, di-

rectors, president, etc.) and represented when agents

play those roles. Currently, OMNI does not provide

a solution for the implementation and integration of

its specifications in a given MAS.

Another important line of research, based on or-

ganizational models for MAS, is mainly proposed by

Sichman, Boissier and their colleagues with their

work started with

OSE (Hannoun et al., 2000).

OSE is an organizational model for MAS based on

three major concepts: the roles which constrain the

individual behaviors of agents, the organizational

links that regulate social exchanges between agents

and the groups which constrain the layout of agents

involved in strong interactions.

In (Hübner et al., 2002), the work on

OSE

evolved resulting in the

OISE

model.

OISE

permits the specification of a MAS organization

along the structural and functional dimensions,

which can be specified independently of one anoth-

er. Furthermore,

OISE

makes explicit the deontic

relation which exists between both dimensions. In

short, the

OISE

organizational model enables the

declaration of the MAS organizational structure

(roles, groups and links), functioning (a set of global

goals and plans), obligations and permissions.

2.2 Electronic Agent-based Institutions

The idea of modeling MAS as institutions came

from the observation that human institutions (North,

1990) have been successfully mediating human

interactions for centuries and, so, EI (meaning E

lec-

tronic Institution(s)) may cope with a similar re-

sponsibility within agent societies. The aim of the

proposal is to promote a natural extension of human

institutions by permitting not only humans, but also

autonomous agents to interact with one another in a

reliable way. This way, EI can be seen as the elec-

tronic counterpart of a human institution in which

interactions between agents are articulated through a

role-based multiagent protocol specification.

ICAART 2009 - International Conference on Agents and Artificial Intelligence

542

The work on formalization of EI has been done

for years and it is extensively presented mainly in

(Noriega, 1997), (Rodríguez-Aguilar, 2001) and

(Esteva, 2003). In (Noriega, 1997), the different

components of an institution are introduced by using

a typical trading institution – the fish market auction

houses – as a motivating example. Noriega proposes

that an institution is defined by: (i) a set of roles and

relationships within them, (ii) a common ontology

and communication language which allow heteroge-

neous agents to exchange knowledge, (iii) the valid

interactions that agents may have structured in con-

versations, and (iv) a set of rules of behavior which

determine the actions that agents must take under

certain circumstances.

In (Rodríguez-Aguilar, 2001), the formalization

of EI presented by Noriega was extended and re-

fined, resulting in the definition of ways of realizing

EI. Rodríguez-Aguilar proposes an infrastructure to

implement EI that can be realized by making use of

a special type of mediator agents, the so called inte-

ragents (Martín et al., 2000). Each agent involved

in a conversation is connected to an interagent,

which mediates the agent’s interactions in one-to-

one conversations.

In (Esteva, 2003), the previous work done by

(Noriega, 1997; and, Rodríguez-Aguilar, 2001) on

the formalization of EI was continued. In his work,

Esteva provides support for the specifications of EI,

their automatic verification and also their realization.

His main concrete result, the ISLANDER graphical

editor, was developed as a generic infrastructure

which could be used for the deployment and verifi-

cation of the specified institutions.

The limitation of the Rodríguez-Aguilar’s work

in which only one-to-one conversations could be

mediated by interagents was improved in Esteva’s

work. There, for each conversation, a governor agent

(an evolution of the interagent one) has two queues,

one for the messages received from its associated

agent and another one for the messages received

from the social layer agents. As a case study, Esteva

evolved the previous examples of Noriega and

Rodríguez-Aguilar on fish markets, now regarding

multi-market institutions instead of only single-

market ones.

Many other publications of EI have appeared re-

cently (e.g., Esteva et al., 2004; García-Camino et

al., 2005 and 2006; Grossi et al., 2007), expanding

the work on the subject.

In (Esteva et al., 2004), the AMELI agent-based

middleware is proposed as an infrastructure that

mediates agents’ interactions while enforcing institu-

tional norms. The combination of ISLANDER and

AMELI supports the design and development of

open MAS adopting a social perspective.

In (García-Camino et al., 2005), a distributed ar-

chitecture for EI is proposed in order to endow MAS

with a social layer in which normative positions are

explicitly represented and managed via rules for

regulation. In (García-Camino et al., 2006), the rule-

based language from the authors is better detailed as

a declarative normative language that can represent

distinct flavors of deontic notions and relationships.

Every external agent from the architecture has a

dedicated governor agent linked to it that enforces

the norms of executed events.

In (Grossi et al., 2007), the work on formaliza-

tion of EI is continued, focusing on both institution

and its components (abstract and concrete norms,

empowerment of agents and roles). Yet, a formal

relation between institutions and organizational

structures is also defined in such a way that institu-

tional norms can be refined to construct – organiza-

tional structures – which are closer to an imple-

mented system. Thus, the gap between abstract

norms and concrete system specifications is better

bridged.

Despite all work done, a MAS implemented as

an EI is still understood as a type of dialogical sys-

tem that simply structures agent interactions by

establishing the commitments, obligations and rights

of participating agents. However, the solution not

only structures interactions, but also enforces indi-

vidual and social behaviours by obliging every agent

to act according to the defined norms.

We agree that the following current limitations

of EI can also be outlined: (i) there are no normative

aspects further than the ones for roles, agent interac-

tions and agents; (ii) the specification of an EI is

often too society-centric

in the sense that it com-

pletely fixes agent interactions in rigid protocols and

interfaces; (iii) external agents have no room for

autonomous behavior, i.e., they blindly follow de-

fined protocols with the only autonomy to accept or

reject them; (iv) all possible interactions among

agents have to be defined; (v) it is difficult, if not

impossible, to describe indirect interactions; this is

due to the fact that all interacting activity taking

place in an EI is purely dialogic by means of direct

communication between the agents; and, (vi) the

structure of an EI is static and, so, cannot evolve at

system runtime.

3 DISCUSSION

The models used to describe or design an organiza-

tion are classically divided into the agent-centered

MODELING OF OPEN NORMATIVE MULTIAGENT SYSTEMS

543

or organizational-centered perspectives (Lemaître

and Excelente, 1998). In the first perspective, system

developers try to analyze and/or design a whole

MAS that shows a non-accidental and non-chaotic

global behavior starting from the agents (parts of the

system).

In the open MAS scenario, the basic problem

with the agent-centered idea is that the system de-

veloper has no control anymore over the creation of

the agents. Thus, at any time, external heterogeneous

agents can join or leave an open MAS and, then,

disrupt the existing order. As long as open MAS are

highly desirable to face today’s increasingly distri-

buted and interconnected computing demands, this

wish poses problems that still need concrete solu-

tions.

In the last few years, one promising path of re-

search and development has been an organizational-

centered analysis and design of MAS (second pers-

pective). In this attempt, system developers proceed

in a top-down fashion, explicitly defining both the

organization entity (external to the agent level) and

the organization statutes that agents must comply

with. The statutes of an organization indicate, at the

most abstract level, the main objectives of the organ-

ization and the values that direct the fulfilling of its

objectives. Moreover, statutes also point to the con-

text in which the organization will have to perform

its activities (Vázquez-Salceda et al., 2005).

Analyzing several organizational-centered mod-

els found in the literature (e.g., OMNI (Vázquez-

Salceda et al., 2005), ISLANDER (Esteva et al.,

2002),

OISE

(Hübner et al., 2002)), we agree

with (Coutinho et al., 2008a) about the two main

sources of difficulties found on organizational-

centered models.

The first source of difficult is that the very notion

of organization admits and is frequently used with

slightly different interpretations. Sometimes, the

organization term refers to “collectivities oriented to

the pursuit of relatively specific goals and exhibiting

relatively highly formalized social structures”

(Scott, 1998). Other times, the term refers to stable

social patterns/structures of joint activity that con-

strains and drives the actions and interactions of

agents towards a purpose.

The second source of difficulty is that the organ-

ization entity can be described in several modeling

dimensions (e.g., in the structural and functional

ones).

These two sources of difficulties of organiza-

tional-centered models are important and should be

considered because each proposal of an organiza-

tional model makes a particular ontological com-

mitment in regard to them.

A proposal for an integrated ontology, which is

developed in a bottom-up manner from the existing

organizational models, is presented in (Coutinho et

al., 2008). The main purpose of such ontology is the

creation of an interoperation mechanism that can be

used by heterogeneous organizational models for

handling interoperability among open organization-

al-centered MAS. However, the proposal is an ongo-

ing work and, therefore, needs to be concluded.

In (Vázquez-Salceda et al., 2005), some draw-

backs of current approaches for MAS modeling also

are pointed out, as follows.

MAS modeling are too agent-centric or too or-

ganizational-centric. Some methodologies (e.g.,

GAIA (Wooldridge, 2000); Prometheus (Winikoff

and Padgham, 2004)) are too agent-centric, in the

sense that they are mainly focused on the model of

single agents, and give limited support to model the

dynamic interactions of the agents in the agent so-

ciety. Other methodologies (e.g., SODA (Omicini,

2001) and ISLANDER) are too society-centric in the

sense that they completely fix agent interactions in

rigid protocols and interfaces. Thus, agents cannot

exercise their characteristic of autonomy.

Roles and agents are usually treated without an

explicit distinction. This distinction is an important

asset in order to establish a difference between orga-

nizational values and individual (agent) values.

Normative aspects are not often considered or,

when considered, they are either too theoretical (i.e.,

the conceptual model of the solution does not have

an implemented solution for it) or too practical (i.e.,

the implemented solution does not have a conceptual

model to guide its specifications). Furthermore, few

agent methodologies cover normative aspects and

they usually do it by trying to model the whole nor-

mative environment in only one level of abstraction,

either too theoretical (by means of computationally

hard logics) or too practical (by means of the usage

of policies or protocols).

Ontologies are often seen as an external (acces-

sory) component, while in fact they should be tightly

coupled with the rest of the system when used to

model most of its elements.

4 CONCLUSIONS

Three main assumptions underlie this research.

Firstly, MAS has emerged as a concrete solution to

develop complex software systems in which mono-

lithic architectures (based on objects) have been

replaced by distributed ones (based on agents). Se-

condly, with the advent of the Semantic Web, agents

will be able to process information from different

ICAART 2009 - International Conference on Agents and Artificial Intelligence

544

sources and, so, they will be able to move around

other MAS looking for resources and/or services not

found locally. In this scenario, openness will be an

intrinsic and mandatory characteristic of upcoming

systems. However, openness without control leads to

chaotic scenarios. The use of norms in MAS is a

promising approach for achieving openness in a

reliable way. So, the final assumption of this work is

that MAS should be normative.

However, despite all efforts made to move theory

and practice of MAS from closed to open agent

societies, current solutions do not yet explicitly

support openness and its consequences. More pre-

cisely, methodologies, modeling languages and tools

(e.g., frameworks, platforms), needed for imple-

menting open MAS, do not conveniently cover the

aspects of regulation and domain representation for

society differentiation.

This paper presents an overview of the existing

works on multiagent organisations and normative

multiagent systems. The study done on solutions for

the modeling of MAS has led to the development of

our DynaCROM methodology. We agree that the

DynaCROM methodology supports the system de-

veloper in his task of implementation and manage-

ment of regulative norms in MAS.

ACKNOWLEDGEMENTS

To our Pronex-Faperj project (E-26/170.011/2008).

REFERENCES

Al-Muhtadi, J.; et al., 2003. Cerberus: a context-aware

security scheme for smart spaces. In 1

IEEE Interna-

tional Conference On Pervasive Computing And

Communications (PerCom 2003). p. 489- 496.

Berners-Lee, T.; et al, 2001. The Semantic Web, Scientific

American. 284(5), p. 34–43, May 2001.

Boella, G.; et al, 2006. Introduction to normative multia-

gent systems. Journal of Comput. Math. Organ.

Theory. v. 12, Issue 2–3, ISSN 1381-298X, p. 71–79.

COIN, URL: <http://www.pcs.usp.br/~coin/>.

Coutinho, L. dos R.; et al, 2005. Modeling organization in

MAS: a comparison of models. In 1

Workshop on

Software Engineering for Agent-Oriented Systems.

Coutinho, L. dos R.; et al., 2008. Modeling Dimensions

for Multi-Agent Systems Organizations. In Multi-

Agent Systems: Semantics and Dynamics of Organiza-

tional Models book, IGI global.

Corkill, D.D. and Lesser, V. 1983. The use of meta-level

control for coordination in a distributed problem solv-

ing network. In 8

International Joint Conference on

Artificial Intelligence, p. 748–756.

Dignum, V., 2004. A Model for Organizational Interac-

tion: based on Agents, founded in Logic. PhD Thesis.

Dignum, F.; et al., 2007. Open Agent Systems. In 8

International Workshop on AOSE, USA.

EDictionary, URL: <http://dictionary.reference.com/>

Esteva, M.; et al, 2002. ISLANDER: an electronic institu-

tions editor. International Conference on Autonomous

Agents and Multiagent Systems, p. 1045–1052.

Esteva, M., 2003. Electronic Institutions: from specifica-

tion to development. PhD thesis, Technical University

of Catalonia.

Esteva, M.; et al. 2004. AMELI: An Agent-based Mid-

dleware for Electronic Institutions. In 3

International

Joint Conference on Autonomous Agents and Multia-

gent Systems, p. 236–243.

Felicíssimo, C.; et al. 2007. Informing Regulatory Dynam-

ics in Open MAS. In LNCS, v. 4386. p. 147–162.

Felicíssimo, C. H.; et al, 2008a. Contextualizing Norma-

tive Open Multi-Agent Systems. In 23

Annual ACM

Symposium on Applied Computing (ACM SAC).

Felicíssimo, C.; et al., 2008b. DynaCROM: An Approach

to Implement Regulative Norms in Normative Multi-

agent Systems. In 3

DEON’08 International Work-

shop on Normative Multiagent Systems (NorMAS).

Felicíssimo, C.; et al. 2008c. How to Concretize Norms in

NMAS? An Operational Normative Approach Pre-

sented with a Case Study from the Television Domain.

In: COIN at AAAI’08.

García-Camino, A.; et al. 2005. A Distributed Architecture

for Norm-Aware Agent Societies. In Declarative

Agent Languages and Technologies (DALT’05).

García-Camino, A.; et al., 2006. Norm-Oriented Pro-

gramming of Electronic Institutions: A Rule-based

Approach. In: COIN@AAMAS2006, Japan.

Grossi, D.; et al. 2007. A Formal Road from Institutional

Norms to Organizational Structures. In AAMAS'07.

Hannoun, M.; et al., 2000. MOISE: An organizational mo-

del for multi-agent systems. LNAI, v.1952, p.152–161.

Henricksen, K. and Indulska, J. 2005. Developing context-

aware pervasive computing applications: models and

approach. Pervasive and Mobile Computing.

Hewitt, C. Open Information Systems Semantics for Dis-

tributed Artificial Intelli-gence. Artificial Intelligence,

v.47, I.1-3, 1991, p. 79–106, ISSN: 0004-3702.

Hübner, J. F.; et al. 2002. A model for the structural,

functional, and deontic specification of organizations

in multiagent systems. LNAI, v. 2507. p. 118–128.

Jennings, N.; et al., 1998. A Roadmap of Agent Research

and Development. In JAAMAS, v. 1, p. 7–38.

Jennings, N. R., 2001. An agent-based approach for build-

ing complex software systems. Communications of the

ACM, 44(4), p. 35–41.

Gasser, L.; et al., 1987. Distributed Artificial Intelligence,

chapter MACE: A flexible testbed for distributed AI

research, p. 119–152, Pitman Publishers.

Lemaître C. and Excelente, C. B., 1998. Multi-agent or-

ganization approach. In 2

Iberoamerican Workshop

on DAI and MAS.

Martín, F. J.; et al. 2000. An infrastructure for agent-based

systems: An interagent approach. International Journal

of Intelligent Systems, 15(3), p. 217–240.

MODELING OF OPEN NORMATIVE MULTIAGENT SYSTEMS

545

Noriega, P., 1997. Agent-Mediated Auctions: The Fish-

market Metaphor. PhD thesis, Technical University of

Catalonia.

North, D. 1990. Institutions, Institutional change and

Economics Performance. Cambridge, U.P..

Omicini, A., 2001. Soda: Societies and infrastructures in

the analysis and design of agent-based systems. In:

LNAI, v. 1957, p. 185–193.

Pattison, H.E.; et al., 1987. Distributed Artificial Intelli-

gence, chapter Instantiating Descriptions of Organiza-

tional Structure, p. 59–96, Pitman Publishers.

Rodríguez-Aguilar, J.A., 2001 On the Design and Con-

struction of Agent-mediated Electronic Institutions,

PhD thesis, Universitat Autonoma de Barcelona.

Scott, W. R., 1998. Organizations: rational, natural and

open systems. Upper Saddle River, NJ: Prentice Hall.

Vázquez-Salceda, J. and Dignum, F., 2003. Modelling

electronic organizations. In LNAI, v.2691, p.584–593.

Vázquez-Salceda, J.; et al, 2005.. Organizing Multiagent

Systems. In JAAMAS, 11(3), 2005, p. 307–360.

Weiser, M., 1991. The computer for the twenty-first cen-

tury. Scientific American, 265(3), p. 94–104.

Werner, E., 1987. Distributed Artificial Intelligence. In:

Chapter Cooperating Agents: A Unified Theory of

Communication and Social Structure, p. 3–36.

Winikoff, M. and Padgham, L., 2004. Developing Intelli-

gent Agent Systems: A Practical Guide. Book. Pub-

lished by John Wiley and Sons. ISBN 0-470-86120-7.

Wooldridge, M.; et al., 2000. The Gaia Methodology for

Agent-Oriented Analysis and Design. In: AAMAS,

3(3), p.285–312.

ICAART 2009 - International Conference on Agents and Artificial Intelligence

546