Social coordination frameworks for social technical systems

Gomez-Sanz, Jie Jiang, and Henrique Lopes Cardoso Part II Social Coordination Frameworks 3 ANTE: A Framework Integrating Negotiation, Norms and Trust.. 1.2 Positioning Models for social

Law, Governance and Technology Series 30

arising from an interdisciplinary approach in law, artificial intelligence and mation technologies The idea is to bridge the gap between research in IT lawand IT-applications for lawyers developing a unifying techno-legal perspective.The series will welcome proposals that have a fairly specific focus on problems

infor-or projects that will lead to innovative research charting the course finfor-or newinterdisciplinary developments in law, legal theory, and law and society research

as well as in computer technologies, artificial intelligence and cognitive sciences

In broad strokes, manuscripts for this series may be mainly located in the fields

of the Internet law (data protection, intellectual property, Internet rights, etc.),Computational models of the legal contents and legal reasoning, Legal InformationRetrieval, Electronic Data Discovery, Collaborative Tools (e.g Online DisputeResolution platforms), Metadata and XML Technologies (for Semantic Web Ser-vices), Technologies in Courtrooms and Judicial Offices (E-Court), Technologies forGovernments and Administrations (E-Government), Legal Multimedia, and LegalElectronic Institutions (Multi-Agent Systems and Artificial Societies)

More information about this series athttp://www.springer.com/series/8808

Virginia Dignum • Pablo Noriega • Julian Padget

Huib Aldewereld

Delft University of Technology

Delft, The Netherlands

Virginia Dignum

Delft University of Technology

Delft, The Netherlands

Pablo NoriegaIntitut d’Investigació en Intel  ligènciaArtificial (IIIA)

Consejo Superior de InvestigacionesCientíficas (CSIC)

Barcelona, Spain

Law, Governance and Technology Series

DOI 10.1007/978-3-319-33570-4

Library of Congress Control Number: 2016949076

© Springer International Publishing Switzerland 2016

This work is subject to copyright All rights are reserved by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed.

The use of general descriptive names, registered names, trademarks, service marks, etc in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use.

The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication Neither the publisher nor the authors or the editors give a warranty, express or implied, with respect to the material contained herein or for any errors or omissions that may have been made.

Printed on acid-free paper

This Springer imprint is published by Springer Nature

The registered company is Springer International Publishing AG Switzerland

Part I Preliminaries

1 Introduction 3

Huib Aldewereld, Olivier Boissier, Virginia Dignum,

Pablo Noriega, and Julian Padget

2 Conceptual Map for Social Coordination 11

Huib Aldewereld, Sergio Álvarez-Napagao, Emilia García,

Jorge J Gomez-Sanz, Jie Jiang, and Henrique Lopes Cardoso

Part II Social Coordination Frameworks

3 ANTE: A Framework Integrating Negotiation, Norms and Trust 27

Henrique Lopes Cardoso, Joana Urbano, Ana Paula Rocha,

António J.M Castro, and Eugénio Oliveira

4 Electronic Institutions: The EI/EIDE Framework 47

Pablo Noriega and Dave de Jonge

5 INGENIAS 77

Jorge J Gomez-Sanz and Rubén Fuentes Fernández

6 InstAL: An Institutional Action Language 101

Julian Padget, Emad ElDeen Elakehal, Tingting Li,

and Marina De Vos

7 The JaCaMo Framework 125

Olivier Boissier, Jomi F Hübner, and Alessandro Ricci

8 ROMAS-Magentix2 153

Emilia García, Soledad Valero, and Adriana Giret

v

9 OperA/ALIVE/OperettA 173

Huib Aldewereld, Sergio Álvarez-Napagao, Virginia Dignum,

Jie Jiang, Wamberto Vasconcelos, and Javier Vázquez-Salceda

10 Specifying and Executing Open Multi-agent Systems 197

Alexander Artikis, Marek Sergot, Jeremy Pitt,

Dídac Busquets, and Régis Riveret

11 Frameworks Comparison 213

Olivier Boissier, Virginia Dignum, and Emilia García

Part III Applications and Challenges

12 Application Domains 231

Julian Padget, Huib Aldewereld, Pablo Noriega,

and Wamberto Vasconcelos

13 Challenges for M4SC 265

Julian Padget, Huib Aldewereld, and Wamberto Vasconcelos

Huib Aldewereld Delft University of Technology, Delft, The Netherlands Sergio Álvarez-Napagao Universitat Politècnica de Catalunya, Barcelona, Spain Alexander Artikis University of Piraeus, Piraeus, Greece

NCSR Demokritos, Athens, Greece

Olivier Boissier Laboratoire Hubert Curien UMR CNRS 5516, Institut Henri

Fayol, Mines Saint-Etienne, Saint-Étienne, France

Dídac Busquets Electrical & Electronic Engineering Department, Imperial

Col-lege London, London, UK

António J.M Castro LIACC – Laboratrio de Inteligência Artificial e Ciência de

Computadores, Porto, Portugal

Dave de Jonge Artificial Intelligence Research Institute (IIIA), Spanish National

Research Council (CSIC), Barcelona, Spain

Marina De Vos Department of Computer Science, University of Bath, Bath, UK Virginia Dignum Delft University of Technology, Delft, The Netherlands Emad ElDeen Elakehal Department of Computer Science, University of Bath,

Bath, UK

Rubén Fuentes Fernández Universidad Complutense de Madrid, Madrid, Spain Emilia García Universitat Politècnica de València, Valencia, Spain

Adriana Giret Universitat Politècnica de Valencia, Valencia, Spain

Jorge J Gomez-Sanz Universidad Complutense de Madrid, Madrid, Spain Jomi F Hübner DAS-UFSC, Federal University of Santa Catarina, Florianópolis

SC, Brazil

Jie Jiang Delft University of Technology, Delft, The Netherlands

vii

Tingting Li Department of Computer Science, University of Bath, Bath, UK Henrique Lopes Cardoso Faculdade de Engenharia, Departamento de Engenharia

Informática, Universidade do Porto, Porto, Portugal

LIACC – Laboratrio de Inteligência Artificial e Ciência de Computadores, Porto,Portugal

Pablo Noriega Intitut d’Investigació en Intel  ligència Artificial (IIIA), Consejo

Superior de Investigaciones Científicas (CSIC), Barcelona, Spain

Eugénio Oliveira Faculdade de Engenharia, Departamento de Engenharia

Infor-mática, Universidade do Porto, Porto, Portugal

LIACC – Laboratrio de Inteligência Artificial e Ciência de Computadores, Porto,Portugal

Julian Padget Department of Computer Science, University of Bath, Bath, UK Jeremy Pitt Electrical & Electronic Engineering Department, Imperial College

Ana Paula Rocha Faculdade de Engenharia, Departamento de Engenharia

Infor-mática, Universidade do Porto, Porto, Portugal

LIACC – Laboratrio de Inteligência Artificial e Ciência de Computadores, Porto,Portugal

Marek Sergot Department of Computing, Imperial College London, London, UK Joana Urbano LIACC – Laboratrio de Inteligência Artificial e Ciência de

Computadores, Porto, Portugal

Soledad Valero Universitat Politècnica de Valencia, Valencia, Spain

Wamberto Vasconcelos University of Aberdeen, Aberdeen, UK

Javier Vázquez-Salceda Universitat Politècnica de Catalunya, Barcelona, Spain

Preliminaries

Huib Aldewereld, Olivier Boissier, Virginia Dignum, Pablo Noriega,

and Julian Padget

1.1 Introduction

As interactive systems and social networking applications increase in reach andcomplexity, there is a growing need to organize interactions between systems oramong their components This is specially the case in socio-technical systems (STS)where interaction between people and artificial systems is essential to the aims

of the system Design of STS must not be limited to the design of the systemitself but must include the design of interactions and enable adaptation to different

situations Socio-technical systems are not monolithic but are systems of systems,

comprised of intricate networks of people, organisations and technical systems, eachwith their own goals, capabilities and requirements which can only be achieved

by pooling and coordinating their resources and capabilities together; the entities

in such systems need to have social coordination (Noriega et al. 2015) Social

H Aldewereld (  ) • V Dignum

Delft University of Technology, Delft, The Netherlands

e-mail: h.m.aldewereld@tudelft.nl ; m.v.dignum@tudelft.nl

O Boissier

Laboratoire Hubert Curien UMR CNRS 5516, Institut Henri Fayol, Mines Saint-Etienne,

158 cours Fauriel, F-42023, Saint-Étienne, France

© Springer International Publishing Switzerland 2016

H Aldewereld et al (eds.), Social Coordination Frameworks

for Social Technical Systems, Law, Governance and Technology Series 30,

DOI 10.1007/978-3-319-33570-4_1

3

coordination is a many-faceted phenomenon that has been the subject of attention in

a number of scientific communities: from economics to social anthropology, frombiology to computer science The arrival of the internet and the massive adoption

of social networks and other web-enabled practices have led the notion of socialcoordination to acquire new meaning and, in reference to such on-line contexts, anunprecedented and substantial economic and social importance In fact, the design

of socio-technical systems is the design of social coordination

Given the complexity of these systems and the dynamic nature of their actions, the technical systems participating in these systems of systems, must

inter-be endowed with socio-cognitive capabilities to understand, reason and decideabout their position in the whole and their dependencies to the other participants.Therefore, these entities need to be able to understand and reason about each others

goals, plans and intentions; the entities need to have social intelligence.

This book is concerned with building socio-cognitive technical systems (SCTS).

An important premise for social intelligence is that of autonomous capability foraction (Castelfranchi1998) Autonomy requires that open environments are able

to accommodate the presence of heterogeneous participants, with opaque internalarchitectures, objectives and plans, allow the participation of such participants indifferent organisations, and their regulation by different regulatory institutions.There is therefore an increasing need for frameworks to describe and analysesocial coordination and collective intelligence phenomena These frameworks helpdecision makers to determine the qualities of related infrastructure, policy, and/ortechnology considerations as an interrelated whole (i.e a socio-technologicalsystem of systems) (DeLaurentis and Callaway2004) Moreover, tools are needed

to specify and analyse the environment, structure, and participants of SCTS Inparticular, these tools must support the “translation” of insights and requirementsabout SCTS into design specifications that result in better systems This bookintroduces models for social coordination (M4SC) and their related tools that fillthat gap

1.2 Positioning

Models for social coordination (M4SC) apply normative, social and organisationalconcepts from human societies to electronic distributed computational mechanismsfor the design and analysis of multi-agent systems (MAS) Using organisationalconcepts as first-class modelling constructs (Miles et al.2003), allows for a naturalspecification of open systems, and can describe both emergent and designedorganisations Just as in human organisations, M4SC describe how participantsinteract with each other and with the environment

M4SC define the formal lines of communication, allocation of informationprocessing tasks, distribution of decision-making authority, and the provision ofincentives That is, M4SC describe objectives, roles, interactions and rules in anenvironment, without considering the particular characteristics of the individuals

involved Organisational objectives are not necessarily shared by any of the ual participants, but can only be achieved through their combined action In order toachieve its goals, it is thus necessary that a SCTS employs the relevant participants,and structures their interactions and responsibilities such that (global) objectivescan be realised The performance of an SCTS is therefore determined both by itsinteraction structures, and by the individual characteristics of the participants.SCTS implies the existence of several coordinating entities that need to worktogether to realise some goal, which none individually has the capacity to achieve.

individ-As such, SCTS also implies the need for rules to indicate how parts must

be put together (i.e., the organisational structure relating roles to each other)

Organisational structure can be defined as that “what persists when components or

individuals enter or leave an [SCTS], i.e., the relationships that make an aggregate

of elements a whole” (Ferber et al.2003)

Inspired by So and Durfee (1998), components of SCTS can be classified into

three broad classes The first are (task) environmental factors, which include the

components and features of the task (such as size, time constraints, uncertainty), theavailable resources and the external conditions under which a SCTS operates The

second are the structural factors, which describe the components and features of

the structure itself (such as roles, dependencies, constraints, norms and regulations)

The third class of factors are agent factors, which describe the characteristics of the

individual participants concerning task capability, intelligence (including decisionmaking and reasoning capabilities), social awareness, etc To sum up, the three mainaspects that must be represented in any model aimed at understanding or specifyingSCTC performance or behaviour are:

1 Environment: this is the space in which SCTSs exist This space is not

completely controllable by the SCTS and therefore results of activity cannotalways be guaranteed Environment can also include the description of tasks andresources (such as size and frequency), and is characterized by properties such

as volatility, scarcity, unpredictability and (task) complexity

2 Participants: are the acting, reasoning, entities in the SCTS which have the

capability (partially) to control the state of some element(s) in the environment.Participants are defined by their capabilities, typically describing their learning,communication, reasoning and decision-making skills

3 Structure: describes the features of the SCTS, such as objectives, roles,

relationships and strategy The roles and relationships holding in the SCTSdetermine control, coordination, and power relations Differentiating dimensions

of structure are size, degree of centralization and formalization

In order to build SCTS, interoperability between the systems based on differentmodels for social coordination is required Among the possible directions, ameta-model at an higher level of abstraction than the existing models for socialcoordination could provide grounds for interoperability while still enabling thecoexistence of the different approaches, each with its own strengths and areas ofapplication

Two important requirements for these kinds of systems are:

1 Opacity/Individuality of participants: the heterogeneity of the participants and

the openness of the environment makes it that one cannot make any assumptionsabout the inner workings of participants Consequently, it is impossible tocreate coordination structures based on knowledge of the internals (beliefs,desires, goals) of participants of the system Rather, external aspects of theparticipants (actions, interactions, etc.) have to be leveraged to create the requiredcoordination structures

2 Institutional flexibility: the design of coordination should take an institutional

perspective, making explicit the ‘rules of the game’, but on a level that modates changes over time, context, actors, and actor preferences

accom-If we start from the premise of autonomy, SCTS as an open environment mustenable and support the presence of heterogeneous participants, with opaque internalarchitectures, objectives and plans Nevertheless, SCTS should enable participants

to act in different organisations and ‘navigate’ between those organisations, even ifthey are based on different models This second statement is the main focus of ourproject, i.e participants, assumed to have their own reasons for participating in anSCTS, such as access to specific services or other participants with certain abilities,must be supported in understanding the possibilities and constraints a SCTS canprovide them, and conversely, SCTS must be supported to control and monitor agentactivities within the SCTS Only then can we truly speak of open environments.Implicit in the definition of SCTS as instruments of purpose, is the ideas thatSCTS have goals, or objectives, to be realised and therefore, the shape, sizeand characteristics of the SCTS affect its behaviour (Horling and Lesser 2004).Objectives of a SCTS are achieved through the action of the individuals in the SCTS,which means that a SCTS should make sure to employ the relevant actors, so that itcan ‘enforce’ the possibility of making its desires happen Note that here, an explicitdistinction is made between the SCTS position, or role, and the actor, or participant

In this way, separation of concerns is possible Furthermore, one of the main reasonsfor creating SCTS is efficiency, that is, to provide the means for coordination thatenables the achievement of global goals in an efficient manner This means thatthe actors in a SCTS need to coordinate their activities in order to achieve thoseobjectives efficiently

Over the last 10 years or so, much research has been undertaken with the aim

of creating suitable models for social coordination to build such socio-cognitivetechnical systems Each of these models differs in focus and, consequently, aretailored for a specific (subset of) applications We believe, however, that the time

is ripe to bring these models together to advance the stage of research on cognitive technical systems, because:

socio-• Work on models for social coordination has now matured sufficiently:

– Several frameworks are available; and

– Each has been used extensively in education and applied to diverse areas inprojects and industrial collaborations

• Nevertheless, the heterogeneity of current approaches to modelling cognitive technical systems hinders a wider uptake of such approaches; and

socio-• Social coordination and collective intelligence need to address the ability problem at the level of the modelling of organisational and institutionalconstructs and richer socio-cognitive agent models

interoper-1.3 Objective

This book presents the collective research effort of the research groups working

on models for social coordination in an effort to bring the various models, withtheir differing focus and applications, together to create an unified approach to thecreation of socio-cognitive technical systems The objectives of this book can besummarised as follows:

• To establish a systematic comparison of the existing models for social tion;

coordina-• To develop an integrative meta-model that can account for these models;

• To identify applications, challenges and opportunities for such models

There has been little work to date, on integrating and comparing the differentapproaches to models for social coordination (a notable exception being Coutinho

et al.2009) The heterogeneity of current approaches to models for social tion make such interaction cumbersome That is, so far, the interoperability betweenmodels for social coordination is mostly lacking In our opinion, social coordinationand collective intelligence can only be fostered by addressing the interoperabilityproblem at the level of modelling and reasoning

coordina-1.4 Overview of Book

The remainder of this book is structured as follows The book contains threeparts; Part I: the front matter or preliminaries (of which this introduction is thefirst chapter); PartIIthe framework descriptions and comparison; and PartIIItheapplications and challenges.The remainder of PartIIconsists of:

• Chapter 2, Conceptual Map for Social Coordination, which presents the

ongoing work on the unification of the different models for social coordinationinto one unifying meta-model for social coordination Chapter 2 includes adescription of the Tendering Use-Case that is used as a common example forthe discussion of the frameworks in PartII

Part II comprises presentations of each of the frameworks, going in depthabout the different models for social coordination developed by various researchgroups The discussions of the frameworks include, among others, a brief history

of the framework, applications, a detailed description of the meta-model, a criticalassessment and key references Part IIconcludes with a comparison between theframeworks presented The chapters in PartIIare:

• Chapter3, ANTE – A Framework Integrating Negotiation, Norms and Trust,

presents the details of the ANTE framework, developed by the Universidade doPorto, Portugal

• Chapter 4, Electronic Institutions – The EI/EIDE Framwork, presents the

details of the Electronic Institutions framework, developed by UniversitatAutònoma de Barcelona, Spain

• Chapter 5, INGENIAS, presents the details of the INGENIAS framework,

developed by the Universidad Complutense de Madrid, Spain

• Chapter6, InsAL: An Institutional Action Language, presents the details of

the InstAL framework, developed by the University of Bath, United Kingdom.

• Chapter 7, The JaCaMo Framework, presents the details of the JaCaMo

framework, developed by École des Mines de Saint-Étienne, France, dade Federal de Santa Catarina, Brazil, Pontíficia Universidade Católica do RioGrande do Sul, Brazil, and Università di Bologna, Italy

Universi-• Chapter 8, ROMAS/MAGENTIX2, presents the details of the ROMAS and

MAGENTIX2 frameworks developed by the Universitat Politècnica de València,Spain

• Chapter9, OperA/ALIVE/OperettA, presents the details of the OperA

frame-work and the ALIVE and OperettA tools, developed by Delft University ofTechnology, The Netherlands, Universitat Politècnica de Catalunya, Spain, andUniversity of Aberdeen, United Kingdom

• Chapter10, Specifying and Executing Open Multi-Agent Systems, presents

the details of the RTEC framework, developed by NCSR Demokritos, Greeceand Imperial College London, United Kingdom

• Chapter11concludes PartIIwith an comparison between the presented works

frame-PartIIIreflects on the applications for models for social coordination and thechallenges and perspectives for future research PartIIIconsists of:

• Chapter12, Application Domains, gives an overview of the characteristics of the

domains which are handled well by models for social coordination Moreover,the chapter presents exemplary cases of applications of the various frameworksincluded in this book

• Chapter 13, Challenges for M4SC, concludes the book with an overview of

challenges for M4SC to be solved in the short-term, and advances a perspective

on their short-comings and the possibilities for the future, to inspire futureresearch in the field of M4SC

Acknowledgements The work presented in this book was initiated by the workshop on Models

for Social Coordination, held in Veldhoven, The Netherlands, March 2–7, 2014 This workshop was made possible by support from the European Network for Social Intelligence, SINTELNET (FET Open Coordinated Action FP7-ICT-2009-C Project No 286370).

Castelfranchi, C 1998 Modelling social action for AI agents Artificial Intelligence 103(12):

com/science/article/pii/S0004370298000563 Artificial Intelligence 40 years later.

Coutinho, L., J Sichman, and O Boissier 2009 Modelling dimensions for agent organizations.

In Handbook of research on multi-agent systems: Semantics and dynamics of organizational

models, ed V Dignum, 18–50 Hershey: Information Science Reference.

DeLaurentis, D., and R.K Callaway 2004 A system-of-systems perspective for public policy

decisions Review of Policy Research 21(6): 829–837 doi:10.1111/j.1541-1338.2004.00111.x,

http://dx.doi.org/10.1111/j.1541-1338.2004.00111.x

Ferber, J., O Gutknecht, and F Michel 2003 From agents to organizations: An organizational

vol 2935, ed P Giorgini, J.P Müller, and J Odell Berlin/New York: Springer.

Knowledge Engineering Review 19(4): 281–316.

Miles, S., M Joy, and M Luck 2003 Towards a methodology for coordination mechanism

selection in open systems In Engineering societies in the agents’ world III, no 2577 in lecture

notes in artificial intelligence, ed P Petta, P Tolksdorf, and F Zambonelli Berlin/New York: Springer.

Noriega, P., J Padget, H Verhagen, and M d’Inverno 2015 Towards a framework for

socio-cognitive technical systems In COIN X, no 9372 in LNAI, ed A Ghose, N Oren, P Telang,

and J Thangarajah, 164–181.

So, Y., and E Durfee 1998 Designing organizations for computational agents In Simulating

organizations, ed K Carley, M Pritula, and L Gasser, 47–64.

Conceptual Map for Social Coordination

Huib Aldewereld, Sergio Álvarez-Napagao, Emilia García,

Jorge J Gomez-Sanz, Jie Jiang, and Henrique Lopes Cardoso

2.1 Introduction

The engineering of applications for complex and dynamic domains withautonomous participants is an increasingly difficult process Requirements andfunctionalities are not fixed a priori, components are not designed nor controlled

by a common entity, and unplanned or underspecified changes may occur duringruntime There is a need for representing the regulating structures explicitly andindependently from the acting components (or agents) Organization computationalmodels, based on Organization Theory, have been advocated to specify such

H Aldewereld (  ) • J Jiang

Delft University of Technology, Delft, The Netherlands

e-mail: h.m.aldewereld@tudelft.nl ; jie.jiang@tudelft.nl

© Springer International Publishing Switzerland 2016

H Aldewereld et al (eds.), Social Coordination Frameworks

for Social Technical Systems, Law, Governance and Technology Series 30,

DOI 10.1007/978-3-319-33570-4_2

11

Fig 2.1 MDE model-to-model transformations

systems Comprehensive analysis of several agent systems has shown that differentdesign approaches are appropriate for different domain characteristics (Dignum andDignum2010)

Work on multi-agent organizations is now sufficiently mature to the extent thatseveral frameworks are available, and have been used extensively in education andapplied to diverse areas in projects and industrial collaborations Nevertheless, therehas been little work done on integrating and comparing such different approaches(a notable exception being Coutinho et al.2009) That is, so far, the interoperabilitybetween organizational frameworks is mostly lacking

In this chapter, we present ongoing work on the integration of models for socialcoordination This results in a preliminary version of a meta-model combining thedifferent models for social coordination, build on the work presented in PartII Thecombination of the models in Chaps.3,4,5,6,7,8,9, and10is a richness of themeta-model, but also limits it since it is built bottom-up from these models Theintegration of the models in an all-including meta-model will allow for seamlessintegration in open environments

In order to build such open environments, interoperability between organizationsbased on different models is required Among the possible directions, we introduce

an organizational meta-model at a higher level of abstraction than that provided byexisting organizational models; while providing grounds for interoperability, thisapproach enables the coexistence of different models, each with its own strengthsand areas of application Such a high-level meta-model would then allow formodel-to-model transformations from one organizational framework to another (seeFig.2.1), following Model Driven Architecture principles (MDA; see Bézivin et al.(2003) for details) with the abstract meta-model as a common basis

This chapter is organised as follows In the next section we present the conceptualmap for social coordination, including a brief description of the concepts, whythey are there, and what the relations between the concepts signify In Sect.2.3weprovide more precise definitions of the concepts and relations of the meta-model InSect.2.4we present a use-case of social coordination (Sect.2.4.1) and provide a firstmodelling of this case using the conceptual map (Sect.2.4.2) We end this chapterwith conclusions and future work.

In this section we present the first conceptual map for frameworks of socialcoordination The model was developed in the first workshop on Models for SocialCoordination (M4SC) in Veldhoven, the Netherlands in March 2014

The aim of the conceptual map is to show the key concepts and their relations in

an interpretation that is largely shared by the frameworks described in Part II In thatway, each of the frameworks can be seen as an extension of this common model; it

thus presents a core model for social coordination (see Fig.2.2) By looking at thecommonalities we have tried to create an abstract model of social coordination Thisabstract model tries to de-emphasise the specific aim of each of the frameworks (e.g.,some frameworks focus more on the normative side of social coordination, others

on the formalisation of runtime systems), and tries to emphasise which elementsshould be included in any model for social coordination The core model will then,

in essence, provide the basis for making the necessary model-to-model relationsneeded to transform models in one of the frameworks to another framework Being

Fig 2.2 Common core

model for frameworks of

social coordination

Fig 2.3 Unified meta-model

the shared core of all the models, the conceptual map presents the opportunity forintegration attempts between the frameworks

In order to integrate, compare and reuse existing agent organization models, wepropose a meta-modelling approach An organizational meta-model abstracts fromthe details of specific models, highlighting the generic properties of organizations

An overview of the meta-model is shown in Fig.2.3

As expected, the main concept of the meta-model is Organization (for an

interpretation of each of the concepts, see the next section) Organizations exist

to fulfil some common objective or Goal, and contain a number of Roles whose players are supposed to interact to achieve this goal Roles are played by Agents

or Organizations The Goal of an Organization might be decomposed into smaller

(more elementary) Goals The fulfilment of Goals is accomplished through ities of the Roles The Activity of an Organization can be composed of smaller, more elementary Activities Atomic Activities can either be Actions, States, or Events Activities may be situated in a Subspace Lastly, the Organization and the organizational Activity are governed by sets of rules (Norms) composing proper Normative Contexts.

Activ-A discussion of the meaning of the concepts, and links to existing literature, ispresented in the next section

that are coordinated for the pursuit of some collective goals

A university, (a department inside) a company, an auction house

Selected interpretations

feature high specificity of structure and coordination

Schein ( 1965): An organization is the rational coordination of the activities of a number of

people for the achievement of some common explicit purpose or goal, through division of labor

or function, and through a hierarchy of authority and responsibility

Pfeffer and Salancik ( 1978): An organization is a coalition of groups and interests, each

attempting to obtain something from the collectivity by interacting with others, and each with its own preferences and objectives

common purpose to achieve certain objectives

Scott ( 1995): Organizations are groups whose members coordinate their behavior in order to

accomplish shared goals or to put out a product

Ostrom ( 2009): Organizations are groups of participants, bound by some common purpose to

achieve outcomes, in situation structures composed of multiple simultaneous and sequential action situations, all constituted by rules as well as by the physical world

Scott ( 1992): Goals are conceptions of desired ends – conditions that participants attempt to

effect through their performance of task activities

Rao ( 1996): A goal is a state of the system which the agent wants to bring about

DeLoach ( 2009): A goal is a desirable state of the world or the objective of a computational

process

(continued)

Concept Explanation Examples

which become endowed with specific rights and are expected to achieve

a particular goal or set of goals

Buyer, dent, author Selected Interpretations

stu-Biddle ( 1986): Roles in organizations are assumed to be associated with identified social

positions and to be generated by normative expectations

Scott ( 1992): Roles are expectations for or evaluative standards employed in assessing the

behavior of occupants of specific social positions

Gasser ( 2001): Roles typically describe an organizationally-sanctioned structured bundle of

activity types

Ostrom ( 2005): Roles/positions are the connecting link between participants and actions In some

situations, any participant in any position may be authorized to take any of the allowable actions

in that situation However, in most organized situations, the capability to take particular actions

in assigned to a specific position

Hindin ( 2009): A role can be defined as a social position, behavior associated with a social

position, or a typical behavior

Shoham ( 1993): An (artificial) agent is an entity possessing formal versions of mental state,

and in particular formal versions of knowledge, beliefs, capabilities, choices, commitments, and possibly a few other mentalistic-sounding qualities

Russell and Norvig ( 1995): An agent is anything that can be viewed as perceiving its environment

through sensors and acting upon that environment through effectors

Wooldridge and Jennings ( 1995): An agent is an encapsulated computer system that is situated

in some environment and that is capable of flexible, autonomous action in that environment in order to meet its design objectives

Franklin and Graesser ( 1997): An autonomous agent is a system situated within and a part of an

environment that senses that environment and acts on it, over time, in pursuit of its own agenda and so as to effect what it senses in the future

Searle ( 2003): An agent is an entity able to make decisions and perform intentional actions on

the basis of reasons, and capable of perception, belief, desire, memory, reasoning, as well as commitment

(continued)

Concept Explanation Examples

the activities of agents and/or constrain

their behaviour in order to achieve shared

expectations

Students should be on time for classes, sellers should deliver the goods once the buyer makes the payment

Selected interpretations

Ganz ( 1971 ) and Ostrom ( 2005): Norms/rules may be defined to be shared understandings by

participants about enforced prescriptions concerning what actions (or outcomes) are required, prohibited, or permitted

Axelrod ( 1986): A norm exists in a given social setting to the extent that individuals usually act

in a certain way and are often punished when seen not be acting in this way

Scott ( 1992): Norms are the generalized rules governing behavior that specify, in particular,

appropriate means for pursuing goals

Jones and Sergot ( 1993): A norm is a prescription of how the agents ought to behave, and specify

how they are permitted to behave and what their rights are

North ( 1994): Constraints that human beings impose on human interaction, consisting of formal

rules (constitutions, statute law, common law, regulations) and informal constraints (conventions, norms and self-enforced codes of conduct)

Searle ( 2005): Rules that either (1) create institutional facts from brute facts, thus substantively

constituting the very behaviour they regulate (constitutive), or (2) regulate activities that exist prior or independently of the norm (regulative)

Activity A set of functionalities offered by one or several agents

that may consume resources and make changes in the

environment and into the agents mental state

Paper review, party ning

plan-Selected interpretations

Bennett et al ( 2010): An activity is a group of one or more actions that may execute as a result

of a triggering event

Object Management Group ( 2007): An activity is the specification of parameterized behavior as

the coordinated sequencing of subordinate units whose individual elements are actions

a system, or the world

Being rich, submitted, evaluated

Selected interpretations

Wittgenstein ( 1922): The world is everything that is the case What is the case (a fact) is the

existence of states of affairs A state of affairs (a state of things) is a combination of objects (things) The possibility of its occurring in states of affairs is the form of an object Objects are what is unalterable and substantial; their configuration is what is changing and unstable

Armstrong ( 1993): States of affairs have as constituents particulars, properties and relations A

state of affairs exists if and only if a particular has a property, or a relation holds between two

or more particulars

(continued)

Concept Explanation Examples

Selected interpretations

McCarthy and Hayes ( 1969): An action, depending on a specific agent and a certain current

situation, produces (or not) a result that may be reflected in a change in the state of the world

Davidson ( 1980): An action, in some basic sense, is something an agent does that was intentional

under some description

Cohen and Levesque ( 1990): In the way that usually the content of beliefs and knowledge is

considered to be in the form of propositions, the content of an intention is typically regarded as

an action

2.4 Tender Request Scenario

To illustrate the use of the conceptual map for social coordination, we now model

a use-case of social coordination: the Request for Tender This use-case is alsomodelled by each of the frameworks presented in PartII Before we model the use-case, we first present a description of the use-case

A request for tenders (RFT) is a formal, structured invitation to suppliers, to bid,

to supply products or services For example, a company or government may put

a building project ‘out to tender’; that is, publish an invitation for other parties tomake a proposal for the building’s construction

The aim of the process is to ensure finding the best supplier possible for therequested service or product, such that no parties will have an unfair advantage ofseparate, prior, or closed-door negotiations for the contract The Actors, or stake-holders, involved in a RFT are the following: contracting authority, bidders consor-tium (possibly consisting of several partners), evaluators, publication body, etc

A system for RFT should handle more that one RFT at the same time; i.e therecan be more than one contracting authority putting out a request to be fulfilled bydifferent bidder consortia, and possibly advertised by different publication bodies

An RFT process consists of (at least) the following stages (Fig.2.4):

1 Tender elaboration: decide on terms, conditions, deadlines, etc for the RFT

2 Publication: publication of the tender and/or distributed to potential bidders

3 Request for information: interested bidders can ask for further information toclarify any uncertainties

4 Bid preparation

(optionally) Consortium formation

5 Bid submission

Fig 2.4 Illustrative example of the request for tender (RFT) process

6 Bid evaluation and decision: an evaluation team will go through the tenders anddecide who will get the contract Each tender will be checked for complianceand if compliant, then evaluated against the criteria specified in the tenderdocumentation The tender that offers best value for money will win the business

7 Notification: When a contract has been awarded, the successful tenderer will beadvised in writing of the outcome Unsuccessful tenderers are also informed.(optionally) may be offered a debriefing interview

8 Contract formation: a formal agreement will be required between the successfultenderer and the contracting authority

The process of requesting for tenders is governed by a number of norms Some

of the norms holding in this scenario are (examples):

• Bids must be submitted before the deadline

• Reviewers have to submit their evaluation on time

• All bids must be written in English

• Bids include at least X and at most Y partners

• Each tender must receive at least Z different bids

• Reviewers and requester cannot participate in any bid consortium

• Bids must be blind

Fig 2.5 Example of meta-model instantiation, identifying organizations in the call for tender case

2.4.2 Use-Case Model

The meta-model defines one single diagram where these entities can be used.Figure2.5shows a possible interpretation of the problem using those entities Thecall for tender involves one administration and one or many external companies Theadministration will publish the call for tenders, will be responsible for clarifyingquestions about the call, registering bids, and will contract only those economicoperators that satisfy the contracting norms One of the norms is that a contractorcannot be a bidder The company acknowledges the existing norms and participates

in the request for tender activity The company pursues to make profit, so it is willing

to participate if it is possible to respect the norms and do the job better than others.The activities involved in the call for tender are several, as shows in Fig.2.6.They have not been fully decomposed The meta-model does not allow to indicatedependencies/ordering among activities beyond the decomposition Some activities

have been identified as atomic actions, like the evaluate tenders This may be a

shortcoming of the current meta-model

This chapter brings together the work that has taken place in the last decade onthe development of conceptual and computational frameworks for open regulatedmultiagent systems and explores their potential for the development of the emergingfield of social intelligence

We are currently further refining and validating this meta-model Evaluation isbeing done by applying the meta-model to describe existing agent organizationmodels included in the next chapters

Acknowledgements The authors would like to present their gratitude to all members of the

meta-model for social coordination work session at Veldhoven The discussions in those work sessions gave the basis for the conceptual map presented in this chapter In particular, we would like to thank the other members of the meta-modelling work session: Antoinne Zimmerman (EMSE Saint- Etienne, France), Andrei Ciorea (EMSE Saint-Etienne, France), and Emad El-Akehal (University

of Bath, UK).

Armstrong, D.M 1993 A world of states of affairs Philosophical Perspectives 7: 429–440 Axelrod, R 1986 An evolutionary approach to norms American Political Science Review 80:

1095–1111.

Bennett, S., S McRobb, and R Farmer 2010 Object oriented systems analysis and design using

UML Maidenhead: McGraw-Hill Education.

Bézivin, J., S Gérard, P.A Muller, and L Rioux, et al 2003 MDA components: Challenges and

opportunities In Workshop on metamodelling for MDA, York, 23–41.

Biddle, B.J 1986 Recent developments in role theory Annual Review of Sociology 12: 67–92 Cohen, P.R., and H.J Levesque 1990 Intention is choice with commitment Artificial Intelligence

42(2): 213–261.

Coutinho, L., J Sichman, and O Boissier 2009 Modelling dimensions for agent organizations.

In Handbook of research on multi-agent systems: Semantics and dynamics of organizational

models, ed V Dignum Hershey: Information Science Reference.

Davidson, D 1980 Essays on actions and events Oxford: Oxford University Press.

DeLoach, S 2009 Omacs: A framework for adaptive, complex systems In Handbook of research

on multi-agent systems: Semantics and dynamics of organizational models, ed V Dignum,

76–98 Hershey: IGI Global.

Dignum, V., and F Dignum 2010 Designing agent systems; state of the practice.

International Journal of Agent-Oriented Software Engineering 4(3): 224–243.

Franklin, S., and A Graesser 1997 Is it an agent, or just a program?: A taxonomy for autonomous

agents In Intelligent agents III agent theories, architectures, and languages Vol 1193 of

Lecture notes in computer science, 21–35 Berlin/Heidelberg: Springer.

Ganz, J.S 1971 Rules: A systematic study Paris: Mouton.

Gasser, L 2001 Perspectives on organizations in multi-agent systems In Proceedings of EASSS

2001, Prague, 1–16.

Hindin, M.J 2009 Role theory In Blackwell encyclopedia of sociology online, ed George Ritzer.

Malden: Blackwell Publishing.

Jones, A.J.I., and M Sergot 1993 On the characterization of law and computer systems: The

specification, ed John-Jules Ch Meyer and Roel Wieringa, 275–305 Chichester/New York:

Wiley.

March, J.G., and H.A Simon 1958 Organizations New York: Wiley.

McCarthy, J., and P.J Hayes 1969 Some philosophical problems from the standpoint of artificial

intelligence In Readings in artificial intelligence, ed Bonny Lynn Webber and Nils J Nilsson,

431-450 Los Altos: Morgan Kaufmann.

North, D.C 1994 Institutions, institutional change and economic performance Cambridge/New

York/Port Chester: Cambridge University Press.

Object Management Group 2007 OMG unified modeling language superstructure specification Object Management Group, version 2.1.1 edn.

Ostrom, E 2005 Understanding institutional diversity Princeton: Princeton University Press Ostrom, E 2009 Understanding institutional diversity Princeton: Princeton University Press Pfeffer, J., and G.R Salancik 1978 The external control of organizations New York: Harper &

Row.

Rao, A.S 1996 Agentspeak(l): BDI agents speak out in a logical computable language In Agents

breaking away Vol 1038 of Lecture notes in computer science, ed W Van de Velde and

J Perram, 42–55 Berlin/Heidelberg, Springer.

Russell, S.J., and P Norvig 1995 Artificial intelligence: A modern approach Englewood Cliffs:

Prentice Hall.

Schein, E 1965 Organizational psychology Englewood Cliffs: Prentice-Hall.

Scott, W 1992 Organizations: rational, natural, and open systems Englewood Cliffs:

Prentice-Hall.

Scott, W 1995 Institutions and organizations Thousand Oaks: Sage Publications.

Searle, J.R 2003 Rationality in action London: MIT Press.

Searle, J.R 2005 What is an institution? Journal of Institutional Economics 1(1): 1–22 Shoham, Y 1993 Agent-oriented programming Artificial Intelligence 60(1): 51–92.

Simon, H.A 1964 On the concept of organizational goal Administrative Science Quarterly 9:

1–22.

Weber, M 1978 Economy and society Berkeley: University of California Press.

Wittgenstein, L 1922 Tractatus logico-philosophicus London: Routledge & Kegan Paul Wooldridge, M., and N.R Jennings 1995 Intelligent agents: Theory and practice Knowledge

Engineering Review 10(2): 115–152.

Social Coordination Frameworks

ANTE: A Framework Integrating Negotiation, Norms and Trust

Henrique Lopes Cardoso, Joana Urbano, Ana Paula Rocha,

António J.M Castro, and Eugénio Oliveira

3.1 Introduction

Social coordination is a broad concept In multi-agent systems, social coordinationcan be seen as related with a number of agreement technologies Negotiation is aform of coordination, whereby a set of agents tries to reach an agreement on somejoint effort or activity Organizational or normative structures are means to regulatethe interaction of a number of agents while enacting roles in the organization.Assessing how trustworthy agents are when executing specific tasks is also a means

of encompassing the social dimension of interactions, eventually enhancing thecoordination of individuals when executing inter-related activities

The ANTE1 framework addresses the issue of social coordination from a prehensive perspective (Lopes Cardoso et al.2013), exploring negotiation (Rocha

com-2001) as a mechanism for establishing some normative coordination infrastructure,based on the notion of a contract Contracts are specified through norms thatare monitored for compliance (Lopes Cardoso and Oliveira2010b) The contractenactment phase enables the collection of behavioural data that can be used by

1 ANTE stands for Agreement Negotiation in Normative and Trust-enabled Environments.

H Lopes Cardoso (  ) • A.P Rocha • E Oliveira

Faculdade de Engenharia, Departamento de Engenharia Informática, Universidade do Porto, Porto, Portugal

LIACC – Laboratrio de Inteligência Artificial e Ciência de Computadores, Porto, Portugal e-mail: hlc@fe.up.pt ; arocha@fe.up.pt ; eco@fe.up.pt

J Urbano • A.J.M Castro

LIACC – Laboratrio de Inteligência Artificial e Ciência de Computadores, Porto, Portugal e-mail: joana.urbano@fe.up.pt ; antonio.castro@fe.up.pt

© Springer International Publishing Switzerland 2016

H Aldewereld et al (eds.), Social Coordination Frameworks

for Social Technical Systems, Law, Governance and Technology Series 30,

DOI 10.1007/978-3-319-33570-4_3

27

Fig 3.1 ANTE framework

computational trust models to infer the trustworthiness of agents (Urbano andRocha2014)

From this perspective, we can say that ANTE’s conceptual framework focuses

on the run-time establishment of norm-governed relationships and their ing (Lopes Cardoso and Oliveira2008) The emphasis is therefore not on modelling

monitor-an existing orgmonitor-anization, but rather on the dynamic processes that support thecreation of specific types of “organizations”, formalized through contracts As itturns out, ANTE is not devoted to the structural aspects as much as it is towards thenormative aspects of organizations’ activity

ANTE’s framework is brought to light through the development of a softwareplatform, encompassing results of research contributions on three main agreementtechnology concepts, namely negotiation, normative environments and computa-tional trust (see Fig.3.1) This platform is the corollary of a long-term researchproject, which has been targeting the domain of B2B electronic contracting,although being conceived as a more general framework having in mind a widerrange of applications

A modelling and specification process (MSP) that explores ANTE’s main cepts is under development, based on the PORTO methodology (Castro and Oliveira

con-2008) ANTE includes a Java API with which use-case specific implementations can

be developed The main benefits of using ANTE result from its modularity: one caneasily replace or enrich any of its components to better address a target applicationscenario Also, it is fairly easy to embed automated reasoning capabilities on user

agents, that is, agents acting on behalf of human (or organizational) principals forfacilitating the use of the different services in the system.

More than 10 years ago, a hot topic on multi-agent systems research was onflexible and adaptive negotiation protocols We, at the University of Porto (LIACC),developed then the Q-Negotiation protocol (Rocha and Oliveira 1999) whichencompassed a learning capability, by using the Q-learning algorithm to improve,through several consecutive rounds and negotiation episodes, agents’ capability toconsecutively make better proposals The protocol was successfully demonstrated

on a kind of B2B scenarios where agents, representing enterprises, tried to compete

to be selected as future partners of a consortium These developments havematerialized in ANTE’s precursor platform, ForEV (Rocha 2001) (standing forVirtual Enterprise Formation)

We then realized that several issues arose that could jeopardize the usefulness andthe application of those flexible protocols in a real-life scenario How can we assureand, up to a certain extent, enforce that what has been negotiated between agentswill in fact take place? How can we assess and exploit agents’ (representing realentities as enterprises, organizations, individuals) ability and goodwill during thenegotiation stage? How can agents make themselves understand each other? Thequest for answering to all these question lead our Distributed Artificial IntelligenceGroup at LIACC- University of Porto, in a many years focused research dedication.The very idea was to develop a software platform – a kind of electronicinstitution – through which agents as enterprise representatives, could negotiateclauses of electronic contracts (Lopes Cardoso and Oliveira2005) to be enactedunder the monitoring capabilities of a normative environment (Lopes Cardoso andOliveira 2009) Moreover, agents’ activity in this environment would be traceddown in order to build up each agent Trust Model (Urbano et al 2010) Eachagent perspective of other agent’s trustworthiness becomes an important factorfor decision-making on agents’ future joint work (Urbano et al.2012) Also, anontology-based translation service (Malucelli et al.2006) has been developed inparallel for the sake of enlarging possible mutual understanding The final ANTEplatform as it has been developed and evaluated, encompasses most of thesefeatures, and aims at being useful for several different application domains otherthan B2B (Lopes Cardoso et al.2013)

As follows from the above description, applications that might benefit the mostfrom the ANTE approach are those that denote open and dynamic environments.Run-time negotiations may be carried out with possibly unknown partners, leading

to the need to monitor compliance and to apply corrective measures Trust is anatural ingredient in these settings, as it enables a more careful assessment of agentscapabilities or benevolence The glove-fitting application scenario that has served as

an inspiration is that of virtual enterprise’s life-cycle in B2B scenarios: in the face

of a new business opportunity, different firms try to engage in fruitful negotiationsthat enable them to configure an appropriate team, in a win-win relationship Theserelationships can be as simple or as complex as we want, ranging from supply orders

to long term consortium agreements

On the other hand, a norm-governed relationship is a formalization of a ously negotiated (or simply adhered to) agreement Unlike constraints, norms arevulnerable to violations: although norms are meant to influence agent behaviourand hopefully to be internalized as goals by the agents addressed by the norms,

previ-it is up to the agents to actually fulfil or deviate from the norms Wprevi-ithin ANTE,norm-governed relationships are coordinated within appropriate contexts inside anormative environment (Lopes Cardoso and Oliveira2009)

Using ANTE for a particular kind of system therefore entails defining whichnegotiation protocols are to be supported, which types of norm-governed relation-ships (contracts) may be formed, and how agents may enact these (that is, whichactions agents have at their disposal and what are the effects of such actions withinthe normative dimension of the framework) Furthermore, different computationaltrust models may be included to address the specific needs of the application inmind, in terms of assessing the perceived trustworthiness of agents within thesystem Agents making use of their own computational trust models are also assisted

by providing contractual evidence as input

It should be noted that the ANTE framework is quite loose in terms of its usageassumptions By this we mean that it is not mandatory to use the whole processcycle depicted in Fig.3.1 For instance, built-in negotiation protocols may be used

simply to assemble a team of agents able to address a specific task, with no attempt

to regulate further interactions among such individuals Conversely, exogenouslyestablished contracts may be handed to the normative environment for monitoringpurposes (Lopes Cardoso and Oliveira 2010b) Also, trust assessments may beentirely neglected or exploited in a number of ways (Urbano et al.2014) This stanceprovides modularity and extensibility to the framework

The overall background scenario of ANTE is the management of run-time governed relationships As such, the framework focuses on the dynamic aspects ofsystems (as opposed to static or descriptive ones), by supporting the specification

norm-of how contracts can be negotiated, established and enacted Instead norm-of addressing

a single specific organization, the goal of ANTE is to provide a facilitatingenvironment encompassing services for negotiation, contract monitoring, and trustassessment Software agents, acting on behalf of their principals, are then developedwhich can benefit from such services to automate some of the associated tasks.Given the fact that ANTE is more concerned with the normative dimension

of social coordination, organizations and their composition through roles are onlyrelevant in light of the norms agents playing such roles are subject to For this reason,

we do not care so much about the structural properties of organizations

Figure 3.2 illustrates the main concepts of ANTE’s meta-model Given the factthat ANTE focuses on the synergies that are obtained when combining negotiation,norms and trust, the meta-model includes a set of concepts that are related with each

of these components A brief explanation of the main concepts follows

A (software or human) user agent interacts in ANTE by participating in a

negotiation protocol A negotiation protocol is an interaction whose aim is to obtain

an agreement between two or more agents When participating in a negotiation

protocol, the agent will make proposals and/or provide feedback regarding a

previous proposal A successfully enacted negotiation protocol leads to a

norm-governed relationship formalized in a contract, which specifies the roles taking part

in it and the agents playing these roles

The normative dimension of the contract is governed by a normative context It

is within this context that the contract will be monitored for compliance Normative

contexts can be nested, and each contains constitutive rules and norms that can be

inherited following the contexts’ hierarchical structure For each normative context,

a state is maintained that triggers both constitutive rules and norms (which also have a rule-based representation) A state is composed of illocutions (statements)

Fig 3.2 ANTE meta-model

issued by agents and also by so-called institutional reality elements, most of

which are generated by a norm monitoring process that provides obligation cycle management (for simplicity, these elements are not shown in the diagram)

life-Another kind of elements of institutional reality are institutional facts, which are

obtained through constitutive rules specifying counts-as relations between states andinstitutional facts Such rules can be defined on a per-context basis

Once fired, norms prescribe obligations that apply to agents playing roles within

a contract A norm is thus a prescription rule, indicating, when a specific state isreached, which fact(s) a specific party is obliged to obtain within a given time-window The semantics of these obligations is implemented through monitoringrules

The monitoring process taking place at each normative context generates

evi-dence that can be used by trust models when computing the trust assessment of

each agent in the system Evidence can also be generated directly by the agents,allowing, for example, reputation to be taken into account

Agents may use trust assessments when participating in negotiation protocols.For instance, such assessments can be taken into account for filtering out agentsconsidered untrustworthy, or for evaluating proposals during the execution of anegotiation protocol

A more general overview of the framework is given now

3.2.3.1 Ontology

Every interaction taking place between agents in ANTE is modelled through based ontologies For instance, negotiations use proposals, and give rise to contracts;contract monitoring manipulates concepts such as obligations, facts, deadlines,violations and fulfilments; trust models use reports (evidence) sent by a contract

JADE-monitoring service and provide trust assessments All these concepts are explicitlydefined in ANTE When using the framework in a specific application domain,

a domain-specific ontology must be defined that touches a number of elements,namely: the contents of proposals within negotiation protocols; contract templates;illocutions and institutional facts

3.2.3.2 Atomic Actions

Within the system, agents can participate in negotiation protocols by submittingproposals, can sign contracts, and can execute illocutions (statements sent to thenormative environment) Agents may also use services of the platform The nor-mative environment maintains a normative state by employing a set of monitoringrules

3.2.3.3 Activities

Negotiation protocols are enacted by following the sequence of messages implied

by the protocol Typically, this involves the exchange of proposals and evaluations.Contract execution is a combination of the enactment of the contract by the involvedagents (executing specific illocutions) and of the monitoring process carried out bythe normative environment

3.2.3.4 Subspaces

Different activities take place in conceptually different spaces, while such spaces arephysically undistinguishable Negotiation is coordinated through protocols, whilecontract enactment is coordinated by a normative environment that monitors norms.Agents can participate simultaneously in several negotiations and contract enact-ments Negotiation, contract enactment and trust assessment take place concurrentlyand independently

3.2.3.5 Social/Organizational Arrangements

Contracts specify roles and how agents relate in the underlying “organization”,

by specifying applicable norms Within the normative environment, contracts arereified as contexts that can be nested

3.2.3.6 Interaction with External Environment

Entities in the external environment are agentified into the system so that otheragents can interact with them In particular, certain external entities are regarded as

trusted third parties through constitutive rules that are employed by the normativeenvironment User agents can be regarded as either software, human or hybridagents (accommodating human-in-the-loop approaches).

or proper violation of norms can be handled through norms activated in deviatingstates

3.2.3.8 Social Devices

From a micro perspective, the norms applicable to each contract may include tions (norms applicable in undesirable states) From a macro (society) perspective,social sanctions are also supported through trust models Such models are fed byevidence that include information from the norm monitoring activity

sanc-3.2.3.9 Run-Time Changes

The run-time negotiation of contracts can be seen as introducing changes to thenormative structure that is in place Furthermore, besides adhering to specificcontract templates (for which the normative background is already established),agents can also negotiate or adapt the normative contents of contracts

3.2.3.10 Types of Agents

Platform agents provide infrastructural services, which agents can make use of Thisincludes negotiation mediators, the normative environment itself, or a centralizedimplementation of a trust model User agents, either human or software, use theplatform’s infrastructure

3.2.4 Languages

At this moment, a modelling and specification process called ANTE-MSP is beingspecified This process inherits from PORTO (Castro and Oliveira2008) (a GAIA-Based Agent-Oriented Software Engineering methodology) the modelling phases,notations and diagrams

ANTE-MSP has three phases as follows:

1 Requirements phase that identifies the requirements from the point of view ofstakeholders and users

2 Analysis phase that identifies the environment roles, interactions, normative andorganizational rules

3 Design phase that identifies the agents and services that will implement the roles,functions and interactions identified in the analysis phase

ANTE-MSP does not have a specific description and/or specification language Ituses diagrams based on UML 2.0 with specific stereotypes as the notation to expressthe modelling and concepts used, plus several text and table descriptions to supportthe modelling

In terms of implementation, ANTE is mostly based on Java Negotiation cols are specified as state machines as provided by the JADE API Services provided

proto-by the ANTE infrastructure are accessed through FIPA interaction protocols.Norms are deontic formulae based on directed obligations with timewindows (Lopes Cardoso and Oliveira 2010a), whose life-cycle has beenmodelled using linear temporal logic (LTL) and implemented through monitoringrules (Lopes Cardoso and Oliveira2010b) A rule-based inference engine (Jess) isused to specify norms and reason about their application according to the state ofthe system

3.3 Tools and Platform

ANTE comprises a Java application programming interface (API) built on top

of JADE (Bellifemine et al 2007) Based on this API, ANTE can be extended

to address use-case specific implementations, e.g for negotiation protocols orcomputational trust handling, or automated agent reasoning capabilities, eitherfor negotiation or contract enactment processes By extending appropriate ANTEclasses, developers benefit from reusing intra-platform protocols that glue togetherthe main concepts of ANTE It is also possible to instantiate a given domain byparametrizing the agents in the system through appropriate XML configuration files.ANTE is agnostic as far as agent architectures are concerned, including the human-in-the-loop paradigm for more complex and real-world applications