European Agent Systems Summer School

Accepted courses for EASSS'09

The handouts of the slides used during the lectures of EASSS 2009 are available for download at this URL: http://www.di.unito.it/~argo/easss09.pdf

Plenary session: Introduction to Multi-Agent Systems (4 hours)
Andrea Omicini (University of Bologna)
download slides
Abstract. This course will provide an introduction to the field of agents and multi-agent systems, spanning both the micro and macro level, including: basic concepts; agent architecture; interaction and multi- agent systems; applications; and future challenges
Designing and Programming Agents' Environments in Multi-agent Systems (4 hours)
Alessandro Ricci (University of Bologna-Cesena) and Michele Piunti (ISTC-CNR)
Abstract. The notion of environment is a main concept in the characterization of the agent abstraction and multi-agent systems (MAS), being what makes it possible to define aspects concerning agent perception, actions and therefore interactions. This is true both considering MAS situated in physical environments, and for MAS situated in virtual (software) environment. Actually, such a notion can play a key role also in the design and development of Multi-Agent Systems, as a first class abstraction where MAS developers can encapsulate critical functionalities for the system to be built, such as agent organisation and coordination. In this perspective, the availability of proper models, architectures and technologies for developing and executing environments, integrated with those adopted for agents, becomes a relevant issue in the context of MAS design and development. The tutorial will provide, first, a general overview of this topic, with a brief survey of the main approaches in the state of the art; then it will focus on the role of environment in programming multi-agent systems, by considering concrete programming models and technologies and simple examples in practice. For this part, CArtAgO will be used as reference model and technology to design and program computational environments, and Jason agent programming language to program agents working together inside the computational environments. Finally, some research directions and open issues will be presented.
Agent Oriented Software Engineering (6 hours)
Massimo Cossentino (ICAR-CNR, Palermo) and Ambra Molesini (University of Bologna)
download slides

Abstract. At present, most of existing agent applications are developed in an ad-hoc fashion; the most common problems that affect the development activities are: little or no rigorous design methodology, limited specification of the requirements, ad-hoc design of the agents/the multi-agent system as a whole, and little attention to non-functional requirements such as mobility, scalability, performance issues, security, standards. These problems turn to be a limitation for the widespread appliance of the agent-oriented (AO) approach as they would be for any new software technology. In fact, a good quality software product calls for a clear and formal defined development process. However, in the software engineering (SE) field, it is widely recognised that an ideal general-purpose design methodology does not exist. The construction from scratch of a new methodology is a very complex and time-consuming task. Conversely, when developing a new methodology, several agent designers would like to reuse phases, models or elements coming from existing design processes for their own problem. This issue can be faced by adopting the situational method engineering paradigm (SME). SME provides means for constructing ad-hoc software processes following an approach based on the reuse of portions of existing design processes, the so called process fragments. So, the reuse of fragments of well-known methodologies has become a crucial issue for developing new design processes. In this tutorial we aim at introducing basic concepts of SME in the context of agent-based systems and at introducing the field of AO methodologies and processes. In particular we focus on the techniques for composing process fragments in order to create a new AO design processes.
Service Oriented Agents (4 hours)
Benjamin Hirsch (DAI-LABOR TU Berlin) and Stefan Fricke (DAI-LABOR TU Berlin)
Abstract. The goal of the tutorial is to present the field of services and service oriented architectures to the students, and to show in how far agents and SOA and friends can benefit from each other. We believe that this topic is very relevant to aspiring agent researchers as service orientation is of great importance in the industry, and the aims of agent research and SOA are (partially) overlapping. It is therefore important to create awareness of the implications of SOA and the potential benefits the two areas can have from each other. The proposed tutorial will consist of the following three parts: (1) Introduction to SOA. Students will be given a short introduction to service orientation and SOA in general. The pro's and con's of SOA will be presented, as well as developed by the students using handouts and small presentations. (2) How agents can contribute to Services and vice versa. In the second part, we will focus on the common ground of agents and services, and point out different shortcomings of the current SOA approach as well as possible solutions that agents provide. Possible topics here are (formal) semantics in general, ontologies, coordination and autonomous planning, goal oriented behaviour and matchmaking. We also present possible lessons that agent research can learn from SOA and friends. (3) Current Approaches. The final part of the tutorial will deal with work at the interface between agents and services. In particular, we plan to present our agent framework, as well as work to incorporate services, BPEL, and process management into agents.
Designing Intelligent Agents (4 hours)
Brian Logan (University of Nottingham)
download slides
Abstract. This course aims to bridge the gap between abstract models of agency and work on agent programming languages. Starting from the position that a key factor in the successful implementation of a multi-agent system is the selection of an appropriate architecture for both the individual agents and the multi-agent system as a whole, it explores the relationship between the task environment of an agent (or multi-agent system) and the types of architectures that are appropriate for that task environment. The course will define what is meant by an agent architecture and what it means for an agent program to implement a particular architecture. It then presents a classification of task environments which can be used to characterise the key features of (multi-)agent programming problems. Building on this foundation, it surveys the wide range of agent architectures which have been proposed in the agent literature, and develops principles which can be used to select (or develop) an appropriate architecture for a particular task. Finally, it briefly relates architectures to classes of existing agent programming languages, allowing the selection of an appropriate implementation technology for a particular architecture. The course assumes familiarity with basic programming concepts but not of any particular agent programming language.
Fair Division (4 hours)
Ulle Endriss (University of Amsterdam)
Abstract. Fair division is the problem of dividing one or several goods amongst two or more agents in a manner that satisfies a suitable fairness criterion. This is an important topic in multiagent systems, because much research in the field is, either directly or indirectly, concerned with mechanisms for the allocation of resources to groups of agents. This tutorial will provide an introduction to the axiomatic foundations of fair division and present a variety of procedures for different instances of the general problem. See also http://www.illc.uva.nl/~ulle/teaching/easss-2009/.
Agent-based negotiations and auctions (4 hours)
Maria Fasli (University of Essex) and Nicola Gatti (Politecnico di Milano)
Abstract. Agents are well-suited for dynamic, constrained, and real-time environments such as electronic marketplaces. In such environments agents representing their users negotiate for goods and services following negotiation protocols. Bargaining and auctions are the principal negotiation protocols for buying and selling goods based upon competition among the interested parties. This tutorial will introduce participants to agent-based negotiations. The tutorial will start by introducing agent-based negotiation and negotiation protocols in general. Bargaining and auctions will then be described in detail. Essential concepts that are required for following the tutorial will be introduced along the way. The bargaining problem will be introduced and the principal non-cooperative bargaining protocol (Rubinstein's alternating-offers) will be discussed in detail. Variations and applications Rubinstein's protocol in computer science will be presented. The four single side auction protocols (English, Dutch, FPSB, Vickrey) will be discussed in depth along with their relative advantages and disadvantages. Double auctions and the M-th and (M+1)-st clearing rules will also be covered. A brief exposition into more advanced auction formats such as multi-attribute and combinatorial auctions will follow.
Agent-Based Modeling for the Simulation of Complex Systems (4 hours)
Giuseppe Vizzari (University of Milan-Bicocca)
download slides
download example

Abstract. Simulation represents a way to exploit a computational model for the purpose of understanding the behavior of a system and/or evaluating various strategies for its operation. The possibility to realize 'in-machina' experiments supports the study, analysis and evaluation of situations that would not be directly observable in the reality, for practical or even ethical reasons. Several simulation scenarios are characterized by the presence of autonomous entities whose action and interaction determines the evolution of the overall system. Agent-based models are particularly suited to represent these situations, and to support the design and implementation of simulators. This approach has been successfully adopted for the modeling of complex systems in most various contexts: from social sciences, to urban planning, biology, logistics and production, and many other more, both for scientific research and for business/industrial applications. This tutorial is structured in three parts: first of all, computer simulation and the peculiarities of an agent-based approach will be introduced, then the various phases, roles and models involved in a simulation project will be introduced by adopting the case of pedestrian modeling and simulation; finally, a sample simulation scenario will be analyzed and a prototype of MAS based simulator will be designed adopting the Repast Agent Simulation Toolkit.
Normative Multi-agent Systems (6 hours)
Guido Boella (University of Torino), Leon van der Torre (University of Luxembourg) and Xavier Parent (University of Luxembourg)
Abstract. Normative multiagent systems (NorMAS) combine theories and frameworks for normative systems with multiagent systems. These systems provide a promising model for human and artificial agent co-ordination, because they integrate norms and individual intelligence. The course will be split into three. The first half will introduce students to the area, and its main challenges. The second half will be devoted to deontic logic, the logic of obligations and permissions, which aims at providing a formal foundation to normative multi-agent systems. We will start with an introduction to the semantics and axiomatic foundations of deontic logic, with a special emphasis on so-called dyadic deontic logic. Issues related to its meta-theory will be briefly discussed. Some problems have been closed, others are still open. We will then look at (what is certainly) the main problem encountered when applying deontic logic to agent reasoning and communication: the problem of contrary-to-duty reasoning. The third half will be devoted to [roles and organizations?].
New Trends in Multi-Agent Planning (4 hours)
Humbert Fiorino (Laboratoire d'Informatique de Grenoble), Damien Pellier (Centre de Recherche en Informatique de Paris5) and Cyrille Martin (Laboratoire d'Informatique de Grenoble)
Abstract. Multi-agent planning has been a very efficient tool for coordinating autonomous systems (e.g. robots, drones, underwater autonomous vehicles). In this tutorial, we propose a progressive introduction to multi-agent planning and its most important applications. The traditional approaches of multi-agent planning emphasize the problem of controlling the actions of agents in a shared environment and solving their conflicts. New approaches, as composition, focuse on the distributed generation of shared plans as composition patterns for ambient computing platforms. First of all, we introduce the fundamental concepts of automated planning, complexity results and essential algorithms: state-space planning, plan-space planning, propositional satisfiability and planning-graph techniques, and planning under uncertainty. We explain how to use the corresponding planners and illustrate their respective advantages and drawbacks with short exercices. In a second part, we present the key mechanisms of distributed plan- ning: plan merging, Partial Global Planning and coordination protocols. The main issues are illustrated with state-of-the-art distributed robotics applications. Finally, we present recent researches and applications based on multi-agent planning systems for automating web services composition or Human Computer Interface on-the-fly generation.
Agents and Arguments (4 hours)
Sanjay Modgil (King's college London)
Abstract. In recent years there has been a rapid growth of interest in the use of logic based models of argumentation for modelling agent reasoning and dialogue. Its promise resides in its intuitive, modular nature, and the potential for tractable implementations. In particular, argumentation provides: 1) a general framework for inference and decision making in the presence of inconsistent, uncertain and incomplete information; 2) a principled way in which to structure rational discussion and enable exchange of, and reasoning about, arguments for proposals and or statements between human and or automated agents. The aim of the course is to introduce students and researchers to the fundamentals of logic based models of argumentation, and recent developments of these models for application in agent reasoning and communication. The course will begin with a review of Dung~Rs abstract argumentation framework, and recent extensions to the framework developed to facilitate flexible and adaptive agent reasoning over conflicting beliefs, goals, and decision making over action. The course will then review argument game proof theories for abstract frameworks, and discuss how such games can be generalised in order to define frameworks for argumentation based dialogues.
Introduction to Game Theory and Mechanism Design (4 hours)
Paul Harrenstein (Ludwig-Maximilians-Universität München) and Mehdi Dastani (Utrecht University)
Abstract. In multi-agent systems multiple agents with individual preferences interact in a common environment. For the design and proper formal understanding of the interaction in multi-agent systems appropriate mathematical concepts are required. Game theory provides and analyzes such concepts as equilibrium outcomes and strategies. Roughly speaking, mechanism design is concerned with the development of multi-agent systems with desirable game-theoretic properties even if the preferences of the agents are unknown to the designer or to the other agents. It is also concerned with what is formally possible and impossible in this respect. Put this way, the relevance of game theory and mechanism design for multi-agent research might seem obvious. Nevertheless, game theory nor mechanism design are commonly part of a standard curriculum in computer science. This course aims to fill this gap. This introductory course is aimed at postgraduate students in computer science and multi-agent systems who have no previous acquaintance with game theory or mechanism design. As such only some elementary mathematical training is assumed. The main goal of the course is to make the students familiar with the most fundamental concepts and results of game theory and mechanism design and to make clear how game theory and mechanism design can be applied to their own research.
Coalitional Games (4 hours)
Stéphane Airiau (University of Amsterdam) and Wojtek Jamroga (University of Luxembourg)
Abstract. Over the last decade, coalition formation has received increased attention in the multiagent systems community: forming dynamic coalitions may lead to more efficient agent societies. Game theory prescribes ways to share a payoff obtained by a coalition in a stable manner, but it does not describe how to form efficient coalitions. Representing coalitional games, and reasoning with such representations are other key issues. In the first part of the course, our first goal is to introduce the basic solution concepts from game theory. Then, we discuss the challenges of coalition formation in the context of multiagent systems: communication, dynamic environments, uncertainty about knowledge or tasks, protocols and manipulation, etc. We survey proposed solutions to these issues, and finally we discuss the problem of searching for efficient coalition structures. The focus of the second part is descriptive. We introduce some modal logics of strategic ability, and show how abilities of coalitions can be specified in that language. Finally, we discuss the algorithmic side of checking such specifications. The course requires some elementary knowledge of logic and game theory. Familiarity with basic modal logic will be an advantage.
Programming Multi-Agent Systems (4 hours)
Koen Hindriks (Delft University of Technology) and Joao Leite (New University of Lisbon)
Abstract. With the significant advances in the area of autonomous agents and multi-agent systems in the last few years, promising technologies have emerged as a sensible alternative for the design of systems that can operate in complex and dynamic scenarios. However, in order for this technology to become accessible to the multi-agent research community in Academia and practitioners in Industry, it is necessary that programming languages and tools that are appropriate for developing such systems become widely known and thoroughly understood. This course aims at introducing novices, researchers, and developers (from both Academia and Industry) who already have basic notions of multi-agent systems to some of the languages, techniques, and tools that are currently available to support the effective implementation of multi-agent systems.

COST Related Course

ABSS4NORMS: Agent based social simulation for norm emergence (4 hours)
Giulia Andrighetto (ISTC-CNR, Rome) and Marco Campennì (ISTC-CNR, Rome)
Abstract. Norm emergence has been the object of a wide amount of studies in several different academic disciplines. Apart from law sciences, there are also interesting contributions from Philosophy, Game Theory and Economics. In the scientific literature, two major views of norm emergence co-exist. In the formal social scientific field, that is in utility and (evolutionary) game theory, the spread of new norms and other cooperative behaviors is not explained in terms of internal representations. The object of inquiry is usually the conditions for agents to converge on given behaviors, which proved efficient in solving problems of coordination or cooperation, independent of the agents~R mental representations.In the field of multi-agent systems, instead, norms are explicitly represented. However, they are implemented as built-in mental objects. When norm emergence is addressed, the starting point is some preexisting norms, and emergence lies in integrating them. No apparent contamination and integration between these different directions of investigation has been achieved so far. In particular, it is unclear how something more than regularities can emerge in a population of intelligent autonomous agents and whether agents~R mental capacities play any relevant role in the emergence of norms. The tutorial is aimed to show how agent-based simulation can represent an opportunity for innovation in addressing these questions, because it provides both an observatory for visualization and a laboratory for experimenting on the dynamics of norms. The course will be split in two parts. The first part will introduce students to the area, and its main challenges: what are social norms? How do social norms emerge? What is the role of agents, their interaction and their internal mechanisms and representations, if any, in the emergence and spread of norms? More generally, what type of agent architecture should be designed and implemented in order to observe normative processes and phenomena on a computer? In the second part, using Net-Logo, students will carry out some simulations on the emergence of norms in different contexts.
Trust and reputation in multiagent systems (4 hours)
Jordi Sabater-Mir (IIIA-CSIC) and Laurent Vercouter (École Nationale Supérieure des Mines de Saint-Étienne)
Abstract. course proposal
Agent-Oriented Modelling (4 hours)
Kuldar Taveter (Tallinn University of Technology, Estonia)
Abstract. As the boundaries blur between systems and their users, new modelling abstractions and notions are required for modelling and designing socio-technical systems in the distributed world. These are exactly the kinds of abstractions we intend to introduce and analyse in the tutorial, the most fundamental ones of them being "agent" and "system". In the tutorial, we first introduce a conceptual space within which to view systems. The conceptual space consists of three layers: a deployment layer, a system design layer, and a motivation layer. Orthogonal to the three horizontal layers are vertical concerns that cross the layers. They are needed for a clear understanding of the issues to be addressed when designing and implementing systems. The vertical concerns proposed by us for systems are information, interaction, and behaviour. The types of models required for domain analysis, design, and implementation lie at the intersections of abstraction layers and cross-cutting concerns. In the tutorial, we first introduce a conceptual space within which to view systems. The conceptual space consists of three layers: a deployment layer, a system design layer, and a motivation layer. Orthogonal to the three horizontal layers are vertical concerns that cross the layers. They are needed for a clear understanding of the issues to be addressed when designing and implementing systems. The vertical concerns proposed by us for systems are information, interaction, and behaviour. The types of models required for domain analysis, design, and implementation lie at the intersections of abstraction layers and cross-cutting concerns. The material to be covered in the tutorial forms the basis for an upcoming book by Leon Sterling and the tutor entitled "The Art of Agent-Oriented Modeling" (MIT Press, August 2009). Copies of the book will be used in the tutorial.