COST Action IC0801

PROGRAMME:

Minutes wg2-cost-at-cyprus

15 December 2009:

14:30 - 16:00: Session 1: Normative reasoning and norm based programming (10+5 min per talk)

14:30 -14:45 Juan Manuel Serrano “Dealing with incomplete normative states”

14:45-15:00 Moser Fagundes “Compliance with Norms in Autonomous MDP Agents”

15: 00-15:15 “John-Jules Meyer “Normative Systems Programming”

15:15 -16:00 Discussion

16:30-18:00 Session 2: Norm enforcement mechanisms (10+5 min per talk)

16:30-16:45 DanielVillatoro “Distributed multilateral sanctioning”
16:45-17:00 Giulia Andrighetto “On norm Internalization”
17:00-18 Discussion

16 December 2009:

9:30-11:00 Session 3: Applications and norms in agreement- technologies (10+5 min per talk)

9:30-9:45 Simon Miles: “Norms and Contracts”
9:45 -10: 00 Maite Lopez Sanchez “Empirical norm adaptation in a P2P saring network scenario”
10:00-11:00 Further inputs for the panel discussion

11:30 - 13:00: Panel on Norms, Organisations, and Semantics (Moderator: Michael Luck, King’s College, UK)

14:30 - 16:00: Panel on Trust, Argumentation and Semantics (Moderator: Francesca Toni, Imperial College, UK)

16:30- 18:00: Agreement Technologies Community Meeting

PARTICIPANTS SLIDES:

Normative System Programming (John-Jules Meyer, Mehdi Dastani, Nick Tinnemeier, Davide Grossi)

Reasoning about Norm-Compliance with MDP Agents (Moser Fagundes, Holger Billhardt, Sascha Ossowski)

Dealing-with-incomplete-normative-states (Juan Manuel Serrano & Sergio Saugar)

Contracts-between-software-agents (Simon Miles, Nir Oren, Michael Luck)

Distributed punishment (Daniel Villatoro)

ABSTRACT:

Giulia Andrighetto ISTC-CNR Rome

Title: On norm Internalization

Abstract:

Internalization is at study in social-behavioural sciences and moral philosophy since long; of late, the debate was revamped within the rationality approach to the study of cooperation and compliance since internalization is a less costly and more reliable enforcement system than social control. But how does it work? So far, poor attention was paid to the mental underpinnings of internalization. This paper advocates a rich cognitive model of different types, degrees and factors of internalization. In future work, it will be implemented on EMIL-A, and simulated on EMIL-S, in order to check the individual and social effect of internalization.

Moser Fagundes Artifical Intelligence Research Group, University Rey Juan Carlos, Spain

Title: Compliance with Norms in Autonomous MDP Agents

Abstract:

Open regulated multiagent systems (RMAS) implement norms to persuade agents to adapt their behaviour. In a world composed by uncountable open RMAS, it is practically impossible to know all norms and sanctions beforehand. In this sense, the autonomous agents should be endowed with knowledge to adapt themselves to the norms that govern these environments. In this work, we propose a model of agent capable of estimating the benefits of being (non-)compliant with norms. Based on such assessment, the agent is able to decide which norms to comply with and which RMAS to join. To develop the agent model we employ Markov Decision Processes (MDP). In this framework, the adaptations take place by adjusting the MDP’s parameter, and the benefits brought by the (non-)compliance with norms are computed taking into account the expected rewards for a given policy.

John-Jules Meyer University of Utrecht

Title : Normative System programming.

Abstract:

In this talk I show some of the work done in Utrecht on the design and development of a programming language to support the implementation of coordination mechanisms in terms of normative concepts. Special attention is paid to the ‘role’ of roles.

Simon Miles King’s College London, UK

Title: Norms and Contracts

Abstract:

In this presentation, I will outline some of the (hopefully) interesting ideas on normative systems which have come from our
group’s recent work in the area of electronic contracts.  These aim to begin addressing some practical problems of ensuring or checking
compliance with norms in systems with distributed authority and bounded resources.  Problems considered include how agents know when
they do not need to consider a norm in deciding their behaviour, how to handle mitigating circumstances in violation not accounted for in
the norms, how to model norms with deadlines where time is measured on distributed clocks, and so on.

Maite Lopez Sanchez Volume Visualization and Artificial Intelligence Research Group. Departament de Matemàtica Aplicada i Anàlisi, Universitat de Barcelona, Spa

Title: Empirical norm adaptation in a P2P saring network scenario

Abstract:

Within MAS, organizational approaches are designed to constrain the scope of possible interactions among agents. In addition to roles and protocols, norms have proven to be effective means of regulating participants’ activities. Nevertheless, changes in the environment and/or agents’ behaviours can cause a decrease in their effectiveness. Thus, a run time norm adaptation may help the system to better fulfill its social ojectives. Empirical studies on specific scenarios may represent a first step towards a formalization of the norm adaptation process. With this perspective, our research focuses on a empirical study over a P2P multi-agent based simulation scenario

Juan Manuel Serrano Artificial Intelligence Group, School of Engineering, University Rey Juan Carlos, Spain

Title: Dealing with incomplete normative states

Abstract:

This paper puts forward a normative framework which enables the handling of incomplete normative states. In particular, attempts to perform a social action are evaluated as permitted, prohibited (i.e. not permitted) or “pending” for execution (i.e. neither permitted nor prohibited). The framework lets designated agents to resolve this latter category of attempts through the speech acts “allow” and “forbid”. We make use of the support for incompleteness of action language K in the formalisation of the framework.

Daniel Villatoro IIIA-CSIC, Spain

Title: Distributed multilateral sanctioning

Abstract:

Distributed norm enforcement is a vital element to take into account when designing a normative multiagent system. Enforcement techniques such as sanctioning are twofold: (1) improving the norm internalization task, (2) control the undesired behaviours to spread amongst the members of the system. My research is focused on multilateral and distributed sanctioning. In order to achieve such general objective we first have to go through some milestones that are analyzed on this work.