Laurent Vercouter

A specification of the Agent Reputation and Trust (ART) testbed: experimentation and competition for trust in agent societies

A diverse collection of trust-modeling algorithms for multi-agent systems has been developed in recent years, resulting in significant breadth-wise growth without unified direction or benchmarks. Based on enthusiastic response from the agent trust community, the Agent Reputation and Trust (ART) Testbed initiative has been launched, charged with the task of establishing a testbed for agent trust- and reputation-related technologies. This testbed serves in two roles: (1) as a competition forum in which researchers can compare their technologies against objective metrics, and (2) as a suite of tools with flexible parameters, allowing researchers to perform customizable, easily-repeatable experiments. This paper first enumerates trust research objectives to be addressed in the testbed and desirable testbed characteristics, then presents a competition testbed specification that is justified according to these requirements. In the testbeds artwork appraisal domain, agents, who valuate paintings for clients, may gather opinions from other agents to produce accurate appraisals. The testbeds implementation architecture is discussed briefly, as well.

A Peer-to-Peer Normative System to Achieve Social Order

Social order in distributed descentralised systems is claimed to be obtained by using social norms and social control. This paper presents a normative P2P architecture to obtain social order in multi-agent systems. We propose the use of two types of norms that coexist: rules and conventions. Rules describe the global normative constraints on autonomous agents, whilst conventions are local norms. Social control is obtained by providing a non-intrusive control infrastructure that helps the agents build reputation values based on their respect of norms. Some experiments are presented that show how communities are dynamically formed and how bad agents are socially excluded.

A logic of trust and reputation

The aim of this paper is to present a logical framework in which the concepts of trust and reputation can be formally characterized and their properties studied. We start from the definition of trust proposed by Castelfranchi & Falcone (C&F). We formalize this definition in a logic of time, action, beliefs and choices. Then, we provide a refinement of C&F’s definition by distinguishing two general types of trust: occurrent trust and dispositional trust. In the second part of the paper we present a definition of reputation that is structurally similar to the definition of trust but moves the basic concept of belief to a collective dimension of group belief.

MADCAR: an abstract model for dynamic and automatic (re-)assembling of component-based applications

Dynamicity is an important requirement for critical software adaptation where a stop can be dangerous (e.g. for humans or environment) or costly (e.g. power plants or production lines). Adaptation at run-time is also required in context-aware applications where execution conditions often change. In this paper, we introduce MaDcAr, an abstract model of dynamic automatic adaptation engines for (re-)assembling component-based software. MaDcAr aims at being a conceptual framework for developing customizable engines reusable in multiple applications and execution contexts. Besides, MaDcAr provides a uniform solution for automating both the construction of application from scratch and the adaptation of existing component assemblies.

Instrumenting Multi-agent Organisations with Artifacts to Support Reputation Processes

Reputation is often cited as an instrument to enforce norm compliance: agents that do not follow the norms have their reputation decreased. Conceiving reputation as a collective process, i.e. a kind of shared voices as proposed by Conte & Paolucci, is not a simple task. In this paper, we propose a first step in this direction by instrumenting multi-agent organisation with an artifact that publishes some objective evaluations of the performance of the agents with respect to their behaviour within the organisation. The members of the organisation can then read these evaluations and build up their reputation of others. The artifact serves thus as an instrument that aid in the building of the reputation of the agents. We propose that the evaluation of the agents is not simply based on their obedience to norms, but also considers their pro-activeness and their contribution to the success of collective tasks that are being executed in the organisation. This proposal is detailed and exemplified in the context of the

The agent reputation and trust (ART) testbed

\mathcal M

Handling privacy as contextual integrity in decentralized virtual communities: The PrivaCIAS framework

oise + organisational model supported by a set of organisational artifacts as proposed in the ora4mas approach.

SOARI: A Service Oriented Architecture to Support Agent Reputation Models Interoperability

The Agent Reputation and Trust (ART) Testbed initiative has been launched with the goal of establishing a testbed for agent reputation- and trust-related technologies. The art Testbed serves in two roles: (1) as a competition forum in which researchers can compare their technologies against objective metrics, and (2) as a suite of tools with flexible parameters, allowing researchers to perform customizable, easily-repeatable experiments. In the Testbeds artwork appraisal domain, agents, who valuate paintings for clients, may purchase opinions and reputation information from other agents to produce accurate appraisals. The art Testbed features useful data collection tools for storing, downloading, and replaying game data for experimental analysis.

Decentralized monitoring of agent communications with a reputation model

Contextual Integrity has been proposed to define privacy in an unusual way. Most approaches take into account a sensitivity level or a “privacy circle”: the information is said to be either private or public and to be constrained to a given group of agents, e.g. “my friends”, when private. In the opposite, Contextual Integrity states that any information transmitted can make this transmission a privacy violation depending on its context. We use this theory to develop a novel framework that one can use in an open and decentralized virtual community to socially enforce privacy. This paper proposes the PrivaCIAS framework, in which privacy constraints are formally described to be used to detect privacy violations according to the Contextual Integrity theory. This PrivaCIAS framework provides social control to agents that handle the information, so that deceiving agents are excluded from the system.

Detection of Multiple Identity Manipulation in Collaborative Projects

Interaction among agents about reputation is essential to accelerate the convergence of the reputation evaluation. Semantic heterogeneity of different reputation models makes agent interaction a problem, especially in open MAS. We propose in this paper a service oriented architecture, SOARI , to deal with this problem. Namely, SOARI is composed of an external Ontology Mapping Service ( OMS ) and an agent internal Translator module. This architecture supports concept mapping and translation among reputation model ontologies to a common ontology and vice-versa, thus allowing heterogeneous agents to interoperate about reputation. We present an example of the OMS s use, by showing the mapping of two reputation models to a common ontology. Finally, we illustrate the use of the whole SOARI architecture in a simplified electronic commerce scenario. These experiments have enabled us to analyze the limitations and advantages of a hybrid approach for heterogeneous agents interoperation.