Abstract. The scientific research in the area of computational mechanisms for trust and reputation in virtual societies is a recent discipline oriented to increase the reliability and performance of electronic communities. Computer science has moved from the paradigm of isolated machines to the paradigm of networks and distributed computing. Likewise, artificial intelligence is quickly moving from the paradigm of isolated and non-situated intelligence to the paradigm of situated, social and collective intelligence. The new paradigm of the so called intelligent or autonomous agents and Multi-Agent Systems (MAS) together with the spectacular emergence of the information society technologies (specially reflected by the popularization of electronic commerce) are responsible for the increasing interest on trust and reputation mechanisms applied to electronic societies. This review wants to offer a panoramic view on current computational trust and reputation models.

The Semantic Web enables intelligent agents to “outsource ” knowledge, extending and enhancing their limited knowledge bases. An open question is how agents can efficiently and effectively access the vast knowledge on the inherently open and dynamic Semantic Web. The problem is not that of finding a source for desired information, but deciding which among many possibly inconsistent sources is most reliable. We propose an approach to agent knowledge outsourcing inspired by the use trust in human society. Trust is a type of social knowledge and encodes evaluations about which agents can be taken as reliable sources of information or services. We focus on two important practical issues: learning trust and justifying trust. An agent can learn trust relationships by reasoning about its direct interactions with other agents and about public or private reputation information, i.e., the aggregate trust evaluations of other agents. We use the term trust justification to describe the process in which an agent integrates the beliefs of other agents, trust information, and its own beliefs to update its trust model. We describe the results of simulation experiments of the use and evolution of trust in multi-agent systems. Our experiments demonstrate that the use of explicit trust knowledge can significantly improve knowledge outsourcing performance. We also describe a collaborative trust justification technique that focuses on reducing search complexity, handling inconsistent knowledge, and avoiding error propagation. 1.

As collaborative repositories grow in popularity and use, issues concerning the quality and trustworthiness of information grow. Some current popular repositories contain contributions from a wide variety of users, many of which will be unknown to a potential end user. Additionally the content may change rapidly and information that was previously contributed by a known user may be updated by an unknown user. End users are now faced with more challenges as they evaluate how much they may want to rely on information that was generated and updated in this manner. A trust management layer has become an important requirement for the continued growth and acceptance of collaboratively developed and maintained information resources. In this paper, we will describe our initial investigations into designing and implementing an extensible trust management layer for collaborative and/or aggregated repositories of information. We leverage our work on the Inference

work contained in this document has been submitted in support of an application for a degree or qualification of this or any other university or other institution of learning. All verbatim extracts have been distinguished by quotation marks, and all sources of information have been specifically acknowledged. Signed: Date: January 2005 This thesis describes a model of norm-governed practical reasoning agents and demonstrates the implementation of this model in the form of the NoA Normative Agent language and architecture. The introduction of normative concepts such as obligations, permissions and prohibitions into the practical reasoning of an agent is motivated by the need for effective coordination mechanisms in open systems. These are typically “multi-vendor ” scenarios, where independently designed agents, as representatives of human organisations and individuals, form short to medium term coalitions and collaborate in the performance of specific tasks. Electronic commerce is one of the most prominent examples of such scenarios. The use of autonomous software agents provides the

In the near future, billions of entities will be connected to each other through the Internet. The current trend is that an increasingly number of entities, from smart personal devices to legacy databases, are controlled by software agents. Such agents often also posses a large amount of information about both the entity and its owner. Thus, a likely scenario is that the Internet will be populated by millions of information agents, all potentially able to communicate with each other. Unfortunately, we cannot assume that these agents are benevolent and are willing to cooperate in an altruistic fashion. As the amount of money transferred via the Internet is rapidly increasing caused by the breakthrough of e-commerce, we should actually expect a similar increase in the number of malicious agents. Another aspect that contributes to the complexity of agent interaction on the Internet is a desired openness, making it difficult to engineer agent societies in a top-down manner. Rather, we will here investigate the prerequisites necessary to form stable and trustworthy societies of information agents, and discuss some open problems and methodologies for studying them. The general conclusion is that more research is needed that takes into account the presence of malicious agents.

In negotiation the exchange of contextual information is as important as the exchange of specific offers. The curious negotiator is a multiagent system with three types of agent. Two negotiation agents, each representing an individual, develop consecutive offers, supported by information, whilst requesting information from its opponent. A mediator agent, with experience of prior negotiations, suggests how the negotiation may develop. A failed negotiation is a missed opportunity. An observer agent analyses failures looking for new opportunities. The integration of negotiation theory and data mining enables the curious negotiator to discover and exploit negotiation opportunities. Trials will be conducted in electronic business. 1.

Abstract. In this paper we apply Boella and van der Torre’s normative multiagent systems to analyze a model of transaction trust introduced by Tan and colleagues. We discuss the role of constitutive and regulative norms in party trust and control trust. Moreover, we use so called contract negotiation games to indicate when controls are needed. Though sometimes regulators focus exclusively on adding controls, a scenario analysis can be used to show when this is not needed. We illustrate these issues of transaction trust by a case study from international trade, namely the Letter of Credit procedure. 1

In some rescue or emergency situations, agents may act individually or on the basis of minimal coordination, while in others, fullfledged teamwork provides the only means for the rescue action to succeed. In such

Abstract. Information systems must establish trust to cooperate effectively in open environments. We are developing an agent-based approach for establishing trust, where information systems are modeled as agents that provide and consume services. Agents can help each other find trustworthy parties by providing referrals to those that they trust. We propose a graph-based representation of services for modeling the trustworthiness of agents. This representation captures natural relationships among service domains and provides a simple means to accommodate the accrual of trust placed in a given party. When interpreted as a lattice, it enables less important services (e.g., low-value transactions) to be used as gates to more important services (e.g., high-value transactions). We first show that, where applicable, this approach yields superior efficiency (needs fewer messages) and effectiveness (finds more providers) than a vector representation that does not capture the relationships between services. Next, we study trade-offs between various factors that affect the performance of this approach. 1

Though communication is a vital ingredient of Cooperative Problem Solving (CPS) in multiagent systems (MAS), an in-depth analysis of different types of communication has been missing in the MAS literature. This paper presents an investigation into the role that five specific types of dialogue play during consecutive stages of CPS in BDI systems, i.e., potential recognition, team formation, plan formation,andteam action. The presented approach is based, on the one hand, on our theory of social and collective motivational attitudes in BDI systems, with collective commitment as a central concept. On the other hand, the typology of dialogues given by Krabbe and Walton is exploited. The relevant dialogue types are: persuasion, negotiation, inquiry, deliberation,andinformation seeking. The whole process of dialogue among computational agents needs to be transparent. Thus, at each level of CPS the agents' internal attitudes need to be established, and then updated and revised when needed.