Catálogo de publicaciones - libros

In this paper we propose a new role for the agent metaphor in the definition of the organizational structure of multiagent systems. The agent metaphor is extended to consider as agents also social entities like organizations, groups and normative systems, so that mental attitudes can be attributed to them – beliefs, desires and goals – and also an autonomous and proactive behavior. We show how the metaphor can be applied also to structure organizations in functional areas and roles, which are described as agents too. Thus, the agent metaphor can play a role similar to the object oriented metaphor which allows structuring objects in component objects. Finally, we discuss how the agent metaphor addresses the problems of control and communication in such structured organizations.

Virtual Enterprises are a major trend within the B2B scenario. Technological support towards enabling this cooperation model includes the multi-agent systems paradigm. In this paper we identify requirements of Virtual Enterprise contracts, developing a normative framework for contract validation and enforcement. Furthermore, we enclose this conception within the structure of an Electronic Institution, which governs and supports the interaction of agents in business scenarios, providing specific services such as brokering, reputation, negotiation mediation, and contract related services. We focus on electronic contracting as a means of establishing cooperation agreements, and we describe the institution’s role on the e-contracting life-cycle.

Corporate knowledge consists both of information that is available throughout a company and of information technology frameworks and paradigms. Considering an enterprise as a distributed computational paradigm, multi-agent systems can be proposed to address knowledge management issues within a company. We consider in this paper a new approach for corporate knowledge based on the agent oriented abstraction paradigm. This paradigm provides a high level of abstraction. We investigate here the concept of virtual knowledge communities, which is a convenient concept for addressing dynamical distributed knowledge management. It allows improved simulation and support for knowledge management processes, and therefore to innovate with new methods in this field. Our approach is well-suited for instance to filter the amount of knowledge that is transmitted throughout a company.

We consider the deep venous thrombosis (DVT) as case study for the specification and implementation of a multi-agent system. The DVT is an application with low clinical accuracy, needing objective tests, some of them satisfactorily accurate in experienced hands and others more definite but invasive. Whether one or more decision makers are involved in this activity is a matter of context, but the main events are decided by a process that has in itself some forms of argumentation. Our approach is an argumentative multi-agent system specified by rules capturing various roles in the diagnosis activity. Although the DVT scenario is a real one, more aspects of health care than the ones presented in this paper can conveniently be accommodated in this framework by extending the set of roles and refining the set of rules.

An autonomous agent in a MAS involves in a protocol – more exactly in a conversation following the rules of a protocol – in order to reach objectives, some ones shared with all other participants, some others specific and private. We assume a MAS architecture where each conversation is monitored by a middleware component – a – that guarantees that the shared objectives will be reached. This paper addresses the means an agent requires to be able to exercise its autonomy and reach its own objectives in the course of conversations. The first step is to define these objectives and this leads to distinguish the strategic and tactic levels in agents’ behaviours. The strategic level must be handled by the agent itself; the required capabilities are abstract and relevant for larges categories of similar protocols. Once a strategy is set for a conversation, its application at the tactic level can be delegated to a middleware component, called a , that intervenes in the conversation on the behalf of the agent. This component is specific to the role held by the agent and it is tailored to make the best use of the subtleties of the protocol’s rules. This approach brings many engineering benefits.

We believe that successful co-operation between multi-agent systems providing services in a specific domain can be realized by constructing an infrastructure that supports the semantic interoperability between them. In this paper , we introduce a conceptual architecture for the semantic interoperability of multi-agent systems in the large-scale. The most important element of the proposed conceptual architecture is the federation. A federation established for a particular domain specifies the common characteristics of multi-agent systems in that domain. Hence, a special ontology is required at the federation level for defining the common characteristics of each multi-agent system in a particular domain as its own meta-knowledge. This meta-knowledge is then used in the semantic discovery of the multi-agent systems with which to co-operate. Since different multi-agent systems may use different ontologies, an ontology translation service is also defined at the federation level.

Overhearing is an indirect interaction type that enacts agents to listen to direct interactions among other agents without taking explicit part in the exchanges. In this paper, we propose a formal model of overhearing named T-compound and a methodology to describe generalised interaction networks in Multi-Agent Systems. The compound is defined with the -calculus as an interaction composite. It is handled as an interaction primitive distinct from the traditional point-to-point one, so that our methodology can treat both cases homogeneously.

The development of multi-agent systems (MAS) implies considering both the social and individual levels of these systems. However, the elements in these levels are not necessarily consistent. Conflicts can arise between the goals of the community and those of individual agents. These contradictions are potentially very complex, given the inherent intentional nature of agents and the interactive features of their societies. Developers can face these problems about contradictions with traditional software engineering verification techniques. Nevertheless, these techniques always depend on the understanding of developers about what properties to check. Abstractions of the agent paradigm offer a new possibility for verification: the use of expert knowledge from social sciences to detect and solve this kind of problems. Social sciences and MAS research share a similar view about their objects of study: societies of intentional actors. With this basis, it is possible to adapt the expertise of social sciences to the study of MAS, providing a new source of knowledge for the verification of MAS focused in their social and intentional features. Based on our previous research, we have developed a method to solve motivational contradictions with one of these social theories, the Activity Theory. This method is explained with a case study about the management of a bookstore.

If we are to enable agents to handle increasingly greater levels of complexity, it is necessary to equip them with mechanisms that support greater degrees of autonomy. This is especially the case when it comes to agent-to-agent interaction which, in systems of selfish agents, often follows the format of negotiation. Within this context, a problem which has hitherto received little attention is that of identifying appropriate negotiation opponents. Furthermore, the problem is particularly difficult in dynamic systems where the need to negotiate over issues and the evaluation of resources may change over time. Such dynamics demand high degrees of autonomy from agents so that such factors can be handled at run-time and without the aid of human controllers. To that end, this paper draws inspiration from biological organisms and theories of motivation, and describes a motivation-based architecture comprising a number of motivation-based classification and selection mechanisms used to evaluate and select between negotiation opponents. Opponents are evaluated in terms of the likely issues they will want to negotiate over and the amount of conflict this might entail. Additionally, the expected cost of a negotiation with an opponent is examined in relation to the agent’s current motivational evaluation of its resources. The mechanisms allow prioritisation between each method of evaluation dependent upon motivational needs. Some preliminary evaluation of the model is also presented.

For over a decade, agent research has shown that social commitments support the definition of open multiagent systems by capturing the responsibilities that agents contract toward one another through their communications. These systems, however, rely on the assumption that agents respect the social commitments they adopt. To overcome this limitation, in this paper we investigate the role of sanctions as elements whose enforcement fosters agents’ compliance with adopted commitments. In particular, we present a model of flexible social commitments to which sanctions are attached, and where the enforcement of sanctions act as a social control mechanism for the satisfaction of commitments.