Catálogo de publicaciones - libros

Social commitments are developed for multi-agent systems according to the current practice in law regarding contract formation and breach. Deafeasible commitments are used to provide a useful link between multi-agent systems and legal doctrines. The proposed model makes the commitments more expressive relative to contract law and it stresses the representational rather than the operational side of the commitment life cycle. As a consequence, the broader semantics helps in modeling different types of contracts (gratuitous promises, unilateral contracts, bilateral contracts, and forward contracts) and negotiation patterns. The semantics of higher-order commitments is useful in deciding whether to sign an agreement or not and to represent a larger variety of protocols and legal contracts.

Goals are used to define the behavior of (pro-active) agents. It is our view that the goals of an agent can be seen as a knowledge base of the situations that it wants to achieve. It is therefore in a natural way that we use Dynamic Logic Programming (DLP), an extension of Answer-Set Programming that allows for the representation of knowledge that changes with time, to represent the goals of the agent and their evolution, in a simple, declarative, fashion. In this paper, we represent agent’s goals as a DLP, discuss and show how to represent some situations where the agent should adopt or drop goals, and investigate some properties that emerge from using such representation.

In this paper, we propose a collaborative framework to realize a Virtual Enterprise (VE) for the domain of Micro Assembly. The framework is developed using 3APL technologies [7] and employs the idea of viewing WebService composition as a planning problem [8]. We describe the implementation of the framework and experiment with two micro assembly work cells.

Agent-UML (AUML) extended UML in order to facilitate the modeling process for agent based systems. It offers several graphical notations, including protocol diagrams which represent agent interaction protocols. In this paper, we describe an AUML-based framework to specify generic protocols. We call generic protocols, agent interaction protocols where only a general behavior of the interacting entities can be described. From AUML protocol diagrams, we identified five fundamental concepts on top of which we defined formal specifications of generic protocols. Through our specifications, we addressed a lack in generic protocol representation by emphasizing the description of actions performed in the course of interactions based on such protocols. The framework we developed is formal, expressive and of practical use. It helps decouple interaction concerns from the rest of an agent’s architecture. As an application, we used this framework to publish the specifications of generic protocols for agent interactions in several multi-agent system applications we developed. Additionally, the framework helped us address two issues faced in the design of agent interactions based on generic protocols, protocol configuration and their dynamic selection.

This paper presents two approaches for generating and executing the plans of cognitive agents. They can be used to define the semantics of programming languages for cognitive agents. The first approach generates plans before executing them while the second approach interleaves the generation and execution of plans. Both approaches are presented formally and their relation is investigated.

We outline the , a prototype language implemented in Haskell (a declarative programming language) in which we model agents executing abstract actions relative to deontic concepts derived from Standard Deontic Logic and Dynamic Deontic Logic. The concepts of abstract actions are derived from Dynamic Logic. The logics are declarative, while the implementation is operational. Actions have explicit action preconditions and postconditions. We have deontic specification of complex actions. We implement a case. We distinguish from obligations on sequences, which has not previously been accounted for in the literature. The central innovation is the expression of complex violation and fulfillment markers. The language can be used to express a range of alternative notions of actions and deontic specification.