Catálogo de publicaciones - libros

The paradigm of argumentation has been used in the literature to assign meaning to knowledge bases in general, and logic programs in particular. With this paradigm, rules of a logic program are viewed as encoding arguments of an agent, and the meaning of the program is determined by those arguments that somehow (depending on the specific semantics) can defend themselves from the attacks of other arguments.

Most of the work on argumentation-based logic programs semantics has focused on assigning meaning to single programs. In this paper we propose an argumentation-based negotiation semantics for distributed knowledge bases represented as extended logic programs that extends the existing ones by considering sets of (distributed) logic programs, rather than single ones. For specifying the ways in which the various logic programs may combine their knowledge we make use of concepts that had been developed in the areas of defeasible reasoning, distributed knowledge bases, and multi-agent setting. In particular, we associate to each program a cooperation set (the set of programs that can be used to complete the knowledge in ) and the argumentation set (the set of programs with which has to reach a consensus).

In this paper, we introduce the notion of communication channel into a multiagent system. We formalize the system in term of logic with Belief modality, where each possible world includes CTL. We represent the channel by a reserved set of propositional variables. With this, we revise the definition of of FIPA; if the channel exists the receiver agent surely learns the information whereas if not the action fails. According to this distinction, the current state in each world would diverge into two different states. We have implemented a prover, that works also as a model builder. Given a formula in a state in a possible world, the system proves whether it holds or not, while if an action is initiated the system adds new states with branching paths.

Global Abduction (GA) is a recently proposed logical formalism for agent oriented programming which allows an agent to collect information about the world and update this in a nonmonotonic way when changes in the world are observed. A distinct feature of Global Abduction is that in case the agent needs to give up one plan, it may start a new one, or continue a suspended plan, while its beliefs learned about the world in the failed attempts persist. This paper describes an implementation of GA in the high-level language of Constraint Handling Rules (CHR). It appears to be a first attempt to a full implementation of GA, which also confirms CHR as a powerful meta-programming language for advanced reasoning. The construction gives rise a discussion of important issues of the semantics and pragmatics of Global Abduction, leading to proposals for a specific procedural semantics and architecture that seem well suited for real-time applications.

This paper reports on a fertile marriage between madAgents, a Java and Prolog based multi-agent platform, and EVOLP, a logic programming based language to represent and reason about evolving knowledge.

The resulting system, presented with a formal semantic characterisation and implemented using a combination of Java, XSB Prolog and Smodels, provides an improvement of madAgents, allowing for the implementation of a richer agent architecture where agents’ beliefs and behavior, as well as their evolution, are specifiable in EVOLP. It inherits the merits of Answer Set Programming (e.g., default negation for reasoning about incomplete knowledge, a semantics based on multiple answer-sets for reasoning about several possible worlds, etc.) on top of which we add all the specific merits of EVOLP for specifying evolving knowledge. At he same time, the resulting system provides a proof of principle that EVOLP can easily be adopted by existing MAS, to represent an evolving belief base, or also to represent the agent’s evolving behavior.

The second edition of the contest on was held in conjunction with the CLIMA ’06 workshop in Hakodate, Japan. Based on the experiences from the first edition of this contest ([8]), we decided to improve the setting of the first edition. In particular, we built a server to simulate the multi-agent system environment in which the agents from different groups can sense the environment and perform their actions. In this way, different multi-agent systems can compete with each other for the same resources. This allows for much more objective evaluation criteria to decide the winner. Three groups from Brazil, Spain and Germany did participate in this contest. The actual contest took place prior to the CLIMA workshop and the winner, the group from Brazil, was announced during CLIMA ’06.

Antimodels constitute a goal-oriented technique to reformulate a software system at the requirements stage mainly for security concerns. This technique recognizes external goals that conflict with the system’s goals. In addition Tropos is one of the most relevant agent-oriented software development methodologies which uses the modelling language. Tropos covers the software development from the early stage of requirements to implementation. In this paper we propose the use of antimodels to identify risk situations and we use an antigoal resolution taxonomy to reformulate agents’ roles and goals. We test our approach in the context of the Second Computational Logic on Multi-Agent Systems contest where we have used Tropos to implement the proposed collecting agent problem. Combining antimodels, the antigoal resolution taxonomy and the Tropos extension to obtain Prolog implementations, we have obtained a competitive solution.

FLUX is a declarative, CLP-based programming method for the design of agents that reason logically about their actions and sensor information in the presence of incomplete knowledge. The mathematical foundations of FLUX are given by the fluent calculus, which provides a solution to the fundamental frame problem in classical logic. We show how FLUX can be readily used as a platform for specifying and running a system of cooperating FLUX agents for solving the Gold Mining Problem.

This paper describes a team of agents that took part in the second CLIMA Contest. The team was implemented in a logic-based language for BDI agents and was run in a Java-based interpreter that makes it easy for legacy code to be invoked from within the agents’ practical reasoning. Even though the implementation was not completely finished in time, the team won the competition, and the experience also allowed us to improve various aspects of the interpreter.