Donald Steiner

The conceptual framework of MAI2L

This paper describes the key concepts behind the Multi-Agent Implementation and Interaction Language, MAI2L. The three major design requirements of rationality, reactivity and generic cooperation are explained and the architecture to support them is detailed.

Collaboration of Knowledge Bases via Knowledge Based Coordination

To take advantage of the growing number of databases and knowledge bases, we present a system that allows for cooperation among spatially distributed agents. Cooperation is supported by providing a uniform agent model and a framework for instantiating various cooperation strategies. The knowledge base functionality forms an agent’s body, while the cooperation functionality is achieved via the agent’s head, which is a highly specialized coordination and communication expert system. The power of the cooperation arises from the ability to combine highly specialized knowledge bases and powerful reasoning engines to “expert system task forces”. These task forces are created only for the duration of a certain project and modify themselves as the need arises. This flexibility brings the benefits of coordination technology from the area of computer supported cooperative work to distributed knowledge bases. We demonstrate these ideas using the domain of linguistic knowledge representation.

Cooperation Structures in Multi-Agent Systems

We propose a multi-agent environment (MECCA) for supporting cooperation between humans and machine systems in weakly structured task domains. The two basic ingredients of this environment are the generalised structure of an individual agent and the representation of a variety of available cooperation strategies and their use by the agents. After a characterisation of the targeted application domains, the global architecture and the integration of cooperation structures within this framework will be discussed.

Resource Adaptation for a Scalable Agent Society in the MoTiV-PTA Domain

Agent-based approaches are becoming more and more mature for applications distributed over large networks, such as the Internet. Although the types of applications vary greatly, ranging from information retrieval to electronic commerce, all these systems can be supported by a common infrastructure provided by generic agent development environments. In this chapter, we describe the MECCA system (Multi-agent Environment for Constructing Cooperative Applications), a class library written in Java which supports the high-level implementation of agent systems. Implementing the FIPA inter-agent communication and agent management standards, MECCA allows for inter-operability among agents of different architectures, coded by different providers.

Multi-Agent Cooperation - Concepts and Applications

In the current discussion the notion of an agent and its descendents like agent-based systems, agent software etc. are used in manifold ways exhibiting both similarities and divergencies. Our approach to this area stems from Distributed Artificial Intelligence (DAI). The central problem underlying the design of agent based systems from a DAI point of view is characterized by the following question: How can individually motivated, independently designed computational artefacts, agents, act together to achieve some (at least partial) common goal in a genuinely distributed problem space?

On The Evaluation of Agent Architectures

By now, intelligent agents have been on the research agenda of the computer science community for roughly one decade. Still, control architectures for autonomous intelligent systems have been an important research issue for an even much longer time, going as far back as James Watt’s steam engine control mechanism based on mechanical feedback. More recent work includes the development of mathematical models for control in the field of cybernetics (most notably, Wiener). Also, arguably one of the biggest contributions of more than forty years of research in Artificial Intelligence were methods and architectures aiming to describe, control, and adopt intelligent autonomous systems.

Wissensbasierte Unterstützung von Gruppenarbeit oder: Die Emanzipation der maschinellen Agenten

Die fortschreitende Entwicklung in den Gebieten Verteilte Kunstliche Intelligenz und computergestutzte Gruppenarbeit beinhaltet hohe synergetische Potentiale fur die Modellierung und Entwicklung komplexer Mensch-Maschine Kooperationsszenarien. In diesem Artikel stellen wir ein theoretisches Rahmenwerk fur die Modellierung kooperativer Arbeit dar, an der sowohl Menschen wie auch intelligente Systeme teilnehmen konnen. Die Sichtweise von CSCW (Computer Supported Cooperative Work) und VKI (Verteilter Kunstlicher Intelligenz) wird dadurch zur Mensch-Maschine-Gruppe Kooperation (engl. Human-Computer Cooperative Work).