Network distributed decision-support systems as multi-agent constructs

The paper introduces some of the problems raised when researching and designing computer- and network-mediated tools intended for supporting decision making processes and the underlying processes of social learning. Some currently- and potentially-available tools from agent-oriented analysis and design, a sub-field of Artificial Intelligence (AI) are also introduced. These tools can be used to represent gaming processes involving potentially large number of participants and interested parties and 'played' within an electronic environment which serves as a common platform. They can be perceived as distributed over a number of human and adaptive artificial agents. Thus the strong support for the research and design of these new kind of computer-based tools that comes from agent-oriented research opens up the development of 'network distributed decision support systems' as Multi Agent Systems (MASs). Different types of possible agent architectures are presented and related to their best mode of functioning within specific environments including other agents. Since these new kinds of decision support tools are envisaged as multi-agent systems consisting of heterogeneous agents, the emphasis is placed on inter-agent communication. It is argued that the interoperability of the agents may be greatly enhanced if they are built with mental states consisting of sets of mental attitudes. The kinds of environments within which these agents might be expected to operate are discussed, especially within the context of the aquatic environment and its relation to the human-social environment.