Good Old Fashioned Automata Theory and the Agents World.
Leona F. Fass · 2008
Our architectural theory is automata theory and our processes combine knowledge representation and automata-based reasoning, producing the “best possible ” agent configurations that may realize specified behavioral goals. We establish that if a goal can be realized by a configuration of Web-dwelling agents then a distinguished Manager Agent, with suitable Local Closed World knowledge of agents ’ properties, may apply automata-based reasoning algorithms, selecting and configuring agents into a realizing device. The Manager also may reconfigure, adapt and optimize devices, relative to available resources. Real world and logical constraints may mean that a “best possible” result is an approximation; this is confirmed as acceptable by other researchers using a variety of reasoning processes we describe. We generalize our results to local groups of agents, not just those Web-dwelling, and describe both theoretical (“Personal Travel Assistant”) examples and applications to practice.