Abstract
The first and the important phase of distributed problem solving is problem decomposition. However, most of the approaches in Distributed AI or Multi-agent systems proceed with the assumption that the problem is already decomposed and concentrate on the other issues. In this paper we present a step towards a methodology for decomposing a given problem either spatially, temporally or by abstraction. Features of these different types are compared and technique for implementing the process is given. Still a lot remains to be done before we can arrive at an application independent formalism to decompose a problem. However, the paper attempts to describe tasks and their interdependencies through the formalism.