Extended breadth-first search algorithm in practice

Abstract

Classical AI search algorithms are not applicable in practice when the state space contains even only a few tens of thousands of states. Extended breadth-first search (EBFS) is an algorithm developed to give remedy to some problems related to the classical state-space representation used in artificial intelligence. This algorithm was initially intended to give us the ability to handle huge state spaces and to use a heuristic concept which is easier to embed into search algorithms. However, the base idea of EBFS – i.e., running more explorations of parts of the representation graph starting from several distinct nodes – also implies the possibility of parallelization. In our paper, we show some real-life examples of problems that can be used to illustrate the advantages of EBFS over the classical search algorithms and the use of extended state-space model (ESSM), which was introduced as a possible problem representation technique for EBFS.