Removing Redundancies in Constraint Networks

Book One

Volume

Proceedings of the AAAI Conference on Artificial Intelligence, 6

Automated Reasoning

The removal of inconsistencies from the problem’s representation, which has been emphasized as a means of improving the performance of backhacking algorithms in solving constraint satisfaction problems, increases the amount of redundancy in the problem. In this paper we argue that some solution methods might actually benefit from using an opposing strategy, namely, the removal of redundancies from the representation. We present various ways in which redundancies may be identified. In particular, we show how the path-consistency method, developed for removing inconsistencies can be reversed for the purpose of identifying redundancies, and discuss the ways in which redundancy removal can be beneficial in solving constraint satisfaction problems.