Assume a set of variables pulling values from respective domains and a set of constraints they are supposed to satisfy, and you have a constraint satisfaction problem (CSP). Solving CSPs is not as easy as defining them. Due to the generality of the problem, efforts to provide approaches that would give insights into techniques to help us effectively solve them are starting to emerge.
According to Stergiou, this paper proposes “a number of simple lightweight heuristics for switching between different constraint propagation methods applied on individual constraints during search.” He proposes several heuristics, such as generalized arc consistency; modification of maintaining arc consistency (MAC), called max restricted path consistency (max RPC); semi-automated domain wipeout (DWO) monitoring; fully or semi-automated deletion monitoring; and a couple of versions of fully or semi-automated hybrid heuristics. These heuristics are shown at work on two-variable problems, and their success is compared.
Due to these and other parallel efforts in the research community, we are sure to soon have methodologies that will make the identification of efficient solutions of CSP in the general case applicable to fields that constantly face these problems and need solutions quickly.