The conceptual development of the rule-based component of the CROQUE query rewrite and optimization system led to the derivation
of an ordering of the rules present in the rule base according to their “optimization potential” in order to increase the
efficiency of the logical term rewriting [5]. This heuristic may indeed be used for any other rule-based optimizer, too.
The major contribution of our approach is a combination of three ideas: (1) limit the search space of query optimization by
grouping and ordering rules (for rule-based optimizers), (2) use of “offline” pre-optimization ordering instead of dynamic
ordering during the optimization process, (3) taking into consideration more than one (“n-best”) alternatives for further
evaluation.