ORDERING AND SELECTING PRODUCTION RULES FOR CONSTRAINT MAINTENANCE - COMPLEXITY AND HEURISTIC SOLUTION

We present a technique for analyzing the run time behavior of integrit y constraints repair actions, i.e., active database rules specifically designed to correct violations of database integrity. When constraint s become violated due to an incorrect user transaction, rule computati on is started to restore the database to a correct state. Since repair actions may be numerous and may conflict with each other, an automate d support to the analysis of their run-time behavior is necessary. The proposed technique helps the rule base administrator define a repair rule selection strategy such that the computation terminates for every input transaction, the final database state satisfies all the constra ints, and the user's preferences among different ways to restore integ rity are taken into account. In addition, it can be used by the rule d esigner to spot ''dangerous'' rules that may be subject to redesign. T his problem is formulated as an optimization problem on directed hyper graphs, which we demonstrate to be NP-hard and solve by means of a heu ristic algorithm.