A strategy for increasing the efficiency of rule discovery in data mining

Increasing the efficiency of rule discovery is currently a major focus of research interest in data mining. Strategies available to the data miner include data sampling, knowledge-guided discovery, attribute reduction, parallelisation of the discovery process, and focusing on the discovery of a restricted class of rules, or those which appear most promising according to some measure of rule interest. This paper presents a new approach which combines the strategies of focusing on rules which appear most interesting, exploiting structural features of the data set when possible, and decomposition of the discovery process into sub-tasks which can be executed independently on parallel processsors.