Incremental induction interesting knowledge based on the change of attribute values in incomplete information systems

In incomplete information systems, this paper proposes an algorithm of incremental induction interesting knowledge based on the change of attribute values. We can obtain new interesting knowledge for partially modify the original accuracy matrix and the original coverage matrix when attribute values have changed in incomplete information systems, which can increase the efficiency. First, some concepts related to the similarity relation, the similarity classes and interesting knowledge are presented. Then we analyze the change of attribute values which have three cases in incomplete information systems: the change of missing attribute values, the coarsening and refining of condition attribute values and decision attribute values. Next, this paper introduces the algorithm of incremental induction interesting knowledge in details. Finally, an example illustrates the algorithm from three cases in the change of attribute values.