A new mechanism of rule acquisition based on covering rough sets

Rule acquisition, known as knowledge acquisition, is an important and topical issue in granular computing theory. Granules are not only composed of objects but also have feature values. However, In the granule associativity rules, the traditional rule extraction methods fail to consider the influence of granules on the decision, thus the method is not well adapted in reality. On the other hand, the existing methods lack a rule-based measure for information systems. In this paper, the action parameters are first introduced to establish a more realistic granule associativity rule in covering information system. Further, we present the rule-based data potential to address the measurement problem. In addition, rule-based scale selection in multi-scale covering rough sets is explored, followed by a scale combination integrating generalization capability, data potential, and lower approximation. Finally, algorithm is designed and experiments are conducted.