Learning a coverage set of multiple-level certain and possible rules by rough sets

Most of the previous studies on rough sets focused on deriving certain rules and possible rules on a single concept level. Data with hierarchical attribute values are, however, commonly seen in real-world applications. In this paper, we thus propose a new algorithm to deal with the problem of producing a set of maximally general rules for coverage of training examples with hierarchical attribute values using rough sets. A rule is maximally general if no other rule exists that is both more general and with larger confidence than it. All the coverage rules gathered together must cover all the given training examples. The rules derived can then be used to build a prototype knowledge base.