As an important method of data mining, rough set theory is a powerful mathematical tool to deal with uncertainty and noise. By use of the taxonomies of data, a rough set approach to mining multiple level rules is proposed. The mining procedure consists of 4 paragraphs: finding reduct, constructing equivalent classes, building difference matrix and mining rules.
Data mining is a human-machine interactive process, domain knowledge plays an important role in data mining system. We present an ontology-based data mining algorithm (OBDM), which makes the seamless integration of knowledge bases and databases. This algorithm can find interesting multiple levels rules automatically.