CRD: A New Data Mining Method in Deductive Databases.
Yunming Wang, Chun Tao, Yonggang Zhao, Yang Yang · 1997
Data mining, also known as knowledge discovery in databases, has been popularly recognized as an important research issue with broad applications. Many kinds of data mining methods have been developed in previous studies. However, in order to serve the representation and implementation of deductive databases and intelligent query answering, there have been increasingly pressing calls on more general-purpose data mining techniques. In this paper, we present a new data mining method called CRD(Common Rule Discovery). It is a general-purpose method that is able to extract a wide spectrum of rules, including recursive ones. We also introduce abstract constants, user-defined predicates, built-in predicates in the extracted rules, which greatly enhance the flexibility and representation power for knowledge. CRD is applicable not only in data mining area, but also in inductive logic programming. This method has been implemented in our CRDS with satisfactory results. Some experiments using CRDS are also provided.