Abstract
Owing to the rapid growth in the sizes of databases, potentially useful in formation may be embeded in a large amount of data. Knowledge discovery is the searc h for semantic relationships in databases. One of the main problems for knowledge di scovery is that the number of possible relationships is very large, thus reducing the search complexity is important. The relationships can be represented as rules which can be used in efficient query processing. We present a knowledge discovery techniq ue to analyze relationships and to derive compact rules. Data are first generalized to reduce their sizes, which makes them easier to be characterized in terms of rules. A mechanism and some heuristics are then proposed to alleviate the computational co mplexity of the rule derivation process. Finally, an evaluation model is presented t o evaluate the quality of the derived rules.