A Granular Computing Approach to Knowledge Discovery in Relational Databases

The main objective of this paper is to present granular computing algorithms of finding association rules with different levels of granularity from relational databases or information tables. Firstly, based on the partition model of granular computing, a framework for knowledge discovery in relational databases was introduced. Secondly, referring to granular computing, the algorithms for generating frequent k-itemsets were proposed. Finally, the proposed algorithms were illustrated with a real example and tested on two data sets under different supports. Experiment results show that the algorithms are effective and feasible. Moreover, the meanings of mining association rules based on granular computing are clearly understandable.