Mining fuzzy association rules using partial support

The paper presents a new approach of mining fuzzy association rules. Most existing methods need to perform multiple scans of the database to get frequent itemsets and work poorly if the data are densely populated and duplicated. Our approach only needs one scan to build the fuzzy P-tree, which is a variant of a set enumeration tree. The tree is stored with partial fuzzy support values of candidate itemsets, which facilitate the calculation of total fuzzy support values. We describe the implementation of our algorithm derived from a priori algorithm. The fuzzy P-tree can be combined with many existing methods and significantly improve their efficiency.