An efficient algorithm for mining association rules for large itemsets in large centralized databases

The proposed algorithm is derived from the conventional a priori approach with features added to improve data mining performance. These features are embedded in the encoding and decoding mechanisms. It has been confirmed by the preliminary test results that these features can indeed support effective and efficient mining of association rules in large centralized databases. The goal of the encoding mechanism is to reduce the I/O time for finding large itemsets, and to economize memory usage in a predictable manner. The decoding mechanism contributes to speed up the process of identifying different items in a transaction. The performance of three different decoding methods is compared to demonstrate the potential gain delivered by any ingeniously devised decoding approach