Efficient Mining of Maximal Frequent Itemsets with the Closed and Intersection Approach

Abstract

The application of association rules becomes more and more popular in the field of data mining. In this research we propose a new approach of mining rules, the Bottom-up Closed and Top-down Intersection algorithm by discovering maximal frequent itemsets efficiently. Based on the Pincer-Search’s bottom-up and top-down process, our method combines the closed set with the intersection operation. In the bottom-up process, our algorithm can accelerate the search speed by using the closed itemsets. In the top-down process, we use the intersection operation to generate minimal number of possible candidates. After scanning the source database at the beginning, we use a transaction frequency table to record the frequency of each transaction appearing in the database. To improve the efficiency of the intersection operation, the transaction frequency table is ordered by length and lexicographic sequence. Our method can not only find possible maximal frequent itemsets much faster and reduce the unnecessary database scanning time, but also use less memory space than the algorithms using the tree-structure. When the difference of the number of 1-itemsets and the length of transaction is larger, our algorithm has much better performance.