Fast Frequent Pattern Detection Using Prime Numbers

摘要

Finding all frequent itemsets (patterns) in a given database is a challenging process that in general consumes time and space. Time is measured in terms of the number of database scans required to produce all frequent itemsets. Space is consumed by the number of potential frequent itemsets which will end up classified as not frequent. To overcome both limitations, namely space and time, we propose a novel approach for generating all possible frequent itemsets by introducing a new representation of items into groups of four items and within each group, items are assigned one of four prime numbers, namely 2, 3, 5, and 7. The reported results demonstrate the applicability and effectiveness of the proposed approach. Our approach satisfies scalability in terms of number of transactions and number of items.