Fast Mining of Global Frequent Itemsets

Fast mining of global frequent itemsets is an important data mining problem in a distributed database environment Conventional mining algorithms employ the same framework as Apriori for global frequent itemsets However, candidate set generation is still costly, and the algorithms need repeatedly scan the database, especially when there exist prolific patterns and/or long patterns And communication overhead is costly by transmitting local frequent itemsets for global frequent itemsets In this paper, an algorithm FMAGF（fast mining algorithm of global frequent itmesets） in distributed database is proposed The idea of FMAGF is to only transmit conditional frequent pattern trees or conditional pattern bases but not to transmit a lot of local frequent itemsets; therefore, the algorithm uses far less communication overhead and improves efficiency of mining global frequent itemsets Theory analysis and experimental results show the feasibility and effectiveness of the algorithm