Abstract: Mining frequent itemsets is a major problem in the data mining field with numerous practical appilcations. Efficient incremental updating of discovered frequent itemsets is the key problem. The existing incremental updating algorithms employ the same framework as Apriori, which need to generate a lots of candidate sets and repeatedly scan the database, especially when there exist many patterns and/or long patterns. In this paper, we propose an incremental updating algorithm of frequent itemsets (FIUA) mainly for the updating of frequent itemsets as a time when the minimum support threshold adjusts dynamically. In worst case, FIUA only scans transaction database once and need not generate candidate sets, which saves effectively storage space. Therefore, FIUA is efficient and effective.

Key words: frequent pattern tree, data mining, frequent itemsets, incremental updating