IMBT--A Binary Tree for Efficient Support Counting of Incremental Data Mining

In the real world application, databases are updated continually. Most data mining approaches face the efficiency problem of repeating the mining process when the database is updated. Therefore, developing efficient approaches of incremental data mining is a critical issue for the real world data mining application. If we could use the previous analysis to incrementally mine the frequent itemsets from the updated database, the cost would be minimized. In this research, we propose a novel mining method with a data structure called IMBT (incremental mining binary tree) which is used to record the itemsets in an efficient way. Furthermore, our approach needs not to predetermine the minimum support threshold and scans the database only once. The results of our research indicate that our method not only performs incremental data mining more efficiently, but also finds frequent itemsets faster than the Apriori and FP-growth algorithms.