While exploiting popular episodes were only interested in the appearance of the items in the transaction (i.e. they have or don't have in the trades) then the file extraction utility (HUI-High utility itemset) are interested in profits when selling the items together. There have been many methods are developed to improve the efficiency of extraction HUI but this method assumes the profit of the item is fixed, meaning that they do not change over time. But in fact, the profit of the item (for example, the selling price of the item) back often change (due to change in price, promotion, etc.). In this paper, we develop a method to efficiently exploit HUI on the database change profit of the item. First, we are addressing the problem of exploitation HUI on the DATABASE change profit. The calculated weight of useful trades, trading useful weight of itemset also as useful property of the itemset also will be presented back to match the performances of the new DATABASE. Due to a change in the formula, we need to modify the algorithm EFIM (mining algorithm is the most effective current HUI) to exploit the new DATABASE on the HUI. Finally, we propose the structure P-set, based on which to develop the iMEFIM algorithm to reduce the number of transactions that need consideration in the extraction process and thus, reduce the time to exploit the HUI. Experimental results showed improved iMEFIM algorithm significantly reduces the number of times the transaction, the time and memory use, especially on the basis of sparse data.
đang được dịch, vui lòng đợi..