IHUIMiner: 一個能在漸增式資料庫中高效探勘高效益物品集的演算法

Abstract

隨著近年來電子商務的興起 ，如何在消費紀錄中找出有用的消費資訊是電子商務公司的一大課題 ，電子商務公司認為商品的獲利情況比銷售量還要重要 ，因此 ，high utility itemset mining 比 frequent itemset mining 更適用於這類情況 。在現實中 ，資料庫的更新頻率相當頻繁 ，而且一直對整個資料庫重新 mining 會很沒效率 。有鑒於此 ，我們在本論文中提出了一種能在漸增式資料庫中高效探勘 high utility itemsets 的演算法 ，我們做了很多實驗以測試我們演算法的效能 。實驗結果指出 ，相較於以前的漸增式資料庫演算法 ，我們的演算法用了較高的記憶體但擁有更高效率探勘 high utility itemsets 的能力。

Electronic commerce is getting popular in recent years. It is important for electronic commerce companies to find useful patterns from purchase records. Companies usually pay more attention in the profit (utility) earned rather than number of items sold. Thus, high utility itemset mining is more suitable for such cases than frequent itemset mining. In practice, the database is updated continuously, and re-mining the whole database is inefficient. In view of this, we propose in this thesis an algorithm to efficiently mine high utility itemsets in an incremental manner. In order to measure the performance of the proposed algorithm, several experiments have been conducted. Experimental results show that our algorithm is able to mine high utility itemsets more efficiently than prior algorithms on an incremental database at a cost of higher memory usage.