挖掘關聯規則之階段搜尋演算法－－GSA

GRA: A Gradational Scanning Algorithm for Mining Association Rules

隨著交易、文件、日常處理資料的電子化、各種型式的資料被大量的累積下來，也隨著資訊科技的進步，資料探勘的技術變得日益重要。而關聯規則探勘在資料探勘的領域中也扮演相當重要的地位。許多的資料探勘演算法不斷被提出來，並針對舊的演算法加以改進，以增進其效能或更節省其所使用的記憶體，本論文主要是針對關聯規則的領域提出效能及記憶體方面改進的演算法。本論文提出一個新的關聯規則演算法GSA (Gradational Scanning Algorithm)，GSA演算法主要是利用Ck=Ck-1*Ck-1的概念產候選項目集，且加入階段搜尋的概念，並配合過濾機制，使得候選項目集之數量逼近高頻項目集之數量，有效增進探勘的效能，而且GSA演算法最少只需掃描4次資料庫，最多掃描6次資料庫便可完成所有探勘。

Due to transactions, documents and data were transformed into electronic types. The huge mass of data has been accumulated. Today, the science and technology make a great progress. Therefore, data mining technology becomes more important than before in recent years. ft is generally applied to forecast in commerce and supports the decisions. In data mining territory, mining association rules plays a quite important position. Many of data mining algorithms were proposed continuously to improve performance of the old algorithms. They try to improve efficiency of the algorithms or to save the memory. In this paper, our study focuses on association rules and proposes a new algorithm-GSA (Gradational Scanning Algorithm) which improve performance and memory utility rate of mining association rules. GSA basically uses a method which is similar to scan reduction method of SWF algorithm. Besides, it also uses the concept of gradational scanning and the filtration mechanisms to reduce the number of candidates. The GSA needs to scan the Database four times at least and at most six times to finish the mining process.