基于IRSM模型的频繁模式挖掘  

MINING FREQUENT PATTERNS BASED ON IRSM MODEL

作　　者：刘爱娣[1] 胡运发[1] 朱洁[1] 

机构地区：[1]复旦大学计算机科学与工程系,上海200433

出　　处：《计算机应用与软件》2008年第7期4-5,27,共3页Computer Applications and Software

基　　金：国家自然科学基金(60473070)

摘　　要：后继矩阵是一种新型的全文存储索引模型。根据warshall理论和区间求交的性质提出一种基于间接后继矩阵模型的频繁模式挖掘算法。和FP-growth方法一样,算法直接构造频繁项集,不进行Apriori算法所采用的代价很高的候选集产生与测试操作。然而它比FP-树模型具有更多的优点:只需扫描一遍事务库;挖掘任务只局部关联于后继矩阵的一行。实验表明,其具有与FP-growth算法相当甚至更高的效率。更重要的是,IRSM模型同时是一种事务库的良好索引形式,具有高效支持事务查询的能力。IRSM is a novel mathematical model,which has been successfully applied to full-text index and storage in text database. According to the theory of warshall and interval intersection ,an algorithm based on IRSM is presented for frequent patterns mining. The algorithm builds frequent patterns directly as FP-growth algorithm does. However,it has several advantages over FP-tree model. Firstly,the transaction database is scanned only once. Secondly,the mining process is only associated with one row of the Matrix. The performance study shows that the efficiency of the algorithm is equal to or even higher than that of FP-growth. Above all,IRSM is a good model for transaction database index, and it can support query on transactions with high efficiency.

关 键 词：Warshall 区间求交 频繁模式 APRIORI算法 FP-GROWTH算法 IRSM 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论] TQ051.84[自动化与计算机技术—计算机科学与技术]