基于项目集知识库的关联规则挖掘与更新的高效算法

通过对已有的诸关联规则挖掘与更新算法进行深入的分析和研究，指出了其共同存在的问题与不足，提出了一种基于项目集知识库的关联规则挖掘与更新方法。该方法既适应当数据库D中数据不变而用户指定的最小支持度和最小置信度这两个阈值变化的情况，也适合事务数据库D中数据发生变化的情况。当事务数据库D中数据不变时，仅需扫描数据库一次，便可建立项目集知识库KBD，然后可反复调整最小支持度和最小置信度进行关联规则挖掘与更新。而当事务数据库D中数据发生变化时，仅需扫描数据集d 和d-各一次；通过对项目集知识库KBD的更新来达到对频繁项目集和关联规则的更新。

Method for mining and efficiently updating association rulesbased on item sets knowledge database

Mining association rules in transaction database is an important aspect of data mining field. Although several algorithms have been proposed for mining and updating association rules recently. However these algorithms have to scan a large database many times and a large quantity of I/Os for scanning lowered the efficiency to solve this problem, a method for mining and updating efficiently association rules based on itemsets knowledge database is provided. The proposed method only needs scanning a database one time to create an itemsets knowledge database comparing with the existing algorithms and this new method is more efficient and can be used not only in the case that the data are not changed while the mining support and confidence given by users are changed, but also in the case that the data in a transaction database are changed and the method is especially applied to the updating association rules interactively and is of practical use.