关联规则算法的演化和进展

Evolution on Association Rules Algorithm

关联规则挖掘是数据挖掘的一个重要应用。从1993年Agrawal等首先提出了用基于频集理论的递推方法来解决挖掘顾客交易数据库中项集间的关联规则问题以来,关联规则的算法已经进行了很多改进。这些改进集中在三个方面:减少候选集和压缩事务总数;减少数据库扫描次数;引入并行技术。近年来,研究人员引入更能代表人类决策思维模式的模糊集理论,并将研究目标对准更复杂的时序和空间数据,挖掘的范围也从单维事务扩展为多维事务,但相应的要面临更加复杂的计算。

Association rules is an important application of Data Mining. The association rules algorithm has improved a lot after 1993 when Agrawal el al proposed to solve the mining association rules of itemsets in the trade database of customers by using deduction method based on frequent itemsets theory. These improvements focus on three aspects reducing candidate itemsets and transaction amount reducing scan times of database inducing parallel computation. Recently researchers induced the fuzzy sets theory which can presents human decision model better and they focused their study on the more complicated sequence and spatial data and the mining confine was extented from single dimension to multiple dimension. But correspond to that we now face more complicated computation.