一种改进的关联规则的增量式更新算法

增量关联规则挖掘的主要思想是在原有规则的基础上,去除那些不满足条件的旧规则,发现满足条件的新规则,目的是尽量减少计算量.增量规则算法主要解决两类问题,即最小支持度的更新和数据库的更新.目前大多数算法对上述两个条件只更新其中一个,另一个保持不变,而实际应用中往往需要两者都更新.通过对数据挖掘中的IUA算法和FUP算法的分析和研究,提出IFU算法,用于解决数据库和最小支持度均发生改变时关联规则的增量式更新问题.相对于IUA算法和FUP算法以及基于他们改进的算法,该算法不仅扩展了更新条件,而且减少了对事务数据库和新增数据库的扫描次数.模拟实验表明IFU算法提高了更新效率.

AN IMPROVED INCREMENTAL UPDATING ALGORITHM FOR ASSOCIATION RULES

The main idea of the incremental association rules for mining are to base on original rules to eliminate those old rules that do not meet conditions and to find the new rules that meet conditions.Their purpose is to minimize the amount of calculation.The incremental rule algorithm mainly solves two problems:the minimum support degree update and the database update.At present most algorithms update only one of the above while keeping the other one intact.In practice,usually both of them should be updated.By analyzing and studying IUA algorithm and FUP algorithm in data mining,the paper presents IFU algorithm to solve the incrementally update problem when both the database and the minimum support degree are modified.Compared with IUA algorithm,FUP algorithm as well as their improved algorithms,IFU algorithm not only extends the updating conditions,but also reduces the scanning times for both the transactional database and the newly added database.Simulation experiment shows that IFU algorithm improves the update efficiency.