摘要
将二进制引入关联规则求解中,充分利用二进制操作方便、运算速度快、节省空间的优势。在求解事务项集真子集和支持度时,对事务数据库中相同事务只求解一次,并给出了真子集的具体求解算法。本算法一次扫描数据库可以挖掘出所有频繁集,而且可以根据需求对最小支持度和最小置信度进行修改,修改后不需要再次扫描数据库即可求出频繁项集,大大提高了挖掘效率。
This paper proposed a novel idea of binary-based association rules mining algorithm. There some transaction items were the same in a database. It was not necessary to work out the pure subsets for everyone but only to aim at every kind of them. With this algorithm all frequent itemsets could be minded by scanning the database only one time. The algorithm was very efficient and effective compaxed to the traditional methods. When the rain-support and rain-confidence were modified, it could get the frequent itemsets easily without scanning the database again.
出处
《计算机应用研究》
CSCD
北大核心
2007年第8期79-80,115,共3页
Application Research of Computers
基金
国家自然科学基金资助项目(60564001)
关键词
二进制
关联规则
频繁集
真子集
binary
association rules, frequent itemset
pure subsets
