摘要
知识发现是指对原始数据进行分析，提取出隐含的、有用的规则，是当前快速发展的研究领域，是知识获取的重要方法．关联规则是知识发现的重要研究内容之一．本文提出了一种新的多层次关联规则挖掘算法ML-AR．算法ML-AR在挖掘过程中，只对最低概括层次上的候选频繁模式进行模式的匹配计算，求解出简化的频繁模式集合，最后再求解各个概括层次上的频繁模式集合、算法ML-AR有效地利用了概括的层次关系，减少了模式的匹配计算和对存储空间的占用，提高了挖掘的速度．试验结果表明，算法ML-AR比算法Cumulate的执行速度约快15％．
Knowledge discovery from database is the non-trivial process of identifying potential useful and valid regularities from raw data. It is a important and fast developing field, and one of the important methods of knowledge acquisition. Mining association rules is one of the important aspects of knowledge discovery. A new multiple-level association rules mining algorithm ML-AR is proposed in the paper.In mining procedures, algorithm ML-AR only matches the candidates of the frequent patterns of the least generalized levels,then calculates the frequent patterns of all generalized levels from the frequent patterns of the least generalized levels.Algorithm ML-AR makes use of relations of items between the generalized levels,decreases the cost of pattern matching and saves the storage, speeds up the mining procedures. By the experiments, algorithm ML-AR is proved to be efficient and effective, and outperforms the traditional multiple-level association rules mining algorithm Cumulate, the execution time of algorithm ML-AR is less than algorithm Cumulate about 15%. Some topics associated with multiple-level association rules are discussed in the paper.
出处
《计算机学报》
EI
CSCD
北大核心
1998年第11期1037-1041,共5页
Chinese Journal of Computers
基金
国家自然科学基金
关键词
知识发现
关联规则
数据挖掘
人工智能
数据库
Knowledge discovery,association rules,generalization,multiple-level
