针对关联规则挖掘中因频繁扫描数据库导致内存负担重、挖掘频繁项集效率不高的问题,提出一种基于有向无环图的挖掘算法DAGFIM.对候选项集构建二进制表,计算构建二进制表支持度作为有向无环图边权值;采用人工设置阈值判断计算出的边权值是否需要保留,若待处理的变权值大于阈值,则在图中予以保留,反之,对该边做"剔除"处理,整个构建过程只需扫描一次数据库,不会产生候选项集.实验结果表明,该算法具有较好的性能,有效提高了频繁项集的挖掘效率.%Aiming at the problem that the classical Apriori algorithm leads to a lot of candidate itemset and scans database for many times in mining frequent itemset, the algorithm called DAGFIM based on directed acycline graph was proposed.The database was projected to binary table, directed acycline graph was generated for each frequent itemset using binary table, and the counts of frequent itemset were used as weight for graph that was generated in previous step.The edge which was larger than the support threshold was reserved to recursively mine frequent itemset using directed acycline graph.Experimental results show that the proposed algorithm, with different support and data size, is more efficient than Apriori in mining frequent itemset.
展开▼
