基于复合粒度计算的频繁模式挖掘研究

针对经典频繁模式挖掘算法存在的不足，提出了一种基于复合粒度计算的频繁模式挖掘算法。该算法借助复合粒度计算方法双向搜索频繁模式，即首先通过二进制的按位取反运算获得复合粒度内涵的像，然后构建复合粒度计算发现频繁模式；虽然该算法需要产生候选项，但它只需扫描一次数据库，减少了I/O开销；算法通过线性数组存储复合信息粒度减少内存使用。理论分析和实验比较表明，其效率优于经典的频繁模式挖掘算法，且内存利用率比较高。

The research of frequent patterns mining based on composite granular computing

Aiming to the shortcomings existing in the typical algorithms of frequent patterns mining, this paper proposes an algorithm of frequent patterns mining based on composite granular computing. The algorithm doubly searches frequent patterns by composite granular computing, namely, it firstly gets the image of the intension of composite granules via the complementer of binary number on each bit, and then constructs composite granular computing to discover frequent patterns. The algorithm needs to generate candidate, but it only needs to scan the database once to reduce the I/O overhead. The algorithm uses the linear array to save composite information granules to reduce the usage of memory. The theoretical analysis and experimental comparison report that the efficiency of the algorithm is better than present typical algorithms of frequent patterns mining, and its utilization of memory is higher.