【摘要】 串行直方图是基于频度排列对关系进行优化分区构造而成的,其连接结果大小估计是最优的,并可用于等值和范围查询结果大小估计。但是,串行直方图的构造算法复杂,影响了实际应用。本文从实用的角度出发,设计了一种构造优化串行直方图的算法BOS,该算法的时间复杂度大大降低,且估计精度接近最优直方图,从而使其具有较高的实用价值。更多还原

【Abstract】 A serial histogram is constructed based on relation frequencies and its join query size estimation is optimal, but its complex algorithm prevents it from application to practical estimations. Here we assume a new algorithm named BOS to solve this problem, and presents both theoretical and experimental results on the new algorithm.更多还原