流数据上的频繁项挖掘算法

摘要

提出了一种流数据上的频繁项挖掘算法(SW-COUNT).该算法通过数据采样技术挖掘滑动窗口下的数据流频繁项.给定的误差ε,SW-COUNT可以在O(ε-1)空间复杂度下,检测误差在εn内的数据流频繁项,对每个数据项的平均处理时间为O(1).大量的实验证明,该算法比其他类似算法具有较好的精度质量以及时间和空间效率.%A frequent items mining algorithm of stream data (SW-COUNT) was proposed, which used data sampling technique to mine frequent items of data flow under sliding windows. Given an error thresholdε, SW-COUNT can detect ε-approximate frequent items of a data stream using O(ε-1) memory space and the processing time for each data item was 0(1). A lot of experiments show that SW-COUNT outperforms other methods in terms of the accuracy, memory requirement, and time and space efficiency.