流数据频繁项算法研究

流数据频繁项挖掘是一项重要的研究课题,是其他流数据挖掘任务的基础。Lossy counting 算法是第一个近似的流数据频繁项挖掘的算法,并且具有空间和时间的高效性。详细分析该算法,尤其是它不能回答关于时间的查询的不足后,对其进行改进,提出了一个在多时间粒度上挖掘流数据频繁项的设想,加入时间维度。改进后的算法在时间倾斜窗口保存与合并频繁项,可以应用于各种对时间敏感的流数据查询和挖掘应用中。

Research on arithmetic frequent datasets over data stream

Mining frequent items over data stream is an important problem of research,which is the foundation of several other researches.Lossy counting algorithm is the fast algorithm proposed to solve the problem of mining frequent item sets over data stream,and it is efficient about space and time.We analyze this algorithm, especially that can not answer the query about the time,and modify it,and present a proposal for mining time-sensitive item sets by incorporating time dimension.Then we can contain and add frequency item sets to the frame of logrithmal tilted-time window to do the time-sensitive query and mining.