有关数据流挖掘技术的研究是当前国际数据库研究领域的一个热点,数据流的特点在于数据规模宏大,并快速、持续地到达,对应的挖掘算法只能在内存中单遍扫描样本子集就可以获取相应的知识结构,还需要在一定时间内对学习的结果进行更新以适应数据分布的变化.本文对现有数据流上的挖掘算法进行综述,最后给出了数据流挖掘今后的一些研究方向.
The study on mining data streams is one of the hot topics among the database circle all over the world recently. Data streams are continuous, unbounded, rapid,time-varying streams of data elements. Mining algorithms on data streams are concerned with extracting knowledge structures by one-pass scan in memory, updating the results to suit the change of the distribution. This article introduces some data stream mining algorithms and summarizes the main ideas. Finally, this paper presents some research trends in this area.