增量式CURE聚类算法研究

Research on Incremental Clustering Algorithm Based on CURE

聚类是一种非常有用的数据挖掘方法 ,可用于发现隐藏在数据背后的分组和数据分布信息 .目前已经提出了许多聚类算法及其变种 ,但在增量式聚类算法研究方面所作的工作较少 .当数据集因更新而发生变化时 ,数据挖掘的结果也应该进行相应的更新 .由于数据量大 ,在更新后的数据集上重新执行聚类算法以更新挖掘结果显然比较低效 ,因此亟待研究增量式聚类算法 .通过对 CURE聚类算法的改进 ,提出了一种高效的增量式 CU RE聚类算法 .它能够很好的解决传统聚类算法在伸缩性、数据定期更新时所面临的问题 .

Clustering , in data mining , is useful for discovering groups and identifying interesting distributions in the underlying data.There have been many algorithms proposed for clustering.However , very little work has been done on incremental clustering.Presented an efficient incremental clustering algorithm based on CURE. A performance evaluation of the algorithm on a spatial database is presented, demonstrating the efficiency of the proposed algorithm. The result of experement shows that this algorithm is an efficent incremental clustering algorithm.