【摘要】 该文提出一种利用最优K相异性算法(OptiSim)的密度聚类算法,通过代表性子集选择技术与DBSCAN算法的结合,可显著降低I/O耗费和内存需求,使之能够有效地处理大规模空间数据库,并提出一种扩展的OptiSim代表性子集选择方法(EOptiSim),它在处理组合数据库方面能弥补基本的OptiSim方法的不足。实验结果表明文章的算法是可行、有效的。更多还原

【Abstract】 In this paper,a density-based clustering algorithm for using optimiazable K-dissimilarity selection is proposed to reduce the cost of I/O and memory usage via integrate the representative subset selection with DBSCAN algorithm.Furthermore,An extending representative subset selection method of OptiSim algorithm fetch up the lack of OptiSim when processing the combinatorial Library.The experimental results show that the new algorithm is effective and efficient in clustering large-scale databases.更多还原