Research of Density-Based Clustering Algorithm Based on Dissimilarity Selection

In this paper,a extending Optimizable k-Dissimilarity selection method,EOptiSim is proposed based on the original OptiSim algorithm.EOptiSim can fetch up the weakness of the original OptiSim processing the combinatorial library and distributed database.The application of the diverse representative subset selection method of OptiSim or EOptiSim before DBSCAN can deal with not only a single large-scale database but also the combinatorial library or distributed database while reducing the cost of I/O and memory.The experimental results show that the novel algorithm is effective and efficient.