DBSCAN-M: An Intelligent Clustering Algorithm Based on Mutual Reinforcement

Abstract

An intelligent clustering algorithm named DBSCAN-M is proposed for the purpose of data mining, which improves the recognition rate of noise under the circumstance of high noise density forming new clusters. The proposed algorithm is a synthesis of density clustering theory from DBSCAN Density-Based Spatial Clustering of Applications with Noise and mutual reinforcement from HITS Hypertext Induced Topic Search within search engine technology. The core points and clusters in the data set are mutually reinforced, thereby the capability of accurate identification of the noise is enhanced beneath high noise density. An algorithmic model was established, and simulations are taken by the WEKA software with real data sets from the University of California. Results showed that the proposed algorithm can obtain a more accurate recognition of the noises contrasting with the usual DBSCAN algorithm.