Byung-Joo An, Eun-Ju Kim, and Yill-Byung Lee, Yonsei University of Korea, Korea
Clustering is a discovering process of meaningful information by grouping similar data into compact clusters. Most of traditional clustering methods are in favor of small datasets and have difficulties handling very large datasets. They are not adequate clustering methods for partitioning huge datasets in data mining perspective. We propose a new clustering technique, HRC(hierarchical representatives clustering), that can be applied to large datasets and find clusters with good quality. HRC is a two phase algorithm that take advantage of a hybrid approach that combine SOM and hierarchical clustering. Experimental results show that HRC can discover better clusters efficiently in comparison to traditional clustering methods.