期刊名称：Journal of Theoretical and Applied Information Technology
印刷版ISSN：1992-8645
电子版ISSN：1817-3195
出版年度：2013
卷号：49
期号：2
出版社：Journal of Theoretical and Applied
摘要：This paper proposed a improved semi-supervised clustering algorithm, aiming at the limitations of the traditional k-means algorithm in the irregular clusters division. This paper mainly discusses two problems: the closure replacement and the traction between data points. First of all, made use of the closure center to replace the original sample point, and to calculated the sample clustering center combining the semi-supervised clustering ideology with a small amount of the tagged data; next, in the process of clustering is introduced into the definition about the traction distance of the clustering center and the traction between the data points, given a complete description of the relationship between data points; finally, fully consider the effect about the tagged data points, especially isolated point, on untagged data point, implement clustering data in the sample space. Experimental results indicate that the cluster is much more efficient than the other existing algorithms when dealing with the irregular cluster.
