Adaptive clustering method and adaptive clustering system on basis of clustering coefficients
暂无分享,去创建一个
The invention provides an adaptive clustering method on the basis of clustering coefficients. The adaptive clustering method includes determining classes of data points and the numbers of the classes according to classes of extracted data point pairs and degrees of associations between extracted data points in each procedure, and creating associations between the data points and associations between the classes; preliminarily dividing each class, computing an intra-class similarity degree of each two sub-classes and an inter-class similarity degree of the two sub-classes, judging whether each two preliminarily divided sub-classes meet dividing conditions or not, and receiving the preliminarily divided sub-classes if the preliminarily divided sub-classes meet the dividing conditions; canceling preliminary dividing of the classes if the preliminarily divided sub-classes do not meet the dividing conditions; computing an intra-class similarity degree of each two classes with the corresponding association and an inter-class similarity degree of the two classes, judging whether each two classes with the corresponding association meet combining conditions or not, and combining each two classes with each other to generate a new class if the two classes meet the combining conditions; abandoning combining of each two classes if the two classes do not meet the combining conditions. The adaptive clustering method has the advantages that pairwise relations between the data points in the classes are taken into consideration for the first time, the intra-class similarity degrees and the inter-class similarity degrees are computed on the basis of the clustering coefficients, and accordingly data can be accurately clustered.