中文摘要：

为了克服单一聚类算法的不足，将粒度计算与聚类算法相结合，提出基于聚合网络的变粒度二次聚类方法（twiceclusteringmethod basedonthevariable granularity and clustering network，VGTC）。首次聚类的目的是寻找合适的聚合粒层，以发现数据的局部结构，二次聚类在此基础之上完成对论域的聚类操作。创新之处在于依据聚类算法参数的改变来调整聚类的粒度，通过粒度计算将两种聚类算法的优点融合在一起。以基于七均值与层次聚类算法的变粒度自适应二次聚类方法 （ Twice clustering adaptive method of variable granulation based on k-means andhierarchical clustering algorithms, KHVGTC）为例，从理论和实验验证了VGTC算法的准确率和效率。

英文摘要：

In order to make up the deficiency of single clustering algorithm, a new twice clustering method based on the variable granularity and clustering network （VGTC） was presented, which combined granularity computing with cluste- ring algorithms together. The aim of the first clustering was to find local data structure through searching an appropriate clustering layer. On this basis, the secondary clustering could complete clustering operation for domain. The creativity of VGTC was that the granularity of clustering could be adjusted by changing clustering algorithm parameters, and the advantages of two clustering algorithms could be combined together through granularity computing. The twice clustering adaptive method of variable granulation based on k-means and hierarchical clustering algorithms （ KHVGTC was an ex- ample of VGTC） verified the accuracy and efficiency of VGTC algorithm by theory analysis and experimental results.