基于模糊聚类的高维划分策略研究

Splitting Strategy for High Dimensional Data Based on Fuzzy Clustering

数据集的划分策略是影响高维数据库索引性能的一个关键因素。金字塔技术是一种较好的高维索引方法,但它只对均匀分布的数据集具有良好的性能。为此,提出了一种改进的基于模糊聚类的金字塔技术,并将其用于高维划分策略,先对数据集进行模糊聚类处理,然后针对每个聚类进行金字塔划分,从而较好地实现了对非均匀分布数据的高维划分。

The splitting strategy for high dimensional data set is important for the performance of the indexing of high - dimensional database. The pyramid technique is a good indexing method for high dimensional data, but it is only efficient for uniform data sets. In order to solve this problcm, an improved pyramid technique based on fuzzy clu sets at first, be available stering and th is put forward. This new strategy uses a certain en it applies pyramid technique on each cluster. only uniform data sets but also ununiform data fuzzy clustering scheme on the original data By this means, the pyramid technique can sets.