分布式网络混合云数据分类捕获方法研究

Research on Classified Capture Method of Hybrid Cloud Data in Distributed Network

由于分布式网络中存在多条链路,易产生数据堆积现象,影响数据分类捕获效果,现有方法不能对不同类型数据进行有效分类,为此提出分布式网络混合云数据分类捕获方法。给出混合云数据分类捕获总体架构,构建海量动态数据的传输链路模型,实现对数据的自适应调度,并采集不同链路中的数据。将线性树与浓密树方法结合,查询数据采集结果中的有效数据,形成训练集,降低数据分类捕获的复杂度。采用粒度计算方法对训练集中的信息进行切分,直到得到可接受的粒度,并从上至下搜索粒度值,将相似度较高的粒度进行聚类,从而实现对不同类型混合云数据的分类捕获。实验结果表明,该方法可以最大程度地抑制多链路对数据分类的影响,快速获取较为精准的数据分类结果。

Due to the existence of multiple links in the distributed network, the phenomenon of data accumulation is easy to occur, which affects the effect of data classification and capture. The existing methods can not effectively classify different types of data. Therefore, a classification and acquisition method for hybrid cloud data in distributed network is proposed. The overall framework of hybrid cloud data classification capture is given,and the transmission link model of massive dynamic data is constructed to realize the adaptive scheduling of data and collect data in different links. The linear tree and dense tree method are combined to query the effective data in the data acquisition results to form a training set and reduce the complexity of data classification and acquisition. The granularity computing method is used to segment the information in the training set until the acceptable granularity is obtained. The granularity value is searched from top to bottom, and the granularity with high similarity is clustered to realize the classification and capture of different types of hybrid cloud data. The experimental results show that this method can restrain the influence of multi link on data classification to the greatest extent, and obtain more accurate data classification results quickly.