移动互联网中特征数据准确提取仿真研究

摘要

In this paper,we proposed an accurate extraction method of feature data in the mobile internet based on the granular computing and separating capacity.Firstly,the stratified sampling technique of statistics was used to split the initial data set into multiple sample subsets and the separating capacity of data attribute on each granule was worked out.Then,the classification approximation criterion of attribute set was introduced as the immune optimization affinity of data attribute reduction integrated with the Niche immune optimization theory.Moreover,the Niche immune shared mechanism was generated and the attribute reduction was made to the mobile internet data.Finally,the accurate extraction model of feature data in the mobile internet was built.The simulation results show that the method has better extraction precision.It can lay a sound foundation for better improving the service quality of mobile internet.%对移动互联网中特征数据准确提取,可减少移动互联网的运行负荷.进行特征数据提取时,应分析不同数据属性的区分能力,对移动互联网数据进行属性约简,减少特征数据提取的工作量,但是传统方法是通过获取移动互联网数据集合的模糊粗糙近似,构造移动互联网特征数据属性集提取的目标函数,但是不能有效对移动互联网数据进行属性约简,导致特征数据提取耗时长,效率低下的问题.提出一种基于粒计算与区分能力的移动互联网中特征数据准确提取方法.首先利用统计学中的分层抽样技术将移动互联网初始数据集拆分为多个样本子集(粒),并计算出每个粒上数据属性的区分能力,融合于小生境免疫优化理论,引入属性集合的分类近似标准作为数据属性约简免疫优化的亲和度,然后生成小生境免疫共享机制,对移动互联网数据属性约简,最终建立移动互联网中特征数据准确提取模型.仿真结果表明,所提方法移动互联网中特征数据提取精确度高,为更好地提升移动互联网服务质量奠定了坚实的基础.