基于网络聚类与自适应概率的数据库缓冲区替换  

Database buffer replacement based on network clustering and adaptive probability

作　　者：贺红艳[1] 李光明 张慧萍[1] 

机构地区：[1]湖北工业大学工程技术学院,武汉430068

出　　处：《沈阳工业大学学报》2018年第1期65-69,共5页Journal of Shenyang University of Technology

基　　金：湖北省教育厅科研计划资助项目(2014277);湖北工业大学工程技术学院项目(X201328)

摘　　要：针对传统替换方法存在替换不准确、效率低的问题,提出模糊kohonen网络聚类算法与自适应概率相结合的大型数据库缓冲区替换方法.采用Broder理论并基于Jaccard相似性度量对缓冲区重复数据进行消除,建立缓冲区数据检测模型,并采用模糊kohonen网络聚类算法对缓冲区数据进行聚类处理.采用复小波法提取缓冲区的特征,引入自适应概率对大型数据库缓冲区进行替换.结果表明,改进的缓冲区替换方法可有效的实现对大型数据库缓冲区的替换,提高替换效率,增加大型数据库存储的整体性能.Aiming at the inaccurate replacement and lowefficiency existing in the traditional replacement method,a large database buffer replacement method in combination with both fuzzy kohonen network clustering algorithm and adaptive probability was proposed. Through adopting Broder theory and based on Jaccard similarity measurement,the repeated data in the buffer were eliminated,and the buffer data detection model was established. In addition, the buffer data were clustered with the fuzzykohonen clustering algorithm. The buffer features were extracted with the complex wavelet method,and the adaptive probability was introduced to replace the large database buffer. The results showthat the improved buffer replacement method can effectively realize the replacement of large database buffer,increase the replacement efficiency,and enhance the overall performance of large database storage.

关 键 词：大型数据库 缓冲区 替换方法 改进 特征 网络聚类 自适应概率 数据存储 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]