基于最大差异延展算法的Web资源描述算法  

Algorithm for Web Resource Description Based on Maximum Variance Unfolding

作　　者：侯臣平[1] 易东云[1] 吴翊[1] 

机构地区：[1]国防科技大学理学院数学与系统科学系,湖南长沙410073

出　　处：《系统仿真学报》2008年第20期5553-5557,共5页Journal of System Simulation

基　　金："九七三"重点基础研究发展规划项目(2005CB321800);国家自然科学基金(60673090)

摘　　要：针对虚拟计算环境下Web资源特性的描述问题,提出资源空间模型,采用流形学习的方法提取Web资源特征。首先根据资源空间模型,有效地将Web资源抽象为高维空间中的数据集;然后,采用流形学习中的最大差异延展算法。此方法不仅能有效地提取Web资源的特征,而且能够挖掘隐含在Web资源内部的本征信息;此时,描述Web资源特征的数据位于低维空间,有利于资源的进一步处理。基于最大差异延展算法的Web资源描述方法有效地解决了Web资源的描述问题。通过仿真实验证明了此方法的有效性。Resource space model was proposed and manifold learning approach was applied to solve the problem of Web resource feature description in virtual computing environment. Firstly,Web resources are translated into a data set in high dimensional space by applying resource space model effectively. Then features of the resources are extracted by employing a manifold learning algorithm-Maximum Variance Unfolding. It can not only efficiently extract the features of Web resources,but also can discover the latent information. Since representations of Web resources are now in a low dimensional space,it is of great advantage for the further processing of resources. Simulated experiments illustrate the validity of proposed method.

关 键 词：Web资源描述 资源空间模型 流形学习 维数约减 最大差异延展算法 

分 类 号：TP39[自动化与计算机技术—计算机应用技术]