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

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.