基于EDDG关联图的Web预测模型

Web predictive model based on exponential descendent dependency graph

DG关联图预测模型的预测准确度较低,PPM树预测模型的空间复杂度巨大。为了解决以上两个问题,在对Web对象的浏览特征以及用户浏览深度特征进行研究的基础上,对传统的DG关联图预测模型进行优化,采用指数级下降算法修正传统DG关联图预测模型在计算转移概率方面的缺陷,提出基于EDDG关联图的Web预测模型。实验结果表明,该预测模型可获得与PPM树预测模型相近的预测准确度,同时算法的空间复杂度也得到了较大的改进。

The predictive accuracy of DG（dependency graph） predictive model is lower,although PPM model has higher predictive accuracy,it occupies large storage space.In order to resolve those problem,the standard DG model is optimized according to Zipf＇s law and web surfing characteristics,and the EDDG（exponential descendent dependency graph） model is proposed in order to fix the defects which lie in the traditional DG model by exponential descendent algorithms.Experimental results show that the EDDG model can save the storage space while get the similar predictive accuracy with PPM model.