PAPPM：一种自适应Web预测模型

PAPPM:AN ADAPTIVE WEB PREDICTION MODEL

提出了一种自适应PPM(Prediction by Partial Match)预测模型:PAPPM。该模型能在预测过程中使用基于熵的自适应选阶策略选择最优阶,降低了预测开销。而且,它能根据当前用户访问的Web序列实时地更新预测模型,保证了预测模型的新鲜度。实验表明,PAPPM提高了预测精度和预测命中率,适用于在线Web预取。

A new adaptive PPM （Prediction by Partition Match） model PAPPM is proposed. In the prediction process, EOOE （Entropy based Optimal Order Estimation） is applied to choose the most optimal order to decrease the cost of prediction, and real-time update is carried out,which makes the prediction model always fresh. The experimental results show that PAPPM model can improve the predictive accuracy and achieve a good hit ratio. The prediction model can be used in on-line environment.