一种用于Web服务预测的改进PPM模型

An Improved Web Service Prediction PPM Model

针对现有PPM模型预测准确率不够高,且无法处理好预测准确率与预测代价之间平衡的问题,提出了一种新的改进PPM模型(PD-PPM)用于Web服务预测,以便更实时地向用户提供个性化、智能化的服务。为了建立该模型,定义了Web服务编码结构、错误补偿矩阵(ECM)、子类服务PPM模型(SW-PPM)和大类服务PPM模型(PW-PPM),并利用控制领域的PD控制算法,在SW-PPM模型与PW-PPM模型之间进行切换控制和参数调整;此外,还在ECM的基础上建立了错误补偿机制。经实验证明,PD-PPM模型能有效平衡预测准确率和预测代价之间的矛盾,并可以对预测准确率进行精准地跟踪控制,从而达到了预期的效果。

A new improved PPM model（PD-PPM）is proposed to predict users Web Service visiting,aiming at the problem that the prediction accuracy of the existing PPM model is not high enough,and that it is unable to deal with the balance between prediction accuracy and cost.This model can provide users with individual and intelligent services in more real-time way.To build the model,the encoding structure of Web Service,error compensation matrix（ECM）,sub Web Service PPM model（SW-PPM）and prime Web Service PPM model（PW-PPM）are defined,and PD control algorithm is mades use of to control the switch between SW-PPM and PW-PPM with parameters adjusting.Besides,the error compensation mechanism is constructed based on ECM.Experiments show that PD-PPM can balance the prediction accuracy and cost effectively,track and control the prediction accuracy precisely.PD-PPM has achieved expected effect.