基于非压缩后缀树的在线PPM预测模型  

On-line PPM Prediction Model Based on Non-compact Suffix Tree

作　　者：班志杰[1] 古志民[1] 金瑜[1] 

机构地区：[1]北京理工大学计算机科学技术学院

出　　处：《计算机工程》2008年第10期70-72,共3页Computer Engineering

基　　金：北京理工大学基础研究基金资助项目(0301F18)

摘　　要：PPM模型适合预测用户的下一个请求,但已有的PPM模型不具备在线性,更新通过重构来实现,不能满足实时更新的要求。该文提出基于非压缩后缀树的在线PPM预测模型,采用非压缩后缀树实现增量式在线更新,提高了模型的更新速度。该模型的优点是具备在线性。Prediction by Partial Matching（PPM） models are appropriate for predicting the user＇s next request, but these models are not on-line and their updates are implemented by rebuilding models which can not meet the real-time update. This paper presents an on-line PPM prediction model based on non-compact suffix tree. The model makes use of non-compact suffix tree to implement the incremental on-line update, and its update speed is improved. This model has the important property of being on-line.

关 键 词：WEB预取 PPM模型 非压缩后缀树 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]