基于动态矩阵预测的网页实时推荐研究  

Study Of Real-Time Web Pages Recommendation Based On Dynamic Matrix Prediction

作　　者：亓俊红[1] 

机构地区：[1]山东省莱芜职业技术学院,莱芜271100

出　　处：《微计算机应用》2006年第6期760-763,共4页Microcomputer Applications

摘　　要：通过对Web日志的预处理,构建动态矩阵,该矩阵能够反映用户访问路径的先后顺序,利用相似度计算进行网页推荐。提出的动态矩阵预测算法具有较快的响应速度,可以满足实时页面推荐的需要,同时该算法无需事先训练,还可以对动态矩阵进行增量更新,提高了预测性能。Dynamic matrix is firstly constructed by means of preproeessing to web log and it can represent the order Of user＇s browsing path on web. Whereafter, desirable web pages are recommended by calculation of similarity between sessions. The algorithm possesses preferable speed of response to satisfy the need of real time web page recommendation and also can be executed without training in advance. Meanwhile , the algorithm can implement incrementally update of dynamic matrix , the performance of recommendation can be improved growingly.

关 键 词：WEB日志 数据预处理 动态矩阵 网页推荐 

分 类 号：TP393.092[自动化与计算机技术—计算机应用技术]