Improved Next Web Page Recommendation using Multi-Attribute Weight Prediction

Abstract: Web page prediction is the web usage mining by performing pre-processing of the data from a web site. The needfor predicting the user�s needs in order to improve the usability and user maintenance of a web site is more thanmarked now a day�s lacking proper guidance, a visitor often wanders aimlessly without visiting significant pages,loses attention, and leaves the site earlier than expected. When they access the network, a largeamount of data is generated and is stored in Web log fileswhich can be used efficiently as many times user repeatedlysearched the same type of Web pages recorded in the log files. These series can be considered as a web access pattern, helpful to find the user behaviour. Through thispersonalized information, it�s quite easy to predict the nextset of pages user might visit based on the previouslysearched patterns, thereby reducing the browsing time of auser.

Keywords: Web usages mining, recommendation, web log analysis, session based predication, K-NN algorithm.