IMPROVING THE INTERESTINGNESS OF WEB USAGE MINING

IntroductionWeb usage mining is the application of datamining to the web server's log in order to discoverthe behavior patterns of the web site visitors.Thebehavior patterns found should be highly interest-ing.That is,they should be valid,novel,poten-tially useful,and ultimately understandable1] .Finding frequently visited page groups is animportant topic in web usage mining.Intuitively,the frequently visited page group ( FVPG) is a setof web pages thatare often requested together by anumbe…

IMPROVING THE INTERESTINGNESS OF WEB USAGE MINING

Improvement on mining the frequently visited groups of web pages was studied. First, in the data preprocessing phrase, we introduce an extra frame filtering step that reduces the negative influence of frame pages on the result page groups. Through recognizing the frame pages in the site documents and constructing the frame subframe relation set, the subframe pages that influence the final mining result can be efficiently filtered. Second, we enhance the mining algorithm with the consideration of both the site topology and the content of the web pages. By the introduction of the normalized content link ratio of the web page and the group interlink degree of the page group, the enhanced algorithm concentrates more on the content pages that are less interlinked together. The experiments show that the new approach can effectively reveal more interesting page groups, which would not be found without these enhancements.