Web Log Mining With Applications Based On Concept Lattices

With the rapid development of the Internet，accumulated data are exponentialincreasing. New information and new products are continuously put on the web everyday. Meanwhile, users’ categories, quantities and concerns are increasing continuously.To many users, it is getting harder and harder to find the information they need fromthe vast body of information data. On the other hand, service providers of the webs aretrying to improve their service quality through acquiring their users’ interests.Recommender system is an efficient way of solving the two problems mentionedabove.The Concept Lattices theory and the key techniques of Weblog mining arediscussed in this thesis. The main work in this thesis is summarized as follows.1．Against the fact that traditional incremental mining algorithms are incapable whendatabase and minimum supports change simultaneously, the Concept Lattices theory isapplied to pattern discovery of Weblog mining in this thesis. The incremental miningof the Association Rule based on the Concept Lattices is studied and an optimizationmethod, which generates all strong Association Rules from the maximum frequent itemsets, is proposed. The method decreases the discovery time of the frequent item sets inpattern discovery and avoids successfully the generation of unnecessary rules, whichimproves the efficiency of pattern discovery.2．The Weblog mining method based on the Concept Lattices theory is applied to theelectronic commerce, which provides personalized recommendation for different usersto improve the development of the electronic commerce websites. In this paper, thesystem structure of electronic commerce personalized recommender system is set upand the function of each module in the system is realized, by doing which can providea effective way for solving the personalized recommendations problem on e-commercesite.