The Research Of Digital Library Personalized Service System Based On Web Usage Mining

With the rapid development of the digital library in all countries, the related research workhas also rose to an upsurge, in which the personalized service of the digital library has becomean important research topic. It uses the Web usage mining technology, through analyzing andmining the user's access log data and subscription information to find in the user's interest, thentargeted providing the personalized information resources service for different levels anddifferent preferences. Personalized service takes a great significance for the digital library'soverall service decision-making.This paper firstly introduced the digital library personalized information technology, thenelaborated the basic principles of the Web usage mining technology, and conducted in-depthstudy of the FP-growth algorithm which basing on association rules, for when it's used in therecommendation system will generate repeat recommending questions, We improved thealgorithm with a FP-growth_s algorithm. This algorithm can produce the largest frequentitemsets avoiding the redundant frequent itemsets'generation and reducing the number ofgenerating association rules, effectively soluted the possible that producing duplicaterecommendation, when using in the recommended system it can improve the system efficiency.After further studying We found that the association rules generated by minimum support andminimum credibility can't provide an accurate recommendation for user in some cases. For thisproblem We proposed an IN_FP recommended algorithm that basing on web pageinterestingness, this algorithm firstly built the web page interestingness model, then weightedthe interestingness for credibility, after using FP-growth_s algorithm generating the maximumfrequent itemsets, the weighed credibility will generate the associate rules, if using thisrecommended algorithm in recommended system will mine more accurate results to providemore value and meaning recommended service for users. In the end, this paper built the entiredigital library personalized service system and used the IN_FP recommended algorithm in thissystem achieving the digital library personalized recommended service.