Research On E-Commcrcc Personalized Recommendation Based On Web Log Mining

Today, with the development of Internet, e-commerce is wildly prevalent by its convenient,fast, no geographical restrictions and other advantages in the global. Recently, e-commerce website has accumulated a large number of web data, how to mining potentially and valuableinformation from these data is becoming a hot issue concerned by enterprises. Web mining isgoing to discover useful patterns from web data by data mining techniques. In this paper, thoughanalysis and study e-commerce sites web server log, we found users’ access patterns by datamining techniques in order to support and help enterprises to understand customers’ needs andmeet their needs.Paper describes the basic knowledge of web mining and e-commerce, and analysis theadvantages and disadvantages of the FCM clustering algorithm and the hierarchical clusteringalgorithm, and propose an improved hierarchical clustering algorithm to make up theShortcoming of hierarchical clustering algorithm saving storage space and improving theexecution speed; in addition, it proposes a new value function, considering the effect of thedistance between classes and the distance within classes on the clustering results; finally, itproposes a new algorithm named NHMF algorithm by combine the improved hierarchicalclustering algorithm and the FCM algorithm, and makes an example application obtaining a userclustering model and providing support for personalized recommendation.