Research And Application Of E-commerce User Type Identification

Along with the explosive growth of e-commerce user scale,e-commerce has become a new driving force for economic growth.Based on the heterogeneity of different consumer groups in e-commerce platforms,it is of great significance to conduct in-depth research and delineate the characteristics of e-commerce user types.In this research background,this paper stands in the perspective of complex networks,and substitutes e-commerce users into complex networks,focusing on the identification of user types of social e-commerce platforms,specifically from the following three aspects of research and practical testing.(1)Introduction of density peak clustering algorithm for user type identification research.In view of the shortcomings of the original density peak clustering algorithm,the algorithm is improved by combining the characteristics of social e-commerce network users,including improving the selection of density indicators in the initial stage of the algorithm,and reflecting the relationship between user nodes by distance measurement;introducing the nearest neighbor KNN method in the clustering stage to find the peak density of the sample;in the community division stage based on the allocation of user label propagation,the remaining node assignment in the community division stage based on user label propagation,and finally user type division.And after conducting the algorithm performance test,it shows that the F-Measure value of the algorithm in this chapter is improved by 14.06% compared with the original DPC algorithm,which has a considerable advantage in accuracy.(2)The classical Kuramoto model in the theory of complex network synchronization is introduced,and the algorithm is optimized to obtain the Kuramoto nearest neighbor model based on the characteristics of social e-commerce user networks with interaction,and the similarity of user behaviors is measured,and similar nodes are selected and coupled to replace the phases.The artificial dataset and Taobao e-commerce dataset are used as the simulation dataset of the algorithm to test the algorithm performance and user type identification.The experimental results show that the optimized model runs faster and the clustering effect,i.e.,cluster class division,is more obvious and more applicable to the real social e-commerce user network.(3)To further optimize the Kuramoto model,we set the coupling role of key nodes in the original Kuramoto model based on the characteristic that there are usually a few leader nodes influencing the majority of consumers in social commerce networks,so that the changed model is more applicable to the network form,and discuss the role of leader nodes in terms of driving range and effect.Finally,the applicable effect of the optimized model is discussed with the artificial dataset and the advertising and marketing-user network simulation dataset,and the results prove that adding the leader driving effect has a significant effect on different types of consumers.Moreover,the experimental results show that the optimized Kuramoto simultaneous clustering algorithm proposed in this chapter runs faster and has better cluster class segmentation than the original algorithm.The research results and conclusions of this paper can complement studies related to user type identification and provide reliable support for personalized recommendation in e-commerce,aiming to provide practical tools for enterprises’ operation management and marketing strategies.