Analysis And Research On User Behavior Based On Social E-commerce Platform

In several years,with the continuous rise and development of Internet economy and social platform,social e-commerce is more and more active in front of the public and occupies an increasingly important position in China's economic society.The analysis and research on the user behavior of social e-commerce,an emerging group,is conducive to promoting the continuous development of economic model of social e-commerce,which is of great significance to the development of economic society.With the help of big data mining technology,based on previous research on user behavior analysis,this paper analyzes and studies the different behaviors of social e-commerce.The main work of this paper includes the following points.(1)The improvement of clustering algorithm based on user preference behavior.Aiming at the preference behavior data of social e-commerce,this paper studies the existing excellent clustering algorithms,then integrates and improves them according to their disadvantages and advantages.Therefore,this paper proposes an improved K-means clustering algorithm combined with density Canopy algorithm to divide social e-commerce groups.(2)A classification algorithm model based on user social behavior is proposed.Aiming at the behavior of social e-commerce publishing social texts,this paper summarizes the characteristics of social texts,studies the existing text classification algorithms,and constructs a composite network model based on BERT+TF-IDF+full connection layers.In this model,the BERT language framework is applied to the classification of social text,which adopts multi-head self-attention to comprehensively consider the relevance of context.Then TF-IDF technology is used to combine the statistical features of word frequency.Finally,full connection layers is used to further extract semantic features for text classification.(3)This paper proposes the user behavior analysis process based on the social e-commerce platform.It uses different methods such as statistical analysis,clustering and classification to mine the social e-commerce data from multiple perspectives,and displays the results in the form of charts.The results are extended to reveal the characteristics and regular patterns of social e-commerce and provide constructive suggestions for promoting the economic development of social e-commerce.The main content of this paper is to adopt different research methods for different types of social e-commerce behaviors.By improving the clustering algorithm and constructing the text classification model,the behavior of social e-commerce can be deeply mined and analyzed,and the process and framework of user behavior analysis based on social e-commerce platform can be formed.