Research On Personalized Recommendation Method Of E-commerce Platform Based On RFM Model

With the rapid development of Internet technology,e-commerce has gained wide attention because of its advantages of convenience and speed.However,with the growth of the number of commodity resources,it is difficult for customers to easily and quickly find satisfactory products in a short period of time when shopping online.In order to help customers find the products they need quickly and bring higher profits to enterprises,personalized recommendation technology has emerged.The performance of personalized recommendation technology can directly affect the effect of recommendation,so the research on personalized recommendation technology has theoretical reference value and application value for e-commerce operators and developers of e-commerce products.Collaborative filtering recommendation algorithm is currently the most successful personalized recommendation technique used in e-commerce recommendation systems,which has the advantages of being able to handle complex unstructured objects and high degree of personalization.However,there are some problems in the practical use of this recommendation technology: the scoring data has high sparsity and subjectivity;it needs to process a large amount of data,with low efficiency and poor scalability;it does not consider the difference of customer value and uses a single recommendation method.To address the above problems,this paper integrates customer relationship management into the personalized recommendation of e-commerce wholesale platform,thus introducing a customer value assessment model,classifying groups according to customer value,and making subgroup personalized recommendation.This paper adopts the RFM model as the basis of customer value segmentation,and improves it according to the characteristics of e-commerce wholesale platform customers,taking into account the purchase quantity of customers,and proposes the RFMQ model.Then,on the basis of RFMQ model,two methods,K-means algorithm and index segmentation,are used to segment customers of different values.Finally,this paper introduces the commodity dimension into the RFMQ model and uses the entropy value method to obtain the weights of each indicator,calculates the customer-commodity value preference,and constructs the customer’s rating matrix accordingly,and combines the collaborative filtering recommendation algorithm to make personalized recommendations.In this paper,the proposed personalized recommendation method is tested and verified using the historical transaction data of customers in an e-commerce wholesale platform,and finally the following results are obtained by comparing three groups of tests:(1)The segmentation-based customer segmentation method is more suitable for practical application of recommendation than the K-means clustering algorithm.(2)The recommendation effect of various groups after segmentation is greatly improved than before segmentation.(3)The recommendation quality of collaborative filtering recommendation algorithm can be improved by analyzing the purchase preferences of customers through the historical transaction data of purchased customers.