Research On Collaborative Filtering Recommendation Method Based On RFM Model And Clustering

The rapid growth of E-commerce offers more choices of commodities to customers, but at the same time costs them more to search for specific commodities they need. E-commerce recommendation system can provide with product recommendations for customers, helping them find target commodities to meet their personalized demand, thus it can turn browsers into buyers, improve the customer loyalty through the interaction between the websites and the customers, therefore, increase the benefit of E-commerce websites.At present, collaborative filtering recommendation is the most successful personalized recommendation technology. It can handle complex and unstructured objects, and boasts higher degree of personalization. However, in practice, this recommendation technology has some deficiencies: the source of data for analyzing customer preferences is the customers'evaluation of commodities, requiring customers'coordination and data of evaluation is subjective; on-line processing of massive data lower efficiency of the system; impossible to provide recommendation for new customers; does not take into account differences in customer value, using a single way of giving recommendation.To solve these problems and improve the traditional collaborative filtering recommendation, this paper introduces RFM (Recency, Frequency, Monetary Value) model in the field of customer relationship management and clustering technology in the field of data mining, then designs a collaborative filtering recommendation method based on RFM model and clustering. The main contributions of this paper are as follows:This paper designs a kind of procedure based on RFM model and K-means clustering, which analyzes customers'buying preferences through historical transactions (RFM data), therefore, make improvement in data source of the traditional collaborative filtering recommendation algorithm, besides, narrow the scope of on-line searching the nearest neighbors by clustering, and improve the efficiency of algorithm.This paper makes use of RFM model which can reflect customer value to distinguish high value customers and low value customers, and then realizes customer segmentation through k-means clustering, develops differentiated way of giving recommendation for different types of customers, thus, organically combining E-commerce personalized recommendation service with customer relationship management and obtain the highest customer loyalty to improve profitability for the websites. This paper makes use of fuzzy c-means to fuzzy cluster customers in accordance with customer attributes, and then make recommendations to new customers through cluster membership of the customer and recommendation of the cluster, therefore, solves the problem that the traditional collaborative filtering recommendation algorithm is unable to make recommendations to new customers due to lack of new customer preference information.