In order to solve the problem of flexibility brought by the existing recommendation algorithm which only considers the recommendation of the same category of products, and improves the sales volume of products and the shopping experience of users, this paper proposed an accurate marketing recommendation algorithm based on customer preference analysis, which could not only accurately recommend products for customers, but also provide potential customers to businesses. Specifically, at first, based on the purchase information of customers and their neighbors in the product purchase network, it expanded the customers' purchase information. Then, it designed a method for calculating the customers' product preference weight, analyzed the customers' purchase preferences, and provided customers with personalized recommendations for products under the guidance of customers portraits. Finally, based on the sample customers provided by businesses, it mined the community formed by customers who were similar to the sample customer to provide accurate customer maintenance and potential customer recommendation for businesses. Experiments on real datasets show the effectiveness of the proposed algorithm. This algorithm pays attention to both customers and businesses, and realizes two directions of product and customer recommendation. This algorithm provides useful help for the research of personalized recommendation.

[2021-09-05] Printed Article /doc/2021.01.0009/v2