Research on Personalized Recommendation System for E-Commerce Products Based on Collaborative Filtering Algorithm

Abstract

Research from aspects such as user consumption habits can help merchants better understand consumers' consumption needs. This article takes a large e-commerce website as the research object and analyzes user shopping behavior by studying shopping data over the past year. A hybrid recommendation model based on product recommendation and friend recommendation is proposed to achieve personalized recommendations for enterprise groups. This paper calculates the user's location similarity based on the user's location and the distance between locations, and then weights it with the user's interest similarity to obtain a new user similarity. This paper predicts the products that users are likely to purchase this month. Taking the fast-moving consumer goods e-commerce platform as an example, an empirical analysis of the proposed method was conducted. The performance of the algorithm was simulated using offline experiments. Experiments show that compared with the collaborative filtering method based on cosine similarity, the accuracy of this method can be 3.74% higher, and the recall rate reaches 3.91%. This method cannot only improve the user's selection efficiency, but also improve the performance of collaborative filtering.