Research And Implementation On Intelligent Recommendation System For Agricultural Products Based On Collaborative Filtering

With the rapid development of Internet technology,the sale and purchase of agricultural products on e-commerce platforms has become an important trend to facilitate people’s lives.However,due to the refinement of the sales types of agricultural products,the number and types of agricultural products on the e-commerce platform are diversified and complex,making it difficult for consumers to quickly and conveniently find the agricultural products they need when facing massive agricultural product information.To address the issue of"information overload"in agricultural products mentioned above,exploring more intelligent recommendation systems has become an urgent need.Based on the collaborative filtering recommendation algorithm,this paper incorporates the content-based recommendation algorithm to improve the recommendation effect.As a result,an optimized intelligent recommendation system for agricultural products is designed and realized.The details are as follows:（1）Improvement of collaborative filtering recommendation algorithm.When traditional collaborative filtering recommendation algorithms recommend items to users,they are often influenced by popular items,which further affects the similarity calculation between users and items,resulting in poor recommendation results.In order to solve this problem,this paper adopts the method of penalizing popular items to reduce the impact on the similarity calculation.Furthermore,considering that users’habit transfer will change over time,this paper adds a time factor to the algorithm to better reflect users’current interests.Finally,experiments verify that the MAE value of the improved collaborative filtering recommendation algorithm is optimized.（2）Hybrid recommendation based on collaborative filtering and content.In the collaborative filtering recommendation system,when a new user or a new item is added to the system,it is difficult for the collaborative filtering algorithm to make personalized recommendations using historical data,i.e.,there is a cold start problem.To overcome this problem,a content-based recommendation algorithm is introduced in this paper.The algorithm calculates the weights of each feature attribute in the corresponding item feature document and user preference document based on the feature attributes of the acquired items,and compares and calculates the similarity between the two documents to make recommendations to users.In this paper,these two algorithms are weighted and fused to form a hybrid recommendation strategy that can complement each other’s advantages,thus improving the accuracy and effectiveness of the recommendation system.Finally,experiments are used to demonstrate that the hybrid recommendation algorithm outperforms the single recommendation algorithm in three metrics:accuracy,recall and F₁ score.（3）Design and implementation of an intelligent recommendation system for agricultural products.The intelligent recommendation system designed in this paper is separated from the front and back ends,the front end is developed by Vue.js framework and the back end is developed by Spring Boot framework.In addition,combined with the requirement analysis,this paper has completes the design of recommendation process,system functional modules,database,and other aspects.Finally,an agricultural products e-commerce recommendation platform was implemented,which provides users with more convenient and effective functions such as recommendation and purchase of agricultural products.