Design And Implementation Of Electronic Commerce Recommendation System Based On Data Mining

The rapid development of the Internet has changed the way people live,online shopping has become the most popular trend of the moment.More and more people choose to buy what they want on the e-commerce website,which greatly promote the commodity number and the user number to have the explosive growth.The surge in the number of goods has caused trouble for users,Electronic commerce recommendation system appears to solve the problem of customers.It's like a virtual assistant,recommend the goods to the user that they may enjoy,help users quickly buy the goods you want and also increase sales of electronic commerce website.Most of the existing E-commerce Recommender systems use collaborative filtering recommendation algorithm,but the traditional collaborative filtering algorithm has sparse problem,which leads to the reduction of the accuracy of the recommendation.At the same time,with the expansion of the scale of the electricity supplier website,the recommended real time is also significantly decreased.Around the above problems,the main work done in this paper is as follows:(1)In this paper,a collaborative filtering algorithm based on item clustering for ALS is proposed,ALS-IC(Alternating Least Squares with Item Clustering)algorithm.The main idea of ALS-IC algorithm is to reduce the complexity of the traditional ALS algorithm by using the Least Squares Alternating algorithm by clustering analysis and mining,improve the efficiency of ALS collaborative filtering algorithm on the basis of ensuring the accuracy of recommendation.At the same time,the distributed implementation of ALS-IC algorithm based on Spark big data processing platform is discussed,and the implementation and optimization details are analyzed.(2)Based on the above proposed algorithms,this paper presents a general design for the Spark based ALS-IC online recommendation system,which is based on the recommendation of the film.The performance and functional requirements of the system are analyzed,and the architecture of the proposed system is proposed.Among them,the focus of the Spark based on the core of the large data platform recommended procedures and the application of the ALS-IC algorithm detailed steps are described.(3)The proposed ALS-IC algorithm and Spark based prototype implementation are simulated to verify the performance and efficiency of the proposed algorithm and system.Through the experimental results,the optimal algorithm parameters are determined in the experimental data set,and it is proved that the ALS-IC algorithm has better recommendation accuracy than the same type algorithm.At the same time,Experimental and simulation is realized by a prototype of spark ALS-IC algorithm and verify the proposed distributed ALS-IC algorithm is relatively common algorithm has better efficiency of the algorithm,can be recommended to calculate process can be accomplished in a shorter period of time.