The Research On Commodity Recommendation Algorithm Based On Users' Consumption Level

In recent years,the electronic commerce has become a kind of very popular new shopping channel as the new shopping pattern along with the rapid progress of the Internet technology gradually rise.E-commerce has so many advantages compared with the traditional way of shopping,on the one hand because of its vast network of users,it can give an enterprise to bring more business profits,on the other hand because users can feel free to browse a variety of dazzling commodity information all over the world when they stay home,e-commerce can provide users with more convenient and more comfortable shopping experience.At the same time,we still can't ignore some drawbacks the existing e-commerce recommendation system shows.Users often can't find their favorite goods from a very large commodity at times.In this case,e-commerce for the user's personalization recommendation become a very effective way to deal with this problem.At present,there are many kinds of research about the recommended methods.Even so there still exist some problems such as data sparse,cold start,poor scalability algorithm and so on.How to break through the bottleneck of these technologies has become the focus and difficulty in the research.Collaborative filtering algorithm is one of the most widely used technology in the personalization recommendation,and the research based on collaborative filtering algorithm is mainly based on the perspective of user project scoring.This paper argues that,in addition to the evaluation of this aspect to measure the similarity between users,but also can use some of the user's own factors to analyze the user's shopping habits.Therefore,the idea of this paper is to integrate the existing collaborative filtering recommendation process and the users' consumption level,and think that users with different consumption levels have different commodity orientation.In this paper,we use the user's background information and shopping records to establish the user's two level of consumption level model,to reduce the dimension of the scoring matrix and the vacancy project score prediction score.And then combined with the consumer level and score data obtained comprehensive user similarity,determining the nearest neighbors of target user set according to the user set screened by the consumption level.Finally,we get the recommended set of items based on the target user nearest neighbor set.This article finally compare the improvement and the traditional collaborative filtering by observing the data results obtain by experimental verification.The experimental results shows that the improvement recommendation algorithm based on users' consumption level can recommend commodity more accurate to people than traditional recommendation algorithm,and alleviating data sparsity problem and the problem of new users to some extent,promoting a lot for the electronic commerce recommendation system improvements.