Research And Implementation Of Group Purchase Recommendation System Based On User Transaction Data

With the rapid development of the Internet and the progress of information technology,"online group buying" this new e-commerce model is also growing rapidly,attracting a large number of stores and customers.The merchandise presentation group purchase website still exists the problem of information overload,customers quickly find the goods or information they are interested in is still a difficult thing,along with the rapid development in the field of group purchase,how to further explore the potential customer's idea,provide better service for users to recommend becomes an important research topic.The electronic commerce group purchase mode and traditional has several different,its impact on consumer behavior by time,location,date and other factors,the single traditional recommendation algorithm has shortcomings and are not well suited to the group purchase system.This paper analyzes different point of group purchase mode and traditional e-commerce,based on user transaction data,combined with the user based collaborative filtering algorithm and stores word vector and content recommendation algorithm based on user preference information into the personalized recommendation system,realized the group purchase website.Group purchase recommendation system proposed in this paper is mainly composed of off-line calculation module and the online recommendation service module consists of two parts,the establishment of a comprehensive and effective evaluation system,and the hit rate,line CTR by contrast experiment in recommendation(Click Through Ratio),CVR(Click Value Rate)online index to verify the effectiveness of this proposed method the.The personalized recommendation system is designed to improve the user experience,reduce the user's choice of costs,and increase user stickiness,improve sales.In this paper,the recommendation system in the test performance is good,the content of the recommended content than the content of random recommendation has been greatly improved.This shows that the personalized recommendation system can be based on user transaction data to give users interested in the recommended results,to achieve the desired level.