Research And Implementation Of Shopping Recommendation System Based On User Characteristics

With the rapid and efficient development of the Internet,the rise of big data and cloud computing,the field of e-commerce is paying more and more attention to personalized recommendation.This is not only because the continuous improvement of material living standards makes people more and more pursue spiritual satisfaction,but also because the personalized recommendation system can realize the “triple-win” situation of users,system platforms and manufacturers.In this thesis,the development status of recommendation systems and related recommendation technologies are studied.Considering that the traditional recommendation algorithms do not involve the study of user characteristics,the sparse data of user/item scoring matrix is difficult to solve,and user characteristics play a decisive role in user interest model to a large extent,based on the traditional user-based collaborative filtering algorithm,this thesis incorporates user characteristics into the new algorithm,and designs the user characteristics-based recommendation algorithm.In this thesis,13 user characteristics are summarized and parameterized,on the basis of clustering users twice based on key eigenvalues by K-means technology,user attribute characteristics set and item score are calculated by cosine similarity method in turn.After the final similar user set of target user is obtained,the scores of items marked by similar users and the number of item purchases are processed with optimization rate,finally the Top-N item recommendation list is obtained.Then,the performance of the UCBR algorithm is evaluated through off-line experiment,which proves that the algorithm meets the requirements on precision index and meets the expectations on diversity index.Then,starts from the reality,this thesis studies the shopping recommendation system based on the UCBR algorithm,and realizes the specific system on the basis of demand analysis and system design.In the system testing stage,the function test proves that the shopping recommendation system not only has the basic function of the shopping system,but also can recommend goods according to expectations.