| In the information age,the Internet has brought convenience to people’s lives,but it has also caused the problem of information overload.As an effective way to solve the problem of information overload,recommendation system has been widely studied by researchers in recent years.Recommendation algorithm is an important part of recommendation system.Traditional recommendation algorithms only recommend items to users based on their historical behavior records,but it is difficult to discover the relationship between these items in this way,thus affecting the recommendation effect.In order to solve this problem,knowledge graph can be used as the auxiliary information source of the recommendation algorithm,and with the help of the rich semantic relations in the knowledge graph,the interests of users can be further explored,so as to improve the accuracy of recommendation.The main research content of this thesis is the recommendation algorithm based on knowledge graph,and a movie recommendation system is built on this basis.In this thesis,KUMR is proposed.Based on the existing recommendation algorithm MKR(Multi-Task Feature Learning for Knowledge Graph Enhanced Recommendation),KUMR algorithm adds user preference.KUMR also adopts the second-order-linear-feature combination method with stronger combination ability for feature fusion in the feature interaction layer of its recommendation module.Experimental results show that the improved algorithm KUMR has better performance than the original algorithm and classical algorithms in the same field.In addition,a real-time recommendation algorithm is proposed.Based on the item-based collaborative filtering algorithm,the algorithm integrates reward and punishment factors to make the recommendation results more relevant to user interests.This thesis designs and implements a well-functioning movie recommendation system.Based on the recommendation algorithm KUMR,this movie recommendation system is designed and implemented on the big data platform Spark based on the mainstream application development framework Spring.The system adopts MVC architecture,which consists of view layer,business layer,recommendation engine layer,and data layer.The system is functionally divided into two modules,the basic function module and the recommendation function module.The basic function module includes user registration,user login,movie rating,movie search and other functions.The recommendation function module includes popular recommendation,offline recommendation,real-time recommendation,and detail page recommendation.Through the system testing,all the functions of the movie recommendation system meet the design requirements in the demand analysis.This movie recommendation system has certain practicality. |
