Research On Hybrid Collaborative Filtering Recommendation Algorithm

With the rapid development of Internet technology,human beings have entered the information age.People in the rapid sharing of information at the same time,have to face the information overload caused by the problem of information overload.In order to solve the problem effectively,the recommended system has been rapid development.It analyzes the user's behavior and project attributes,take the initiative to get the user's interest,to provide users with personalized recommendation services,effectively solve the information overload problem.Collaborative filtering recommendation algorithm is one of the most widely recommended algorithms.The algorithm can accurately obtain the recommendation result by analyzing the user 's historical data.However,with the rapid growth of the number of users and projects within the system,the algorithm still faces problems such as user interest drift,Matthew effect and recommendation precision.This paper will study the cooperative filtering algorithm from user interest,Matthew effect,time effect and recommendation precision.The main work of this paper is as follows:Aiming at the problem of user interest,this paper introduces the concept of user activity,user activity attenuation factor and interest degree among users,and proposes a hybrid similarity algorithm that integrates user activity and user interest.Aiming at the problem of Matthew effect and time effect,the paper first introduces the effectiveness factors of project popularity,and proposes a collaborative filtering algorithm based on user behavior and project characteristics.Then,this paper proposes a new user and project influencing factor building process by constructing Usernetworks based on temporal and Item-networks based on temporal.Aiming at the problem of recommendation accuracy,the classic Slope One algorithm is used to improve the model.Firstly,using the hybrid similarity algorithm and the weighted Pearson algorithm proposed in this paper,this paper proposes a hybrid recommendation algorithm with the similarity degree.Then,by constructing the influence factor,this paper proposes a hybrid recommendation algorithm.At the same time,this paper has designed rigorous verification experiments in turn.The results show that,after introducing the factors such as user interest degree and user influence,it can effectively improve the precision,recall,popularity and coverage of collaborative filtering algorithm.Through the above research,this paper provides an effective reference and guidance for the improvement and development of collaborative filtering algorithm in the future,so that the collaborative filtering algorithm can be applied more widely.