Research On Hybrid Recommendation Algorithm Based On Multiobjective Optimization

With the rapid development of the Internet and information technology,the society has already stepped into the era of information overload.As a tool to help users get information automatically,the recommendation system is widely used in various fields.There are two shortcomings in the traditional recommendation system:1)the simplification of the recommendation algorithm;2)the simplification of the recommended target.The simplification of the recommendation algorithm refers that the recommendation system only uses a single recommendation algorithm,and the traditional recommendation algorithm has advantages and disadvantages.In recent years,with the complexity of the recommendation environment improved,using a single recommendation algorithm shows more obvious disadvantages.The simplification of the recommended target refers that the accuracy of recommendation is the only evaluation index,and is committed to providing highly personalized recommendation service for users.In a big data environment,this will make the information recommended to the user extremely similar to the user's historical accessed information,causing the users to gradually fall into the information cocoon room.Moreover,the significant long tail effect makes the profit of information producers affected.From the perspective of user and business interests,the hybrid recommendation algorithm and the recommended single objective optimization are replaced by multiobjective optimization,which can effectively alleviate the above problems.In this paper,a hybrid recommendation algorithm based on multiobjective optimization is proposed on the basis of full investigations of multiobjective optimization algorithms and various recommendation algorithms.The algorithm generates candidate sets through hybrid recommendation algorithms,and based on this,we use multiobjective optimization algorithm to select the most accurate and diverse recommendation list.The research contents of this subject are as follows:(1)Recommendation problems and multiobjective optimization problems are analyzed,and the strategy of combining multiobjective optimization algorithm and hybrid recommendation algorithm to solve the problems currently faced by recommendation is proposed.(2)All kinds of recommendation algorithms are studied,including their algorithm ideas and working principles.According to the advantages and disadvantages of all kinds of recommendation algorithms and the application scenarios,we propose a hybrid recommendation algorithm in this subject.(3)Multi objective optimization problems and multi objective optimization algorithm are resear-ched,including the form of the solution and the evaluation index of the quality of the multiobjective optimization problem,and we use the multiobjective optimization algorithm to encode the actual problem and abstract the objective function.(4)A hybrid recommendation algorithm based on multiobjective optimization is designed.The candidate set is generated by the hybrid recommendation algorithm,and the optimal recommendation list is selected from the multi objective optimization algorithm.The algorithm is applied to the actual data set,and the advantages and disadvantages of the hybrid recommendation algorithm based on multiobjective optimization and the other three sets of single recommendation algorithm based on multiobjective optimization are compared.The experiment shows that the algorithm produced by the algorithm is better than the control group,and can produce a better recommendation list with more accuracy and diversity.