Research And Application Of Weighted Hybrid Collaborative Filtering Algorithm

With the development of Internet and information technology,people have entered the era of large data,there are more and more ways to get information with the explosive growth of information resource.While enjoying the convenience of the information age,people are faced with the challenges brought by information explosion.The emergence of personalized recommendation system effectively alleviates the problem of information overload,as an efficient means of information filtering,widely used in major portals and e-commerce.The recommended algorithm as the core content of the personalized recommendation system has always been a hot topic in the academic and electrical industry research.In many recommendation algorithm,collaborative filtering algorithm does not need to analyze content of the project,can recommend the concept of complex projects,and produce more personalized recommendation results,is the most widely used recommendation algorithm.However,in the face of data sparsity,similarity measurement,cold start and other issues,collaborative filtering algorithm still has many aspects need to be improved.In order to improve the recommendation accuracy of the collaborative filtering algorithm,this paper systematically studied the user-based collaborative filtering algorithm and the item-based collaborative filtering algorithm,tries to alleviate the problem of data sparseness and the problem of similarity measurement,proposes a new type of weighted collaborative filtering algorithm.First of all,the new algorithm will improve the method of similarity calculation,and the number of common items between users and items will add to the calculation of user similarity degree and item similarity degree respectively.Re-measure the similarity size,so that the order of similarity is more in line with the actual situation.Secondly,this paper introduces the concept of similarity quality to weigh the similarity quality of the neighborhood,so as to determine the weight of the user-based collaborative filtering algorithm and the item-based collaborative filtering algorithm in the hybrid algorithm.In order to mitigate the influence of data sparsity at the same time,the control factor is introduced into the weighting factor to optimize the recommendation effect.Then,design experiments to verify the feasibility of the new algorithm with the MovieLens data set.These experiments determine the existence of the optimal value of the control factor and effectively reduces the influence of the sparsity of the data on the recommendation.And the experiments further verify the new algorithm improves the prediction precision and quality of recommendation by comparing with the other three kinds of collaborative filtering algorithms.Finally,based on the weighted hybrid collaborative filtering algorithm proposed in this paper,a film recommendation system is designed with the B/S structure divided into the presentation layer,the business layer and the data layer to achieve the personalized film recommended function.