Personalized Book Recommendation Based On Collaborative Filtering And Association Rules

At present,there are a large variety of books in the university library,and the huge number of books.In the face of massive books and information resources,on the one hand,students need to spend a lot of time and energy to get the books they need.On the other hand,traditional search query or popular recommendation often can not produce satisfactory results.Faced with the ever-changing needs of students,the traditional mode of library work obviously can not meet the individual needs of students.Therefore,the development of university library personalized book recommendation service is the inevitable trend of the future.On the basis of consulting the relevant documents,this paper proposes a personalized book recommendation algorithm based on collaborative filtering and association rules.First of all,this paper analyzes and resolves the problems of system scalability and data sparsity in collaborative filtering based on book borrowed record.Secondly,this paper improves the user similarity calculation model in collaborative filtering algorithm.In the calculation of user similarity,this paper not only considers the user similarity based on the traditional book borrowed record information,but also considers the user similarity based on the characteristics of student attributes for the background of university book recommendation,and obtains the users' comprehensive similarity.In the recommended result set,this paper uses association mining to optimize it.Firstly,through the improved collaborative filtering algorithm,it chooses the TOP-N books which target users are interested in,and then mining the association rules of all the users' borrowed record,finding strong association rules between books and save them.Then through the rules of the database to find out more the associated books about TOP-N books,and these associated books and TOP-N books are recommended for mixing.In the end,this paper conducts experiments on traditional collaborative filtering recommendation algorithm,collaborative filtering recommendation algorithm based on improved similarity and collaborative filtering and association rules recommendation algorithm,respectively,and compares the results with the recommended precision and recall.Experimental results show that the improved algorithm has further improvement in accuracy and recall.At the same time,based on the improved algorithm,this paper implements a set of personalized prototype books recommendation prototype system.