Research On Algorithm And Model Of Book Recommendation System In Universities

Since the 21st century,with the continuous improvement of the technological level of the world,information appears explosive growth,people have become the model from looking for information turns out to useful information model.There are many ways to find useful information from the massive information.Recommended system is one of the most important and most widely used means.The major network operators have achieved good results by using the personalized recommendation system to users.This provides possibilities for recommended system's application in the field of library in colleges and universities.There are many algorithms in the recommended system,and the most classic and widely algorithms is collaborative filtering algorithm.In this paper,we have done an in-depth study for the user-based and project-based collaborative filtering algorithm.For the specificity of university books recommendation,such as borrowing data can't be directly used,similarity matrix is too sparse and can't produce recommendations and other issues,we improved the two algorithm.However,these two algorithms have their own advantages in the field of university books,and also have their own disadvantages.Therefore,through the combination of both,a hybrid recommendation system model is proposed.Then we verify the feasibility of the hybrid recommendation system in the field of university books recommended by the experiment.The main work of this study is the following five parts.In the first part,the principle of the recommendation system and some classical recommendation algorithms are studied in depth,and the feasibility of the application in the field of university books is analyzed.Then,we construct the collaborative filtering algorithm model based on user and object.In the second part,due to the university book recommendation is different from the recommendation of the film or the commodity,it does not include the user's rating on the item.According to this characteristic,we proposed establish the scoring rules,quantify the reader's score,and construct the reader-matrix.In the third part,because of the shortcomings of the reader-book scoring matrix is too sparse,the book is classified by combining the Chinese book classification method,to form a the reader-book category scoring matrix,So that we design user-based and project-based collaborative filtering algorithm.In the last part,the hybrid recommendation system model is generated by combining the improved UB-CF and IB-CF,and then the experimental model is evaluated by experiments.Finally,we make a suggestion for the future of university books recommended.