Research On The Application Of Personalized Recommendation Methods In The Bibliographic Recommendation Of University Libraries

With the development of mobile Internet,the rise of e-commerce,and the expansion of user scale and commodity scale,it has become increasingly time-consuming for users to find items they are interested in.How to help users quickly find their own wants or potential interests has become a hot topic in the academic and industrial research.At present,the personalized recommendation algorithm has achieved certain results in the recommendation,and has been widely used in the fields of e-commerce,film and television,music and so on.But in the field of university library,this technology has not been applied.Different from other fields,the number of university library users is very small,maybe only 10 to 20,000 per year.However,there are millions of books in university library and tens of thousands of books that are often borrowed,which makes the data of university library sparse.Aiming at the scene of university library,this paper puts forward two algorithms which are suitable for the bibliographic recommendation in university library,the recommendation algorithm based on graph structure and the improved collaborative filtering algorithm.The main research contents and achievements of this paper are summarized as follows:1.This paper analyzes the problems in the study of bibliographic recommendation algorithms in the library field.The paper also summarizes the recommendation algorithms,introduces the current commonly used recommendation algorithms in detail,and compares the advantages and disadvantages of various recommendation algorithms.2.Aiming at the problem of data sparsity,this paper proposes personalized recommendation based on graph structure.It shows the data in graph form,and shows the similarity between users in edge.This method can effectively avoid data sparsity.This paper also makes a case recommendation for a user based on the library data of guizhou university of finance and economics.3.Although the recommendation based on graph structure can effectively avoid data sparsity,there are still some problems,for example,it may reduce the items in the candidate recommendation set.In order to avoid this problem as much as possible,an improved collaborative filtering recommendation algorithm is proposed to transform the implicit rating of books into the explicit rating of books.This method can also effectively reduce the sparsity of data and will not reduce the items ofcandidate sets,and can reflect users' interests more effectively.On the other hand,this method also increases the effectiveness of user similarity.