The Research Of Reader's Borrowing Behavior In School Library Based On Data Mining

With the advent of the internet era and the rise of electronic reading,the traditional paper reading is suffering an unprecedented impact.How to provide quality service for the school teachers and students concerns the survival and development of a library.The school library is the soul of a school and an important base of knowledge.It is the focus of all school libraries to understand the readers' reading behavior patterns and to recommend appropriate books for the readers.The application of data mining technology to the library field can achieve the function of book recommendation,which gives promotion to the development of the library.Data mining technology can provide an important technical measure for understanding readers' reading demand and reading behavior.This paper takes book borrowing records of a school library in shanghai in 2015 as an example,introducing various knowledge in many fields such as data mining,program development,library science,maketing management to learn readers' borrowing behavior.First of all,in view of the problems existing in the field of school library,there are great differences in the readers' borrowing behavior,this paper uses the method to divides the readers into groups,the readers are clustered by K-means algorithm,divided into different groups of readers and tagged with different characteristic values.This method can choose different data mining strategies according to different readers' characteristic values to avoid the interference caused by the difference of readers' borrowing records.Secondly,there is a phenomena of many inactive readers in the school library,these readers' borrowing record will cause the date mining to be unable to carry on,the paper regard these inactive readers as a whole model,data mining these readers with innovative Based on Book Classification Matrix Apriori(BBCMA)for association rules.This paper finds borrowing relationship between different categories of books and compares the efficiency between traditional Apriori algorithm and Based on Book Classification Matrix Apriori algorithm.In view of those active readers' group,this paper propose a kind of improved Based on Book Classification Collaborative Filtering algorithm(BBCCF)to establish the model of book recommendation.Using cosine similarity to calculate the neighboring matrix and select reader's borrowing book type with the largest neighboring value as the basis for book recommendation.This can solve the problem of too large score matrix in traditional collaborative filtering algorithm.Finally,according to the results of data mining and improved collaborative filtering algorithm,hybrid books recommendation algorithm is designed and recommendation module based on readers' borrowing behavior is established,which can provide different readers with personalized borrowing service according to readers' various behavior.Proving the function and efficiency of recommendation module by experiments.