Application Of Data Mining In Research Readers' Preference

As an important knowledge base for teachers and students, university libraries cover a very wide field of books. Books are bought annually, so the number of books increases constantly. It is not pleasant for teachers and students to find books relevant to their own needs, so optimizing the layout of books quickly and effectively becomes more and more important for teachers and students. The dissertation is to mine the association information among the borrowed books from the history data quickly and effectively with association rules methods of data mining.Firstly, the theoretical foundation of data mining is described, some classic association rules algorithms of data mining are analyzed and summarized. Then, the history data of borrowed books is preprocessed according to the algorithm, including data input and extraction, the establishment of transaction database. At last, a new association rules algorithm MFP-Miner is applied to the transaction database, association rules of the borrowed books are mined. Experimental results showed that the algorithm is superior to other algorithms in efficiency.In order to raise service efficiency of the recommendation system, an improved collaborative filtering recommendation method based on clustering of readers is proposed. This new method revises the original similarity using readers'interest in item, takes synthetically into account the influence of readers'interest in item and readers rating. The experimental results show that the presented method not only reduces the search space for nearest neighbors but also improves the performance of CF systems in recommendation quality and efficiency.