Research And Application Of Personalized Recommendation Technology In Library Service

In recent years, personalized recommendation technology has become a hot research topic, with great success in e-commerce and other fields. However, it is very difficult for the readers of university library to find books of interest in the vast library resources. So it has become an important research direction to use personalized recommendation technology in library service by learning from the successful experience in the field of e-commerce.This dissertation focuses on the application and research of personalized recommendation technology in library service, the main contents and contributions of dissertation can be described as follows:1. This dissertation first outlines the research status of book recommender system, and then focuses on a variety of recommended techniques and compares their advantages and disadvantages.2. Considering the actual situation of university library, this dissertation chooses collaborative filtering algorithm as the focus of research and make some appropriate improvements:Aiming at the problem of books’score data sparseness, we propose a special solving method, through introducing readers-books classification, we combine the Collaborative Filtering algorithm based on readers’record on borrowing books with the Collaborative Filtering algorithm based on readers-books classification, finally make some improvement in similarity computing and the selection of the neighboring items. It makes the similarity be dynamically resized with the actual situation and improve the accuracy of similarity calculation through adding the weighting factor λ.3. This dissertation uses the improved algorithm to make personalized book recommendation for the readers, and then designs some experiments to do the empirical research, which is based on the readers’real record of borrowing books. The experimental results show that the improved collaborative filtering algorithm is both better in precision and recall than the traditional collaborative filtering algorithm. The recommendation quality can be improved by this algorithm. The result proves that it is feasible to use this algorithm to make personalized book recommendation.4. Finally, this dissertation designs and implements a simple book prototype system of personalized recommendation, which uses improved algorithm and content-based recommendations. It can provide individuation service in the library.