Personalized Recommendation Algorithm with Data Mining for College Libraries using Long Short Term Memory Technique

With the popularization of the internet and information technology, ordinary users in university are also struggling to read books that can really interest them or have some real value in a sea of books by simply using traditional book finding services. Therefore, this research proposed data mining and new book recommendation service by using the semantic classification with deep learning for university libraries. The recommendation algorithm which considers the semantic classification constructs the book feature model, which is title keywords depending model and Reader Preference Model. Based on the classes divided above, combine them by using the book feature model and the reader preference model, then establish the book recommendation list by using collaborative filtering recommendation algorithm and content-based recommendation algorithm. Artificial intelligent system integrated with Long Short Term Memory (LSTM) to design a recommender system for semantic classification of user preferences. The recommendation strategies that can be adopted in the system framework are described. In the context of the experiment featured in this research, the division of the personalized recommendation approach for semantic classification is applied. The proposed method attained the 93.7% accuracy in semantic classification.