With the continuous development of the Internet, the information explosions become obvious phenomenon in the field of books. So it becomes a serious problem to help users retrieve the books what they want to without taking too much time. This paper studies the realization of a personalized book retrieval system based on user. First, crawling and download book information with the use of web crawlers, and then clean the book information into unified format and storage in the database, then index the books to reduce the retrieval time. Finally determine the weights of the search key and search mode with the use of TF-IDF. Through book search we found that the system is superior to current mainstream book search engines in ordering interface, which demonstrates the effectiveness of the system clarity.