information to the user and the information to recommend post-treatment[5]. This technique has been widely used in electronic commerce, mainly due to expanding the number of species and commodities rapid growth of e-commerce model, resulting in users will spend a lot of time to filter product information in order to find their own goals merchandise. In this process, the user can do lots of wasted efforts, and even can not find a suitable target, causing the target user's losses. Digital Library emergence of personalized recommendation in this regard is also the reason, hoping to save the reader's time by the initiative of recommended information to the reader, so that readers can rapidly detect the target group of interest. Personalized recommendation system is a method of using a variety of data mining Nnowledge database to dig out the Nnowledge of applied technology, its goal is to provide real-time, proactive and valuable information to the user. In addition, the system can also be recommended to the satisfaction of the user feedbacN, the statistics of users, in order to verify the effect of recommendation. In totally, we can use specific data mining of big data preprocessing, and some pre-excavation of the library to access the data, mining the library personalized users mode. Big data preprocessing of data mining techniques include data cleaning, user identification, users access identification, pattern recognition and users path complete. The following is the specific process of the pattern of users: