Study On Prototype System Design Of Personalized Digital Books Recommendation Based On Forecasting Users' Like

With the rapid development of computer network technology and mass storage technology, the traditional passive way of information are increasingly difficult to obtain useful information from the vast content of the message. Personalized digital books recommendation services appear to follow the individual needs of users of the product. This paper studies to implement a algorithm of forecasting digital books users' like and design the prototype system of personalized digital books recommendation.Content-based filtering of personalized recommendation system helps people find documents and activities from potentially extractable, meaningful knowledge and models. The personalized recommendation method is researched by using system data modeling based on k-nearest neighbor classification method and Naive Bayesian classifier, a predict recommendation algorithm is designed based on user preferences and content filtering performance, and its experiment performance is tested.Based on Windows operating system, Myeclipse and MySQL development tools, and Java programming, a prototype system of digital library personalized recommendations designed and implemented by using predicting users'like algorithm. The system uses a mixture of k-nearest neighbor and Naive Bayes classification as filtering and search algorithms of books, the algorithm improves the accuracy of the search for this books'management system which can add, delete, change, check as the basic functions. Results show that combining content filtering and personalized Web mining can be better, more actively grasp the user's behavior patterns and habits and hobbies, to improve the accuracy of digital library personalized recommendation and make the system has "self learning" ability.