Study On Recommendation Service Of Data Mining In The Bibliography Of Library

With the development of Internet technology, brought a strong impact to the way of people’s life, rich and convenient ways to obtain information triggering a global information revolution. Under such a background, a lot of commercial websites constructed personalized recommendation system in commodity based on data mining, providing a more intuitive and effective service for the people, but so far, recommendation service had not been given sufficient attention in the library application.This paper introduced the data mining technology into the library system, committing to improve the recommendation service of library bibliographic service. The article compared the domestic and foreign research situation of book recommendation system, pointing out how many aspects the library information recommendation service should be divided into, what technical supports were needed, and introduced the frequently-used recommendation technologies, choosing the most suitable one based on the compare of their advantages and disadvantages. Then introduced the data mining methods needed for bibliographic recommendation in detail:cluster analysis, association rules analysis method, decision tree analysis method, and choose the most suitable algorithm for each kind of data mining method. When introducing the association rule analysis method, this text optimized the Apriori algorithm, drawing the idea of matrix into it, transforming the String operations based on business database into Boolean operation based on matrix. This Improvement not only reduced the frequency of access to the database in the running process, but also released the memory space, improving efficiency of the algorithm operation. Finally, using Clementine software to conduct data mining on the database of North Central University library circulation records, providing reference for bibliographic recommendation service. In data mining to North Central University library records, this article implemented four steps:data preprocessing, data mining implementation, mining results analyzing and propose., during which data mining implementation phase is the key. In this paper, it used cluster analysis, association rules analysis and decision tree analysis to implement data mining in library records. Cluster analysis and association rules analysis undertook data processing from the perspective of the readers, on the other side, decision tree analysis from the perspective of book classification. Decision tree analysis got the readership interested in a class of books, and then decided whether to recommend them to the reader according to whether the reader meets the characteristics of the readership. Among them, implementing the decision tree analysis method is a first attempt in book recommendation service.