Bayesian Network In Library Books Purchase Application

The book acquisitioning work of library is a very important work, how to make good use of the limited library purchasing funds, improve the utilization rate of books, meet the needs of the library readers need for knowledge, is worthy of attention and research topic. Because the readers to borrow books is uncertain, so using Bayesian network to study this topic provides conditions. The use of Bayesian networks for readers to borrow books data contains uncertainty knowledge through learning, representation and processing, so that the library book purchasing is more targeted, to do a good job of book purchasing and using, good service of library readers, improve the service efficiency and level has very important significance. It is based on simple statistical methods to express more clear, can carry out the bidirectional reasoning, also facilitate reconstruction and debugging to work, readers have a better guidance.The main work and contributions are as follows:First of all, the Bayesian network overview. In the introduction to the origin and development of Bayesian network and the associated probability theory based on knowledge, elaborated the definition of Bayesian networks, Bayesian network types, Bayesian network advantages and Bayesian network.Secondly, introduces the Bayesian network structure and reasoning. In the Bayesian network structure and reasoning on the basis of summarizing, the Bayesian network structure and reasoning process and methods are presented in this paper analysis. In the Bayesian network structure part, introduces the manual and machine learning Bayesian network structure for the process and method, pointed out the existing knowledge and experience with the combination of data analysis, the use of manual is constructed with the combination of machine learning method can more effectively and quickly construct Bayesian network. At the same time, focusing on the learning of the parameters in the maximum likelihood estimation and Bias estimates, on structure learning in score on score criterion and algorithm and based on the dependency constraint structure learning method are discussed. In the Bayesian network inference part, introduces the basic process of reasoning and inference and approximate reasoning in two types of reasoning, and focus on the exact inference algorithm of tangent sets conditional reasoning Cutset conditioning algorithm and approximate reasoning algorithm of Monte Carlo sampling algorithm is discussed.Finally, in the Bayesian network theory of serious study and research, its application in library use and in procurement. On reader information, books information and reader library information collection, collation, organization based on Bayesian network structure, using the theory and the corresponding software, is constructed based on reader information, books information and reader borrow return book information of Bayesian network, and on this basis, using Bayesian networks inference and the relevant reasoning software and the weighted average principle, established a library book purchase model, which is how to improve the utilization of library books purchase and provides reference and an intuitive and scientific methods, but also for books purchasing decision to provide powerful decision support.