Research On Book Purchasing Model Of The University Library Based On Hybrid Intelligent Algorithms

In the construction of university libraries,besides the basic hardware facilities such as desks and chairs,library buildings and service desks,the most critical is the construction of literature resources.Optimizing procurement decision-making always attracts the attention of library and information workers and subject librarians.Especially in recent years,artificial intelligence has become more and more mature,and it has become a hot spot and focus of current research to improve the efficiency and quality of book purchasing by using big data and related intelligent technology.Aiming at the problems of subjective selection and inefficiency of book value evaluation in current research,this thesis proposes a book purchasing model based on hybrid intelligent algorithm,in order to make purchasing decision more scientific,improve acquisition efficiency and maximize the value of limited funds.The specific work is as follows:By investigating the actual business of book purchasing in university libraries,combining with the research of other scholars on book value evaluation and the characteristics of the schools' own discipline construction,in this thesis,we probe into the selection of book value attributes and related evaluation characteristics,and construct the new book value evaluation system using eight key disciplines as research indicators including subject category,publishing house,reader demand,publishing time,book unit price,text type,relationship with library collection and whether key disciplinary fields or not.Then,using BP network and SVM algorithm with good non-linear mapping and predictive ability,the book purchasing models of university libraries Based on BP network and SVM algorithm are establishedrespectively.In the book purchasing model of university library based on BP network,aiming at the problems of BP algorithm that easily falling into local optimum and low training efficiency,GA algorithm with good global optimization ability is introduced to improve the threshold and weight of BP network,and then improve the training efficiency and optimization ability of the algorithm.Examples show that the prediction accuracy of the book purchase model based on GA-BP algorithm is improved by about 8% compared with the standard BP algorithm model,reaching91.90%,and it shows good prediction stability.In the book purchasing model of university library based on SVM algorithm,the same as the GA-BP model mentioned above,the data and algorithm characteristics are integrated.In order to improve the search efficiency of SVM,GA algorithm is introduced in order to find the best kernel function type,penalty factor and the combination parameters of kernel function,and then improve the prediction effect of the model.Experiments show that the prediction accuracy of SVM optimized by GA is significantly higher than that of other optimization methods,reaching 94.18%.Finally,the prediction results of the two models under different parameterseeking methods are compared and analyzed.It is found that the prediction effect of the model based on GA-SVM algorithm is the best.It can provide theoretical guidance and decision-making support for the book procurement work of university libraries,and provide a new exploration idea for the follow-up research.