Application Research Of Personalized Recommendation Technology For The Digital Library Of University

In the construction of the University Library, the digital library as a new form,integrates with the library’s online resources and offline resources, to provide users with unprecedented service. However, literature resources in the library system increase every year, it is a hot topic for the reader to find the information they need efficiently and accurately. In fact, system functions still dominate the literature search in the most of the domestic digital library, personalized service,which is passive approach, provides for services to readers with less limit, resulting in library literature resource utilization is not high. To address this issue, personalized recommendation technology,which is change from passive to active service, have been proposed in the application of digital libraries.Personalized recommendation technology is an important technology to achieve personalized service, it has been widely used in e-commerce. On the one hand,the recommendation technology is to provide users with high-quality services; on the one hand,it is help businesses to obtain greater profits. In the digital library, user preference information that is the basic information and behavior information will be collected by technical means, forecast information resources which the readers like,and recommendation systems provide active and diversified information services.Therefore, the digital library personalized recommendation system already gets more concern.In the massive data environment for digital library, the recommendation accuracy and efficiency are two important indexes to measure the recommendation performance.At present,In the recommendation systems, recommendation algorithm that is the most widely used is technology of collaborative filtering.But collaborative filtering has some question, such as cold start, sparse matrix problems, new products can not be recommended because of no user rating or only a few score; At the same time, the traditional collaborative filtering algorithm is often unsatisfactory when it is running in a very large amount of data. Collaborative filtering algorithm based on clustering, can alleviate these problems. This paper proposes a collaborative recommendation algorithm based on clustering about book information recommendation. The basic idea is that divided all the users into different clusters according to the similarity clustering,users who belongs to the same cluster have the similar interest or similar types. When system recommends book information for active user, predict the user score according to the similarity of the user with each cluster centers, and then recommends theinterested information. Based on the proposed algorithm, this paper presents a digital library based on recommendation system, through the application of collaborative filtering algorithm based on clustering, can be more efficient, accurate recommendation information for readers, so as to enhance the digital library service quality and improve the utilization of the literature resources. Specifically, the main work of this paper are as follows:(1) discuss a personalized service model of digital library in universities based on the facts. Investigate and analysis on the present situation of university digital library service, reader demand, resource utilization and so on, discuss concept, mode, content and implementing technology of the personalized service, propose the personalized service mode of Libraries in colleges and universities conform the actual situation, the concrete content includes the design goal, principle and realization architecture.(2) research on the key technology of personalized book recommendations.Analysis the algorithm principle, application status of the classic personalized recommendation, propose a collaborative filtering recommendation algorithm based on clustering, realize this algorithm, and apply it to the personal digital library.(3)develop personal digital library based recommendation system based on the Apache Mahout component library, according to software engineering, analysis and design system structure, data table and UI, and introduction function and realization ideas and the key function module.