The Design And Application Of Recommendation System In Liberary Information System

The traditional university information can only provide information of books for readers. Contents are monotonous and boring. Information flow is not fully activated. The system proposed in this paper incorporates readers’ comments and book scores into the information system, which deeply socialize the traditional library information systems.Almost all e-commerce sites such as Amazon and Jingdong have their own recommendation system. The vast majority of entities will also blend recommend functionality into their site. The main recommending methods include making recommendation based on user’s personal information, making recommendation based on transaction history and browsing record. In this paper, the proposed system will make recommendation based on readers’ records and reviews and ratings histories with sophisticated recommendation algorithm to deeply socialize the university library information system. Besides the aforementioned advantages, socialized library information system also provides guidance for the procurement staff.Unified Modeling Language was started in 1997 with an OMG standard. It is a software system to support the development of modeling and graphical languages, provide the model for all stages of software development and visualization support including from the requirements analysis to the specifications, the structure and configuration. With the help of UML class library information systems and graphics involved were introduced. In addition, key parts of the system will be materialized on the ground of ASP.NET framework and SQL Server 2005.Usually, the whole application can be dissected into three-tiers which are the presentation layer(UI), business logic layer(BLL), the data access layer(DAL). Based on the three-tier architecture a structure of "high cohesion, low coupling" can be achieved.The widely used recommendation algorithms in several popular business websites include content based recommendation, memory based collaborative filtering, model-based collaborative filtering, and recommended methods of complex networks bipartite network. The article will propose a mixed algorithm based on UBCF and UICF, which is also implemented in the thesis.Testing is indispensable in development of software and would better be started as earlier as possible. Definition of requirements entails testing. In the last part of this thesis, the details of testing are illustrated including functional testing, testing for recommendation module and performance testing.