Research On University Library Recommender System Based On Resource And User Characteristics

Nowadays,with the rapid development of Internet technology,the university libraries and the Internet environment are both facing the problem of "information overload".However,different from the situation that personalized recommender system is widely used in Internet information resource platforms,the application of recommender system in university libraries are not common at present,and it is difficult for university teachers and students to find information resources they really interested in in libraries’massive resources only by traditional retrieval system.Therefore,it is necessary to apply recommender system and data mining technology in university libraries to mine users’potential preferences and needs according to their behaviors,so it can help the university teachers and students to work better on their learning,teaching,scientific research and other work by reducing the cost of information retrieval while improving the utilization rate of library resources.This paper compares and analyzes the resources and user characteristics of university libraries with general commercial recommendation platforms,and finds that compared with general commercial recommendation platforms,university libraries have a much higher number of resources than users,and its users and resources have clearer research directions and subject areas.This paper proposes that the recommender system of university library should be optimized on the basis of the general recommendation algorithm in view of the difference,so as to better play the performance of the recommender system under the resource environment of university library.This paper proposes an optimization scheme for a recommendation algorithm that combines clustering calculation and collaborative filtering.The K-Means clustering algorithm and the user-based collaborative filtering recommendation algorithm are integrated.Users’professional background is added to the extracted features,so that the algorithm can take into account the influence of the user’s professional background on their literature needs.So that relatively more accurate recommendations can be provided in the situation of university library system where user data is not abundant,After the users are clustered,there is no need to find the nearest neighbor users in the range of all users.The similarity calculation within the cluster reduces the calculation time and improves the efficiency of the recommender system to a certain extent.This paper uses the 2019 yearround borrowing records of the Central China Normal University Library as example data to compare the recommendation results of the general collaborative filtering recommendation algorithm and the optimization algorithm proposed in this paper to verify the actual effect of the optimization scheme.Finally,by comparing the recommendation results of the two recommendation methods,the accuracy and computational efficiency of the recommendation results of the optimization algorithm proposed in this paper are higher than those of the general collaborative filtering recommendation algorithm,which proves the feasibility of this optimization scheme.There are two main points of innovation in this research:(1)Taking into account the difference between the resource environment and users of university libraries and general commercial recommendation platforms,and considering the influence of this difference on the needs of users of university libraries,and a targeted recommendation system optimization method is proposed.(2)The fusion of clustering algorithm and collaborative filtering algorithm can not only integrate the subject domain characteristics of users and books into recommendation to improve the accuracy of recommendation,but also can reduce the calculation time required for recommendation to a certain extent and improve the performance of the recommendation system.