Construction of Personalized Recommendation System of University Library Based on SOM Neural Network

Unreasonable resource classification and imperfect resource retrieval mechanism are important issues in the digital construction of university libraries. Based on the characteristics of SOM neural network clustering algorithm with no parameters, high accuracy and strong objectivity, the article firstly clusters and optimizes the Web access behavior of the library users of the Nanfang College of Sun Yat-sen University. Secondly, based on the output of user analysis results, the user's personal feature information, user behavior data, and literature databases and other related data resources are filtered and integrated to form a related data set with higher reliability and availability, combined with semantic retrieval and attribute value matching. Technology to build a personalized recommendation service system for university library users. Finally, the effectiveness of the system is verified, and the coordination of the three subsystems of library theme recommendation, book recommendation and expert recommendation is realized. Through the calculation of the correlation between the user and the characteristics of the document resources, the user's points of interest and the cluster set where they are located are further identified.