Library academic resource recommendation and association analysis using GraphSAGE

In the context of the rapid development of digital academic resources, it is a great challenge to effectively provide researc hers and students with appropriate academic resources. In this paper, we intend to build a graph network for academic resources based on GraphSAGE (Graph Sample and Aggregated), and use it as a basis to study the node-based feature learning method improve the accuracy and relevance of the recommendation system. First, the graph network structure is initially established the basis of analyzing the existing electronic information resources such as books and theses in the library. Secondly, based on the null domain-based algorithms such as GraphSAGE, the corresponding feature vectors are extracted from the information such as the association relationship and similarity between data sources. Meanwhile, this paper will also improve the node expression in the graph network by combining the behavioral data such as user habits and online interactions, so as to achiev e more personalized and accurate recommendations. Users with ser_ID (Identification) of 2 demonstrated 80 engagement on a resource with Resource_ID of 2, although the popularity of the resource was only 19. The research results in this paper not improve the efficiency of library recommendation of academic literature, but also provide deep mining of literature The research results of this paper will realize more efficient and accurate academic support for university libraries in the of digital transformation.