Academic Literature Recommendation Experimental Dataset

The study includes three datasets, with details as follows:WoSCite Dataset: This dataset is constructed based on 70 core journals in Web of Science (WoS) related to statistics, econometrics, and computer science. It includes metadata such as article titles, abstracts, journal names, publication years, and citation information, but does not contain user–item interaction data.CiteULike Dataset: A widely used public dataset in personalized recommendation research, CiteULike captures real user interactions with academic publications. The dataset collected from both CiteULike and Google Scholar, and includes titles, abstracts, user-assigned tags, and user–item interaction records.ICF Dataset: This is a custom-built dataset developed based on the "Intelligent Data Center for Cryosphere Science and Frozen Ground Engineering" platform. It provides comprehensive bibliographic metadata and records diverse user–item interactions with explicit timestamps.