Ranking academic institutions by means of institution–publication networks

Ranking academic institutions is a crucial aspect of scientometric research and has been an attractive topic. However, existing methods for measuring the reputation of institutions do not adequately consider the interconnected relationship between multiple scientific agents, such as papers and institutions, which limits the robustness and accuracy of the evaluation results. To address this issue and accurately identify influential academic institutions, we propose a novel heterogeneous ranking method by means of interconnected institution–publication networks. Firstly, we construct an institution–publication network consisting of an institution layer and a paper layer to capture the interconnected relationship between institutions and papers. And then, we propose a novel ranking method based on random walks on top of the institution–publication network. Each layer has its own random jump probability, and there is an additional interlayer jump probability to depict the interdependence between collaboration and citation. Finally, we conduct extensive experiments on large-scale empirical data from American Physical Society journals. The results demonstrate that the proposed method, HRank, performs well in identifying influential institutions, predicting the increment of citations, and improving robustness against malicious manipulation.