This seminar explores the application of hyperlink prediction on directed hypergraphs constructed from the ”see also” references within encyclopedias focused on protests and social movements. We introduce a methodology for transforming the directed graph of citations into a directed hypergraph, where each encyclopedia entry is a node, and the set of its predecessors and successors form the tail and head of a hyperedge, respectively. We then employ the CHESHIRE (Complete Hypergraph Embedding for Link Prediction) algorithm, a deep learning framework utilizing Chebyshev spectral graph convolutions, to predict missing hyperlinks. This task is framed as a machine learning classification problem. We evaluate the performance of CHESHIRE on six different encyclopedias, presenting and discussing the results of 5-fold cross-validation. Our findings demonstrate the viability of using hypergraph-based deep learning models for uncovering latent relationships in the complex, interconnected knowledge structures of specialized encyclopedias. 

- § - 

Short Bio: Moses Boudourides is a retired Professor of Applied and Computational Mathematics from the University of Patras, Greece. Since his retirement in 2017, he has held visiting appointments at Linköping University in Sweden (Institute for Analytical Sociology, 2019), New York University Abu Dhabi (Science Division, 2019–2020), Haverford College (Department of Computer Science, 2021–2022), and Arizona State University (School of Public Affairs and School of Computing and Augmented Intelligence, 2022–2023). He has also been serving as an instructor in the Data Science Program at Northwestern University’s School of Professional Studies from 2019 through 2025. Earlier in his career, from 1982 to 1997, he was Associate Professor at the Department of Electrical and Computer Engineering of the Democritus University of Thrace in Xanthi, Greece, and Visiting Professor at the University of California, Irvine (Department of Mathematics, 1989–1990). He received his PhD from Johns Hopkins University, following undergraduate studies in Chemical Engineering at the National Technical University of Athens. In 2007, he organized the Sunbelt Social Networks Conference in Corfu, Greece. His current research focuses on higher-order social network analysis, computational social science, hypergraph analysis and hyperedge prediction, as well as the study of fair and unbiased algorithmic systems and responsible AI.