Hypergraphx-data launches to fill gap in multi-body network datasets

Researchers led by Quintino Lotito have launched hypergraphx-data, a curated repository of real-world hypergraph datasets designed for higher-order network analysis. Traditional repositories like SNAP and Netzschleuder focus primarily on pairwise interactions, leaving a gap for systems where multiple nodes interact simultaneously—such as co-authorship groups, protein complexes, or financial trading clusters. Hypergraphx-data addresses this by offering datasets from diverse domains including social networks, biology, and finance, and supports advanced configurations: weighted, directed, temporal, and multiplex hypergraphs.

Each dataset is stored in an open JSON format (plus a binarized version for the Hypergraphx library), ensuring accessibility and ease of use. Integrity and reproducibility are guaranteed through hash-based verification and data versioning. The repository also provides a user-friendly interface for browsing and filtering, making it easier for researchers in network science and machine learning to benchmark models, validate algorithms, and conduct empirical studies on complex many-body interactions. The work is published in the Journal of Complex Networks.