bnRep: A repository of Bayesian networks from the academic literature

TL;DR

bnRep addresses the scarcity of documented Bayesian-network repositories by delivering an open-source R package with a large, well-documented collection of networks. It integrates with bnlearn to provide interoperable BN objects and includes a Shiny app for interactive exploration, facilitating benchmarking, replication, and education in BN research. The repository currently houses 214 BNs from 150+ publications across discrete, Gaussian, and CFL types, accompanied by a detailed bnRep_summary and export capabilities. Looking ahead, bnRep plans to expand to additional BN types (e.g., copula, additive, dynamic, continuous-time) and to introduce benchmarking tools through community contributions, strengthening cross-domain BN research and practice.

Abstract

Bayesian networks (BNs) are widely used for modeling complex systems with uncertainty, yet repositories of pre-built BNs remain limited. This paper introduces bnRep, an open-source R package offering a comprehensive collection of documented BNs, facilitating benchmarking, replicability, and education. With over 200 networks from academic publications, bnRep integrates seamlessly with bnlearn and other R packages, providing users with interactive tools for network exploration.