Prioritizing motion pictures for archival preservation using a decision-aiding Bayesian network

Abstract

The National Archives and Records Administration (NARA) preserves and provides U.S. federal government records of historic and cultural value, including approximately 200,000 reels of 35 mm black-and-white motion picture film. Of immediate concern is allocating finite resources over a 5-year program to preserve the most at-risk, frequently used, and valuable films. To aid this need, we developed a decision-modeling approach based on film deterioration and acidity conditions, known or anticipated access requests, archival value, and other attributes. We used a Bayesian network parameterized with NARA’s holdings inventory, a database compiled through laboratory inspection of individual films, and expert judgement of NARA archival and preservation staff. The model aids identifying film series, elements, and reels with high risk ratings and high priority levels, and priorities and risks given film series and element categories, and provides ancillary information for further prioritization decisions and actions, and can be incrementally improved with new data.

Citation

Marcot, Bruce G.; Ormsby, Mark. 2025. Prioritizing motion pictures for archival preservation using a decision-aiding Bayesian network. npj Heritage Science. 13: 376. https://doi.org/10.1038/s40494-025-01954-x.