Teaching machines to think like conservators: Improving decision making in strategic preservation planning through machine learning algorithms

A Danish Research group is looking into the use of machine learning to improve the preservation of cultural heritage. Our hypothesis is that we can utilize machine
learning to improve the analysis of preservation related data, thereby creating a better basis for decision making and preservation planning in archives, libraries and museums. Machine learning is the science of enabling computers to learn without being directly programmed. It applies algorithms – a kind of recipe for solving mathematical problems – that are trained on existing data, and then used to make predictions for new datasets. For example, machine learning is used to support diagnosis of diseases in medicine. Development of preservation strategies for cultural heritage objects requires deep insight into the interactions between the materiality of objects, and the way the surrounding environment affects the chemical and physical degradation. Likewise, it requires a good understanding of the social, political and economic context in which the actual preservation planning takes place – the significance of the objects, who the stakeholders are, and intended use scenarios. Finally, it requires solid knowledge about the utility and costs of possible preservation efforts. In order to strengthen this complex analysis of multiple data sources, we want to bring in machine learning, and explore how it can help improve decision making related to the preservation of cultural heritage. As the project is in its upstart phase, we hope this presentation will enable us to reach out to potential collaborators and organisations holding relevant data.