Cultural heritage and disasters risk: A machine-human coupled analysis

Natural hazards represent a major threat to cultural heritage. Literature has analyzed this nexus using different approaches depending on their focus. To provide a comprehensive understanding of the core pillars structuring the field, we use a machine-human methodology that combines bibliometric and machine-learning text analysis. We focus on a sample of 565 peer-reviewed documents published between 1988 and 2020. Results prove there is increasing interest in the topic, covering different types of hazards depending on the area of interest and its most frequently associated risks. To enhance the granularity of the analysis we apply machine learning to the publications abstracts and we classify documents based on their core topics. We find that the field is highly diverse and includes conservation, restoration and management of historical sites and cultural heritage. Scholars use sophisticated tools and innovative methodologies to account for this heterogeneity. We highlight the need for stronger interdisciplinarity in the field and we call for further progresses in spatial-explicit analysis. Finally, we point towards more inclusion of humanities in the area to account for the cultural aspects of heritage protection.