The seismic assessment of existing unreinforced masonry structures is particularly complex. Defining the correct modelling assumptions is essential when using global models to ensure valid results. Achieving this often requires the collaboration of a group of stakeholders with diverse backgrounds who can thoroughly study the structure under consideration. Field-collected data must then be compared with existing literature and regulations before proceeding to the computational model. This phase is particularly labour-intensive, and errors, data loss, or duplication are common pitfalls. The advent of new digital data management methods can improve this methodology. Specifically, a linked data approach based on web ontology language can enhance interoperability between different research areas and enable the formal and comprehensive representation of data to facilitate informed decision-making. This article presents a new method based on linked data for defining modelling assumptions for analytical models used in the seismic analysis of existing unreinforced masonry buildings. Two complementary ontologies are proposed: the Historic Masonry Ontology and the Failure Masonry Ontology. The former defines the mechanical properties of masonry material, while the latter defines the most plausible collapse modes evidenced by earthquakes. In particular, this is achieved through Semantic Web Rules Language (SWRL), which interprets geometric and material data introduced into the ontology. The methodology is successfully applied in a real case study.