This paper presents an approach to semantic modelling of large-scale trust ecosystems to improve the interpretations of trust claims. The problem of interpreting trust claims is described and relevant types of reasoning are analysed. A model based on $$\mathcal {SROIQ}$$ and OWL Description Logic is proposed. The novel elements are the creation of classes and properties on the basis of legal and regulatory sources that extend existing vocabularies (W3C, Dublin Core), and the use of these classes and properties to create assertions that represent information harvested from on-line information sources. The resulting model allows automated classification via a reasoner, as well as queries that support use cases from various actors. A general approach is presented, as well as results from a prototype implementation based on the European eIDAS and US FICAM trust ecosystems.