A Typed Model for Linked Data

The term Linked Data is used to describe ubiquitous and emerging semi-structured data formats on the Web. URIs in Linked Data allow diverse data sources to link to each other, forming a Web of Data. A calculus which models concurrent queries and updates over Linked Data is presented. The calculus exhibits operations essential for declaring rich atomic actions. The operations recover emergent structure in the loosely structured Web of Data. The calculus is executable due to its operational semantics. A light type system ensures that URIs with a distinguished role are used consistently. The main theorem verifies that the light type system and operational semantics work at the same level of granularity, so are compatible. Examples show that a range of existing and emerging standards are captured. Data formats include RDF, named graphs and feeds. The primitives of the calculus model SPARQL Query and the Atom Publishing Protocol. The subtype system is based on RDFS, which improves interoperability. Examples focus on the SPARQL Update proposal for which a fine grained operational semantics is developed. Further potential high level languages are outlined for exploiting Linked Data