Semantic Web applications often require the partitioning of triples into subgraphs, and then associating them with useful metadata (e.g., provenance). This led to the introduction of RDF datasets, with each RDF dataset comprising a default graph and zero or more named graphs. However, due to differences in RDF implementations, no consensus could be reached on a standard semantics; and a range of different dataset semantics are currently assumed. For an RDF system not be limited to only a subset

more »

... of online RDF datasets, the system would need to be extended to support different dataset semantics-exactly the problem that eluded consensus before. In this paper, we transpose this problem to Notation3 Logic, an RDF-based rule language that similarly allows citing graphs within RDF documents. We propose a solution where an N3 author can directly indicate the intended semantics of a cited graphpossibly, combining multiple semantics within a single document. We supply an initial set of companion N3 rules, which implement a number of RDF dataset semantics, which allow an N3-compliant system to easily support multiple different semantics.