On the Synthetic Generation of Semantic Web Schemas

In order to cope with the expected size of the Semantic Web (SW) in the coming years, we need to benchmark existing SW tools (e.g., query language interpreters) in a credible manner. In this paper we present the first RDFS schema generator, termed PoweRGen, which takes into account the morphological features that schemas frequently exhibit in reality. In particular, we are interested in generating synthetically the two core components of an RDFS schema, namely the property (relationships between classes or attributes) and the subsumption (subsumption relationships among classes) graph. The total-degree distribution of the former, as well as the out-degree distribution of the Transitive Closure (TC) of the latter, usually follow a power-law. PoweRGen produces synthetic property and subsumption graphs whose distributions respect the power-law exponents given as input with a confidence ranging between 90− 98%.