A design space for RDF data representations

Abstract RDF triplestores’ ability to store and query knowledge bases augmented with semantic annotations has attracted the attention of both research industry. A multitude systems offer varying data representation indexing schemes. However, as recently shown for designing structures, many design choices are biased by outdated considerations may not result in most efficient a given workload. To overcome this limitation, we identify novel three-dimensional space. Within space, map trade-offs between different representations employed part an triplestore unexplored solutions. We complement review empirical evaluation ten standard SPARQL benchmarks examine prevalence these access patterns synthetic real workloads. find some patterns, be prevalent workloads under-supported existing triplestores. This shows capabilities our model used designers reason about allow (possibly artificially intelligent) designer evaluate fit system