Triple pattern join cardinality estimations over HDT with enhanced metadata
Elena Wössner, Chang Qin, Javier D. Fernández, Maribel Acosta · Repository KITopen (Karlsruhe Institute of Technology) · 2019
In this work, we present HDT-stats, an extension to the HDT operations, to compute further metadata when evaluating triple patterns over RDF graphs represented with HDT. Then, we propose a novel model that relies on the HDT-stats metadata, as well as the distinct position of SPARQL variables, to estimate the cardinality of joins between triple patterns. Our preliminary results suggest that our approach produce smore accurate cardinality estimations than existing solutions.