Holistic and Compact Selectivity Estimation for Hybrid Queries over RDF Graphs

Many RDF descriptions today are text-rich: besides struc- tured data they also feature much unstructured text. Text-rich RDF data is frequently queried via predicates matching structured data, combined with string predicates for textual constraints (hybrid queries). Evaluating hybrid queries eficiently requires means for selectivity estimation. Previous works on selectivity estimation, however, sufer from inherent drawbacks, which are reflected in eficiency and efectiveness issues. We propose a novel estimation approach, TopGuess, which exploits topic models as data synopsis. This way, we capture correlations between structured and unstructured data in a holistic and compact manner. We study TopGuess in a theoretical analysis and show it to guarantee a linear space complexity w.r.t. text data size. Further, we show selectivity estimation time complexity to be independent from the synopsis size. In experiments on real-world data, TopGuess allowed for great improvements in estimation accuracy, without sacrificing eficiency.