On Constrained Open-World Probabilistic Databases


Nov 17, 2018 AKBC 2019 Conference Blind Submission readers: everyone Show Bibtex
  • Keywords: probabilistic databases
  • TL;DR: We propose adding constraints to open-world probabilistic databases
  • Abstract: Increasing amounts of available data have led to a heightened need for useful waysof representing large-scale probabilistic knowledge bases. One such approach is to use aprobabilistic database, a model with strong assumptions that allow for efficiently answeringmany interesting queries. Recent work on open-world probabilistic databases strengthensthe semantics of these probabilistic databases by discarding the assumption that any in-formation not present in the data must be false. In this paper we build upon open-worldprobabilistic databases, strengthening the semantics further by using constraints to restrictthis open world in a reasonable way. We establish a basic hardness result subject to theseconstraints, showing that the addition of constraints causes a fundamental change in thedifficulty of query evaluation and opening an interesting new theoretical direction. Finally, we propose an efficient approximation for a large class of interesting queries, providingbounds with their guaranteed tightness.
  • Archival status: Non-Archival
  • Subject areas: Databases, Reasoning
0 Replies