Extended caching, backjumping and merging for expressive description logics

  • Authors:
  • Andreas Steigmiller;Thorsten Liebig;Birte Glimm

  • Affiliations:
  • Ulm University, Ulm, Germany;derivo GmbH, Ulm, Germany;Ulm University, Ulm, Germany

  • Venue:
  • IJCAR'12 Proceedings of the 6th international joint conference on Automated Reasoning
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

With this contribution we push the boundary of some known optimisations such as caching to the very expressive Description Logic $\mathcal{SROIQ}$. The developed method is based on a sophisticated dependency management and a precise unsatisfiability caching technique, which further enables better informed tableau backtracking and more efficient pruning. Additionally, we optimise the handling of cardinality restrictions, by introducing a strategy called pool-based merging. We empirically evaluate the proposed optimisations within the novel reasoning system Konclude and show that the proposed optimisations indeed result in significant performance improvements.