Efficient Model Construction for Horn Logic with VLog

Jacopo UrbaniJacopo Urbani, Markus KrötzschMarkus Krötzsch, Ceriel JacobsCeriel Jacobs, Irina DragosteIrina Dragoste, David CarralDavid Carral

Jacopo Urbani, Markus Krötzsch, Ceriel Jacobs, Irina Dragoste, David Carral
Efficient Model Construction for Horn Logic with VLog
In Didier Galmiche, Stephan Schulz, Roberto Sebastiani, eds., Proceedings of the 8th International Joint Conference on Automated Reasoning (IJCAR 2018), volume 10900 of LNCS, 680--688, 2018. Springer

Details
Bibtex

KurzfassungAbstract
We extend the Datalog engine VLog to develop a column-oriented implementation of the skolem and the restricted chase -- two variants of a sound and complete algorithm used for model construction over theories of existential rules. We conduct an extensive evaluation over several data-intensive theories with millions of facts and thousands of rules, and show that VLog can compete with the state of the art, regarding runtime, scalability, and memory efficiency.
Bemerkung: Note: Erratum: This version fixes two typos that were present in the publication in Algorithm 1: line 1.5 now also includes a set difference to avoid adding existing facts (necessary since heads can be multi-atom); line 1.8 now uses the correct index (i instead of i+1)
Projekt:Project: Cfaed, DIAMOND, HAEC B08
Forschungsgruppe:Research Group: Wissensbasierte SystemeKnowledge-Based Systems

The final publication is available at Springer via http://dx.doi.org/10.1007/978-3-319-94205-6_44.

@inproceedings{UKJDC2018,
  author    = {Jacopo Urbani and Markus Kr{\"{o}}tzsch and Ceriel Jacobs and
               Irina Dragoste and David Carral},
  title     = {Efficient Model Construction for Horn Logic with {VLog}},
  editor    = {Didier Galmiche and Stephan Schulz and Roberto Sebastiani},
  booktitle = {Proceedings of the 8th International Joint Conference on
               Automated Reasoning (IJCAR 2018)},
  series    = {LNCS},
  volume    = {10900},
  publisher = {Springer},
  year      = {2018},
  pages     = {680--688},
  doi       = {10.1007/978-3-319-94205-6_44}
}