Expressivity of Planning with Horn Description Logic Ontologies
Authors: Stefan Borgwardt, Jörg Hoffmann, Alisa Kovtunova, Markus Krötzsch, Bernhard Nebel, Marcel Steinmetz5503-5511
AAAI 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We therefore conclude the paper with an experimental evaluation that combines an existing practical implementation of an (exponential) Datalog -rewriting for Horn-SHIQ (Eiter et al. 2012) with our generic compilation scheme, compares this against previous approaches for DLLite e KABs on existing benchmarks (Calvanese et al. 2016; Borgwardt et al. 2021), and also introduces new benchmarks exploiting the newly increased expressivity. All experiments were run on a computer with an Intel Core i5-4590 CPU@3.30GHz, and run time and memory cutoffs of 600s and 8GBs, respectively. |
| Researcher Affiliation | Academia | 1Faculty of Computer Science, Technische Universit at Dresden, Germany, 2Saarland University, Saarland Informatics Campus, Germany 3Faculty of Engineering, University of Freiburg, Germany |
| Pseudocode | No | The paper does not contain structured pseudocode or algorithm blocks. |
| Open Source Code | Yes | The compiler and the benchmarks with a detailed description are available online.4 [Footnote 4: https://gitlab.perspicuous-computing.science/a.kovtunova/pd dl-horndl] |
| Open Datasets | Yes | Our benchmark collection consists of 125 instances adapted from existing DL-Lite e KAB benchmarks (Calvanese et al. 2016; Borgwardt et al. 2021), and 110 newly created instances. The compiler and the benchmarks with a detailed description are available online.4 [Footnote 4: https://gitlab.perspicuous-computing.science/a.kovtunova/pd dl-horndl] |
| Dataset Splits | No | The paper does not provide specific training/validation/test dataset splits. It evaluates a planning system on benchmark instances, which are treated as problems to be solved rather than a dataset to be partitioned for model training and validation. |
| Hardware Specification | Yes | All experiments were run on a computer with an Intel Core i5-4590 CPU@3.30GHz, and run time and memory cutoffs of 600s and 8GBs, respectively. |
| Software Dependencies | Yes | For the experiments, we use the Fast Downward (FD) planning system (Helmert 2006) version 20.06 (the newest version as of August 2021)... The Datalog rewriting is generated with CLIPPER (Eiter et al. 2012). |
| Experiment Setup | Yes | We ran FD with a dualqueue greedy best-first search using the h FF heuristic, a commonly used baseline in the planning literature. |