Answering Conjunctive Queries over EL Knowledge Bases with Transitive and Reflexive Roles

Authors: Giorgio Stefanoni, Boris Motik

AAAI 2015 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Our preliminary evaluation suggests that the algorithm can be suitable for practical use. To evaluate the feasibility of our approach, we implemented a prototypical CQ answering system and we carried out a preliminary evaluation. We implemented our algorithm in a prototypical system, and we conducted a preliminary evaluation with the goal of showing that the number of consequences of DK is reasonably small, and that the nondeterminism of the filtering procedure is manageable.
Researcher Affiliation Academia Giorgio Stefanoni and Boris Motik Department of Computer Science, University of Oxford Wolfson Building, Parks Road, Oxford, OX1 3QD, UK
Pseudocode Yes Algorithm 1: is Sound(q, DK, τ)
Open Source Code Yes Our system, the test data, and the queries are all available online.1 http://www.cs.ox.ac.uk/isg/tools/EOLO/
Open Datasets Yes We tested our system using the version of the LSTW benchmark (Lutz et al. 2013) by Stefanoni, Motik, and Horrocks (2013). ... We used the data generator provided by LSTW to generate KBs U5, U10, and U20 of 5, 10, and 20 universities, respectively. Our system, the test data, and the queries are all available online.1 http://www.cs.ox.ac.uk/isg/tools/EOLO/
Dataset Splits No The paper mentions generating KBs U5, U10, and U20 for testing, but it does not specify any training/validation/test dataset splits, percentages, or absolute counts.
Hardware Specification Yes We ran our tests on a Mac Book Pro with 4GB of RAM and a 2.4Ghz Intel Core 2 Duo processor.
Software Dependencies No The paper states 'Our prototype uses the RDFox (Motik et al. 2014) system', but it does not specify a version number for RDFox or any other key software components.
Experiment Setup No The paper describes the overall evaluation setup including the use of RDFox and the LSTW benchmark, but it does not provide specific details such as hyperparameters, optimizer settings, or other fine-grained experimental configuration details typically found in an 'experimental setup' section for training models.