Radon Ð Rapid Discovery of Topological Relations

Authors: Mohamed Sherif, Kevin Dre§ler, Panayiotis Smeros, Axel-Cyrille Ngonga Ngomo

AAAI 2017 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Our evaluation shows that we outperform the state of the art significantly and by several orders of magnitude. 4 Evaluation
Researcher Affiliation Academia a Department of Computer Science, University of Leipzig, 04109 Leipzig, Germany {sherif|dressler|ngonga}@informatik.uni-leipzig.de b EPFL, BC 142, Station 14, CH-1015 Lausanne, Switzerland panayiotis.smeros@epfl.ch
Pseudocode Yes Algorithm 1: Radon
Open Source Code No A more complete list of results can be obtained from the project website8. http://aksw.org/Projects/LIMES. This link is to a project website (LIMES) for obtaining results, not explicitly stating that the source code for Radon (the methodology described in this paper) is available there.
Open Datasets Yes We evaluated our approach using two real-world datasets. (1) NUTS9 is manually curated by the Eurostat group of the European Commission. (2) CORINE Land Cover or simply CLC is an activity of the European Environment Agency... http://nuts.geovocab.org/data/0.91/ ... https://datahub.io/dataset/corine-land-cover
Dataset Splits No The paper evaluates Radon on the NUTS and CLC datasets, but does not provide specific train/validation/test split percentages, sample counts, or mention cross-validation. It uses
Hardware Specification Yes All experiments were carried out on a 64-core 2.3 GHz PC running Open JDK 64-Bit Server 1.7.0 75 on Ubuntu 14.04.2 LTS. Each experiment was assigned 20 GB RAM and a timeout limit of 2 hour.
Software Dependencies Yes All experiments were carried out on a 64-core 2.3 GHz PC running Open JDK 64-Bit Server 1.7.0 75 on Ubuntu 14.04.2 LTS. For Silk experiments, we ran our experiments using its latest version (v2.6.1)... For Strabon, we also used the latest version (v3.2.10) with the accordingly tuned Postgre SQL (v9.1.13) and Post GIS (v2.0)
Experiment Setup Yes Based on these results, we used the average heuristic as the granularity selection policy in the rest of the experiments. For load balancing in Radon, we used the simple round robin load balancing policy (Shreedhar and Varghese 1996) with chunks size of 1000. For Silk experiments, we ran our experiments using its latest version (v2.6.1) with a blocking factor of 10 as in (Smeros and Koubarakis 2016).