VLog: A Column-Oriented Datalog System for Large Knowledge Graphs
Jacopo Urbani, Ceriel J. H. Jacobs, Markus Krötzsch
VLog: A Column-Oriented Datalog System for Large Knowledge Graphs
Abstract. We present VLog, a new system for answering arbitrary Datalog queries on top of a wide range of databases, including both relational and RDF databases. VLog is designed to perform efficiently intensive rule-based computation on large Knowledge Graphs (KGs). It adapts column-store technologies to attain high efficiency in terms of memory usage and speed, enabling us to process Datalog queries with thousands of rules over databases with hundreds of millions of tuples—in a live demonstration on a laptop. Our demonstration provides in-depth insights into the workings of VLog, and presents important new features such as support for arbitrary relational DBMS.
Published at ISWC2016 Posters and Demos (Workshop paper)
Download PDF (last update: 11 Nov 2016)
Citation details
- Jacopo Urbani, Ceriel J. H. Jacobs, Markus Krötzsch. VLog: A Column-Oriented Datalog System for Large Knowledge Graphs. In Proceedings of the 15th International Semantic Web Conf. (ISWC-16), Posters and Demos. CEUR-WS.orgProperty "Publisher" has a restricted application area and cannot be used as annotation property by a user. 2016.
author = {Jacopo Urbani and Ceriel J. H. Jacobs
and Markus Kr\"{o}tzsch},
title = {{VLog}: A Column-Oriented {Datalog} System
for Large Knowledge Graphs},
editor = {Takahiro Kawamura and Heiko Paulheim},
booktitle = {Proceedings of the 15th International
Semantic Web Conference (ISWC'16),
Posters and Demos},
series = {CEUR Workshop Proceedings},
volume = {1690},
publisher = {CEUR-WS.org},
year = {2016}
}
Remarks
This is a short demo of the Datalog reasoner VLog. The underlying research is described in detail in the AAAI 2016 publication Column-Oriented Datalog Materialization for Large Knowledge Graphs, which should be cited instead.
