Trigger Graphs

Reasoning Over Billions of Triples Under Datalog

C++ engine to support sound and reason over knowledge graphs using Datalog rules with existentials. GLog, the Trigger Graphs engine, outperforms commercial and open-source engines, such as VLog, RDFox, Vadalog, WebPIE, and Inferray, by several orders of magnitude in terms of runtime, while incurring the same or less main memory overhead. Regarding RDFox, GLog substantially outperform it in terms of time and memory efficiency despite the fact that GLog reason using a single thread only: in on-device reasoning, GLog is more than 18x faster (1s vs 18.7s) than RDFox when RDFox uses 1 thread, and up to 6x faster (1s vs 6s) when RDFox uses 16 thread. Furthermore, the memory requirements of GLog are usually lower than those of RDFox. In terms of scalability, GLog allows materializing knowledge graphs with 17B facts in less than 40 min using a single machine with commodity hardware.

Other knowledge graphs in which GLog outperforms the state-of-the-art are ChaseBench, LUBM, UODBM, DBpedia, Claros, Reactome, and Yago.

GLog can be deployed on mobile phones under the Android NDK for reasoning running exclusively on-device.

Repository

GLog

Relevant publications

2021

  1. VLDB
    Materializing Knowledge Bases via Trigger Graphs
    Efthymia Tsamoura, David Carral, Enrico Malizia, and Jacopo Urbani
    Proceedings of the VLDB Endowment, 2021

2017

  1. PODS
    Benchmarking the Chase
    Michael Benedikt*, George Konstantinidis*, Giansalvatore Mecca*, Boris Motik*, Paolo Papotti*, Donatello Santoro*, and Efthymia Tsamoura*
    In Proceedings of the 36th ACM SIGMOD-SIGACT-SIGAI Symposium on Principles of Database Systems (PODS), Chicago, IL, USA, May 14-19, 2017