TWed Talk: SQuARE, the SPARQL Query Agent-based Reasoning Engine

Printer-friendly version

When: April 7 2021
Where: Webex

We describe SQuARE, the SPARQL Query Agent-based Reasoning Engine, a deductive inference engine implemented by encoding each OWL Description Logic (DL) axiom as a SPARQL CONSTRUCT query. While earlier work has proposed SPARQL extensions for various purposes, we find that the complete queries used in SPARQL-based reasoning methods are not publicly available. We make the queries, software, and validation ontology that we use for our methodology openly available. Encoded in this validation ontology are RDF examples for each OWL-DL axiom, allowing for the testing of our reasoning engine at a per axiom basis. We describe our approach in terms of implementation, query formulation, and validation. We evaluate the DL complexity of our implementation versus existing reasoners. We further discuss future research directions related to the application of this approach for clinical reasoning. Our approach allows for inference engine rule customization as well as the ability to reason over inconsistent knowledge. This work is motivated by potential applications involving hybrid reasoning, distributed reasoning, and explainability in reasoning by embedding provenance in the form of nanopublications.