Ontological Reasoning over Shy and Warded Datalog+/- for Streaming-based Archite...
summary
Claims novel streaming-friendly chase variants for Shy and Warded Datalog+/- and implements them in Vadalog, replacing full materialization with parsimonious chase using path-matching to avoid redundant null-generated facts and an isomorphism check via null canonicalization hashed into a set. Introduces a per-fact resumption by freezing nulls up to the theoretical bound to recover completeness. Experiments show streaming resumption adds little overhead on a real-world supply-chain benchmark, is essential to avoid exponential harmful-join elimination on the ancestor example, and matches or narrows the gap with materialization engines as data grows.
AI-written summary · about
chapters (12)
- 0:00Introduction
- 1:26Datalog+/- with existential quantification
- 3:12Chase procedure in ontological reasoning
- 5:15Shy Datalog and parsimonious chase
- 6:41Warded Datalog and isomorphic chase
- 7:25Streaming-based approach contributions
- 9:53Hash-based isomorphism check
- 11:57Vadalog Reasoner streaming architecture
- 13:40Resumption mechanism for completeness
- 18:19Experimental evaluation and results
- 20:34Conclusion and future directions
- 21:21Q&A
related talks
52mEnhancing Datalog for Datomic
13mAccelerating Program Analyses in Datalog by Merging Library Facts
52mI See What You Mean
32mReactive Datalog for Datomic
01h 01mLogic and Algebras for Distributed Computing: Applications and Open Questions
36mOptimizing Relational Programs II
15mIncremental Processing of Structured Data in Datalog
42mData in Motion: Streaming Static Data Efficiently in Akka Persistence