Title: Optimizing Temporal Datalog Materialisation for Composite Event Detection

Abstract: Numerous applications require the prompt identification of critical scenarios, including safety hazards and transparency issues, within high-speed streams of symbolic events. This need has driven the creation of (i) event specification languages that define complex events through temporal patterns built on simpler ones, and (ii) stream reasoning frameworks designed to evaluate these patterns. However, because event specification languages are often examined in isolation, comparing their expressivity becomes difficult, and the full capabilities of their corresponding stream reasoners remain unclear. To address this challenge, we translate practical fragments of leading event specification languages into Temporal Datalog->-, a variant of temporal Datalog featuring stratified negation and excluding future dependencies. To facilitate efficient stream reasoning over this logic, we introduce Streaming Trigger Graphs, an enhancement of a current best-practice method for Datalog materialisation. This strategy provides a standardized mechanism for composite event recognition, offering the potential to extend across a broad spectrum of practical event specification languages.

Source: arXiv Generated at: 2026-06-03 00:00:00 UTC