Aggregates are all you need (to bridge stream processing and Complex Event Recognition)

Paper in proceeding, 2024

Emerging as an alternative to databases for continuous data processing, stream processing has evolved significantly since its inception in the early 2000s, leading to the emergence of numerous Stream Processing Engines (SPEs). Two main approaches exist to define streaming applications: to explicitly define graphs of common operators (Filters, Maps, Joins, and Aggregates) as the Dataflow model prescribes, or to express patterns of interest based on observations of low-level events within the domain under analysis, known as Complex Event Recognition (CER). Motivated by SPEs' semantic overlap, recent research has shown Aggregates suffice for an SPE to be as semantically expressive as other SPEs. However, a question remains open: Do Aggregates possess the semantic expressiveness required to cover CER too? We address this question formally demonstrating they indeed hold such semantic expressiveness.