Discovering Hierarchical Patterns in Event-Based Systems

Expressive and powerful event-pattern rules are key to successful applications of Complex Event Processing (CEP). However, the definition of complex pattern-detection logic may place heavy demands on business users. In this work we present a model for hierarchical event-patterns, allowing users to compose complex patterns from encapsulated, fully-abstracted sub-pattern definitions with a visual decision graph. Pattern definitions can be reused across business scenarios, thereby encouraging the creation of pattern libraries. In the course of this paper we give a brief overview of the event-based system SARI and describe a number of extensions to its ruling model. We present an algorithm for evaluating hierarchical ruling logic and demonstrate our approach with a use case from the fraud-detection domain.