Monitoring business constraints with the event calculus

Abstract

Today, large business processes are composed of smaller, autonomous, interconnected subsystems, achieving modularity and robustness. Quite often, these large processes comprise software components as well as human actors, they face highly dynamic environments and their subsystems are updated and evolve independently of each other. Due to their dynamic nature and complexity, it might be difficult, if not impossible, to ensure at design-time that such systems will always exhibit the desired/expected behaviors. This, in turn, triggers the need for runtime verification and monitoring facilities. These are needed to check whether the actual behavior complies with expected business constraints, internal/external regulations and desired best practices. In this work, we present Mobucon EC, a novel monitoring framework that tracks streams of events and continuously determines the state of business constraints. In Mobucon EC, business constraints are defined using the declarative language Declare. For the purpose of this work, Declare has been suitably extended to support quantitative time constraints and non-atomic, durative activities. The logic-based language Event Calculus (EC) has been adopted to provide a formal specification and semantics to Declare constraints, while a light-weight, logic programming-based EC tool supports dynamically reasoning about partial, evolving execution traces. To demonstrate the applicability of our approach, we describe a case study about maritime safety and security and provide a synthetic benchmark to evaluate its scalability.

An Operational Decision Support Framework for Monitoring Business Constraints

“…More specifically, the formal semantics of Declare can be specified by using LTL [5,6], abductive logic programming with expectations [6], or the EC [7,8]. These characteristics are crucial for two reasons.…”

“…A comprehensive description of how the EC can be used in the Declare setting can be found in [8]. Here, we consider one of the constraints mentioned in Fig.…”

“…1 refer to the extension of the Declare language. Metric temporal aspects have been already incorporated into the Mobucon EC provider [8]. Metric temporal logics and timed-automata will be investigated to improve the LTL-based approach in this direction.…”

“…For the Mobucon EC provider, some complexity results are inherited from the seminal investigation by Chittaro and Montanari [13]. An initial investigation of the performance of this approach (with YAP Prolog as underlying reasoner) can be found in [8]. Differently from the LTL-based approach, whose most resourceconsuming task is the generation of the automaton, which is done before the execution, the EC-based approach triggers a reasoning phase every time a new event is acquired.…”

“…The language is declarative and graphical. Moreover, Declare has been formalized using a variety of logic-based frameworks, such as Linear Temporal Logic (LTL) with a finite-trace semantics 1 [5,6] and the Event Calculus (EC) [7,8]. Section 3 describes the architecture of our proposed framework.…”