Flexible Process Compliance with Semantic Constraints Using Mixed‐Integer Programming

An adaptive process management system (APMS) allows for flexible, dynamic, and even ad hoc adaptation of processes based on case data, context, and events. These processes may arise in various domains such as business, healthcare, etc. In knowledge‐intensive environments, it is important that APMS technology ensures error‐free process execution and compliance with semantic constraints. However, most process design tools handle only syntactic constraints. This restricts their value in real‐world applications considerably. This paper proposes a novel approach to check the compliance of process models against semantic constraints and the validity of process change operations using mixed‐integer programming (MIP). The MIP formulation allows us to describe existential, dependency, ordering, and various other relationships among tasks along with business policies in a standard way. In addition to incorporating the semantic constraint specifications into an MIP formulation, we introduce three novel ideas in this paper: (1) the notion of degree of compliance of processes to constraints based on a penalty function, (2) the concepts of full and partial validity of change operations, and (3) the idea of compliance by compensation. Thus, compensation operations derived from compliance degree can transform a noncompliant process into a compliant one both at design and execution time. We illustrate our approach in the context of a healthcare workflow as a way to reduce medical errors and argue that it is more elegant and superior to a pure logic‐based approach. Complex scenarios with multiple concurrent processes (and constraints across them) for a single patient are also considered.