The Design of Graphical Process Modeling Languages: from Free Composition to Modular Construction

A graphical Process Modeling Language (PML) is a language tailored for modeling software systems by means of process models. It is said to be graphical because the primary representation of models are diagrams obtained combining visual constructs and previously defined components. Graphical PMLs are interesting as they open the design space to new geometric representations of complex interrelated aspects like concurrency and interaction. A Process-Aware Information System (PAIS) is a software system driven by explicit process models with the aim to coordinate and support agents in performing their activities. It is responsible for managing several process model instances at the same time balancing the available resources. A PML is the primary interface of a PAIS and a main concern in its design, because it is used by end-users, consultants, and developers for understanding, implementing and enacting complex processes. The adoption of PAIS technology may be severely limited by the weakness of PMLs in describing complex use cases. The overall aim of this thesis is to improve the design of graphical PMLs in order to engineer more effective PAISs. This goal is pursued following three intertwined paths: firstly, mainstream PMLs and their theoretical foundations are analyzed for exposing their features and limits; secondly, a widespread PML verification method is consolidated and then extended with a novel technique for automating process correction; finally, an alternative PML design solution is explored through a proof-of-concept language, called NestFlow, that improves both modularity and comprehensibility by providing a more structured modeling approach. A modular approach is only possible if data-flow dependencies are accepted as a main concern in PML design. NestFlow tries to ease the modeling activity by providing a comprehensive set of tightly integrated control-flow and data-flow constructs, promoting the latter as first-class citizens in process modeling.