A Formal Framework for Process Interoperability in Dynamic Collaboration Environments. (Un cadre formel pour l'interopérabilité des processus dans les environnements collaboratifs dynamiques)

Designing complex products such as aircrafts, helicopters and launchers must rely on well-founded and standardized processes. These processes should be executed in the context of dynamic cross-organizational collaboration environments. In this dissertation, we present a formal framework that ensures sustainable interoperability for cross-organizational processes in dynamic environments. We propose a declarative modeling language to define contracts that capture the objectives of each partner in the collaboration. Contract models built using this language under-specify the objectives of the collaboration by limiting the details captured at design-time. This under-specification decreases the coupling between partners in the collaboration. Nevertheless, less coupling leads to the creation of mismatches when partners’ processes will exchange messages at run-time. Accordingly, we develop an automatic mediation algorithm that is well adapted for dynamic environments. We conduct a thorough evaluation of this algorithm in the context of dynamic environments and compare it with existing mediation approaches which will prove its efficiency. We then extend our framework with a set of management operations that help realize the modifications on the collaboration environment at run-time. We develop an algorithm that assesses the impact of modifications on the partners in the collaboration environment. Then, this algorithm decides if the modification can be realized or should be postponed to wait for appropriate conditions. In order to figure out how to reach these appropriate conditions, we use the planning graph algorithm. This algorithm determines the raw set of management operations that should be executed in order to realize these conditions. A raw set of management operations cannot be executed by an engine unless its operations are encapsulated in the right workflow patterns. Accordingly, we extend this planning algorithm in order to generate an executable workflow from the raw set of operations. We evaluate our extension against existing approaches regarding the number and the nature of workflow patterns considered when generating the executable workflow. Finally, we believe that monitoring contributes in decreasing the coupling between partners in a collaboration environment. Accordingly, we extend our framework to support monitoring queries. We define a new execution plan for these queries that is more efficient. We conduct a thorough evaluation for this plan and compare it with existing industrial implementation. We implement a prototype for our framework based on standardized software components and illustrate it through a real world use case from the European project IMAGINE1.