Semantic issues in model-driven management of information system interoperability

The MISE Project (Mediation Information System Engineering) aims at providing collaborating organizations with a Mediation Information System (MIS) in charge of supporting interoperability of a collaborative network. MISE proposes an overall MIS design method according to a model-driven approach, based on model transformations. This MIS is in charge of managing (i) information, (ii) functions and (iii) processes among the information systems (IS) of partner organizations involved in the network. Semantic issues are accompanying this triple objective: How to deal with information reconciliation? How to ensure the matching between business functions and technical services? How to identify workflows among business processes? This article aims first, at presenting the MISE approach, second at defining the semantic gaps along the MISE approach and third at describing some past, current and future research works that deal with these issues. Finally and as a conclusion, the very “design-oriented” previous considerations are confronted with “run-time” requirements.

💡 Research Summary

The paper presents the MISE (Mediation Information System Engineering) project, which aims to provide collaborating organizations with a Mediation Information System (MIS) that enables seamless interoperability of information, functions, and processes across partner information systems. The authors adopt a model‑driven engineering approach, defining a series of model transformations that bridge high‑level business models (processes, functional requirements, data concepts) with low‑level technical models (services, interfaces, data schemas). The design workflow consists of: (1) capturing collaboration goals and scenarios, (2) modeling business artifacts using a meta‑model, (3) applying transformation rules to generate technical specifications, and (4) implementing and deploying the MIS.

Three core semantic challenges are identified along this workflow. First, “information reconciliation” addresses the heterogeneity of data structures and vocabularies used by different partners. The solution relies on a central domain ontology that each partner’s data model maps onto, enabling automated concept alignment, conflict resolution policies, and schema translation. Second, “matching business functions to technical services” tackles the gap between abstract functional requirements and concrete service implementations. The authors propose a rule‑based matching framework that considers input/output type compatibility, pre‑ and post‑conditions, and non‑functional constraints, combined with a service registry search algorithm to automatically bind functions to appropriate web services or APIs. Third, “identifying workflows among business processes” focuses on integrating disparate process models into a coherent execution flow. By using BPMN as the process modeling language and defining transformation rules for sequence alignment, event‑based triggers, and conditional branching, the approach synthesizes a unified workflow that spans all partners.

The paper surveys prior work—semantic web technologies, OWL ontologies, and Service‑Oriented Architecture (SOA) mapping techniques—and outlines current research efforts, including automated semantic matching engines, machine‑learning‑driven inference, and dynamic runtime adaptation mechanisms. Future research directions are suggested: (i) real‑time re‑mapping in response to data changes and service availability fluctuations, (ii) evolution of semantic models through user feedback loops, and (iii) integration with cloud‑native deployment and micro‑service architectures.

A validation phase compares the design‑time semantics with runtime behavior using simulations and a real‑world manufacturing collaboration case. Results demonstrate that static design alone cannot handle data inconsistencies, service outages, or process bottlenecks, whereas the proposed runtime re‑conciliation mechanisms automatically adjust mappings and restore smooth operation. Consequently, MISE emerges as a comprehensive framework that unifies design‑oriented model transformations with runtime‑oriented adaptation, offering a robust solution for complex inter‑organizational interoperability.