Semantic interoperability is the ability of systems to exchange information with unambiguous meaning. This is an outstanding challenge in the development of Industry 4.0 due to the trend towards dynamic re-configurable production processes with increasingly complex automation systems and a diversity of standards, components, tools and services. The cost of making systems interoperable is a major limiting factor in the adoption of new technology and the envisioned development of production industry. Therefore, methods and concepts enabling efficient interoperation of heterogeneous systems are investigated to understand how the interoperability problem should be addressed. To support this development, we survey the literature on interoperability to identify automation approaches that address semantic interoperability, in particular in dynamic cyber-physical systems at large scale. We find that different aspects of the interoperability problem are investigated, some based on a conventional bottom-up standardization approach, while others consider a goal-driven computational approach; and that the different directions explored are related to open questions that motivates further research. We argue that a goaldriven machine learning approach to semantic interoperability can result in solutions that are applicable across standardization domains and thus is a promising direction of research in this era of the industrial internet of things.

Modern large-scale automation systems integrate thousands to hundreds of thousands of physical sensors and actuators. Demands for more flexible reconfiguration of production systems and optimization across different information models, standards and legacy systems challenge current system interoperability concepts. Automatic semantic translation across information models and standards is an increasingly important problem that needs to be addressed to fulfill these demands in a cost-efficient manner under constraints of human capacity and resources in relation to timing requirements and system complexity. Here we define a translator-based operational interoperability model for interacting cyber-physical systems in mathematical terms, which includes system identification and ontology-based translation as special cases. We present alternative mathematical definitions of the translator learning task and mappings to similar machine learning tasks and solutions based on recent developments in machine learning. Possibilities to learn translators between artefacts without a common physical context, for example in simulations of digital twins and across layers of the automation pyramid are briefly discussed.