Big Data Conceptual Modelling in Cyber-Physical Systems

Abstract

Management of large volumes of data, collected from modern Cyber-Physical Systems, is calling for models, tools and methods for data representation and exploration, in order to capture relevant properties of physical objects, and manage them in the cyber-space. In this context, the impact of big data disruptive characteristics (namely, volume, velocity and variety) on data modelling and information systems design
needs further investigation. In particular, data exploration is assuming an ever growing relevance, being a way users/operators can learn from data by inspecting it according to different perspectives. In this paper, we use conceptual modelling for (big) data exploration in a dynamic context of interconnected systems. We rely on a multi-dimensional model, that is suited for properly providing data organization for exploration. Furthermore, we propose a model-driven approach that guides the design of multiple exploration strategies according to different objectives. The model-driven approach exploits a model of relevance, aimed at focusing the attention of the users/operators only on relevant data that are being explored. We describe the instantiation of the proposed concepts through some scenarios in the smart factory context, in order to show how conceptual modelling helps abstracting from implementation details and focusing on semantics of explored data.