Umeå universitets logga

Analyses of products during manufacturing are essential to guarantee their quality. In complex industrial settings, such analyses require to use data coming from many different and highly heterogeneous machines, and thus are affected by the data integration challenge. In this work, we show how this challenge can be addressed by relying on semantic data integration, following the Virtual Knowledge Graph approach. For this purpose, we propose the SIB Framework, in which we semantically integrate Bosch manufacturing data, and more specifically the data necessary for the analysis of the Surface Mounting Process (SMT) pipeline. In order to experiment with our framework, we have developed an ontology for SMT manufacturing data, and a set of mappings that connect the ontology to data coming from a Bosch plant. We have evaluated SIB using a catalog of product quality analysis tasks that we have encoded as SPARQL queries. The results we have obtained are promising, both with respect to expressivity (i.e., the ability to capture through queries relevant analysis tasks) and with respect to performance.