Collecting Data and Metadata by Transforming between Differently Expressive Query Languages

Recent trends lead to an ever-increasing data availability while gathering and linking valuable metadata is still an open research topic. Most industrial applications use vendor-specific data access and extract metadata either hard-coded or not, mainly because the used communication protocols are fast but have limited expressiveness. More powerful and convenient query languages (QLs) exist but are not yet applied to these scenarios. This paper crosses that bridge by transforming queries formulated in highly expressive QLs into lower-expressive ones to enable flexible metadata access in existing setups. We apply our methodology to an POC Unified Architecture (OPC UA) application in manufacturing and implement an open-source mapping from the expressive QL XPath. We export OPC UA information models to XML, design XPath expressions, transform these to OPC UA, and finally execute them in OPC UA with a patched client. An evaluation in a real-world injection molding application demonstrates how we, in addition to raw sensor values, dynamically capture associated metadata such as unit, scale, and precision. Our approach replaces individual, machine-dependent solutions with universal ones that drastically reduce the time and complexity for experts and can be easily transferred to other machines.