Automating Sensor Data FAIRification with the Sensor Interfacing Language

The successful implementation of the FAIR Guiding Principles heavily relies on standardized, interoperable data models using technologies from the Semantic Web, such as ontologies and standardized vocabulary. However, fully semantically interoperable data models or ontologies bear high complexity, hindering their application by researchers, such as engineers, who are not familiar with the concepts from the Semantic Web. To lower the barrier for implementing the FAIR Guiding Principles, we developed the Sensor Interfacing Language (SOIL), a light-weight domain-specific modeling language for describing the data model of any arbitrary sensor or measuring system. Based on a textual description with an intuitive syntax and grammar, inspired by JSON, Turtle, and Python, researchers can quickly document the features of a sensor or measuring system via the publicly accessible SOIL Editor. Employing model-driven software engineering (MDSE), the SOIL model can be automatically translated into a full data model based on Semantic Web standards, such as RDF and OWL. Thus, researchers do not have to deal with the complexity of the semantic sensor model, as they never come into direct contact with it. Parsing, validation and translation of SOIL models is carried out using the SOIL Workbench, deployed and running in the background of the SOIL Editor. Metadata profiles (MPs) are automatically generated out of a SOIL model, which enables automated compliance tests of acquired data. Moreover, MDSE enables the generation of interfaces for various different communication protocols and data formats, which is required to significantly increase the accessibility of the measurement data provided. This streamlines the development and implementation of cyber-physical measuring systems (CPMSs). Thereby, Separation-of-Concerns is realized, which effectively decouples the implementation of the three layers of the FAIR Sensor Service. Finally, semantic interoperability, i.e., "use of formal, accessible, shared, and broadly applicable language for knowledge representation", is ensured by the SOIL-based FAIRification. Technical interoperability, i.e., accessibility, is guaranteed by the unified network interface of the FAIR Sensor Service using standardized, open, free and universal communication protocols, namely HTTP, MQTT, and OPC UA, and data formats, such as JSON and XML or JSON-LD and Turtle for data modeled according to RDF. The developed tools reduced the deployment of new measuring equipment at our institute significantly and increased the accessibility and (re-)usability of acquired data. This improvement has been achieved by the implementation of an automated data acquisition pipeline based on the standardized data models and communication interfaces of FAIR Sensor Services. The FAIRness of sensor data acquired with FAIR Sensor Service has been assessed using FAIRness assessment frameworks, namely F-UJI, FAIR Checker, and FAIR enough. The individual FAIRness scores for acquired metadata, data, and metadata profiles have been normalized to a percentage scale for comparison. Compared to publicly available assessments using FAIR enough of foreign research data, SOIL and FAIR Sensor Services significantly increase the FAIRness of acquired sensor data by a factor of 1.7. Lastly, SOIL and FAIR Sensor Services are applicable beyond engineering in any domain using sensors or measuring systems, such as natural and life sciences, bridging the gap between distinct consortia with interoperable data models and enabling cross-domain re-use of research data.