Collaboratrices et collaborateurs

The "Principles of Fertilisation of Agricultural Crops in Switzerland" (PRIF) serve as the Swiss reference for all aspects of fertilisation and nutrition of agricultural crops. They contain around 150 interlinked tables, providing standard and correction values for crop- and site-specific fertilisation assessments. Currently, these tables are only available in paper and PDF format, limiting fertiliser requirement calculations to an analogue process. To support the digital transformation in agriculture and align with federal data strategies, the pilot project WebGRUD aimed to provide the content of selected PRIF tables in a machine-readable format. This project sought to facilitate access to these tables using standard programming languages, ensure consistent fertilisation recommendations across different users, and meet Open Government Data criteria. A linked data service for storing and linking the PRIF tables and the development of a test application programming interface (API) for calculating and visualising fertilisation recommendations was considered the most suitable approach. For a proof of concept, we selected the PRIF tables for calculating nitrogen (method of the corrected norm), phosphorus, potassium, and magnesium fertilisation recommendations for arable crops. The Linked Data Service (LINDAS) of the Swiss Federal Archives was chosen for data storage and linking, as it utilizes the open W3C-standardised Resource Description Framework (RDF) technology and is accessible via the standard SPARQL query language. In preparation, the PRIF tables needed significant modification to convert them from wide format to normal form. The transformation from CSV to RDF, along with the augmentation and annotation of the RDF data cubes, was facilitated by the Cube Creator interface. Finally, the self-descriptive, hypermedia-driven test API webgrud-app was programmed to calculate fertilisation recommendations based on the conditional relationships between the PRIF tables. A simple web user interface enhanced human accessibility to the functionality of the test API. The pilot project WebGRUD succeeded as a proof of concept, providing a selection of the most frequently used PRIF tables in machine-readable format and demonstrating the correct implementation of their conditional links in data models. However, it does not cover the full functionality of the PRIF or all crop groups included. Additionally, several constraints hindered the conversion of the PRIF tables to the required format for RDF transformation, such as purely descriptive or poorly defined information, essential data spread across table headers, footnotes, and main text, and inconsistencies within the PRIF contents. Suggestions for improvement and future work are provided.