Data semantics for the agrifood industry

Data semantics for the agrifood industry

From sensory perception to product quality and process control; the agrifood industry has access to a huge amount of data. This data is often widely distributed, making it difficult to interpret and re-use. Wageningen Food & Biobased Research develops semantic web solutions that allow agrifood companies to organise and integrate data, and so maximise its value.

Adding context to data

Whether agrifood data is useful depends on its content and on the way users can explain and use it. By adding meaning, context and source information to data, its value becomes clearer. Only then is it possible to combine different datasets. For example, to predict the quality of fresh fruits – and subsequently take well-considered decisions on storage, transport and markets –  a product manager should be working to an agreed meaning of ‘quality’ and how it relates to post-harvest conditions. In practice the metadata is often incomplete and not machine-readable. This downgrades the value of otherwise useful data and limits its re-use. Adding formal semantics solves this problem.

Semantic web technology for agrifood

Wageningen Food & Biobased Research combines profound knowledge about advanced information technology with expertise, experience and a broad network in agrifood. This unique skill-set is essential for acquiring high-quality and reliable data semantics: machine-readable metadata that clearly describes the meaning of the data used.

We apply several methods from the World Wide Web Consortium (W3C), including:

  • RDFS/OWL (data modelling languages that allow data sharing)
  • SPARQL (a QUERY language to retrieve semantic data)
  • SKOS (a data modelling language comparable to thesauri in libraries)

Our in-house developed ROC+ tool allows product developers, R&D managers and other non-IT experts in the agrifood industry to collaboratively create data vocabularies. We also developed OM: an ontology of units and quantities. It captures thousands of units and quantities as length, weight and energy in a formal model and is globally recognised as the most extensive units ontology in existence. Other vocabularies we have developed, for example of food products, can be found at FoodVoc.

From document selection to consumer app

Wageningen Food & Biobased Research is involved in a broad range of projects and initiatives around organising and integrating data. Ask-Valerie (2014-2017) for example, enables smart selection of documents about innovations in agriculture and forestry. In this project, involving 15 European partners, we annotated a large set of documents present on the web with concepts from the vocabulary we created.

In the Personalised Nutrition and Health project, semantic technologies are used to add information to food products that supports consumers to devise healthy diets. We relate data on these products to lifestyle, eating schedules and personal health parameters. We also investigate how machine-learning can improve the quality of the data. This will provide a basis for the development of smart and trustworthy apps.