Use case

Linked Data in Agrifood

Promoting a healthy and sustainable diet and reducing food waste are major challenges in today's society. This requires consumers and producers to make informed choices about food products and ingredients. Information technology plays an important role in supporting this decision making process. This requires the availability of reliable and meaningful data, together with evidence-based knowledge rules. In 2018, we investigated how semantic enrichment of data can be achieved in the food domain and how it makes smart applications possible.

Solution / approach

Practical decision making requires data from multiple sources to be combined. For example, while shopping, the consumer needs to know the price of a product, but also the impact on health and sustainability, how it fits with his or her personal preferences and circumstances and how suitable it is for a specific occasion. This means that the app that assists in making the choice must have access to nutritional values, preference values, alternative products, recipes, etc., all from different sources. These different data must then be harmonized and merged.

To share data in a meaningful way, we must be able to express the associated metadata. This requires the development and application of digital vocabularies and ontologies. Therefore, we have performed a project on how semantic enrichment can be achieved in the food domain. This strategic project was closely linked to the PPS
Personal Nutrition and Health (WUR and TNO), PPS
Trusted Source and the NWA project 'Intelligent measurements of food
intake'. A selection of the Food Taxonomy, extended with nutritional and
preference properties has been applied to define product replacers for
increasing fibre intake. An important lesson is that sharing data requires
continued interaction between data suppliers and data users at the level of
semantics in order to avoid misunderstanding and misuse.

(Expected) impact of the approach

The concept of Linked Open Food Data was proposed in 2018 and translated into a number of new collaborations with, for example, GS1, RIVM and the Netherlands Nutrition Centre. To support nutrition researchers, a data and model library has been further filled and new
vocabularies have been developed. In 2019 a new public-private collaboration will be proposed on Linked Open Food Data in a wider sense. Such data can for example be used in the context of personalised health advice. This will help finding healthy and sustainable alternatives for ingredients of one’s individual diet, given personal health status, preferences and situation. The same concept would be transferable to industrial food product developers.

Our platform for sharing food ontologies and vocabularies FoodVoc portal, version 2 has been completed, now also presenting the latest version of OM. OM, the ontology of units and quantities is again considered as the best ontology in this field world-wide. The number of users is still growing.

Next steps

The next question is how to move from general and average values of food parameters (such as nutritional values, CO2-footprint, taste) to  data specific for each product. Secondly, we should extend the set of properties, including sustainability, taste and texture, processing options, eating moments, etc. and make this data openly and semantically available for any developer of food software applications. One can see an increasing willingness to share data in the supply chain. This leads to the notion of true transparency, contributing to consumer trust and operational
efficiency.

Tools used:

  • W3C semantic standards RDFS/OWL
  • ROC+ for creating vocabularies by experts
  • OM ontology of units and measure
  • Java for application developmentore

Jan Top

jan.top@wur.nl

+31 317 480 212