RIKILT at Wageningen University & Research is conducting research in food safety. Big data approaches are being used in numerous applications, however in food safety it still has not been fully implemented. RIKILT has been stepwise testing and implementing big data elements into its food safety research, starting in 2015 with the consideration of its use.
What is big data?
The definition of big data as provided by the European Commission:
Four different characteristics can be identified in big data:
- Volume: refers to the vast amounts of data generated every second.
- Velocity: refers to the increasing speed of which data is created and the speed at which it can be stored, processed and analyzed.
- Variety: refers to the different types of data including structured data, semi-structured data and unstructured data.
- Veracity: refers to the trustworthiness and accuracy of the data.
Conclusions from a literature study show that Volume and Velocity are generally not an issue within big data applications in food safety, however for Variety and Veracity applications were found. Future trends include smartphone and social media applications, as well as making data from public research projects available. These trends will stimulate the big data approach and open new opportunities, but require the presence of infrastructures and the knowledge about big data tools.
Typically, the big data workflow follows the steps as seen in Figure 1.
Big data at RIKILT
RIKILT initially focused on the data analysis phase, which resulted in the creation of a preliminary Bayesian network model for predicting different types of food fraud on a global level, linking 36 data sources. The Bayesian network method was chosen over other systems and networks as it proved to give the most consistent and correct results. Bayesian networks are new in the field of food safety and prediction results as found by RIKILT in various applications were generally >90% using this method. Forecasting is also possible, which for example shows interesting transitions in the type of food fraud in the near future for countries like China, where the food safety index is expected to improve due to policies that are being implemented.
European development: DEMETER project
The DEMETER project is an initiative on European level, which aims to support the sharing of data, knowledge and methods on emerging risk in a rapid and effective manner between EFSA and EU Member State authorities.
The project is intended to create automated data retrieval, data validation and data mining pipelines in order to identify emerging issues. These products will be made available on the Risk Knowledge Exchange Platform (ERKEP), which also will be developed by DEMETER.
Read more about the DEMETER project.
About WDCC lunch meetings
The Wageningen Data Competence Center organizes a lunch meeting every two or three weeks, where topics are presented and discussed along the overall theme of (big) data.
Contact the WDCC for more information on these lunch meetings.