Prediction of food fraud type using data from Rapid Alert System for Food and Feed (RASFF) and Bayesian network modelling

Abstract

Because food fraud can harm human health and erode consumer trust, it is imperative that it is detected at an early stage. Therefore the aim of this study was to predict the expected food fraud type for imported products for which the product category and country of origin are known in order to target enforcement activities. For this purpose we used a Bayesian Network (BN) model that was developed based on adulteration/fraud notifications as reported in the Rapid Alert System for Food and Feed (RASFF) in the period 2000–2013. In this period 749 food fraud notifications were reported and were categorised in 6 different fraud types (i) improper, fraudulent, missing or absent health certificates, (ii) illegal importation, (iii) tampering, (iv) improper, expired, fraudulent or missing common entry documents or import declarations, (v) expiration date, (vi) mislabelling. The data were then used to develop a BN model. The constructed BN model was validated using 88 food fraud notifications reported in RASFF in 2014. The proposed model predicted 80% of food fraud types correctly when food fraud type, country and food category had been reported previously in RASFF. The model predicted 52% of all 88 food fraud types correctly when the country of origin or the product-country combination had not been recorded before in the RASFF database. The presented model can aid the risk manager/controller in border inspection posts in deciding which fraud type to check when importing products.