Food safety ontology and text mining strategies as a tool in (re)emerging risk identification

Industry and government are held responsible for the safety of food and feed products. Therefore actual and relevant information concerning emerging safety risks is crucial. But how is it possible to filter relevant information from the fast growing volumes of information produced by science and the media? In recent history several incidents with contaminated food have occurred. For example, the surprising detection of dioxin in the milk of cows fed with potato peelings in 2004 in the Netherlands, resulted in high costs and loss of confidence in the food production chain. The origin of this contamination turned out to be the kaolinitic clay that was used in the new sorting process of the potatoes in the plant (2004). However, in 1997 it had already been published on the internet that kaolinitic clay can be naturally contaminated with dioxin. This example illustrates that information is crucial for preventive risk management. In order to enable filtering of relevant information from large volumes of information TNO developed a smart information system. This system combines the TNO Food Safety Ontology with text mining techniques. This information system supports the filtering of large volumes of electronic information produced by science and the media. By means of this dedicated system, documents from websites and databases are processed, which results in an overview of information on potential risks. The system can be customized in such a way that the resulting filtered information can be matched with specific needs for government and industry