The long term competitiveness of food companies as well as the general health and wellness of citizens depend on the availability of products meeting the demands of safe, healthy and tasty foods. Therefore there is a need to merge heterogeneous data in order to develop the necessary decision support systems. Aims: The objective of this paper is to propose an approach for durum wheat chain analysis based on a knowledge management system in order to help prediction. Material and Methods: The approach is based on an information system allowing for experimental data and expert knowledge representation as well as reasoning mechanisms, including the decision tree learning method. Results: The results include the structure of the knowledge management system for durum wheat process data, statistics and prediction results using decision trees. The use of expert rules for decision support is introduced and a method for confronting expert knowledge with experimental data is proposed. Different case studies from the durum wheat process are given. Discussion: Our specific original contributions are: the design of a hybrid system combining both data and knowledge, the advantage of not requiring an a priori model, and therefore, an increased genericity, and the potential use for both risk and benefit analysis. Conclusion: The approach can be reused for other purposes within the chain, and can also be transferred to other domains. Such a project is a starting point to integrate new knowledge from multidisciplinary fields, and constitutes a tool for structuring the international cereal research community.