Knowledge Engineering: a tool for integrating knowledge about food chain processes and supporting decision-making

At the intersection of cognitive science, computer science and applied mathematics, knowledge engineering offers concepts, methods and techniques for modelling and/or acquiring knowledge in fields where formalisation is difficult or where understanding of the phenomena is incomplete. Knowledge acquisition emerged as a discipline with its own research focus in the late 1980s. Following developments in expert systems during that decade, the issue of modelling and acquiring knowledge for these systems emerged as both crucial and problematic. Work in knowledge engineering is therefore based on the need to model knowledge explicitly. This involves constructing models suited to the nature of the knowledge to be described, so that it can then be represented using appropriate formalisms (Charlet, 2002). Knowledge engineering can be broadly defined in three stages: the acquisition of available knowledge, its computational representation, and its use for simulation, prediction, validation and optimisation to aid decision-making. A wide range of methods exists to address this issue. Each is more or less suited to the problem at hand, and specific expertise is required to select and implement the most appropriate methods. The approach and certain methods are illustrated for various applications in the food industry, such as the durum wheat processing chain, bread dough kneading and cheese ripening. These examples draw on specific approaches, such as decision trees and concept maps, qualitative reasoning, Bayesian networks and constrained optimisation, respectively. The application of these approaches opens up new possibilities for the virtual design of food products within the framework of a sustainable food supply chain.