Decisions and lack of precision in crop management: The role of processing both objects and procedures through semantic networks

Abstract

This paper adresses the central issue of technical agricultural decision in crop management intercourse which is a complex task where decision of action is the most crucial topic in regard to environment characteristics, in regard to the objects of the task and in regard to the procedures that have to be applied. Moreover, many object properties are variables which take imprecise values due to: the dynamic character of the cultivation system, more or less accessibility to these values, the presence of intermediary variables and, finally, imprecise instruments.

Starting with a functional diagram of [plant population*soil*weather*techniques] system and a set of observations for sugar beet settling, we describe agronomic and ergonomic knowledge integration in task decomposition. We then show that relevant knowledge can be represented into semantic networks of the objects of the task (where classes factorize common structure of objects through associated procedures with simple and multiple inheritance principles). In such semantic networks, objects are classes and subclasses instances by an inclusion relation which is defined by the procedures (considered as properties of classes of objects) which they may be applied to these objects. Then, we suggest how such a representation involves some farmers' decision making mecanisms in spite of uncertainty.