The use of expert knowledge by manufacturing companies to support everyday activities has become an emerging practice thanks to the new knowledge management tools. A big set of knowledge is available in the organizations but its profitable use to solve problems and assist decision making is still a challenge. This is the case of CAM operations or preprocessing steps for which various works have been led to involve experts’ knowledge in the decision-making based on qualitative principles. However, so far, there is no methodology to the quantitative representation of that knowledge for more assistance. This paper introduces an approach for the conversion of knowledge into quantitative mathematical models. The main idea is to go from elicitation data in the form of action rules to simple unitary mathematical images; here desirability functions. The whole process carried out to extract the useful information that help building the desirability functions is exposed and different useful mathematical considerations are proposed. The resulting methodology identifies the categories of concepts in action rules and translate them into codified action rules. Then, through a mathematization process, the desirability functions are built. In short, this new approach allows evaluating the satisfaction level of the rules prescribed by the experts. As an illustration, the model is applied to action rules for CAM operations in additive manufacturing and more precisely to the definition of part orientation. This has shown the robustness of the methodology used and that it is possible to translate elicitation data into mathematical functions operable in computation algorithms.