A cognitive map is a graphical model that represents influences between concepts. This model is used in decision-making. It provides an operation, called propagated influence, that computes the global influence of any concept of the map on any other one. Our first contribution is to provide some quality criteria in order to validate a map. These criteria verify if the propagated influence of a concept on another one is not ambiguous. Our second contribution is to synthesize a set of cognitive maps provided by different designers into a unique map. To do so, we use a taxonomy of concepts that expresses relations between the concepts of the different maps and a preference order on these designers to manage contradictory information. Our third contribution is the study of the semantics of a cognitive map by comparing this model to other knowledge representation models. In that study, we define the probabilistic cognitive map model such that the influence values are close to probabilities. We have also developed a software that implements the notions previously described.
