Stakeholders involved in the defi nition of managerial problems perceive and internalize the complexity of these problems as mental models. These models are not always made explicit, which can lead to misunderstandings and confl icts. This article presents Ulysse, a tool that enables stakeholders to progressively elicit their mental models as causal loop models. Ulysse uses web technologies that store information in a database and interactively display the modeling results. It is based on matrix calculation that facilitates statistical operations and provides better understanding of the model structure. Analysis and comparison of individual models reveal the most important variables and relationships among them as a basis for identifying key issues, divergent and convergent opinions between stakeholders. Ulysse has been tested in the context of regional economic development in Atlantic Canada. The tool was designed to support managers, during interviews, in building qualitative models of indicators. stage of a model-building process. Ulysse supports interviews with experts for eliciting from them and structuring mental models. The process aims at helping experts build their own view of a system, draw out a causal loop structure from their mental models and sketch out strategies. Ulysse is implemented into a computer interface, based on web technologies. It allows users to store the externalized mental models in a database, analyze them statistically (on content and structure), and display them for communication purposes. Moreover, by comparing and merging individual mental models, a common understanding can be built and it serves as a point of departure for further group discussion. The tool is theoretically grounded in the system dynamics approach, and particularly in conceptual modeling, but it also benefi ts from other fi elds like cognitive and strategy mapping, graph theory and soft system methodology.This article fi rst exposes some theoretical concepts related to conceptualization, mental models and tools based on causal mapping to elicit and analyze models. The Ulysse interface, its components and the stages of using the tool during an eliciting process are then presented. Ulysse is illustrated with the building of a system of indicators for regional development in Atlantic Canada. The last part of the paper discusses the use of Ulysse in a second round which includes further interviews with individuals as well as a group meeting.
Conceptualization, mental models and learningGilles Desthieux et al.: Ulysse 165 analyzed considering the number of relationships that the variables generate or receive (see Hall et al., 1994). A high number of relationships reveals a variable that is signifi cant in one of two ways:• Driving-force variables. A high number of relationships originating from a given variable confi rms its importance as a policy or action lever. In other words, such a Fig. 5. Rules for determining the order of variables and relationships (feedback loop case) Fig. 4. Rules for determining the ord...