UMR CNRS 7253

In order to evaluate the dependability parameters of systems, part of components and subsystems reliability data is provided by experts. Eliciting expert opinions is therefore a major and decisive task in the process of the quantitative evaluation of systems dependability. A critical part of the elicitation is to choose the questions to ask. Those need to be simple (i.e., do not require high cognitive effort) and in terms and format experts are familiar with. Furthermore, the process of information acquisition need to be optimal for the elicitation to be effective and the least possible time or effort consuming. In other words, we want the querying strategy to be optimal. For the particular problem of estimation and uncertainty reduction, we developed a querying strategy that is adaptive and optimal, i.e., that selects, at each stage of the elicitation, the optimal questions based on the answers to the previous ones. Then we use this optimal elicitation method in estimating the dependability parameters of systems. For this, we implemented the elicitation procedure into a Matlab interface.