Adaptive systems can be optimised at runtime due to errors. One can use the model-driven engineering methodology to implement them. But, gathered information may not be known with high confidence. And effects of optimisations actions may suffer from a delay incompatible with rapid detection of errors. These come with a global challenge for software engineers: how to represent uncertain knowledge that can be efficiently queried and to represent ongoing actions in order to improve adaptation processes? To tackle this challenge, this thesis defends the need for a unified modeling framework which includes, besides all traditional elements, temporal and uncertainty as first-class concepts. I will present two contributions towards this vision: a temporal context model and a language for uncertain data.