摘要：Self-awareness holds the promise of better decision making based on a comprehensive assessment of a system's own situation. Therefore it has been studied for more than ten years in a range of settings and applications. However, in the literature the term has been used in a variety of meanings and today there is no consensus on what features and properties it should include. In fact, researchers disagree on the relative benefits of a self-aware system compared to one that is very similar but lacks self-awareness. We sketch a formal model, and thus a formal definition, of self-awareness. The model is based on dynamic dataflow semantics and includes self-assessment, a simulation and an abstraction as facilitating techniques, which are modeled by spawning new dataflow actors in the system. Most importantly, it has a method to focus on any of its parts to make it a subject of analysis by applying abstraction, self-assessment and simulation. In particular, it can apply this process to itself, which we call recursive self-reflection. There is no arbitrary limit to this self-scrutiny except resource constraints.
