Smart CPSs (S-CPSs) have been evolving beyond what was identified by the traditional definitions of CPSs. The objective of our research is to investigate the concepts and implementations of S-CPSs, and more specifically, the frameworks proposed for the fuzzy front end of their reasoning processes. The objectives of the paper are: (i) overview of the various framework concepts and implementations in the context of S-CPS, and (ii) analyze the presented frameworks from the points of view of reasoning processes of S-CPSs that included the concepts of structuring knowledge, building awareness, situated reasoning, decision making, and system adaptation. Our major findings are: (i) model-based and composability approaches do not support a development of S-CPSs; (ii) awareness and adaptation behaviors are considered as system level characteristics of S-CPSs that are not achieved by traditional design approaches; (iii) a new framework development should support a compositional design for reasoning in S-CPS. Based on the findings above, we argue that a development of S-CPSs should be supported by a proper framework development for compositional design of smart reasoning and coping with the challenges of compositionality requires both software-level integration and holistic fusion of knowledge by means of semantic transformations. It needs further investigation if a compositionality enabling framework should appear in the form of a meta-framework (abstract) or in the form of a semantically integrated (concrete) framework.