The paper considers the problem of providing rational data selection from heterogeneous information space for adaptive-heuristic intelligence embedded in the mechanism of robust management of socio-economic systems. The relevance of solving the presented problem is determined by the need to adapt the work of the socio-economic system to the paradigm of modernity -sustainable functioning and development in conditions of accelerated changes. Systems move from complex to complex-adaptive, diffuse forms with the development of the ability to self-regulate like living, biological systems. To survive and develop, they try to make flexible, anticipatory adaptations within the limits of acceptable deviations of physiological parameters from some parameters -constants of health, which represent the conditions of maintaining homeostasis. The prototype is the automatic robust control of closed systems, in which the leading parameters are given a certain degree of freedom and the limit of their permissible changes is set. The effectiveness of such control strongly depends on the reliability of the data -facts entered the information granules of adaptive and robust homeostasis. We propose a technology for obtaining reliable information about the state of the studied objects for robust management based on the use of rational choice methods.