Soft computing approach is expected to play a major role in data assimilation in the field of computational nuclear physics so that multiple or competing model-based data such as nuclear reaction cross sections can be aggregated in the presence of both aleatory (random, stochastic) and epistemic (systematic, originating from lack of knowledge) uncertainties. The paper deals with the aggregation of multi-model generated nuclear data using the probability-box approach. Random set theory provides the fundamental guidelines for data aggregation in terms of probability bounds. High energy neutron induced production cross sections of radio-nuclides in different elements constitutes important nuclear data in the field of nuclear power production, transmutation of nuclear wastes and basic nuclear physics research.