The Cyber Crime Investigation is challenged by large and complex data as a key factor of emerging Information and Communication Technologies. The size, the velocity, the variety and the complexity of the data have become so high that data mining approaches are no more efficient since they cannot deal with Big Data. As a result, it can be infeasible to represent specific evidences found in such data in a Court of Law in a human-perceivable manner. Moreover, majority of computational methods result in complex and hardly explainable models. However, Soft Computing, a computing with words, can be beneficial in such case. In particular, hybrid Neuro-Fuzzy is capable of learning understandable and precise fuzzy rule-based model. This paper presents novel improvements of NF architecture and corresponding results.