This study addresses the challenge of identifying key-hackers in darkweb forums, a task complicated by the presence of many unskilled or transient users and a lack of validated ground truth information. The research proposes a systematic method based on reputation to validate the results and compares three approaches—content analysis, social network analysis, and seniority-based analysis—both individually and in combination. The study tests two hypotheses: that a hybrid approach yields better results than individual methods, and that models developed for one forum can be generalized to others with inadequate reputation systems. The results show that optimization metaheuristics outperform traditional machine learning algorithms in identifying key-hackers.