Recently, there is an interest in studying cyber crime from a hacker-centric perspective, whose insight is to locate key-hackers and use them to find credible threat intelligence. However, the great majority of users present in hacking environments seem to be unskilled or have fleeting interests, making the identification of key-hackers a complex problem. Moreover, as ground truth information is rare in this context, there is a lack of a method to validate the results. Thus, previous work neglected this validation step or had it done manually-by hiring qualified security specialists. In this work, we address the key-hacker identification problem including a systematic method based on reputation to validate the results. Particularly, we study how three different approaches-content, social network and seniority-based analysis-perform individually and combined to identify key-hackers on darkweb forums, aiming to confirm the following two hypotheses: 1) a hybridization of these approaches tends to produce better results when compared to the individual ones; 2) a model conceived to identify key-hackers in one forum can be generalized to other forums that lack a user reputation system or have a deficient one. We conduct our experiments using a carefully selected set of features, showing how an optimization metaheuristic obtains better performance when compared to machine learning algorithms that attempt to identify key-hackers.