Formal concept analysis revolves around the notion of formal concept. Many tools have been developed to compute these concepts, based on the fact that the extents and intents are closed under the closure operators given by the context. However, concepts are also dual to the complement of the incidence relation of the context, a fact that is rarely discussed in the community.
In this paper, we compare tools developed in the FCA community against dualization-based tools on runtime in both real and artificial datasets. The experimental observations show that dualization-based tools are competitive and more resistant to increases in context density.