MicroRNAs (miRNAs) are small RNA molecules that bind messenger RNAs (mRNAs) to silence their expression. Understanding this regulation mechanism requires the study of the miRNA/mRNA interaction network. State of the art methods for predicting interactions lead to a high level of false positive. Roughly, the score distribution of interactions is a mix of two overlapping gaussian laws that need to be discriminated with a threhold. To improve the discrimination between true and false interactions, we allow to keep or discard some interaction edges that are above or below the threshold. This \repair" process is founded on the hypothesis that the graph is formed by interaction modules represented by formal concepts, i.e. set of miRNAs having the same regulation pro le, a plausible biological structuration. Our assumption is that a network consisting only of true interactions has a simpler concept topology. It allows to discriminate between true and false interactions on the basis of concepts that can be simpli ed by edge addition or deletion. To validate our hypothesis and method, we have extracted parameters from a biological miRNA/mRNA network and use them to build random networks with xed concept topologies and true/false interaction ratio. Each repaired network can be evaluated with a score balancing the number of edge changes and the conceptual adequacy in the spirit of the minimum description length principle.