This article addresses multi-robot map merging and the problem of acquiring reliable merging results. The conceptual model of the multi-robot map merging in the context of mapping is presented. The model can be used to dynamically propose and reject map merging hypotheses without losing the information acquired after the map merging. It consists out of three modules (Global map merging module, Local map merging module and Hypothesis verification module), and each module fulfills an important role to ensure that the map merging is reliable and easily reversible.