There should be a special command to show a sample graph for each importer such that users can easier understand the model an importer is building, as there are slight deviations between models.
As an alternative, there should be a standard way of modeling the data, but that'd cost some efficiency (efficient modelling could be optional, though, by using a specific flag).