Genome-scale metabolic models are complex systems that describe thousands of reactions thought to participate in the organism's metabolism. They are tailored for a computer simulation, and can be too complicated for a human. To help a human expert to analyze these detailed models, we developed a method for knowledge-based generalization that provides a higher-level view of the model. The generalization process groups biochemical species present in the model into semantically equivalent classes, based on their hierarchical relationships in the ChEBI ontology, and merges them into a generalized chemical species. For example, '3-oxo-decanoyl-CoA', '3-oxo-lauroyl-CoA' and '3-oxotetradecanoyl-CoA' species can be generalized into '3-oxo-acyl-CoA'. After the species generalization, reactions that share the same generalized reactants and products, are factored together into a generalized reaction. To represent the model generalization in SBML we use the 'groups' package, that allows to encode the grouping of similar species and reactions as well as to annotate the species groups with their generalized ChEBI identifiers. The choice of a visual representation is harder. In this talk/poster we compare SBGN submap solution with quotient graph nodes.