A meta-model of modeling abstractions is presented and discussed in the context of using simulation as a means for system verification and validation activities. Extending the classical results of presenting modeling as a relevance reasoning problem, abstractions are classified to build an ontology based on the concepts of teleological modeling and relevance reasoning. The domain model is built in the standard ontology tool of Protégé to exploit the reasoning and inference capabilities to build a model abstraction library. Lattice structures of model instances are formed and a recursive algorithm is implemented as an activity diagram in SysML for automated and consistent model selection. The approach is presented with a battery model example from the literature. Challenges and future work in implementation of such a semi-formal approach in model selection is briefly presented in the context of improving fidelity of simulation in the industry.