DeepTelos: Multi-level Modeling with Most General Instances

Multi-level modeling aims to reduce redundancy in data models by defining properties at the right abstraction level and inheriting them to more specific levels. We revisit one of the earliest such approaches, Telos, and investigate what needs to be added to its axioms to get a true multi-level modeling language. Unlike previous approaches, we define levels not with numeric potencies but with hierarchies of so-called most general instances.