Multi-level metamodeling is a promising alternative to traditional two-level modeling
approaches as it enables classification and instantiation across multiple abstraction
levels. Many existing solutions remain constrained by rigid level boundaries, limited
runtime adaptability, or limited support for semantic evolution. Dynamic Multi-Layer
Algebra (DMLA) addresses these challenges through a formal framework in which structural
elements, relations, behavioral definitions, and validation mechanisms are represented
within a single executable model space.
After more than a decade of development, DMLA has reached a stage where the primary
focus is shifting from core language design toward systematic testing, practical use
cases, and the identification of conceptual and usability issues arising in real-world
applications. The process has resulted in several substantial refinements of the framework.
This paper presents the current interpretation of the fundamental concepts underlying
DMLA models and describes the updated principles. The paper provides a consolidated
reference point for future development, evaluation, and application of the framework.
