Research into knowledge and expert system engineering has seen numerous attempts to streamline conceptual modeling, especially in the field of so-called Multi-Level Modeling (MLM). In this paper, we will refine and augment the approaches made there, so that these can also be used for conceptual modeling in general. To this end, we introduce a formal methodology for the definition of attributes (data properties) and relational relationship types (object properties). Rules are then defined as instantiation regulations in this connection. As a new methodology, we introduce the so-called Value Assignment Propagation (VAP), and then show how this streamlines conceptual modeling, how layer-mistakes can be avoided and how previously unfulfilled requirements for MLM can be satisfied.