Database models provide concepts and tools to establish data-based models of complex real-world applications. The currently prevalent set-oriented and record-based data models provide powerful and quite adequate structural tools for modelling formatted data objects in data intensive applications. Limitations of traditional database models become obvious, however, when modelling such advanced 'non-conventional' database applications as in CAD/CAM, computer graphics, or information retrieval which require management of large sets of variously structured data objects (e.g., part representations, drawings, or texts). For a more adequate representation of such 'complex' data objects, generalized 'RECURSIVE DATA MODELS' are proposed. They allow for the definition and manipulation of objects of not necessarily the same structure, but only a similar structuring concept. This work integrates experiences from the areas of programming languages and classical data(base) models.