During the 1980's the flat relational model (relational model), which was initiated by Codd in 1970, gained immense popularity and acceptance in the market place. One of the main reasons for this success is that the relational model provides physical data independences i.e. changing the physical organization of the database does not require alteration of the database at the conceptual level. However, the relational model does not provide logical data independence, since users must navigate amongst the flat relations in the database when posing queries to the database. logical data independence would imply that changing the database at the conceptual level does not have an effect on the user's view of the database.

The universal relation model (UR model) endeavours to achieve logical data independence in the relational model by allowing the user to view the flat database as if it were composed of a single flat relation. To this end, the user is provided with a UR inter- face - with all the semantics embedded into the attributes - encapsulating the user's view of the flat database at the external level, on top of the conceptual level. The UR model was firmly established in mainstream relational database theory during the mid 1980's with the introduction of the weak instance approach. In the weak instance approach to the UR model, the representative instance becomes the underlying data structure of the UR model, which is suitable for storing all the data in the flat database in a single flat relation. Although the application areas of the UR model are slightly restricted by several underlying assumptions it could serve well as the foundation of a natural language interface to a database.

In recent years there has been a growing demand to use databases in non-business applications, such as: office automation, computer aided design (CAD), computer aided software engineering (CASE), image processing, text retrieval, expert systems and geo- graphical and statistical analyses. The flat structure of relations imposed by the first nor- mal form assumption on relational databases poses a severe restriction on the modelling capabilities of the relational model for such non-business applications.

In order to facilitate the modelling of the above non-business applications the nested relational model was developed during the 1980's as an extension of the relational model. The nested relational model achieves this wider applicability by allowing hierarchically structured objects, also referred to as complex objects, to be modelled directly, whilst maintaining the sound theoretical basis of the relational model.