Are there arguments against ER models and their extensions


The entity-relationship models has been proposed by P.P. Chen in 1975/76. This model had a large impact on database modeling. The model is very simple, allows an easy capturing of the main structure and is also formally precise. Since the main constructs of the original models have shown some inefficiency the model "obtained" a large number of extensions. Some of the extensions have gained a wide popularity. For this reason, the entire complex of extensions has been taken for granted and used together without thought on their compatibility. Theoretical research has however now clarified the mismatch and the compatibility. Not all extensions can be considered to be useful or to have a theoretical basis. In the book the main achievements have been compiled into a consistent and at the same time well-based model.

Thus, the higher-order entity-relationship model does not have all the stupid pitfalls which are observed for object-oriented models. The number of pitfalls published so far on oo models reached now 88. What is a lot.

UML uses the binary ER model and falls back to the situation of database modeling observed in the end of the 70ies. Thus the model must be extended by useful and well-based concepts.

This site is developed as a discussion forum. Anybody who want to send me an email with proposals, complaints or with critics is very welcome. I will answer him directly and on this page.

  1. ER modeling is not rich enough
  2. UML is more natural than ER
  3. Binary relationship types cope with anything what is necessary
  4. Modeling does not require so much
  5. Weak types are very useful in modeling
  6. Problems related to cardinality constraints
  7. Relationship types using relationship types as their components are non-sense
  8. More readings

Please send your comments to Bernhard Thalheim. This document has been updated at 01.01.04.