Page en développement
[P10-02] Jean-Luc Hainaut, Conceptual interpretation of foreign keys, DB-MAIN Technical report, May 2010, 60 pages, [full text]
- Description. Foreign keys form a major structuring construct in relational databases and in standard files. In reverse engineering processes, they have long been interpreted as the implementation of many-to-one relationship types. Though one could naively think they are useless, or at least unnecessary, in hierarchical and network models, foreign keys also appear very frequently in IMS, CODASYL, TOTAL/IMAGE and even in OO databases. Besides the standard version of foreign key, according to which a set of columns (fields) in a table (file) is used to designate rows (records) in another table, a careful analysis of existing (both modern and legacy) databases puts into light a surprisingly large variety of non standard forms of foreign keys. Most of them are quite correct, and perfectly fitted to the requirements the developer had in mind. However, their conceptual interpretation can prove much more difficult to formalize than the standard forms.
- The aim of this study is to classify, to analyze and to interpret some of the most frequent variants of foreign keys that have been observed in operational files and databases.