Deriving Identity from Extensionality.pdf

Deriving Identity from ExtensionalityA.H.M. ter Hofstede and Th.P. van der WeidePublished as: A.H.M. ter Hofstede and Th.P. van derWeide. Deriving Identity from Extensionality. Techni-cal Report CSI-R9416, Computing Science Institute,University of Nijmegen, Nijmegen, The Netherlands,December 1994.Abstract|In recent years, a number of proposals have beenmade to extend conventional conceptual datamodeling tech-niques with concepts for modeling complex object struc-tures. Among the most prominent proposed concepts isthe concept of collection type. A collection type is an ob-ject type of which the instances are sets of instances of another object type. A drawback of the introduction of sucha new concept is that the formal denition of the techniqueinvolved becomes considerably more complex. This is a re-sult of the fact that collection types are populatable typesand such types tend to complicate updates. In this paperit is shown how a new kind of constraint, the extensionaluniqueness constraint, allows for an alternative treatmentof collection types avoiding update problems. The formaldenition of this constraint type is presented, other advan-tages of its introduction are discussed, and its consequencesfor, among others, identication schemes are elaborated.Keywords| Conceptual Data Modeling, Uniqueness con-straint, Extensional uniqueness constraint, Identication,Collection type, ER, NIAM.Classication: AMS 68P15, CR H.2.1I. IntroductionNowadays it is commonly accepted that information sys-tems are best specied at the conceptual level rst. Usu-ally, two perspectives are distinguished at this level, thedata and the process perspective. It is generally recog-nized that the data perspective tends to be more stablethan the process perspective, and that a precise model ofthis perspective is of vital importance for successful imple-mentation of the system under consideration.Many techniques for conceptual data modeling exist ([12;16]). Among the most well-known is the Ent