On the specification of behavioural constraints

When designing information processing systems, the key problem is to find what symbols and constructs (available in some given language) should be used so as to be able to answer the questions that the users want to ask and to perform the state transitions that the users have to perform in order to keep in step with some reality that is being modelled. If the language that is being used cannot be interpreted directly by the software of a computer (e.g. set theory, algebra, semantic networks), the resulting model is called an abstract model (e.g. a specification, a data base schema). If all of the symbols and constructs used to express the model can be interpreted by a computer, the model is a concrete implementation (e.g. a data base system). Both the abstract and concrete models capture some aspects of the reality that is modelled. They differ with respect to the languages in which they are expressed. The choice of the languages which are appropriate for writing such models is a subject of intensive debate. The recent workshop on Data Abstraction, Data Bases and Conceptual Modelling has shown that this debate is not likely to diminish in the near future. In the context of this debate, however, a common opinion was expressed several times by data base participants: the data models used to specify data base schemas are lacking to support the specification of operations tailored to particular application environments. It is therefore expected that in the context of data base modelling, the integration of operations (behaviour) with data (structure) will be one of the most fertile research areas for the next few years.