Traditionally, relational databases have emphasized static aspects of data. For instance, data modeling has been done using just static constraints, and data manipulation has emphasized queries over updates. This was due in part to the nature of early database applications, which were mostly business oriented. Consequently, the relational model has not reflected semantics relating to data behaviour. This feature has been widely regarded as a drawback of the relational model, particularly at a time when new applications have led to increased emphasis in conceptual modeling on behavioural aspects of data. However, many basic aspects concerning data behaviour and its connection with structure are not well understood. The primary focus of the present proposal is the study of update languages and specification methods for behaviour. A variety of update languages are studied and their expressive power characterized. One tractable update language of special practical importance, called domain-based, is studied in detail with respect to optimization, parallelization, and concurrency control. The other languages range over a spectrum of procedural and declarative update languages. Next to update languages and specification methods for behaviour, of special concern is the connection between behaviour (expressed by valid updates or dynamic constraints) and structure (specified in relational databases by static constraints). This will be investigated under the present proposal subject to time and resource availability. It is expected that this research will lead to better database design methodologies integrating structural and behavioural aspects, better means for checking the soundness of a specification, and a more efficient approach to database integrity. Some of the results will be relevant to adjacent areas, such as Programming Languages and Artificial Intelligence.
