A STRUCTURED APPROACH FOR THE DEFINITION OF THE SEMANTICS OF ACTIVE DATABASES

Active DBMSs couple database technology with rule-based programming to achieve the capability of reaction to database (and possibly external ) stimuli, called events. The reactive capabilities of active database s are useful for a wide spectrum of applications, including security, view materialization, integrity checking and enforcement, or heterogen eous databases integration, which makes this technology very promising for the near future. An active database system consists of a (passive ) database and a set of active rules; the most popular form of active rule is the so-called event-condition-action (EGA) rule, which specifi es an action to be executed upon the occurrence of one or more events, provided that a condition holds. Several active database systems and prototypes have been designed and partially or completely implemented. Unfortunately, they have been designed in a totally independent way, without the support of a common theory dictating the semantics of ECA rules, and thus often show different behaviors for rules with a simila r form. In this article we consider a number of different possible opt ions in the behavior of an active DBMS, based on a broad analysis of s ome of the best known implemented systems and prototypes. We encode th ese options in a user-readable form, called Extended EGA. A rule from any existing system can be rewritten in this formalism making all the semantic choices apparent. Then an EECA rule can be automatically tran slated into an internal (less readable) format, based on a logical sty le, which is called core format: the execution semantics of core rules is specified as the fixpoint of a simple transformation involving cor e rules. As an important premise to this research, a semantics for dat abase updates and transactions has also been established, with respect to a notion of state that comprises both data and events. The article also presents an extensive bibliography on the subject of active data bases.