Formal Characterizations of Active Databases: Part II

1 I n t r o d u c t i o n a n d M o t i v a t i o n The core concept which makes a database system active is the concept of an active rule. The origin of the active rules is the production rule paradigm from the field of Artificial Intelligence with languages like OPS5 [8], used in expert systems. Typically, a production rule is of the form condition--+ action, where an inference engine cycles through all the rules and matches the condition part with the data in the working memory. Active rules, on the other hand, typically follow the event-conditwn-action (ECA) paradigm which specifies an action (possibly a sequence of actions) to be executed when a given event occurs, provided that certain conditions hold. The reactive capabilities of active databases are useful for many applications, such as views [10, 11], integrity constraints [37, 9], and workflows. Several active database languages [34, 23, 22] have been proposed, and many systems and prototypes have been designed and, partially or completely, implemented [26, 29, 12, 20, 35, 41] (many systems are presented in the collection [42]). Each system has some active features, expressed in its own syntax and (operational) semantics. However, it can be noticed that sometimes rules with a similar form will behave differently in different systems ([25] and [14] present surveys of several systems functionalities and operational semantics). The dissimilarities arise because there are many different functional features [31] tha t make a database system active and the existing systems have taken different choices among appropriate alternatives. This has resulted in recognizing * Supported by the National Science Foundation, grant Nr. IRI-9211662 and IRI9501577. ** Supported by Argonne National Laboratory, contract Nr. 963042401.