In the recent years, Event-Condition-Action (ECA) rules have been successfully used for developing on-line transaction processing systems (e.g., warehouse management systems). ECA rules are based upon the principle that when an event occurs and a specified condition is satisfied, then a particular action will be carried out without requiring any interaction with the user. The field of application of ECA rules in on-line transaction processing systems ranges from internal tasks (e.g., ensuring the integrity of databases) to external tasks (e.g., realizing the business rules of a particular application program). Although practically relevant, ECA rules have not been utilized for automatizing various processes (e.g., routine analysis tasks) within decision-support systems (data warehouses and OLAP systems) so far.
This project proposes an approach to carry out routine decision tasks and semi-routine decision tasks automatically within data warehouses using ECA rules. Such systems are called active data warehouses, since these systems analyze the data warehouse as reaction to occurred events. The "decisions" that are generated by these rules will be realized by executing transactions in on-line transaction processing systems. The emphasis of such ECA rules, which we call analysis rules from now on, is (i) on automatizing analyses originally carried out manually by users (analysts) and (ii) on defining how decisions will be generated.
This basic approach of analysis rules is extended by a framework to specify complex decision tasks. These extensions are (i) flexible modeling of decision criteria, (ii) alternative approaches to specify decision-making models, and (iii) determining the bindings of action parameters. We define the semantics of the basic approach and of the extended approach declaratively and procedurally and propose a basic implementation of analysis rules using off-the-shelf database technolo