H2PTM (1992) Simon

Unifying Deductive and Active Databases

---

 
 

Titre Unifying Deductive and Active Databases Auteurs E. Simon(i), J. Kiernan(ii) et C. de Maindreville(iii) Affiliations (i) INRIA Rocquencourt, 78153 Le Chesnay, France (ii) IBM Almaden Research, 650 Harry Road, San Jose, CA 95120, USA (iii) Alcatel Alsthom recherche, route de Nosay, 91460 Marcoussis, France Dans actes du colloque H2PTM 1992 Paris publié dans H²PTM92 : Les bases de données avancées : du modèle relationnel au modèle orienté objet

Abstract

Active database systems have rules (usually called triggers), consisting of an event that causes a condition to be evaluated, and if true, results in the executon of a predefined action. However, existing trigger languages have a few drawbacks. first, the proposed semantics do not take advantage of well understood and accepted formalisms developed for rule-based systems, and thereby do not capitalize on existing rule-based technology. Second, trigger languages are low-level languages. These languages require that the user provides all triggering conditions associated with rules. This makes difficult the specification of triggers and their maintenance. In this paper, we present an extension of a deductive database language, namely RDL1, towards active rules. By active, we mean rules that react to external events. Rules are expressed at a high level so that triggering conditions are derived from rules by system. the semantics of our rule language is formally described by means of a parrtial fixpoint operator which encompasses the deducative database and active database paradigms. We also present an architecture in which the system responsable for detecting events issued by application programs and triggering rules, is front-ended to a relational DBMS.