Modelling Non Deterministic Queries and Updates in Deductive Databases

Abstract

In this paper, we propose a new formalism to model and hnplement general, rule-based languages for querying or updating deductive databases. We consider as a target language a production rule language for. databases that we introduced in previous papers, namely RDLl. This language can be seen as an extension of a logic-based language for databases to support updates in the head of rules. The model we introduce, named Production Compilation Network (FCN), is derived from Petri-Net based models. A PCN models the two aspects of a rule program: the static aspect which consists of the relationships between the rules and the database predicates and the dynamic aspect which describes the semantics of a rule program. The PCN are shown to have the following features: (i) to provide a formal framework to describe general computation strategies and query optimization algorithms in a clear and concise manner and (ii) to model in a unified way queries and updates in a deductive database context.